Dave Richards AA7EE

September 9, 2016

A Scratch-Build of N6KR and Wilderness Radio’s SST for 20M

Note – all links in this post open in a new browser window. It would be a good idea to clear your cache from time to time, to make sure your browser loads the latest version of this post. As an example, I just found an error on the schematic, and uploaded a newer, correct diagram.

I’ve been wanting to build an SST for a few years now. It’s a plucky little rig, with a lot of character. Designed by Wayne N6KR in the late 90’s, appearing as a full article in QRPp, and a kit sold by Wilderness Radio, it ignited the imaginations of a whole generation of builders for it’s combination of simplicity, performance, and willingness to accept modifications cheerfully. The review from Adventure Radio Society was quite positive. It is a VXO-based QRP CW transceiver, with a simple superhet receiver (SST = Simple Superhet Transceiver), and a TX that puts out up to 3W, depending on your choice of transistor in the final (it can be dialed down for battery-powered outings). It has very fast, clean QSK – so fast, in fact, that it feels as if I can hear the band all the time I’m sending (W6JL would approve). You actually listen to your own signal as you’re sending – there is no separately generated sidetone. The sidetone level does vary with the volume control, as opposed to being a fixed volume, but I only find this to be an issue when I have the AF gain all the way up, in which case I either quickly adjust the volume knob, or partially pull the earbuds away from my ears while sending. By the way, the sidetone on this little rig sounds really nice. It’s a feature which helps to make operating the SST an enjoyable experience.

QRP’ers loved their SST’s. There was a lively discussion about the minimalist rig on QRP-L, with builders reporting back on the frequency coverage and performance of their builds, with details of mods they were trying. The kit came with a light gauge unfinished aluminum enclosure. The raw-finish aluminum was a blank slate which invited many different creative solutions to the age-old question of how to show off your project. Some folk endowed theirs with professional-looking paint jobs, while others used dymo labels, or simply scrawled right onto the panel with a Sharpie® for that authentic home-brew look. All approaches worked admirably well. I saw one SST that had been painted with a US flag, and looked great. Some of them were taken out on the trail many times, and showed many knocks and scratches on the case which, of course, just made ’em look better still.


SST Kit Version – Image from Wilderness Radio


SST Kit Version – Image from Wilderness Radio

JN3-bong;y modified his SST with 2 varacters for extra frequency coverage. He also added a speaker, and a spot button. Click on this photo to go to his site, and read more about his SST.

Koichi JN3DMJ modified his kit version of the SST40 with 2 varacters for extra frequency coverage. He also added a speaker, a spot button, and expanded the LPF for better filtering of the transmitted signal. Click on this photo to go to his site, and read more about his SST. He says that this SST is his favorite rig. He has made over 1300 QSO’s on it so far. Thank you for the permission to reproduce this photo and link to your site Koichi.

Recently, I decided it was time to build my own SST, only to find that I had missed the boat on a kit, as Wilderness Radio had discontinued it at some point in the recent past. I called QRP Bob on the phone, hoping that there would perhaps still be a board and/or enclosure available, but I was out of luck. In fact, it meant that I was in luck, as I would have to scratch-build one, and that’s a good thing.

There is plenty of documentation for this rig online. The initial write-up was in the Spring 1997 edition of QRPp, available from Chuck K7QO here. Note that the preceding link is a PDF of all 4 volumes from that year. The entire QRPp archive is also on Chuck’s site and accessible from the main page. Ken WA4MNT has a copy of the manual for the Wilderness SST kit on his site. Other essential documentation is the mods and information collected from QRP-L messages from 1997-2004, from Ken Larsen AL7FS, here as an HTML file, here as a text file, or here as a .doc file.

Mine differed just slightly from the original. Here’s my hand-drawn version of the schematic, reproduced here with kind permission of N6KR. Don’t rely just on the following schematic, as my drawing is a bit goofy. Best to refer to the original circuit diagram in the manual, and use mine to see how it differs –


Note – I do not recommend using the LPF comprised by C34, C35, C36, L2 and L3. Although it just met FCC specs at the time of release, it is very unlikely to provide enough filtering for this rig to meet current FCC rules regarding suppression of spurious emissions. For more satisfactory filtering, see my post here.

Component designations (e.g. C27, R11 etc) are the same as in the schematic in the Wilderness Radio manual. There are 3 parts in the above schematic that do not have such designations – they were added by me (the series capacitor and resistor from pin 5 of U3 to ground, and the 1N5817 diode in series with the 10-16V DC supply line).

Differences between the stock SST and mine are –

1)  Inclusion of a series  diode in the supply line for polarity protection. I did consider using a P-channel MOSFET so as to avoid the voltage drop, but decided to go with a Schottky barrier diode. Some diodes of this type have a big enough reverse leakage current such that they are not effective in this role, but not so the 1N5817. It’s forward voltage drop is about 0.34V in transmit. I still don’t like losing that much, but it’s an improvement on the higher drop of a regular silicon diode.

2) The use of trimcaps in the RX and TX oscillators in order to place the received signal in the center of the passband, and to put the TX signal on the same frequency as the RX signal. I’m still about 20Hz off due, I think, to touchy trimcaps, but it’s close enough – until I get the urge to tweak and adjust again🙂

3) The use of 68pF capacitors for C19 and C20 instead of the 100pF values specified. With 100pF feedback caps, my VXO wouldn’t oscillate, so I swapped them for 68pF ones, and it sprang into life (thanks to LA3PNA for the help on that). It may well also have oscillated with 82pF caps, and that is an option if you want your frequency coverage to be a little lower.

4) An MRF237 was used for the final instead of a 2N3553. This substitution was suggested in the manual for higher output power, as the MRF237 has higher gain. If you want a cheaper and more modern alternative to the 2N3553, I’m thinking that a BD139 should work as a direct replacement. Joel KB6QVI just told me that W8DIZ has the 2SC5706 at 10 for $4, and I’m wondering how it would work in this application.

5) A Zobel network was added to the output of the LM386. Mine was unstable at high volume settings. A series capacitor and resistor from pin 5 to ground is very commonly used in these circuits, and the inclusion of these 2 parts tamed my instability immediately.

6) Alternate values of the capacitors in the crystal filter were used to widen out the response. The original values were reported to be giving a particularly narrow bandwidth of around 200 – 300Hz at the -6dB points. I wanted something a bit wider. There were several suggestions in the QRP-L archived discussions. K7SZ tried widening his SST20 out with the help of these suggestions, but it still wasn’t wide enough for him. He suggested the use of 47pF for C6 and C9, and 120pF for C7 and C8, which is what I used. Thanks Rich. A few builders went further, and implemented the ABX (adjustable bandwidth crystal filter) mod that was used in the Wilderness Radio version of the NorCal Sierra. EDIT – K7SZ notes in his ARRL book “Low Power Communication” that his mod brought the center frequency of the filter down too low for him on his SST30, so he ended up going back to the stock values. I have found that using the values of C7/C8 = 120pF and C6/C9 =  47pF that Rich suggested on QRP-L for his SST20, I set the sidetone at 400Hz ( a new thing of mine – I’m experimenting with lower than normal sidetone pitch), and the center of the filter passband was still about 20Hz lower, which I guess is pretty close. If you like higher sidetones though, you may be better off with one of the 2 sets of stock values in the SST manual. SECOND EDIT – I plan to tighten the response of the filter, by using the alternate values quoted in the manual. After using the SST for a while, I think my version is a little wide.

Although the SST didn’t come with a keyer, many users added their own, the Wilderness KC1 keyer and frequency readout being popular. At the time of writing, this is still available – the only kit that Wilderness still supplies. I decided that I wanted to build a keyer onto the same board as an integral part of my SST. I didn’t need much in the way of special features, my only 2 requirements of a keyer for this rig being that it will operate in iambic mode B, and that it has a speed pot – a feature I think of as essential. Changing speed on the fly during a QSO is tricky if the option is accessible only via menus, but a piece of cake if all you have to do is reach out and twist a knob. Perhaps I didn’t look hard enough, but the only freeware I found didn’t support a speed pot. I remembered how well the N0XAS Super PicoKeyer that Dar W9HZC had given me had worked, and a light went on in my head. Dale sells spare chips for his PicoKeyer Plus at $6 each. I purchased 4, used one in this rig, and saved the others for future projects. The manual on Dale’s site will give you the info on all the features of this keyer, and how to access them. The surrounding circuitry is simple (the genius is in the coding), so it was easy to incorporate into the SST –

The on-board keyer used a replacement PicoKeyer-Plus chip from Dale N0XAS

The on-board keyer used a replacement PicoKeyer-Plus chip from Dale N0XAS

The keyer uses a piezo transducer to announce the responses to command inputs made via the CMD pushbutton and the paddle. It would be possible to feed this audio into the AF amp of the SST so that it can be heard in the earphones, but I elected to fit a small piezo transducer on the edge of the board. I had intended to punch a small hole in the side of the case to make it easy to hear, but this appeared not to be necessary. Note that in the schematic, I have called it a piezo “buzzer”. It is actually a transducer, but allow me to explain. There are piezo buzzers available to which you apply a voltage, and the unit rewards you with a loud piercing tone, generated by an internal oscillator. Some of the units available are called piezo buzzers, but they don’t contain the audio oscillator – just a transducer, which is often sharply resonant at a specific audio frequency, to enhance the volume of the emitted tone. I bought a 5-pack of so-called “piezo buzzers” on eBay. They looked too small to contain an internal oscillator and I was correct – they consisted of just the transducer, which was exactly what was needed.

This build looked great when it started (as they all do🙂 ). A nice, clean board, with nothing but potential. As projects progress, I tend to become more anxious that in a single unconsidered moment during a late night soldering session, the iron will slip, the odor of burning plastic will waft into my nostrils, and all the hard work will be undone in a careless fraction of a second. In truth, there aren’t that many errors that can’t be corrected, but this early shot of the board was the best it ever looked! As with all my projects, all the Manhattan pads were MePADS (for IC’s) and MeSQUARES (for everything else) from Rex at QRPMe

If you compare the above photo to later pictures, you’ll notice that I ended up changing the layout of the front panel controls.

In the next picture, the AF amp and the VXO have been constructed, as well as the 8V regulated supply line. Temporary DC and headphone jacks were also connected, so the circuit can be plugged in to see if it works. The shielded cable that connects to the tuning pot was installed, but left longer than needed to allow for the final install in the enclosure. Not too much of the circuit had been built at this point, but it was already possible to test the voltages at the input and output of the regulator, as well as ensuring that the LM386 made a nice, loud noise when the input terminals were touched with a screwdriver (a highly controlled and accurate test🙂 ) The VXO was tuned in on a nearby receiver and tested for frequency coverage. With the 20M version, the VXO is in the 18MHz range, and by subtracting the IF of 3.932MHz from the highest and lowest frequencies it oscillates at, you can estimate the final coverage of the SST, and make adjustments at this stage if you wish. The discussions on QRP-L (which are linked earlier) contain a lot of info on tailoring the coverage, so I won’t repeat it all here, but your options involve using different varactors, connecting a second crystal in parallel with the VXO crystal, using different values of rubbering inductor, and adjusting the value of R5. I won’t explain how these all affect the frequency coverage and stability, as this is discussed at length in the QRP-L archive and also, to a certain extent, in the manual. There is plenty of homework reading to be done if you are thinking of building this rig!

Here’s another view, with the VXO in the foreground –

Suddenly, the product detector and BFO burst onto the scene. In the next shot, the 3.932MHz BFO crystal is the one closest to the camera, with the trimcap for centering the passband just behind it. I used 60pF trimmers, as that is what I had the greatest quantity of. Something a little smaller might have made the adjustment less touchy though. I’ll leave you to experiment, if you want to. Things are getting pretty exciting at this point, because when you a touch a wire or metal screwdriver screwdriver to pin 1 of the product detector IC U2, you hear honest-to-goodness atmospheric noise – a distinctly different sound from what you hear when touching the input of the AF amp IC. It’s instructive, not to mention really cool, to hear this progression in the sounds you hear, as you touch the inputs of stages closer and closer to the antenna, as the build progresses. If you have a signal generator, you can inject that into the circuit, and look at the output on a ‘scope. Don’t despair if you don’t have a full stable of test gear though – it’s important not to underestimate the power of touching and listening. Once you’ve done it a few times, you get used to knowing what sorts of things you should be hearing. See the curved red power wire that supplies 8V regulated to the BFO/product detector? You’ll notice in later photos that it was replaced with a different-shaped wire. It’s rarely possible to get everything right the first time you construct something, so one-off builds like this tend to morph somewhat as they progress. It’s OK to change things as you go along –

The next stage to construct was the crystal filter. Do you notice how, on the board for the kit version of the SST, the crystals for the filter were lined up with the short edges parallel to each other, so that the filter takes up a significant length of one side of the board? You usually see filters with the long edges of the crystals lined up parallel to each other. I don’t know why Wayne did it this way, but it did occur to me that with this physical configuration, the input and output of the filter are further apart than they would be with the more conventional placement pattern. Perhaps this was an attempt to decrease the possibility of filter blow-by? It seemed like a good idea, so I replicated it in my Manhattan copy. The crystals in the filter are not yet grounded in this next shot –

As far as matching the 3 crystals for the filter, I placed them into an oscillator circuit, and measured the frequency of oscillation. My cheap Chinese stand-alone frequency counter only had a resolution of 100Hz, but then I remembered that my K2 had a built-in counter with a 10Hz resolution, so I used that. I needed 5 x 3.932MHz crystals total – 3 for the filter, and the other 2 for the oscillators in the TX mixer and the BFO/product detector, so I picked the 5 that were closest in frequency. Out of that group of 5, I took the 3 closest and used them for the filter, while the other 2 were used for the local oscillators (but not the VXO, which required an 18MHz crystal).

To verify that the receiver is working, you’ll need to also build the antenna LPF, consisting of L1, L2, and associated parts. Without it, you won’t be able to peak the antenna input trimcap C1. Notice that if you touch the input of the crystal filter, the noise from the phones sounds much more restricted than when you touch the output of the filter. In fact, you can work your way back through the filter, with the rushing atmospheric noise becoming more and more restricted-sounding as you touch each stage of the filter with your metal screwdriver. These quick checks help to confirm that your project is pretty much on track. Adding the receive mixer means that the receiver is complete. After peaking C1 for maximum band noise, you should be able to receive off-air signals. Congratulations! If you substituted a trimcap for C10, you can also adjust it to place the received signal in the center of the passband, an adjustment that will depend on what pitch of sidetone you like to listen to.

My receiver didn’t work particularly well at first – I was getting very low audio out of it. One or two posts in the QRP-L archive made the same observation. I was beginning to talk myself into believing that the design was deficient in the audio department, and resolving to substitute a different audio chain, when I discovered that the coax which delivered the output of the VXO to the input of the RX mixer wasn’t properly soldered at the output of the VXO, resulting in low drive to the RX mixer. Re-soldering the joint solved the problem, and I can happily report that the audio output is more than adequate to drive a quality set of earbuds or a pair of reasonably sensitive headphones. If you attempt to drive a speaker, you will find that the level is only adequate for monitoring whether a frequency has activity or not, in a quiet room. That’s fine, as this was designed as a trail-friendly rig, with low current consumption in mind, and it certainly achieves that. VK3HN mentioned to me the idea of adding a lower noise AF chain designed to drive a speaker, and retaining the original AF output stage, feeding the inputs of both in parallel. The advantage of this would be that you’d retain the AGC action provided by the LED.

See the VXO in this next shot, with it’s 18MHz crystal? It has a total of just 10 parts, including the tuning potentiometer. I know that it represents old, well-established technology, but I feel that it still has it’s place in ham home-brew. Only 10 parts, and yet it has great stability and signal purity too. As long as you can deal with the fairly limited frequency range, a VXO is still a great choice as the frequency control in a simple rig –

This is always the point, when building a transceiver, where I slow down and spend some time playing with the receiver. I was dead chuffed, as we Brits say, that I had successfully built a little superhet receiver with a narrow crystal filter, that was sensitive, and sounded good.

But at some point, the momentum needs to be capitalized on before it is all gone, and so the build proceeded, with the addition of the transmit mixer. I also added the keying line (the green wires around the edge of the board) so that I could key the TX to see if it worked. If it did, then all that would be required would be to amplify the output of the TX mixer with the driver/buffer, and the PA. We were really getting close at this point! You’ll notice that there is a “channel” of space separating the crystal filter from the rest of the circuit. I did this for two reasons – firstly, as I thought it couldn’t hurt to physically separate the filter a little, to help prevent filter blow-by. Secondly, if there was excessive blow-by, it would give me enough space to erect a screen made from PCB material. C39, the 470uF AGC capacitor, is not present in this shot, nor in the later overhead notated view. I was planning on mounting it off the board, on the inside front panel, but eventually decided to mount it on the board. It ended up occupying the space between the AF amp and the edge of the board –

If you look carefully at these pictures, you may notice one or two components changing position slightly. As the build progresses, I will occasionally move a part or two in order to refine and improve the layout. I’m not going to point out which parts this applies to, as I don’t show these photos with the intention that you follow the layout closely. I started out by following the layout of the stages on the board from the SST manual but as the build progressed, realized I’d be able to move the position of the driver/buffer, thereby freeing up space for the on-board keyer, in one of the back corners of the board. Here’s another view of the board in the same state as in the above photo, with the receiver fully built, as well as the TX mixer. If you have a scope, you can measure the output of the VXO, which should be between 200 and 500mV RMS (that’s 0.565 – 1.414V peak – peak). You can also adjust C28 to peak the signal that will drive the buffer –

This was really the point at which I felt that I was home free. The TX/RX switching was working well, and the rig was putting out a small signal on the operating frequency in the 20M band. All that was left was to amplify it – and even if that didn’t work, I still had a cool little receiver and let’s face it – receivers are cooler than transmitters🙂

The next shot shows the rig fully built, with the exception of the keyer, with the board temporarily mounted in an enclosure. I ended up changing the layout of the front panel controls, which necessitated the use of another enclosure. In both cases, I used the LMB Heeger 143 enclosure in plain aluminum finish (they also have it in black and grey). I used an MRF237 instead of the 2N3553 in the PA, in order to provide a bit more output, and you can see that transistor, wearing it’s heatsink. To the right of the PA transistor is the orange top of R12, the drive control, and to the right of that is the LT1252 driver/buffer stage. The RF input to the LT1252 is carried by a single wire underneath the board. There are only 2 wires under the board – the one just mentioned, and a length of RG174 coax connecting the output of the VXO to the input of the RX mixer –

A view of the completed board, with the N0XAS keyer in the rear left-hand corner (which is actually the rear right-hand corner, if you are looking at the board from the front panel end). I mounted the keyer chip in a machined socket. I run as many of the control cables as possible underneath, and drill holes in the board for them to enter. I think it looks neater that way –

This next view shows the layout. I got carried away and labeled a few too many parts. The side of the board that faces the front panel is the left edge. As with the previous overhead anotated view, C39, the AGC capacitor, is not shown here, though it did end up being mounted on the board. Remember those SMD SA602’s I was giving away for the price of postage a while back (courtesy of KV7L)? I hadn’t used any myself, until now. 3 of them are in this little rig –

Time to get this thing in a case. The LMB Heeger 143 is an ongoing favorite of mine. It measures 4″ x 4″ x 2″ high, and has 2 small lugs at the front and back of the cover that engage with the front and back panels to prevent them from being pushed in. This feature adds rigidity and stoutness. One of the things I don’t like about most clamshell cases is that the front and back panels can be flexed; not so with this model. It comes in grey and a sort of black wrinkle finish, if you don’t want the raw aluminum. All the pots are Alpha brand. The 3 small ones were $1.29 each from Tayda. The larger tuning pot is also an Alpha part, but is very slightly smoother in action, and I wanted to optimize the experience of tuning this rig. It is Alpha part # RV16AF-10-20R1-B10K-LA (I got it from Mouser).  There is a small dummy load plugged into the back in this next shot, because I was having fun using the rig as a code practice oscillator🙂

This SST is quite a triumph for me, as it is the most complex project I have built so far with Rex’s MeSQUARES and MePADS.

The piezo transducer for the keyer was fixed to the edge of the board with a small spot of hot glue (on the high temperature setting of the hot glue gun, as it flows better) –

The keyer CMD pushbutton was a 22 cent cheapie from Tayda. These types are available with plastic and metal shafts. The ones with the metal shafts have a slightly smoother, more positive action. Get those ones. 4 vinyl bumpers from the local hardware store keep the SST from slip-sliding on my desk –

This enclosure is higher than the kit version, at 2″ high. I wasn’t initially planning on such a high case but the advantages are that it supports a larger tuning knob and, as you can see from the next shot, there is room for an internal battery pack, speaker, ATU or other add-on –

Here’s a view of the SST-20 upside-down and from the rear. From left to right – RF gain (rarely used), Antenna, DC power, and paddle –

I have a couple of spare covers for this enclosure, from projects that didn’t go as planned, and am thinking that it would be possible to have different covers with different accessories built in. For instance, one cover could have a speaker and extra AF amplifier, for operation at home. Another cover could contain a battery pack, for portable ops –

The red and yellow knobs look a bit garish, and I’m still getting used to them, but the thinking is that yellow = audio (AF gain and headphones), while red = keyer (speed pot and CMD button). It was also a way of using the cheap knobs I got from Tayda for 49¢ each🙂

So how does it perform? Well, in 2 words – very well. I don’t operate a lot, but I do spend a lot of time listening. I’ve had 6 QSO’s so far with a horizontal loaded dipole (a Buddipole) at 25 feet above ground at my home QTH. 3 of them were with stations in Colorado, about 900 miles distant, one with KE5AKL who was doing a SOTA activation in NM, also 900 miles from me, and one with a mobile station in Hooks, Texas, who was running 25W. He was 1600 miles away as the crow flies. The other was with a local station. The receiver is as sensitive as you’d need a receiver to be, and there’s a good amount of opposite sideband suppression. I haven’t measured it, but you only hear the opposite side of the signal weakly when tuning through a very strong station. The RF gain only needs to be backed down when in the presence of very strong stations, as the use of an SA602 in the front end can cause it to crumble under such circumstances. I haven’t needed to use it yet, and from what I’ve read, it doesn’t need to be used very often – hence the reason it is on the back panel. My frequency coverage, with an MV209 varacter, is approximately 14055 – 14064KHz, a swing of 9KHz, which is about as much as you’d want when tuning with a 1-turn pot. Many users mounted a switch on the front panel to switch in another varacter (usually an MVAM108, which was also supplied with the kit) to extend the coverage downwards. With the MRF237 in the final, my WM-2 wattmeter indicates an output power of about 2.25W with 11.61V at the input to the rig (11.27V after the polarity protection diode). When supplied with 13.8V from a PSU, the power output was about 2.8W. I didn’t measure the current consumption on transmit, but on receive it is between 26 and 27mA. This is low, but somewhat higher than the 15-16mA quoted in the manual. The keyer consumes <1mA, so that isn’t the reason for the difference.

The AGC LED is a rather unique feature. I’ll let you read up about it in the manual but for the addition of a few extra parts, it will save your ears from the worst ravages of sudden loud signals – and the LED is fun to watch too🙂 Most red LED’s have a forward voltage drop of about 1.7 – 1.8V. If you want to raise the AGC threshold, look for a red LED with a higher Vf – some of them go as high as 2.2V.

Although all the parts for this little rig are still available, a few of them are a bit harder to find than others. I purchased the LT1252 from Digi-Key – they have them in both through-hole and SMD versions. Chuck K7QO tipped me off to a supplier on eBay who was selling them in 10 packs. I couldn’t resist purchasing a pack. Thanks Chuck🙂 W8DIZ has the MPN3700 PIN diodes, though see the next paragraph for a worthy substitute.  There are several different choices for the PA transistor. 2N3553’s and MRF237’s were available on eBay when I was looking. Try to buy legitimate parts from a reputable supplier (my gut feelings seem to serve me well in this regard). I’m thinking that a BD139 would work in this position too. All the crystals are still available from Digi-Key. The part numbers are the same as in the SST manual, with the exception of X1-X5 for the 30M version. The manual quotes the Digi-Key part # as X007-ND. It is, in fact, CTX007-ND. Perhaps it changed. It has, after all, been 19 years since the kit was introduced!

Even though this design is now quite old, I think it is still very relevant. An experienced home-brewer can build this into a fairly small case, and take it on the trail with a simple tuner and, say, an EFHW, for a compact and effective portable set-up. All of the parts are still available, though it would be great to see a partial redesign, utilizing more modern and widely available parts. I’m thinking of a redesign of the buffer/driver and PA stages. BS170’s are cheap, and 3 of them in parallel, in class E, could provide close to the full QRP gallon. The original SST had room in the case for a 9V lithium battery, and could be dialed down to lower output powers to help battery life. Nowadays, newer battery technologies make more power available in a light and small package, so running 4 or 5W while portable with a small rig like this is practical. Kenjia JH1PJL used an NPN transistor in his driver, instead of the LT1252 IC. He also used a 1N4004 instead of the MPN3700 PIN diode (you can see pictures of his SST scratch-build here). In fact, all the diodes in the 1N4001 – 1N4007 series have the relatively slow recovery time of 30µS, giving them PIN characteristics. Any of them should work fine in place of the MPN3700. If a 1N4000-series diode is good enough for RF switching in the Elecraft K2 (the 1N4007), then it’s good enough for us!

Here’s a brief video of it in action, with a surprise appearance by Jingles the blind kitty. I fed her just before starting the video and forgot that her routine after eating, is to jump up on the desk to relax and digest her meal for a few minutes. I’ll work on producing a slightly better video, though videos are not my strong point. Apologies for the slightly crackly audio. It’s a combination of operator error and a camera that was designed primarily for stills, and not video –

The yellow knob was making me uncomfortable. It has since been replaced with a black one, and I am feeling much calmer now🙂

10 minutes later, and the red knob has now been changed for a solid and dependable black knob also. I finally feel that I know where I am in the world again🙂

For the near future, the next tasks are to –
a) add an extra stage to the LPF between the antenna and the rig for greater harmonic suppression  and
b) tighten up the crystal filter a bit. I have decided that it’s just a little too wide🙂

Both the above were done here.


Note – as of early Oct 2016, I received this very informative message from Walt K3ASW –

I have an actual SST20, with several mods.

If you’d like more VXO range, try adding a small value NP0 or C0G from the crystal-RFC junction to ground. Mine has a 3 pF and I get about 14042-14064. However, the VXO voltage to the two mixers (RX, TX) drops by about 1/4 to 1/3 over the lower 2-3 kHz. Also, I have a 1u8 RFC in series with the 5u6 below the crystal.

Currently, it has the K8IQY filter mod, but I’m likely to change that per my crystal measurements and modelling (via W7ZOI’s GPLA).

I added a JFET amplifier (MPF-102) between the crystal filter and product detector ‘602. This reduced the RX noise level quite a bit and I don’t have to run the AF gain as high. This simple circuit is from NA5N’s little handbook he published some years ago. If one does this mod, a dual JFET keying switch needs to be added between the 602
‘product detect and LM386; otherwise, the sidetone is way too loud. (My first version of the JFET switch circuit didn’t work, so I’ll have to try again. It is a tight fit! – on the   Wilderness Radio PWB  version.)

I’d also noticed the output of the TX mixer needs another filter section. I’m trying to figure out how to shoehorn it in on the existing PWB. (You won’t have that problem with your HB layout.)

I’ve worked quite a bit of DX on mine over the years. When I lived in a condo (before 2000), the antenna was a `98 foot horizontal loop in the attic above the fourth floor and I regularly worked into EU from here in MD.

Hope you like yours; it’s a neat little rig.

73 Walt K3ASW

– and a few days later. Walt sent these extra SST tips –

A couple other simple mods: Add a 0.033 or 0.039uF between the product detector pins 4 and 5; this reduces some of the high frequency noise. Also, a 10 or 15 uF from the phone jack + to ground will reduce the noise a bit more (22uF is too much – its starts attenuating the desired signal). The latter is a RC low pass filter for audio; the R is the AF gain control.
73 Walt K3ASW
Thanks for the helpful tips Walt!

August 21, 2014

The Sproutie – A General Coverage Regen Receiver with Plug-In Coils

NOTE – Many thanks to Aaron N9SKN and Cliff WA9YXG, who pointed out errors in the schematics. They have been corrected, and N9SKN has built a working Sproutie from the schematics in this post so rest assured that if you follow them, you can too.

If you’re thinking about building this great little general coverage regen, I’d urge you not to print out any of this article to work from. The reason is that whenever I make an improvement or addition to this receiver, I edit this article. By working from a printout, you’ll be missing out on any changes I subsequently make. Having said that, The Sproutie works fine as is, so don’t be scared off from building it.

I’ve mentioned before in posts how one of my first shortwave receivers as a teenager growing up in England (in fact, possibly the first) was a one tube battery-operated regen built from a kit. Many of the popular electronics magazines at the time, including my favorite, Practical Wireless, carried advertisements from a company called H.A.C. (“Hear All Continents”) who sold kits for simple HF regenerative receivers. This was the ad I remember best. To the teenage me, this receiver was the holy grail. With this receiver, there would be no stopping me. I would be the king of the hill, if only I could have this magnificent shortwave receiver –

Ads like this for H.A.C. shortwave receiver kits were common in the UK up until the early 1980’s. Image taken, with permission, from Louis Meulstee at http://www.wftw.nl/

I saved my pennies and eventually sent off for the H.A.C. Model DX Mk. 2. It wasn’t as fancy-looking as the one picitured in the ad, as it didn’t have a front panel or a calibrated dial but hey – those kinds of regens were only for the truly well-heeled, and I was just a kid with a modest allowance. The kit that arrived used an HL23DD valve (or equivalent). It was a battery operated double diode triode, with a coated filament, to maximize emission on the low filament voltage of just 1.5V (2V maximum, with a current consumption of just 50mA). This set didn’t have a front panel or a calibrated dial, sporting just a modestly-sized aluminum chassis with 3 chicken head knobs on the front, but with my 2000 ohm headset and 90V high tension battery, I truly was the king of the shortwave hill. I don’t have any pictures of my H.A.C. Model DX Mk. 2, but featured here are pictures of someone else’s taken from Louis Muelstee’s great website, which is where the ad shown above came from too.

The H.A.C. Model DX Mk. 2 cost me all of £14.50 in the late 70’s. (Photo taken by Philip McNamara and taken, with permission from Louis Meulstee at http://www.wftw.nl/)

This one tube regen used an HL23D double diode triode tube with low current consumption 1.5V filament. The blue lead on the left leads to a connector that plugged into a 90V high tension battery. (Photo taken by Philip McNamara and taken, with permission from Louis Meulstee at http://www.wftw.nl/)

Truth be told, this wasn’t exactly the most sensitive receiver ever created, but it mattered little to a teenager in England in the 1970’s, with plenty of loud shortwave broadcast signals. There was much to keep those high impedance headphones firmly glued to my head. They were fairly cheap quality, and the way the metal headband tensioned the earpieces against my ears made them a little red after 30 minutes of use. Did I care? Not at all – I wore them for hours on end, as I was enthralled by the sounds of Radio Nederland, Radio Prague, Radio Tirana Albania, Radio Moscow, The BBC World Service and many other broadcasters, as well as all the weird-sounding utility stations and the very mysterious numbers station from East Germany. I had no idea back then what it was, but the female voice announcing strings of numbers in German was strangely compelling. Every day, I would rush home from school, eager to get into my bedroom, plug the low-tension battery in, wait for a short while for the tube filament to heat up, then connect the high tension battery and clamp the headphones to my head, cup of tea by my side, as another listening session began. Weekends were heaven. As soon as all my homework was out of the way, there was nothing but blissful hours and hours of potential shortwave listening time stretching ahead. From time to time, the 90V battery would run down, and I would walk the 2 miles into the village of Astwood Bank to buy a new one from the local gas station.  It didn’t take me long to figure out how to power the filament from a transformer in order to save money on low tension batteries. I eventually figured out how to do the same for the high tension supply too. On good days, I could even pick up some local amateurs on 80M SSB. Man, I was indeed the king of the shortwave hill! At the back of my mind, though, was the idea that somewhere out there was still a truly dreamy shortwave receiver – one that had a front panel fashioned from a sheet of aluminum, and a calibrated tuning dial. It only took me until the age of 50 to finally own one of those only-in-your-dreams kind of receivers.

Which is what this blog-post is all about.

I’ve had some encouraging success with regens recently. Both the WBR and my modified version of the the WBR, which I built for the 31M BC band, worked well, with no common-mode hum, instability, or any of the other kinds of naughtiness that sometimes accompany the operation of regenerative receivers. There was one main thing about the WBR’s that limited them for me, and that was the fact that they only operated on a limited range of frequencies. After building these 2 receivers, the next logical step was to build a general coverage regen with plug-in coils. I wanted a set that was built solidly, with a reduction drive and a calibrated tuning scale, so that I could prove to myself something that I already knew – that a regen, properly constructed, can serve well as a shortwave receiver rather than just as a novelty, which seems to be the category most people have placed them into these days. I’m reminded of a comment on a discussion forum I saw recently, in which a gentleman was talking about a regen he had built once. It was sensitive, received lots of stations, and gave him much enjoyment, he said, but he never really knew where he was on the band. “Well of course you didn’t!” I thought to myself, “but that’s not because it’s a regen – its because when you built it, you didn’t build it with a calibrated dial. It’s not the regen’s fault you didn’t know where you were on the band – it’s yours!”

Charles Kitchin had a design for a receiver which caught my attention. It was published in the Feb 2010 edition of CQ Magazine and consisted of an oscillating detector feeding a 2-stage amplifier consisting of a low-noise FET-input opamp acting as the preamp. The preamp had a low-pass filter with a variable cut-off point, as well as an extra capacitor in the audio chain that could be switched in to give a nice lift to the lower frequencies, for those times when you have a nice strong signal and want a bit of bass boost. This preamp drives an LM380, which makes for a much lower noise AF amp chain than the default in these types of receivers that employ an LM386. On top of that, there is provision for a line out jack for recording. I was interested – regardless of the front end I used, this could definitely be the AF amp for a “serious” regen!

Before I had even fully decided on the finer details of this project, I assembled the AF amp on a separate board. I wanted this to be a somewhat modular receiver, with the AF and RF sections built on different boards so that if either section didn’t work out, I could try a different one. I wasn’t entirely convinced that the value of Hammarlund tuning capacitor that I was planning on using was going to be ideal, so I wanted it to be a relatively easily swappable part with other tuning capacitors of different value but with the same form factor (of which I own a few). If I was going to go to the trouble of building something like this on a nice chassis, I wanted to give myself the maximum possible chance of succeeding.

Here is the schematic of the AF board. It is a little different from the version originally designed by Mr Kitchin (though not by much) as I will explain –

In Chuck’s original version, the 2.2K resistor on the input was 5K.  The ratio of this value to the value of the 100K resistor between pins 2 and 6 of the AD820AN determine the gain of the stage, and I wanted a bit more. Also, the LM380 was motorboating when the AF gain pot was set to anything higher than half-volume, so I added the 10 ohm resistor in the supply line and bypassed it with a 470uF electrolytic, which stabilized it nicely. The +ve supply line was connected directly to pin 7 of the AD820AN, but this could also be bypassed if necessary. A series 100 ohm resistor with a 47uF or 100uF bypassing to ground should work nicely. You can also sprinkle a few large decoupling electrolytics in the range of 100 – 470uF at various points on the 12V bus directly to ground. A small issue I experienced was that when the AF gain pot was at absolute maximum volume, rotating the low-pass filter pot caused clicks and “bloops” in the speaker. It seemed to only happen when the slider of the pot had finished traveling over the carbon track and had actually made contact with the metal that formed the hot end of the control. A 47 ohm resistor placed at the hot end of the 10K AF gain pot provided the necessary isolation (this is shown in the schematic), and I was left with a volume control that operated smoothly, and a variable low-pass filter pot that also operated smoothly. I also added a 0.1uF coupling capacitor on the input (pin 2) of the LM380. The 0.1 and 220uF bypass capacitors on the 12V line were placed so as to bypass the 12V supply directly at the point of entry into the chassis. They were soldered directly on the back of the DC power connector and grounded with a solder tag bolted to the chassis. The 1N4001 diode was also placed at the same point.

Here’s the AF board as I first built it, before changing the 5K resistor on the input to a 2.2K resistor – and before adding the 10 ohm resistor in the 12V supply line and the 470uF capacitor to bypass it. The lengths of lavalier mic cable for the variable low-pass filter and the AF gain potentiometers have already been soldered in place, and they exit through holes drilled in the board. The headphone jack and DC power connector are temporary, for the purpose of testing. The Manhattan pads are of course, as always, W1REX’s MeSQUARES and MePADS

The front end is a very standard design. It is the same configuration (and indeed the same circuit) as used in the WBR, with the exception that the tank (unlike that in the WBR) is unbalanced. This same arrangement was used in Nicky’s TRF, as featured in issue 70 of SPRAT, and I believe the original circuit was developed by GI3XZM. I wanted this receiver to be usable over a wide range of frequencies, and in keeping with my “modular” approach, wanted the receiver to be as versatile as possible. A plug-in coil system, with both gangs of a dual gang variable capacitor, as well as the fine tuning capacitor, all available at the pins of the coil base, allows for a lot of flexibility when winding coils for different bands. The user decides, when constructing a plug-in coil, whether to include parallel or series capacitors for the main tuning and fine tuning capacitors, as well as choosing whether to use one, or both gangs of the tuning capacitor. In this way, with some calculations and a bit of trial and error you could, say, wind a coil to cover a large segment of the HF spectrum, or a single narrow band of frequencies. If, after some listening, I decide one day that I am particularly interested in the 16M broadcast band, I can construct a coil to cover just that one band. Neat!

The J310’s in the RF amp and the detector stage could be any similar N-channel JFET such as the MPF102 or the 2N3819. Likewise, the two 2N3904’s could be most any small signal general purpose NPN transistor. I originally fed the output of the J310 “infinite impedance” detector stage directly into the input of the AF amp board, but quickly discovered that the gain wasn’t enough to comfortably drive a loudspeaker. Had I done a few quick calculations beforehand, I would have realized that. I wanted to take advantage of the fact that the output chip is an LM380, by driving it enough to make a loud noise into the speaker! Adding the single 2N3904 preamp stage after the detector solved the problem nicely. I have built enough of these simple receivers that can drive “a small speaker to a comfortable volume in a quiet room”🙂 No more!

As with any circuit of this type, the RF stages, and the frequency-determining part of the circuit especially, should be built with short leads, and stiff wiring. Top quality components will help.  The two 330pF capacitors in the feedback circuit of the 2N3904 oscillator stage should be NPO’s (or C0G’s – same thing), as should the 39pF capacitor. The coils were wound on toroids, and the coil assemblies mounted in octal tube bases. I spent a great deal of time on W8DIZ’ site, using his online calculators to figure out the number of turns required for varying degrees of coverage. Unless you build a receiver with the same variable capacitors, and use a very similar physical layout, you’ll need to do your own calculations, and then be prepared to tweak the final values of inductance and capacitance to get the coverage you want. Incidentally, I used a Hammarlund MCD-35-MX dual gang component for the main tuning capacitor. It was this one that I got a deal on over a year ago. The official specs say that each section has a capacitance of 6 – 31pF, but I also had to make a rough estimate of the stray and circuit capacitance when calculating the required values of inductance and capacitance to cover each band. My fine tuning cap was a Hammarlund MC-20-S, and I had to include the capacitance of that in the calculations too. This is the online calculator on W8DIZ’s site for the T68-6 core. He has similar calculators for all the popular toroids. Very useful stuff.  Note – for some reason, the calculator doesn’t always estimate the correct length of wire that needs to be used. This is easy to work around. Just wind one turn around a toroid measure it’s length, multiply that by the number of turns, and add a few extra inches for good luck (and pigtails).

Here’s a view of the RF board as initially built, before adding the extra (pre-AF board) preamp stage –

Here are the details of the coils wound so far, including the temporary “experimental” coil for 24-29MHz. I didn’t get as far as installing a link winding for this coil, but the main coil was picking up plenty of signal from the proximity of pin 7 of the tube socket to the coil. I have been very pleasantly surprised at how sensitive and stable the set is at these higher frequencies. Soon after winding it, I copied SSB on the 12M and 10M amateur bands, as well as plenty of over-modulated and very loud local signals on 27MHz🙂  Unless you also use the same values of tuning and fine-tuning variable capacitors, and closely copy my layout, your values will be different, but here is the info on my coil set so far. After a little while spent looking at it, it should make sense. Once you get used to figuring out how to wind a coil for a specific set of frequencies, it’s fun.  I have 15 coils so far, with ideas for a few more. I have already filled up my cigar box coil box, and am getting ready to make a second coil box and wind a few more coils. One of the really enjoyable things about a regen with plug-in coils is making coils for new bands. Fun!

If you wind too many turns for the link winding, you may find that you have to turn the regeneration control nearly all the way clockwise in order to reach oscillation, or you may not be able to reach it at all, as the link winding loads down the oscillator. It is particularly easy to do this on the higher frequency bands. If this occurs, remove a turn or two from the link winding. In operation, it is easy to overload the detector (as it is with all regenerative receivers). I use my Sproutie with a large outdoor antenna and find that on the lower bands, I usually only need to operate the set with the RF “gain” control set halfway.

The 15855 – 17850khZ coil stops about 50KHz short of the top of the 16M band, which is nominally 17480 – 17900KHz. However, all these coverage figures are quoted with the fine tuning control set to maximum capacitance. With the fine tuning control, I can tune all the way up to 17900KHz with that coil plugged in.

With the first set of coils I wound for specific bands, I was using significant values of fixed capacitance across L1 in order to reduce the frequency swing caused by adjustment of the main tuning capacitor. I noticed after a while that these specific band coils weren’t giving such good sensitivity as the general coverage coils. I have since discovered that it is best to avoid large values of parallel fixed capacitance, as this seems reduce the performance. Adding a capacitor of a few pF to tweak the coverage is fine, but large values (of the order of 50 or 100pF) will reduce performance. If you want to reduce the frequency swing to cover a narrow band, best to achieve it with the use of a capacitor in series with the main tuning capacitor instead. The performance of the coils in this receiver seems to be maximized by using as much inductance and as little capacitance as possible. This is more noticeable on the higher frequency bands.

The table for specific band coils is a work in progress. I will add to it as I wind more coils –

The coils were constructed in two different ways. The lower frequencies used a larger T68-6 core which I mounted with nylon hardware. I first took a #10 nylon bolt, cut the head off, and epoxied it into the hollow center spigot of the tube base thus –

Before adding the toroid, any jumpers and capacitors were soldered in place (this is going to be the 3050- 3950KHz coil). The soldering’s a bit messy, but it was the first time I had soldered one of these things –

A couple of nylon nuts followed, then a nylon washer, and then the toroid, topped off by another washer and finally, another nut –

The higher frequency coils used T50-7 toroids, and were mounted vertically and secured with a couple of dollops of hot glue from a glue gun. In the following picture, the 3050 – 3950KHz coil is on the left, the 14460 – 15980KHz coil in the middle (in a white ceramic tube base), and the 8040 – 10720KHz coil on the right.  The middle and right coil were pictured before the hot glue was added. Since building this version of The Sproutie, I have started using hot glue for the larger, lower frequency coils too, and it works fine. It is a lot faster than using the nylon bolts and nuts –

Here’s the 14460 – 15980KHz coil with the 2 dollops of hot glue to secure the toroid. I like these ceramic bases and think I’ll use them for all subsequent coils. Incidentally, here’s a quick hot glue tip. I don’t know what temperature the guns that have a single setting use, but if you purchase a dual-temperature gun and use the dual temperature sticks, the hotter setting allows the the glue to flow more freely before it sets, which makes building these coils a bit easier, and gives a better end result. I think the coils in these photographs may have been made with the temperature inadvertently set to the lower setting, giving me headaches while the glue was setting as to whether it was going to flow into all the places I wanted it to before it set! –

The coils for the higher frequency bands need less in the way of a link winding, such as 1/2 a turn, which is simply a piece of wire passing through the toroid, but not even being wound around it. For the 16M/17M coil, I found that a 1/2 turn from pin 1 to pin 7 wouldn’t allow the circuit to oscillate, so I used a simple u-shaped loop of wire between pins 1 and 7 placed near the toroid, as in the photo below. The link winding is the green wire. My attempt at using a 1/2 turn link winding for this coil involved a wire from pin 7 through the toroid to pin 1, and this stopped oscillation. However, it’s possible that a 1/2 turn from pin 7 directly through the toroid to pin 4, which is also at ground potential. might allow oscillation while coupling more signal into the detector (it’s a shorter run of wire). I didn’t try it though, opting instead to go for a loop outside of the toroid. Experimentation is definitely key here, and it’s one of the things I had in mind when building The Sproutie. Once you’ve built the receiver, you can still have plenty of fun designing coils for many different bands and amounts of coverage. Here’s that 16M/17M coil, showing the green link winding –

The coil for the 16M BC and 17M ham bands. This one covers 17400 – 18200KHz.

When designing coils for The Sproutie, here are a few things to bear in mind –

Adding capacitance across the coil will bring the overall frequency down, and limit the range of frequencies that the main tuning capacitor will cover (as the tuning capacitor is now just part of the overall capacitance across the coil). However, if you place too much capacitance across the coil, the circuit will not oscillate. When making estimates and performing calculations, remember to include the capacitance introduced by the circuit, and stray capacitances. Another strategy for limiting the range of frequencies the tuning capacitor covers is to put a capacitor in series with it (the tuning capacitor).  If I haven’t already mentioned it, the online calculators on W8DIZ’s site are great for figuring out resonant frequencies for tuned circuits involving toroids. The calculator for the T50-7 is here, and the menu to the left of the page has links for the pages for each of the other toroid cores. Each page also tells you what range of frequencies that particular material is good for. However, even after you think you’ve figured out what values you need for the inductor and capacitors, whether you’re going to use padders, series caps etc, you’ll most likely still have to do some tweaking of values until you get the coverage for each coil that you want, based on observation and experimentation. Once you’ve got the exact values you want, make sure to hot-glue the toroid to the tube base. If you don’t do that, you’ll experience instability and microphony. It’s amazing what difference a couple of dollops of hot glue will make!

For the chassis, I first looked at what was available in off-the-shelf sizes and couldn’t find anything that fitted the bill. Hammond have a good selection of different sizes, but their enclosures, for the most part, use 0.04″ thick aluminum. I wanted something thicker, for a very sturdy structure, so I decided to look into having a custom chassis made. A bit of searching turned up two businesses that manufacture aluminum chassis’ for homebrew tube amp enthusiasts – Dirty Dawg Amps, a US based business who are temporarily out of business due to a fire, and Seaside Chassis Design, who are located in Novia Scotia. Seaside Chassis use a minimum of 14 gauge aluminum for their enclosures. 14 gauge is about 0.064″, which I knew would make for a nice stout case.

Terry was very communicative and straightforward via e-mail about what he could do and what it would cost. I sent him rough drawings, with dimensions, of the chassis, front panel, and mounting bracket for the main variable capacitor that I was hoping he would be able to fabricate. He was able to make all 3 items and on top of that, he would punch all the main holes for me, leaving me just to drill the smaller holes for mounting screws. This was great news, knowing that I would shortly have a solid and well-made chassis on which to build this receiver.

I dropped the ball somewhat and didn’t take a picture of the chassis when it arrived, but here’s what it looked like with all the main components fitted, before wiring it all up. The 2 biggest factors in making this receiver look so grand are the National “N” dial with Velvet Vernier drive, and the excellent chassis. Does this look inspiring or what?

The controls on the upper row are, from left to right – regeneration, the main tuning knob, and the fine tuning. On the lower row, also from left to right is the headphone socket, RF attenuation, the bass boost switch (down = more bass) , the low-pass filter cut-off control, and the AF gain control.

Here’s a view from the back at this point in the construction. Look at that accurately made chassis, front panel, and capacitor mounting bracket. Terry from Seaside Chassis Design did a great job –

Both the RF board (without the extra AF preamp that was built later) and the AF boards installed but not yet wired up. All cables are tagged for easy identification –

Another view of the underside, before everything has been wired up –

The next task was to begin wiring the boards to each other and to the controls. Looking at this view of the underside, I’m thinking that I perhaps could have put a little more effort into dressing the cables more neatly, but it’s perfectly functional. The schematic shows pin 1 of the octal base being grounded but as I was wiring it up, I decided to also ground pin 4 –

Some more views of the underside from different angles and distances. I only twisted the 12V supply lines together for neatness and not for any electrical reason, though it does rather make them look like tube filament wiring🙂  Just to the left of the antenna socket on the right, is the phono jack for the line out. This is such a useful feature. In fact, as I write this, I am using the line out to record KCBS from Pyongyang on 11680KHz. On the other side from the BNC antenna connector, you can see the DC power jack with the reverse polarity protection diode and the RF bypass capacitors. Vinyl grommets were used for all wiring that needed to pass through the chassis. RG-174/U in the form of Belden 8216 was used for the connection from the BNC antenna connector to the board, and lavalier mic cable with 2 conductors and a shield for all other connections to controls (and to the phono jack) –

A view from the top, with a coil plugged into the octal tube base. The shaft couplers came from different sources. The one on the left, on the main tuning control, is a Jackson Bros part, purchased from Mainline Electronics in the UK through eBay. The coupler on the right was made from all aluminum by John Farnsworth KW2N. He has a small business making these and can also make custom sizes, if you have a non-standard shaft you want to use. For instance he just made a 3/16″ to 1/4″ coupler for my next project. John sells on eBay, but you can also contact him directly through his fledgling website (not yet finished) here. I really like his all-aluminum couplers –

The original intent was to mount the internal speaker on top of the chassis on the side using some kind of simple right angle bracket(s). I didn’t ask Terry from Seaside Chassis to fabricate a bracket for me because at that point, I didn’t know what speaker I was going to use. Looking around my room for something I could use, I noticed an unused LMB Heeger enclosure #143 in the size 4″ x 4″ x 2″ – exactly the same enclosure I used for the 31M version of the WBR. I figured that the top part of the box, being a U-shape, could be used as a bracket. If using that part though, why not use the whole box? There might even be some extra acoustic benefits to housing the speaker in a little case, and having it fully enclosed will protect it from dust and small bits of wire, metal filings etc being attracted to the speaker magnet (which happens here in the shack). The sound was a little “boomy” with the case closed, so I stuffed some foam in with the speaker, and it cleaned the “boominess” right up. Although you can’t really see it in these next shots, the speaker case is bolted to, and spaced off the chassis with 4 vinyl grommets to dampen any unwanted acoustic resonances in the chassis. The speaker wire enters the speaker enclosure through a grommet in the side. It’s a small detail, but the grommet is mounted not in a hole, but in a slot in the side of the cover. That way, when I remove the cover of the speaker enclosure, I can slide the grommet out, leaving the grommet still on the speaker wire, and allowing me to completely remove the cover –

There are a few improvements and modifications I’m considering making to The Sproutie but it is now completely functional, and this is how it looks at this point. I must say that I think it’s looking pretty good –

I was lucky enough to obtain a National “N” Dial in good condition and nice working order – not all of them look or operate this well. I bought several from Gary at Play Things Of Past and used the nicest one. The tuning knob and reduction drive are an important part of the feel of any receiver, and can do a lot to affect the operating experience so I’ll say a few words on that subject if I may. Before, I do, here’s a clip from a page of the 1947 National Radio catalog. I get a kick from seeing vintage parts in old catalogs, then seeing the exact same thing, in really nice condition, in front of me. It’s a bit like meeting a celebrity for the first time🙂

I was initially concerned that the 5:1 reduction ratio of the National drive wasn’t going to be high enough for accurate tuning on the HF bands – it was a good part of the reason why I chose the value of the main tuning capacitor and wound the coils so as to limit the tuning ranges to around 2MHz or less. This approach results in more coils, but really helps in creating a regen that can be set to a particular frequency, and from which you can read the frequency (with the help of a calibration graph – more on that later.) This receiver can be set to within a few KHz of any frequency. This is good enough for finding a particular SW AM broadcast station. I can also read the dial setting and then consult my custom calibration graph to find what frequency I am on to within a few KHz. It’s not much by modern standards, but is pretty good for a regen with an analog dial.

The National “N” Dial is marked from 0 to 100 and thanks to the vernier scale located at the top, it can be read to one-tenth of a point. These dials, when in good condition, have a firm yet smooth action with no backlash that makes tuning a receiver like this a good experience. Another thing to note is that these dials were manufactured with CW (clockwise) and CCW (counter-clockwise) characteristics, meaning that as you rotate the dial clockwise, the numbers either go up (CW type) or down (CCW type). This makes sense when you consider that variable capacitors were made as units that either increased in capacity as you rotated the spindle clockwise, meaning that the frequency went down (CCW type), or as units that decreased in capacity as the spindle was rotated clockwise, meaning that the frequency went up (CW type). The latter type is the convention for variable capacitors today. When we rotate our tuning knobs clockwise, we expect the frequency to increase. Back in the days when our predecessors thought more in terms of wavelength, they would have expected wavelength (instead of frequency) to go up with a clockwise rotation of the knob. This particular regen uses a CCW-type variable capacitor, so I married it up with a CCW-type vernier dial. It does take a while to get used to the fact that the frequency goes down when you turn the tuning knob clockwise, but I am beginning to adjust. There seem to be quite a lot of these lovely old National “N” dials around, if you take the time to look. A fellow homebrewer told me that his local electronics surplus store had a number of them in good condition for a very good price (I believe he bought them all!) Hamfests and swapmeets are also a good place to look. eBay is another possibility but the prices asked are a bit on the high side, in my opinion. I got mine from Gary at Play Things Of Past. He was easy to deal with.

Note – the National N dials have 3 small rubber/fiber bumpers installed on the mounting plate to prevent the front metal “flange” (the part with the engraved dial markings) scraping on it when the dial is turned.  If your dial is in good condition, none of the main parts are bent, and everything is running “true”, you can remove these bumpers. I did, and the result was a dial that rotated very smoothly. With the bumpers in place, there is a very slight scraping sound as the dial is turned. If you do this, make sure to save the bumpers in case you later wish to re-install them.

One downside to each coil only covering a relatively small part of the shortwave spectrum is that you end up with quite a few of them – more if you decide to wind specialty coils for specific bands. A coil box was definitely in order, so I headed to my local cigar and tobacco shop and purchased an empty cigar box. A trip to the local craft store yielded a length of bass wood, which is a little harder than balsa but can still be cut with a sharp craft knife. I cut slots in the lengths of basswood so they would slot together to form dividers to store the coils in –

The dividers installed in the cigar box, with the coils that had been wound so far (at time of writing this, I now have one more, for the 120M BC band) –

In order to know where you are on the band, you’ll need to calibrate your dial.  I accomplished this by plotting a graph for each coil with frequency on the x axis and dial markings from 0 to 100 on the y axis. For frequency references, you can use a crystal-controlled marker, or off-air signals and an online frequency database such as short-wave.info  Bear in mind that the setting of the regeneration control does alter the received frequency. Probably the best way to standardize your results is to keep the regeneration at or just below the point of oscillation at each dial setting that you take a measurement. The following is one of the graphs I am currently plotting. The original was larger and it is a little hard to read the markings on each axis on this smaller version. That’s fine, as your calibration will be different anyway.  In this graph, look at the line formed by the red dots (the black dots are something different – you can ignore them) –

The variable capacitor I used is what Hammarlund called a “midline” type in which the moving plates (the rotor) were mounted off-center so that the relationship between degrees of rotation and the resulting capacitance was non-linear.  The intent was to keep the relationship between degrees of rotation and frequency fairly linear and by looking at the graph, you can see that it is not bad at all. I also found that once a graph was plotted, I was able to set the dial and return to a particular frequency with a good degree of accuracy and repeatability. When listening to AM stations with this receiver and it’s bandwidth, which is of the order of 10KHz, you can be assured of returning to a dial setting and hearing the station you want.

The Sproutie does work on SSB and CW, but SSB reception is trickier due to the need to control the signal input level (with the RF attenuation/gain pot) in order to prevent overloading of the regen detector and pulling of the oscillator, and to adjust the level of regeneration in order to inject the right amount of carrier. If the signal level to the regen stage is too high, the oscillator will pull, resulting in the signal sounding “wobbly” due to FM’ing of the oscillator. This happens very easily, even with moderately strong signals. My preferred method of operating the set when listening to SSB stations is to run the AF gain at, or close to, maximum volume, and to keep the RF “gain” low. Sometimes, I need to keep the RF gain twisted almost to zero in order to achieve a nice stable demodulated signal. When adjusted properly, SSB sounds good on The Sproutie, but it takes a fair bit more work than with a superhet fitted with a product detector. Hams and shortwave listeners who have used older superhets that used BFO injection into a receiver with a diode detector will be familiar with the technique of keeping the AF gain up high, and using the RF gain to control the signal-to-carrier injection ratio. In this case, we are also using the RF gain to prevent the detector from being overloaded and pulling the oscillator. Sound tricky? If you’ve never done it before, it can take time to get used to, but after a while, it becomes almost second nature.

While I’m on the subject of fine tuning, allow me to expound a little more on reduction drives. In contrast to the friction drive on the National “N” Dial I used for the main tuning, the reduction drive on my fine tuning is a Jackson Bros 10:1 ball drive which has a small amount of backlash and feels a bit “spongy”. I don’t like it, and am grateful that for my main intended use of listening to AM stations, I won’t need it. I may change this ball drive for either another friction drive, or a different ball drive.  The Xtal Set Society sell 6:1 ball drives manufactured (I believe) by Oren Elliot) that have a more pleasing feel. EDIT – I have since found that making the 2 screws that hold this mini ball-drive to the front panel very, very tight seems to eliminate the backlash and reduce the spongy feel a bit.  I suppose it increases the pressure on the bearings a little. For the time being, I’ll keep this drive but if I ever change the front panel, that will be the point at which I’ll drill a bigger hole for a more conventionally-sized reduction drive.

For the above reasons, if I were intending to listen to more SSB and CW on this receiver, I would definitely wind coils to spread each entire amateur band over the whole rotation of the dial, and make sure I had a reduction drive for the fine tuning with very little or zero backlash and a better feel (though having said that, the regeneration control works very effectively for fine tuning).

Another thing that has often interested me is the bandwidth of regens. In general, as you approach the point of oscillation, the bandwidth becomes narrower, until it is at it’s narrowest somewhere around that critical point. As you continue to advance the regeneration, the bandwidth broadens out somewhat. I connected the output of a simple noise generator to the antenna socket of The Sproutie, and took screenshots while running Spectrogram, which was being driven from the line out jack of the receiver.  All 3 of these screengrabs were taken just marginally below the point of oscillation (the ideal point for receiving AM). The first one was with the low-pass filter adjusted for maximum bandwidth –

Well, it’s obviously not the kind of brick wall shape we might expect from a good crystal or mechanical filter but if you look closely, the passband is about 25dB down at the 5KHz point and 30dB down at the 10KHz point, That’s not too bad for AM reception, though if you wanted to get really serious about it, a passband of around 5 – 6KHz with a much steeper wall would, of course, be more ideal.

Here’s a grab taken with the low-pass filter pot at the median point, which is looking better –

…and better still with the low-pass filter set to the lowest cut-off point –

It is important to remember that these spectrums represent the response of the entire receiver, and not just that of the front end. SSB and CW signals become higher-pitched as you tune away from the center of the signal, but although the audio frequency make-up of an AM signal tuned off-center does change, the whole signal does not become higher-pitched. Therefore, an audio filter will be more effective at rejecting off-frequency signals for SSB and CW signals than for AM. Nevertheless, the adjustable low-pass filter is very good at cutting down much of the high pitched static that can make simple receivers like this quite tiring to listen to for long periods whatever mode is being received. It makes The Sproutie feel like a “grown-up” receiver!

Here’s The Sproutie, with it’s coil box. I would have felt as if I had died and gone to heaven if I’d had this receiver as a teenager. I’m feeling pretty good about it at my current age of 50🙂

The Sproutie is not completely finished yet (is any homebrew project ever truly finished?) The changes and additions I am considering include –

-Designing a thicker, and custom front panel with Front Panel Express, and making it a little wider than the current one to allow for a pair of instrument handles to be mounted

-Changing the 10:1 Jackson Bros reduction ball drive on the fine tuning control for something with less backlash and a firmer, less “spongy” feel

-Winding more coils for specific bands, so that the bands I am most interested in can be spread out over the entire dial, making tuning using just the main tuning control even easier

-I had also considered finding a local cabinetmaker to make a wooden cabinet for The Sproutie, but am not too sure about the convenience of sliding the chassis out of the cabinet every time I want to change a coil. If you’re thinking about building a receiver like this, completely enclosing it in metal would be quite a good idea – perhaps with a hinged flap or door on top for coil-changing. Regenerative receivers are quite sensitive, and this one picks up signals from my computer and/or my monitor, which are located nearby

It always feels good to build something that works, and The Sproutie certainly does that. It’s a great little receiver for shortwave listening and with an extra tube base, a toroid, some wire and a few extra capacitors, you can add whatever frequency coverage to it you like as you go along.

Oh – and I just realized I didn’t explain that Sproutie is the nickname I gave my 2 1/2 year-old cat Sprout, whose ham radio name is Sprat The QRP Cat. I had already named one of my home-brew radios after a kitty I used to have called Rug, so figured it was time to honor Sproutie in the same way.

Sprat The QRP Cat aka Sprout aka Sproutie, after whom this receiver is named.

Videos of The Sproutie in action are here.

Thoughts on using The Sproutie to receive SSB are here.

Sproutie is ever-curious, just like her regen namesake is always seeking out signals.

I continue to update and re-write this post, as I make additions and improvements to The Sproutie. I would rather incorporate them here, rather than into subsequent, and separate posts. That way, if you are thinking of making your own version of this receiver, you can get all the latest updated information by reading this one (rather long!) post, instead of checking my entire blog for updates. As of February 2015, I have filled the entire cigar box with coils, with one to spare (that one is plugged into the receiver). The latest coil I made was for 17400-18200KHz to cover both the 16M BC band and the 17M ham band. 16M is just about the highest frequency shortwave BC band that is in regular use so I now have all the SWBC bands covered – most of them with coils specific for the individual bands. Here’s what my coil box looks like now. The unmarked coil sitting at the top of the box is the experimental one for 24-29MHz –

This cigar box cost $5 from a local tobacconist and makes an excellent coil box. The unmarked ceramic tube base sitting on top is the experimental coil for 24-29MHz. 4 adhesive vinyl bumpers stuck to the bottom help to protect it from rough surfaces.

May 11, 2013

The NA5N Desert Ratt 2 Regen

EDIT – If you’re thinking of building the Desert Ratt 2, although the pictures in this post are numerous and quite large, I do recommend reading all the text too, as I have included what I thought were relevant details on the construction as part of my narrative. Also make sure to read the comments and replies.  Previous blog-posts have taught me that readers often ask pertinent questions, so you may be able to glean a little more information from them too.  In fact, just before I wrote this, Paul NA5N made a comment which includes a usefiul piece of information about the 2 x 1,000pF (0.001uF) capacitors in the regen stage.

EXTRA EDIT – Please read the update at the end of this post, after the videos.

I’ve been wanting to build NA5N’s Desert Ratt regen ever since I first found his very attractively drawn schematic for it online. I then found the updated version, called the Desert Ratt 2, and a very good description of how the circuit works – all of these documents available on Paul’s website. What more could an avid regen builder want? Not much, it turned out. Late last year, when N2CX and N2APB dedicated an episode of Chat With The Designers to the Desert Ratt (and to the subject of regens in general), I just had to listen and of course, it fueled my interest in building the DR2 even more. The whiteboard for this particular episode of CWTD is here, and the podcast audio is here.

The WBR was a successful regen for me and while it worked well on SSB/CW, it didn’t seem to quite have the gain with AM stations. This makes sense, as a regenerative detector has to be set below the point of oscillation for AM reception, at which point it has less gain than when it is oscillating (which is where you set it for SSB/CW reception.)  Even so, I had read that bipolar transistors tend to work better as regen stages for AM, as they have higher gain when not oscillating. The search was on for such a receiver, and this was one of the key deciding factors in building the DR2 for me. In fact, Paul has mentioned (I forget where I saw it, as I have done so much reading on this receiver) that the Desert Ratt doesn’t do so well with SSB/CW as it does with AM. My experience with it backs up this assertion, thought it’s a pretty neat receiver for AM.

In particular, I wanted a receiver for covering the 49M SW BC band as although my Elecraft K2 covers a few of the BC bands, 49M is not one of them. There were a few things I found interesting about the design. The use of a phase splitter transistor to convert the single-ended output of the detector to a balanced output in order to drive the LM386 in differential mode was novel. Paul talks about how much RF is flying around inside regen receivers, and how the common-mode rejection of the 386 when used in differential mode can be advantageous in such an environment. I was also intrigued by the detector consisting of 2 germanium diodes – I think I was just looking for an excuse to build something with Germanium diodes again to remind me of my crystal-set building days as a kid🙂

If you look at the schematic of the DR2,  you’ll see that one of the changes in the design from the original DR is that instead of a variable capacitor, it uses 1N4004 diodes as varicaps. I have a bit of a “thing” for nice air-spaced variable capacitors, and I had in mind a nice Millen 50pF capacitor that I picked up on eBay for a very fair price last year. Combined with a 6:1 reduction drive, it made a good combination with a very useable tuning rate for tuning in AM stations.

Anyway, I’m getting ahead of myself here. I did make a few changes to the original schematic for my version, so allow me to introduce my rather wobbly circuit diagram –

The differences between my schematic and Paul’s are as follows –

– I added an RF attenuation pot at the antenna input. After building the DR2, I found that using a relatively short piece of wire indoors as an antenna was causing a lot of common-mode hum.  On top of that, I wanted to be able to increase the signal level into the receiver with the use of my regular outside antenna (A 40M dipole fed with 300 ohm balanced feeder.)  Using the attenuation pot allowed me to use the large outdoor antenna without overloading the receiver.  Use of my outdoor antenna created enough separation between the receiver and antenna that the hum problem almost entirely disappeared.

– Earlier versions of the Desert Ratt included instructions for winding the coil on a plastic 35mm film canister and on an IC shipping tube. The DR2 schematic doesn’t include such instructions, but I wanted to use a toroid, so I experimented a bit and came up with a scheme that seems to work OK.  I used a T68-6 former and the turns info is on my schematic above – a T50-7 would take up a little less space. More about this later.

– I had a few 2-position center-off switches that I wanted to use, so I used one of these for a bandswitch instead of the SPST switch in NA5N’s DR2 schematic. I had originally thought that using the 50pF tuning capacitor with no padding would make the upper limit of frequency coverage too high, resulting in too large a frequency swing in one band, but there must have been more stray circuit capacitance than I had anticipated, as the coverage with no extra padding was about 7.3 – 13MHz. This band became the center position.

– I was attempting to power the DR2 from my shack power supply, which is about 45AH of sealed lead acid batteries with a float charger constantly connected.  This also powers my K2, and the DR2 was picking up processor noise from the K2, as well as a low-frequency “burbly” kind of noise of undetermined origin. The problem went away when I powered the receiver from a separate SLA. but I decided to add extra filtering to the power line anyway.  I found that a 1mH choke as well as a 1,000uF electrolytic almost (but not quite) got rid of the unwanted interference on the power line.  For good measure, I added a 0.01uF RF decoupling capacitor across the power line at the input connection.

– I added an AF preamp stage directly after the diode detector to ensure enough power to easily drive a speaker – even with weak signals.

– The inputs to the LM386 are the opposite way around from the way indicated in NA5N’s DR2 schematic.  With the inputs connected as shown in Paul’s diagram, the LM386 emitted a loud screeching sound.  Swapping the inputs cured this. I was not the only person who had this problem, as I discovered from this post in the GQRP Yahoo Group (you need to be a member of the group to read the post).

–  I left pin 7 unconnected. I don’t understand the way that NA5N has it connected to the junction of the series resistor and capacitor connected between pin 5 and ground in his diagram.  Most circuits that use pin 7 call for a decoupling capacitor direct from pin 7 to ground (usually about 10uF).  This helps reduce large signal distortion, though Paul does say that in this application, it may not do a great deal to help and is therefore optional.  I elected to leave it unconnected.

Now for some pictures.  I didn’t want to spend a lot of time constructing an enclosure, so decided to make a simple PCB L-shaped chassis and build the circuit directly onto that.  With the variable capacitor mounting bracket, it still ended up taking quite a while to construct though. All my projects begin like this, with the main components and control being laid out on the front panel, while deciding on the basic layout –

I’ll spare you the words at this point and apologize for all the pictures that are about to come. If you’re living in a remote area and are still relying on dial-up, then I feel a bit sheepish about the sheer number of images to follow!  I’ve talked before about constructing enclosures from PCB material, so won’t repeat that information here. As well as constructing the chassis from PCB material, I also made a mounting bracket for the variable capacitor and a tuning pointer to attach to the reduction drive with 2 small screws – all from double-sided copper-clad laminate.

I applied several thin coats of lacquer from an aerosol spray.  It was sprayed from a distance, resulting in a light, and stippled coating, which you can see in these pictures. I’d rather apply too light a coat than risk overdoing it. The downside of this is that oxidation will being to affect the appearance of the copper fairly soon. Oh well. The capacitor mounting bracket received a thicker coat. You can see the smoother, shinier finish.

I got the 6:1 reduction drive from Midnight Science. A number of others sell them, and one place that springs to mind is Mainline Electronics in the UK. They are the suppliers for Jackson Bros components (I think they have the rights to manufacture and sell the parts).  They sell on eBay using the name anonalouise.

The enclosure looked a little bit different by the time the DR2 was finished, as the hole for the nylon toroid mounting hardware hadn’t been drilled in the base at this point.

Look at that gorgeous variable capacitor!

A close-up view of the Millen 21050 50pF air-spaced variable capacitor and mounting bracket. This component is silver-plated (the vanes are probably brass), and has double bearings and a ceramic base. It is a very nice variable capacitor, and had never been soldered to before being used in this project. It is at least 35 years old – most likely older!

Boy, was I glad to finish the chassis so that I could start work on wiring it all up.  I decided to build the AF amp first and work backwards, my thinking being that the AF amp would be relatively straightforward. The act of touching the input with a metal screwdriver and hearing a hearty buzz in the loudspeaker would give a welcome psychological boost! If I started by building from the antenna end, I’d have to wait until the entire receiver was built before getting any clue as to whether it was working.

Here’s the chassis with the LM386 amp, the 2N3904 phase splitter, and the 2N3904 preamp built. As has been the case with all my projects since I started using then, I used W1REX’s wonderful MePADs and MeSQUAREs to build the circuit –

Here’s a close-up. The 2N3904 preamp is just below the 6:1 reduction drive, and the 2N3904 phase splitter is to the left of the LM386.  The 100uF capacitor that decouples the supply line to the LM386 straddles it. I read that it is best to ground it to pin 4 instead of to some other point on the chassis to avoid instability, hence the reason for this placement. The other electrolytic that is straddling the chip is the 10uF capacitor between pins 1 and 8 that sets it to the maximum gain of 46dB. The black shielded cable connecting the AF gain pot to the circuit on the PCB is lavalier mic cable.  It has 2 conductors, each of them in it’s own shield, which is ideal for wiring up potentiometers. It is fairly thin and very flexible. I use it in all my home-brew projects. I bought it from a local pro-audio store which recently closed down, so will now need to find another supplier.

In this view, you can clearly see the extra DC supply line filtering that I added, consisting of a 1mH choke in series with, and a 1,000uF electrolytic across, the DC supply. After seeing these pictures, I noticed that there wasn’t very much solder on the joint connecting the choke to the power jack, so I re-flowed the joint and melted a bit more solder onto it.

The power indicator LED’s main function is as a voltage regulator. NA5N marked the various voltages on his schematic for the DR2, and I chose an LED with a forward voltage drop to match those voltages as close as I could.  A green LED in a variety pack I got from Radio Shack had a forward voltage drop of 2.1V, which seemed about right.  The 1N4148 had a forward drop of about 0.65V.

The next stages to be built were the detector and impedance converter/buffer stages.  The description of the DR2 on NA5N’s site gives more info on these stages (as it does for the whole receiver). I couldn’t be sure these stages were working, but bringing my finger close to the diodes resulted in a pleasing cacophony of stations in the headphones – and at a louder level than in doing the same to subsequent stages, so I figured there was some detection/amplification going on🙂

I didn’t know how many turns I was going to use on the toroid, but using the calculator on W8DIZ’ site and an online resonant frequency calculator, I figured that 36 turns on a T68-6 should be a good starting point for the whole winding from pin 3 to pin 6. In Paul’s version, with the coils wound “traditional style”, the tickler winding was about 1/3 of the whole winding.  Coupling between windings is tighter with a toroid than a “regular” coil, so I reduced the number of turns on the tickler. I found that regeneration was occuring at only about 25% rotation of the regen pot, so further reduced the number of turns. Using the turns shown on my schematic at the beginning of this post,  the regen stage moved into oscillation at anywhere between 40 and 50% rotation on the pot, so I left it at that. For the same reason of tight coupling, I used fewer turns on the antenna winding too and because I am using an outdoor antenna, could probably have used even fewer turns.

The toroid was fixed to the PCB with nylon nuts, bolts and washers that I got from my local Ace hardware store.

Here are some pictures of my Desert Ratt 2 with the circuit finished –

The red wires running along the back of the front panel are the regulated 2.1V and 2.75V lines.  I would have run them on the main board but ran out of room due to lack of planning, so went vertical.  Incidentally, although I refer to the 2 regulated lines as 2.1V and 2.75V,  the exact voltages aren’t important.  That’s just what they turned out to be in my case.

The RF amp and regen stages can benefit from transistors with high hfe. I got a cheap Harbor Freight DMM that measures hfe from an eBay vendor for under $6 including shipping.  hfe varies depending on the collector current, but I was doing this mainly for comparative purposes rather than absolute values, so the fact that I didn’t know what value of collector current was used to measure hfe in this cheap meter didn’t matter. It just so happened that my 2N2222A’s tended to have higher hfe than my 2N3904’s, so I ended up using a 2N2222A that measured in at hfe = 203 for the RF amp, and a 2N2222A with hfe = 223 for the regen stage.  The other stages don’t require high-gain transistors. NA5N talks about it in this post on QRP-L from 1999. Bear in mind that he was talking about the original version of the Desert Ratt in this post (just so you don’t get confused when he identifies the various transistors).

I did promise that I’d give a bit more detail on the toroid. Mine was wound on a T68-6 former. The main winding was 30 turns tapped at 27 turns from the top (3 turns from the bottom). The antenna coupling winding was 5 turns.  All turns are wound in the same direction. I used 26 gauge wire, but the precise gauge isn’t important. 26 gauge was narrow enough to easily fit all the turns on the former, yet stout enough to lend some stability to the oscillator, as the toroid isn’t sitting close to the board, and the leads are relatively long. When putting taps on coils, I used to not cut the wire i.e. I would simply make a loop in the wire, twist it, tin the twisted part and keep on winding.  Now I find it is easier to treat them as 2 separate windings connected together. If you can get heat-strippable wire, please do – it makes winding toroids so much easier and more pleasurable.  I wound the first winding of 27 turns, stripped and tinned the end, then stripped and tinned the end of another piece of wire, twisted and soldered them together, and carried on winding the last 3 turns in the same direction (this is important).  The separate antenna winding of 5 turns is also wound in the same direction.  I’m afraid I didn’t write down (or if I did, I have since lost it) the lengths of wire used. I did notice that the turns calculator on W8DIZ’ site (linked earlier in this post) was quoting lengths that are too short for the T68-6 former.  All you have to do is wind one turn around your former, measure that length, multiply it by the number of turns you’re going to wind, add an extra inch or two for the leads and, as we say in England, Bob’s yer Uncle and Fanny’s yer Aunt (meaning – you’re home free!)  When winding toroids, I often find that the first 1 or 2 turns aren’t quite as tight as the rest so when I’ve finished winding, I will unwind one turn from the beginning of the coil, then wind an extra one at the end, to keep the total number of turns the same.  Sometimes I will repeat that exercise a few more times until all the turns are nice and tight.  For this reason, I use enough wire to leave several extra inches at each end.

The next picture shows an anti-hiss filter that wasn’t in the earlier pictures, which I tried and ended up removing due to a low-frequency oscillation it was causing at the higher volume settings.   It was a series 0.01uF capacitor and 4.7K resistor connected from pin 1 of the LM386 to pin 5.   From what I have read, too low a value of resistor or too high a value of capacitor can cause the oscillation. I have seen other anti-hiss filters that used a 0.01uF cap and a 10K resistor, so it is very possible those values would have cured my problem. However, I was near the end of the project and itching to move on, so I just removed it. You can also see the 0.1uF capacitors on the inputs of the IC that have been swapped over to stop the uncontrolled oscillation, and are now crossing each other.  You may not have to cross these caps if you plan your layout accordingly –

Other than the problem with the loud screeching that was solved by swapping over the inputs to the LM386 (my schematic reflects the way the inputs were finally connected), the only other problem I had was with what appeared to be a defect in the 0.001uF (1,000pF) capacitor that leads from the tap on the coil to the emitter of the regen transistor.  I wasn’t getting any regeneration at all but on replacing this capacitor, the circuit broke into a nice loud hiss when advancing the regen pot.

I do have one ongoing issue that I hope someone can shine a light on for me, and that is a loud crackling sound when adjusting the tuning capacitor. At first, I thought a dirty rotor connection was the problem, but it only happens when extra padding capacitance is switched in by the band-switch   With no extra capacitance switched in, the tuning is smooth, but on the lower frequency bands, the receiver crackles when being tuned.  I need to try bypassing the band-switch and soldering the padding capacitors into circuit in case the switch is the problem. I’ll report back when I’ve done further work on this.

Incidentally, the main tuning range on mine covers approximately 7250 – 13000KHz.  Switching in a 47pF capacitor changes the range to 5825 – 8050KHz. I’m a bit limited with my receiver and test equipment here, so haven’t yet been able to determine the coverage of the lowest frequency band.

When first listening to the DR2, I had no idea what frequency I was listening to – only that I was probably somewhere between 5 and 12 MHz. I had no antenna connected (and at this point, hadn’t even built the RF amp stage) but started hearing CW. Lo and behold, it was Hank W6SX 180 miles away from me in Mammoth Lakes, CA. His CW signal was coming through well and in fact, this was the only time I have received CW in a satisfactory fashion on the Desert Ratt. There was no antenna – he was being picked up directly by the toroid.  Any concerns I might have had about the sensitivity of this receiver would have been immediately allayed.

I know the main question that is probably on your mind is – how does it sound, and what is it like to use? How does it “handle”? There are some videos of my Desert Ratt 2 in action at the end of this post. Apologies for the poor video quality, but my only video camera is 10 years old (and has a faulty CCD sensor). You’ve probably read articles about regens that describe the many and subtle adjustments that need to be made when tuning a regen in order to coax maximum performance from it. If you haven’t operated a regen before or if it’s been a while, it does take some time to get the hang of getting the best out of it. As you get further away from the setting of the regen pot where it breaks out into oscillation you lose selectivity and gain, so you need to try and keep the control set just under the point of oscillation. Loud stations can overload the detector, resulting in audio distortion, so it’s worth keeping an eye on the RF attenuation pot too. Also, if the attenuation pot is set too high (too little attenuation), you may get breakthrough from stations on other frequencies. There’s quite a bit going on to keep under control, but if you manage to keep all controls adjusted well, you can coax some pretty decent performance out of the set. I think this is why regens appeal to some people – we are incurable knob-twiddlers!

Stability is easily good enough for AM reception and with a logging scale fitted to the front panel, I don’t think it would be hard to find specific frequencies, as the majority of SW BC stations stick to 5KHz channels. In my casual listening so far, I have heard The Voice Of (North) Korea on 9435 and 11710KHz, Radio Habana, Cuba on the 49M band, Radio Australia on the 31M band, coastal station KLB (South Korea) on 8636KHz, the BBC World Service (forget which band or frequency), China Radio International on 9790KHz, WTWW on 5830KHz, and a number of other evangelical Christian stations (sorry, I tune them out and don’t pay them much attention.)

To sum up, you can definitely have a lot of fun and engagement with the bands on this set.  Being a regen, it is not the easiest receiver to operate, but you shouldn’t let that put you off. The best analogy I can think of is to reference the way that although an older British sports car may not have the finesse and performance of a newer sports model, it’s a lot of fun, and it’s lack of suspension gives you an exhilarating feel for the road that the more expensive cars cannot.

The Desert Ratt 2. A logging scale fixed to the front panel would make frequencies in the SWBC bands easy to find. I must do this sometime🙂

Please note that in the following videos, an MFJ-281 ClearTone speaker was used. My understanding is that this speaker has a slight resonant peak at around 700Hz (helpful for CW) and a relatively restricted overall bandwidth that is good for communications applications. This probably means that it’s not optimum for getting the maximum fidelity from an AM SW broadcaster (not that those stations have a lot of fidelity, but they tend to have a bit more than your average SSB transmission). On top of that, the audio was captured with the built-in mic in my old Canon A80 compact. Please don’t judge the quality of the Desert Ratt 2 audio from these clips. It’s better than this! I’m working on a few audio only recordings that will better demonstrate what the DR2 sounds like, and will put them up in the next blog-post (hopefully within a week or so).

Update – It has been about a year since I built my version of the Desert Ratt 2 and I feel compelled to provide an update. Whenever I first build a project I am often so thrilled that it works at all, that I tend to gloss over any shortcomings, particularly in my blog write-ups. Some of this is due to the possibility that any deficiencies are due to my layout and construction, as opposed to a problem with the circuit design. In the case of my DR2, I am still not sure whether the issues arise from the circuit itself or from my construction, as I have only built one of these. I did, however, want to document what I have observed, as my DR2 has laid on my shelf for the past year, largely unused, while I drag my WBR out and take it for a spin on a regular basis. Here are the issues I have observed –

* There is a lot of scratchiness in the speaker when tuning the DR2. This happens on some frequency ranges more than others, but it happens a lot.  At first, I wondered if it was due to inadequate grounding of the rotor plates but I don’t think this is the case. There is a solder tab for both the rotor and the stator, and the rotor is grounded to the chassis by a direct wire. Also, it is a quality Millen variable capacitor, and it is clean (the oxidation has been cleaned off).  I’m still considering the possibility that it as something to do with my variable capacitor, or the way that I have connected it.

* The set does seem to overload very easily on my outside antenna. Breakthrough from other frequencies is a common occurrence. This got me to thinking about the RF amp stage. The instructions call for picking a high hfe transistor to use in this position but thinking about this, I’m not sure why. Surely the purpose of an RF stage in a regenerative set is to provide isolation between the detector and the antenna, with gain actually being undesirable, due to the tendency of the detector to overload? The more I think about it, the more I think that different configuration for this RF stage would be more appropriate.

* Hum, though not always apparent, does still occur from time to time.

A commenter who goes by the name of Mast does mention that the tank circuit is very tightly coupled to the collector of the regen transistor. I’ll cut and paste his comments here, as I now wonder why I didn’t pay more attention to his input at the time,

“A nice schematic for general use. But the tank circuit is tightly coupled to the collector of the regenerative stage. You will suffer a lot from changing internal stray capacitances of the transistor when setting the regen level. And strong SSB signals will change these capacitances too, causing an unintelligible reproduction of SSB signals.”

At this point, the DR2 has gone back on the shelf while I move on to planning other projects, but I’d be very interested to hear what the experiences of others have been with this circuit.  I know there are folk who found N1TEV’s beginner’s regen to be a little hard to tame – and the DR is based in part on that circuit.  In contrast, both my WBR’s are well-behaved, and have been used regularly since I built them.

July 27, 2012

Using the NM0S Hi-Per-Mite Filter From 4SQRP To Make A Simple 40M DC CW RX

When 4SQRP brought out a version of the David Cripe NM0S-designed Hi-Per-Mite Filter, I noticed that it could be configured to give up to 50dB of gain. Like many others, I’m sure, my next thought was that it would make the process of building a simple direct conversion receiver even simpler. I decided that at some point, I needed to use this little filter board as the audio chain in a direct conversion receiver for CW.  I wanted to see for myself how it would sound.

In the meantime, my preoccupation with direct conversion receivers continued, as I embarked on building a G3WPO/G4JST DSB80. The receiver in that DSB rig uses an SBL1-8 DBM. That particular mixer package is out of production, but I used an ADE-1 and was pleasantly surprised by the receiver. I had some problems with the TX section, so put it aside and started building a ZL2BMI DSB transceiver.  I’d known about this little rig for a while from the pages of SPRAT, but a comment on Twitter from NT7S encouraged me to have a go at building it. The receiver is a simple NE602/LM386 combination with just a single-tuned filter on the antenna input. Of course, the receiver is not THAT simple, as each chip contains more complex circuitry, but the schematic at the chip level looks almost too simple to work.  As with the DSB80, I had problems with my ZL2BMI rig transmitting a sizeable residual carrier, but was surprised that the receiver sounded pretty good. The seed was sown…..

The ZL2BMI DSB rig was actually only the second time I had built an NE602/LM386 direct conversion receiver. The first was about 20 years ago when I bought a kit for the Sudden receiver on 20M. At the time, I wasn’t that keen on it and it soon ended up in pieces. I’m not exactly sure why I wasn’t taken by it, but I think it was a combination of factors. Firstly, I was living in an apartment in Hollywood with no outdoor antenna and was using just a short length of wire indoors as an antenna. It’s also highly possible that band conditions weren’t great, but I just remember not hearing many signals and also not liking the fact that a half-turn of the variable capacitor covered the entire 20M band. Because of this one not so great experience, I have since harbored a bias against using NE602’s in direct conversion receivers, an attitude that I now realize was unfair.

After achieving only partial success with both these rigs but enjoying the fact that the direct conversion receivers in both of them worked quite well, I wanted to build a simple receiver and use the Hi-Per-Mite kit that had been sitting patiently on my shelf for the last few weeks.

Look at the schematic for any NE602/LM386 DC receiver, and they are almost all the same, differing only in whether they use a crystal, ceramic resonator, or free-running VFO for frequency control, whether they have a single-tuned or double-tuned antenna input filter, or whether they use a differential or single-ended audio output.  The circuit is so simple that there is only so much room for variation, but there were a couple of things I had in mind for my mine.  While the receiver portion of the ZL2BMI rig only has a single-tuned filter on the antenna input, I wanted to do what G3RJV did with the Sudden, and place a double-tuned filter there. It’s an easy thing to do, and living in a built-up urban area, I have a lot more RF around me than ZL2BMI does when he’s using his rig in the bush, so a bit of extra bandpass filtering certainly couldn’t hurt. The other decision to be made was the method of frequency control. I was curious to see how the internal oscillator in the NE602 would work as a free-running VFO.  My apologies to you folk who have built umpteen simple NE602-based receivers and have-been-there-and-done-that.  I guess I need to get this out of my system! Another thing that I had seen in other circuits but hadn’t tried was the use of a 1N4001 diode as a varactor. Some call it the poor man’s varactor. Sounds like it was custom-made for me!

So that was my configuration – double-tuned input filter (just like the Sudden) and a free-running VFO tuned with a 1N4001 diode.  From what I’d read, I knew that I should be able to get enough capacitance swing from the 1N4001 to cover at least the bottom 30-50KHz of 40M. Here’s what I ended up with:

There’s nothing original in this schematic  but as you’ll see soon, it did turn out quite well, so I decided to name mine “The Rugster” after my cat, whose full name is Chloe-Rug, but who gets called several different variants of that name, of which “The Rugster” is one.  For the input bandpass filter, I used the vales of L and C that are used in the GQRP 40M Sudden Receiver. The pre-wound inductors in that design have a value of 5.3uH. I decided that I wanted to use T37-2 cores, so I used the calculator on W8DIZ’s site to figure out that I’d need about 36 turns on a T37-2 core to give 5.3uH of inductance. I fiddled around a bit with the values of the parallel fixed capacitors before settling on 47pF for those. I had originally thought that a higher value of around 68pF should work, but that didn’t put the peak of the filter somewhere in the mid-range of the adjustment of the trimcaps.  Some experimentation revealed that 47pF fixed caps achieved this.  Your value may vary depending on what value of trimcap you are using.  For the VFO inductor, I found a design for a 40M DC RX on the internet that was tuned by a 1N4001, noted the value of the inductance, decided that I wanted to use a T50-7 core (for stability) and once again, used Diz’ site to calculate the number of turns needed.

When building receivers, I like to start with the audio chain first. It’s quite affirming to be able to touch the input and hear that reassuring loud hum and mixture of AM broadcast stations in the speaker! So I got out the 4SQRP Hi-Per-Mite kit. Close inspection of the PCB revealed a small problem that was easily remedied. There was a very, very narrow copper trace connected to the ground plane that had been bent so that it was contacting the input pad.  A DVM test confirmed that the input pad was shorted to ground. I didn’t take this picture until I was part of the way through removing the offending trace, but you should be able to see the problem:

A minute or two with a sharp craft knife and that pesky little trace was history. Here’s the finished board:

The only thing that prevented me from putting this kit together even more quickly was my own compulsion to solder the components into the board so that they line up as straight as possible. The holes for some of the capacitors were a bit larger that they needed to be so that if you just plug them into the holes and solder, they’ll be a bit off-square. It’s not a big deal, and most builders won’t be bothered by it, but I just can’t control my OCD tendencies sometimes. Anywhere, I finally got there:

Although with modern low-flux solders, there is no need to remove the flux, it sure does make a board look nicer. I recently started using flux remover on my boards so that I can show off the underside too. At the suggestion of W2EAW, I apply it with a cheap plumber’s flux brush obtained (for 19c) from the local hardware store. Look at that nice clean board!

Trader Joe’s sells Green Tea Mints in a clear-top case that fits the Hi-Per-Mite board well if you’re looking to use it as a stand-alone device. There’s also room for power and in/out connectors (though you might need to cut off the top corners of the board in order to make room for the connectors.)  The tin in this shot had been knocking around my bench for a while, so the clear-top is a bit scratched up, but you can see the potential:

The next step was to build the direct conversion receiver with the NE602 and some of W1REX’s MePADS and MeSQUARES. The most frequency-sensitive part of the VFO circuit was built ugly-style, with the help of a 10M resistor for a stand-off, to help the stability. I don’t have any pictures of the board without the Hi-Per-Mite audio filter/amp added, so here’s the finished board. In the earlier shots, you can see the VFO toroid before I decided to slather it in hot glue in a bid to make it a bit more stable. There is only one component in the final receiver that I missed from the schematic, and that is a 1N4001 diode in series with the 12V supply for reverse-voltage protection.  I’ve blown too many fuses in my shack supply to not use these little fellows in the future🙂

Another shot from basically the same end of the board, showing the input bandpass filter on the left, VFO toroid on the left at the back and the DC input circuit nearest you on the right (reading from right to left, you see the reverse-polarity protection diode,  10uF input smoothing capacitor, 78L08 voltage regulator, and the 0.1uF DC output bypass capacitor. I’ve had those green trimcaps for a long time and it was time they were used. They don’t take solder well (I think I got them from Radio Shack about 10-15 years ago) but it seemed a shame not to use them in something:

Here you can see the varactor tuning circuit built ugly-style just to the left of the VFO toroid. The 3 connecting cables are for the AF gain pot (which is connected in place of R11 and R12 in the Hi-Per-Mite, as per the Hi-Per-Mite instructions), multi-turn tuning pot, and AF output jack. I’ve used lavalier mic cable that I bought from a local pro-audio store. It has 2 inner conductors shielded by a single outer braid and is very flexible, which makes it useful for wiring up small radios.  RF connections (which you’ll see in later photos) will be made with Belden 8216 coax:

At this point, I was very gratified to discover that the receiver actually worked, and seemed to work quite well too. That fact gave me enough inspiration to start building an enclosure to house it in.  The next shot isn’t ideal, as the back of the case is a little out of focus. You’ll notice that the front, and the two bolts sticking up on the right side are in focus, but the back of the case (and especially the bolt at the back left) isn’t. When I realized this, I was too far along installing the board to go back and take another shot and besides, sometimes I just have to control myself and keep forging ahead:

A view of the board installed in the case, with the lid:

A closer view from the top, in which you can see the VFO toroid now doused thoroughly in hot glue.  I’m not sure if it helped the long-term stability at all, though it did help to make the VFO more resistant to physical shock. Another way of securing the VFO toroid in a vertical position would have been to drill 2 small holes in the PCB and use a small nylon tie:

The coax that leads from the 1K log RF attenuator pot to the board is a little long, and that is because the one you see in the photos is a 10K linear pot. It was all I had at the time of building.  I am going to swap it out for a 1K log pot when the next order arrives from Mouser in a few days, and kept that lead a little long in case I need to chop some of the end off when changing the pot out.

Here’s another view of the innards. I am quite pleased with the way this little receiver turned out. You can see one of the 4 brass nuts that are used to secure the lid to the rest of the case. I use brass, as I can solder to it with regular solder. When placing the nut, and before soldering, I screw the 4-40 machine screw through the side of the case into it so I know the nut is in the right place.  If I were to get a little over-zealous when applying solder and accidentally get solder on the screw it wouldn’t stick, as the screw is made of stainless steel:

This is what The Rugster looks like in her case with the lid on:

If you look closely at these pictures, you’ll see that the PCB panels don’t line up perfectly and that the edges are just a little rough.  I am somewhat detail-oriented, but I am not a craftsman by any means, and I know my limits.  I made a conscious decision not to spend a lot of time filing and sanding edges. After making deep scores in the board and breaking it, I ran the edges across a file on the bench a few times and left it at that. Good enough is good enough in this case, and it’s certainly good enough for me.

A view of the underside:

I’ve spent a few evenings listening to it and most of the signals that my K2 could hear, this little receiver could hear almost as well. With the very weak signals the K2 won, of course.  A small part  of this could be due to the fact that the DC receiver, even with it’s 200Hz-wide filter, is still at a 3dB disadvantage to a single-signal receiver (such as a superhet), as it is hearing on both sides of the local oscillator.  Naturally, I expected my K2 to be the better receiver (duh!) but was really surprised at what an enjoyable experience listening to this one is for such a simple circuit.  As expected, it does overload in the evenings, but all I have to do to get rid of the breakthrough is to adjust the RF attenuation pot back a little from “full gain”, and the breakthrough disappears.   Elecraft use an NE602 in the front end of the K1, and it has an attenuator for the same purpose. I’m not sure whether the breakthrough is from strong in-band amateur signals or from AM broadcasters – it’s actually hard to hear the content due to the narrow bandwidth of the Hi-Per-Mite. One small adjustment of the RF gain pot and it’s gone – then I just bump up the AF gain a bit to compensate. Volume is enough to drive my MFJ-281 ClearTone speaker.  I normally like to listen to CW with a 500Hz note, but because the response of the ClearTone drops off below 600Hz, I kept the Hi-Per-Mite filter at it’s design center-frequency of 700Hz, and the speaker sounds good. EDIT – If the AM breakthrough is from signals in the 550 – 1700KHz AM broadcast band, which I strongly suspect that it is, a simple high-pass filter with the cut-off set somewhere around 2 or 3MHz should work wonders to eliminate this.

Another point I wanted to mention is that even though the Hi Per Mite employs an LM386 for the AF amp, it is used in the low-gain configuration. Because of this, you don’t hear any of the hiss that is heard in other simple receivers that employ a 10uF capacitor between pins 1 and 8 for high gain. This is a surprisingly pleasant receiver to listen to.

The fact that the tuning rate is quite low helps a lot in the enjoyment of this receiver too.  The 10-turn pot covers 6999KHz – 7051KHz giving an average tuning rate of about 5KHz/rotation. The VFO shifts in frequency very little indeed when  the receiver is knocked or moved. Drift is not great, but manageable.  Drift in the first 10-15 minutes is only slightly worse than after that initial warm-up period,  so I’ll give you the results from switch-on, which were 90Hz drift in the first hour,  140Hz drift in the second hour, and 110Hz in the 3rd hour. All the drift was downward – there was no upward drift at all. The drift was steady, and with a little bit of compensation, I think those figures could be improved upon quite a bit. However, for casual listening (and this is all I am going to use this for) it is adequate.

This receiver is certainly not perfect, but considering it’s just an NE602 and an LM386 (oh – and a quad op-amp IC for the filtering), it’s not bad at all.

Thanks Dave NM0S and 4SQRP for this neat little audio CW filter kit.

Here are a couple of videos that I just made of The Rugster. Please excuse the poor video quality – it’s an old, cheap camera:

This one is a little shorter:

July 24, 2011

The WBR – A Simple High Performance Regen Receiver for 40M by N1BYT

EDIT (July 26th 2014) – If you’re thinking of building the WBR, I strongly suggest you check out my most recent build here, which incorporates a mod suggested by LA3PNA, and a different configuration for the AF amp that I think provides nicer sounding audio. The full schematic is published there also.

A few weeks ago when a sizable order of parts arrived from Dan’s Small Parts And Kits, the plan was to use them (along with the parts from a few smaller orders from kitsandparts.com and Mouser) as the basis for some fun home-brew projects. I got my feet wet by building some small circuits Manhattan style – an audio filter and RF preamp for my VRX-1 DC receiver, and a crystal oscillator in an Altoids tin to check out the MeSQUARES I had just received from QRPMe. These are detailed in recent posts.

Like anyone who builds circuits, I have a mental list of things that I’d like to build which is updated constantly.  Some ideas get pushed to the back of the queue to make way for newer ones, and some stay pretty close to the front for long enough that eventually opportunity and desire collide and magically, it gets made. This is what happened with the WBR regen receiver.

As a teenager growing up in England, I had a one-tube shortwave regen that I built from a kit.  It was an HAC Model DX which used a mere handful of components and used a big rectangular 90V battery for the high tension supply.  It used plug-in Denco coils for band-changing (I think I remember having 3 of them).  I spent countless hours in my bedroom with the high-impedance headphones clamped around my ears, constantly tweaking the regeneration and tuning capacitors and listening mainly to shortwave broadcast stations.  I could figure out which band I was on, but had little idea of the actual frequency.  It really didn’t matter though, because the likes of Radio Australia, Radio Tirana Albania, Radio Moscow, Radio Nederland’s “Happy Station” shows, Radio Prague and many, many more kept me glued to that little set. It was just a small bent aluminum chassis with 3 variable capacitors,  a battery triode, a set of headphones, a coil, 3 fixed resistors and 2 fixed capacitors, but it was pure magic to me.

I think I spent the majority of my teenage years in my bedroom, listening to my record collection and radios. As an adult who recently semi-retired, it feels as if I’ve come full circle.  The chance to spend all the time I want building and listening to radios is an absolute gift, and the WBR Regenerative Receiver (Aug 2001 QST) that I’ve had the pleasure to use these last few days has brought the magic of radio listening back in a big way. It is sensitive, very stable, suffers from no microphonic effects at all, and thanks to the ingenious Wheatstone Bridge tank circuit has very minimal radiation of the local oscillator signal from the antenna port and so no common-mode hum problems.  Oh – and no hand capacitance effects either. I haven’t yet measured the drift, but after I’ve switched it on and let it warm up a little, I can set it to a frequency to listen to a net or QSO, and it stays there. Obviously, there must be some drift, which I hope to measure soon, but I’m not hearing any.  One more thing, not only is there little long-term drift, but my unit is very frequency-stable when subjected to physical shock too.  It’s a great little receiver and has pleasantly surprised me with it’s performance.

I won’t be publishing the schematic for the WBR receiver here, as it’s not mine to publish. (NOTE – Dan has since said he doesn’t have a problem with me publishing the schematic, so I went ahead and published the schematic for the slightly different version I made for the 31M band here.)  It appeared in the Aug 2001 edition of QST, so if you’re an ARRL member, you should be able to download it from their site. The article was also reprinted in the ARRL book “More QRP Power”. There were a few errors in the original QST article and although the parts list has been corrected in the “More QRP Power” reprint, the schematic hasn’t. For the record, when you look at the schematic, R17 should be labelled R7, C22 should be 0.01uF,  C19 (the capacitor connected between R15 and pin 5 of U2) should be 0.01uF, and “the other C19” – the capacitor connected between R16 and ground, should actually be labelled C20. If they’re going to reprint articles (which I’m very glad they do) I wish they’d make sure that all the corrections are included.

This was the first time I had ever attempted to fabricate an enclosure from PCB material and I’m quite pleased with the results.  I won’t fully detail my construction methods, as I got them from K7QO and WA4MNT, and they both already have excellent tutorials available online on how to make PCB cases. Ken WA4MNT’s PDF tutorial is here, and Chuck K7QO’s is here (Chuck’s tutorial has since disappeared from this url and I haven’t been able to find it’s new home, if it has one).

One or two people have asked me how I cut PCB material. If you don’t have a bench shear you can get nice clean cuts by scoring it with a utility blade and breaking it.  After using a sharp pencil to accurately mark the line where I want to cut, I use a metal rule and fresh sharp utility blade to make deep scores on both sides of the board. Expect to go through a few blades, and don’t be shy about replacing them;  cutting with a nice new blade is a pleasure, while an old one will just give you grief and raggedy cuts. Make sure that the cuts on each side are exactly opposite each other. Then sandwich the board in a vice between 2 bits of wood at the score mark thus:

Then, bearing down hard on the whole sandwich, firmly flex the board up and down until it breaks cleanly off.  You can run the board back and forth a few times over a piece of sandpaper, sandcloth etc. on a flat surface to smooth it a little, and you’re done. Easy! If you have a vice, you can use it for the above step.

Work progressed on cutting the panels for the enclosure (the 2 triangular pieces were used later to strengthen the cover) –

and before long I had a drilled front panel, a back panel, a base and 2 strengthening side-struts, which cleaned up with a scotch-brite pad and a little dish soap (for de-greasing). I made sure to rub the Scotch-Brite pad on the board mainly in straight lines as it does give the copper-clad a slight brushed look:

As suggested by WA4MNT in his tutorial, I used an angle iron to ensure 90 degree corners. The one I’m using is actually aluminum angle stock.  One thing that I would have had to learn the hard way were it not for this tutorial is the fact that when you solder the pieces of board together, as the solder cools, it shrinks, and you end up with something closer to 88 or 89 degree corners.  That can really mess up your nice square box, but Ken tells you how to square up your angles. Little by little the enclosure took shape and the thought occurred to me “Wow, I can actually do this!”:

I gave the finished enclosure a final cleaning with Tarn-X and sprayed it with a light coating of lacquer to prevent against oxidation of the copper. It didn’t fully work, as you can see from pictures of the completed receiver at the end of this post.  I’m thinking that I didn’t apply a thick enough coat of lacquer.

I thought it would be fun to fit the controls and stick-on feet to the enclosure to get a feel for what the finished receiver was going to look like:

I have a fondness for quality multi-turn wirewound pots, in the same way that I used to like using high quality air-spaced variable capacitors in my projects. Similarly,  the way that a regular one-turn pot feels when operating is important to me. The pot that I originally was going to use for the AF gain felt a little “scratchy” in the way it rotated, so I bought one from Radio Shack that feels silky smooth.  Even the first BNC connector that I used for the antenna socket (which was left over from an earlier project) was a cheap component, and I noticed that plugging and unplugging the antenna took more force and fiddling than it should have, so I used a new higher quality part (Mouser # 161-9323).  If this seems like a little too much attention to detail, let me explain. The connectors and controls are the way that you, the operator,  interface with the radio.  Higher quality components used here will greatly improve your experience of the radio. What would you rather use – a radio that takes force and fumbling to connect the antenna, has hard to adjust tuning due to the use of a cheaper one-turn pot, and a volume control that turns roughly, or a radio with controls and connectors that operate and rotate smoothly? These kind of things make a big difference. It might seem illogical to spend so much more on connectors and controls than on the rest of the circuit, but given how positively they can impact the user experience, it’s an investment worth making. I was hoping that this receiver would become a permanent part of my station, so spending a little more wasn’t a problem.

You’ll notice in the next picture that I used two 10-turn pots. One was for the tuning, and the other for regeneration. In the QST article, N1BYT details his method of using a one-turn pot with a separate preset to set the regeneration range on the main regen control. I’m sure that this works very well, but I happened to have an extra 10-turn pot available and was wanting an excuse to use as many of these wonderful things as possible, so I omitted the preset (R6 in the article) and used a 10-turn for the regen control. You do get a “whizzing” sound when adjusting the regen if you use a wirewound. To eliminate this, you can bypass between the wiper and ground for both audio and RF with a 10uF and 0.1uF connected in parallel. I’ve left mine as is for the time being, as I quite like the whizzing sound! (Edit: The novelty of the whizzing sound wore off swiftly, so I placed a 33uF cap between wiper and ground at the regen pot. It took care of things nicely.)

Daniel N1BYT’s original design for the WBR didn’t use a volume pot in front of the LM386, relying just on the 1K RF attenuation pot in the antenna lead. I decided that I wanted a little more audio gain in order to drive a loudspeaker, so I used the audio chain from N1BYT’s OCR II receiver, but connected a 10uF electrolytic between pins 1 and 8 of U2 to get the maximum gain of 200 out of the LM386 IC. Running the LM386 with maximum gain, and also fed with a preamp, I wanted to be able to control the gain of the AF stage as well as being able to attenuate the RF input, hence the extra pot. The toggle switch will be used to switch audio filtering at a later stage.

The enclosure build had gone so well that I started hoping that the performance of the receiver, when I built it, would be good enough to justify putting so much time and effort into the case. In retrospect, I probably should have built and tested the circuit first before deciding whether it was worthy of an enclosure.  Luckily it was.

I built the circuit Manhattan style using pads from W1REX at QRPMe.  I talked about the MeSQUARES in an earlier post and liked them so much that I ordered a sheet of MePADS from Rex.  MePADs are Manhattan pads for IC’s, as you can see in this picture. I had already broken off the pad I was going to use in this project:

As you’re building with these pads, if you find that you’ve glued one in the wrong place (I use Super Glue in the gel form), just slip a sharp utility blade underneath the pad and it comes right off. No need to fear a bad circuit layout, as you can change it if you make a wrong move.

The first part of the circuit to be built was the detector and regeneration circuit.  I didn’t have an MV104 dual tuning diode as specified in the original article, so used two back to back MV209’s, which you can see in the following picture, along with Q1 and Q2, the 2N3904 regeneration transistor and the MPF102 regenerative detector respectively:

Now I’ve added a few more components, including the 78L05 voltage regulator, which is hiding behind the grey 2.2uF electrolytic in the foreground:

The completed circuit. At this point, I hadn’t added the 10uF electrolytic between pins 1 and 8 of U2 (LM386).  The 2N3904 audio preamp is just below the LM386 audio IC:

Some more views of the completed circuit.  I knew that I was giving myself plenty of space to work with, but didn’t realize that there would be so much of it left over.  This is rather handy though, as it allows room for some low-pass audio filtering that I want to add later:

All very well and good, but would it work, and how well? The next picture shows the board installed in the case.  I added a 10K trimpot in place of R12 in order to adjust the tuning coverage of R11, the main tuning potentiometer. You can see the orange trim-pot near the front left-hand side of the board. It was only just now looking at this picture, that I realized I had left out C22, the 0.1uF capacitor in the antenna lead:

The same shot from a slightly higher angle:

By adjusting trimmer capacitor C8 and the trimmer resistor in place of R12, I achieved coverage of approximately 6970 – 7311 KHz. The extra 11KHz at the top of the band was important to me so that I can listen to the BBC World Service on 7310.  Vatican Radio broadcasts in English daily at 0250 utc on 7305, so the extra coverage above the top end of 40 allows me to receive that also.  Adjust C8 to set the top end of your coverage, and the trimmer in place of R12 to set the bottom end of the coverage. The tuning rate is a bit higher at the bottom end of the band than the top, being about 50KHz/turn in the CW sub-section.

I spent a few days enjoying the radio in this state before making a top cover for it.  The RF attenuation control doesn’t take the input signal completely down to zero. No doubt this has a lot to do with my decision to use an unshielded piece of copper wire to connect the antenna socket at the back of the rig with the 1K pot.  N1BYT does this in his receiver pictured in the QST article.  I liked it because it’s a reminder of the days when complete radios were wired using this technique, but I think I’ll probably replace that wire with a length of co-ax (after fitting C22 inline with the antenna lead) from the BNC to the 1K pot. That should allow attenuation of the input down to zero (or very close).  (Edit:  I just did, and it does.)

A top cover for the receiver:

The completed WBR Receiver (darn those front-panel fingerprints!) –

The space left at the top right-hand side of the front panel was in case I found a way to hook up a frequency counter to the regenerative detector while in oscillation in order to provide a frequency readout.  I have heard of people achieving this with the WBR by placing a pick-up coil near the main tank coil.  Still not sure whether I’ll pursue this.

Some of the things I’ve heard on this wonderful little radio in the first few days of owning it are the Vatican Radio transmitter in Sackville, New Brunswick on 7305 and Radio Australia (it was great hearing their Waltzing Matilda sign-on melody).  I also heard FO8RZ in French Polynesia on 7001 and immediately switched over to the Norcal 2N2/40 to work him with 4 watts! One of the reasons for building this receiver was in order to listen for AM activity on 40 (amateur, as we all know that there is plenty of broadcast activity there!) and I was happy this morning to hear W6LHQ running 200W of AM from his QTH in Modesto, CA on 7293KHz. This is one great thing about a regenerative detector – it will receive SSB and CW as well as AM. I believe it can receive FM too via the slope detection method, though I haven’t had a chance to try this out.

The Wheatstone bridge circuit in the tank is an elegant way to resolve the problem of oscillator radiation from the antenna that is common in simple regens, without resorting to adding a stage of RF preamplification, and it works very well.  I made no particular effort to match C5 and C6, so either I was lucky, or the circuit is forgiving, because I have encountered no issues with re-radiation. Without re-iterating the summary of this receiver’s performance that I gave in the second part of the 4th paragraph of this write-up I’ll just say this; I was hoping that this regen’s performance would be good enough for me to have it as a permanent part of my station, and it is.

The next step will be to replace the antenna input wire with a length of co-ax, and then build some audio filtering.  At that point, I’ll make some recordings of the audio and post a YouTube video.

I’m also thinking about bread-boarding another version of this receiver to encompass a larger portion of the shortwave spectrum. I have some MV108 varactors, and am thinking that the wider capacitance swing of one of those could give me broad coverage of a large part of the HF spectrum, with a fine tuning control provided by another diode. It would be great to have a regen to give me continuous  coverage of, say, 3 – 10MHz, or even higher.

Lots of fun and experimentation to come. Many thanks to N1BYT for this fine little receiver.

Notes added after above post was written – 

*Current consumption is low too – I measured around 12-13mA in regular use.

*A number of other builders have experienced problems with low sensitivity with their WBR’s.  LA3ZA found that substituting a 0.22uH inductor for Z1, the 1″ length of stiff wire between the center-tap of the coil and the ground plane, did the trick.  In the QRP-tech group on Yahoo, Steve AA7U did some experimenting and found 1uH was the optimum value for him.  Other builders have not had these issues with theirs.  My suggestion would be to build it as in the original design, adding the inductor if it seems necessary.

*As mentioned earlier, I added an audio pre-amp stage to my WBR, as suggested by N1BYT in the original article.  The schematic of the simple stage I added can be seen in this post.

*If I were building this again, I would change the configuration of the LM386 amp stage. Using a 10uF cap between pins 1 and 8 to get maximum gain introduces quite a lot of hiss.  The design that VK3YE used in the Micro 40 utilized a couple of ideas that had been discussed in SPRAT and is lower noise. I would try using the 2N3904 pre-amp as detailed in this post with the LM386 circuit as used in the Micro 40.

July 1, 2011

First Forays Into Manhattan Construction – Crystal Oscillator and a Peaked Lowpass Filter

With the exception of NT7S’ VRX-1 receiver I’ve had no experience with Manhattan construction. Once the majority of my recent flurry of parts orders came in, I figured it was time to start building a few things Manhattan style, and to start with some simple circuits.

I found a good supplier of PCB material. His eBay username is abcfab. He has a lot of PCB listings – different sizes, thicknesses and materials, and based on the shipment I received from him, it’s good stuff. The cuts were clean and the cut corners were square:

Although I wanted to make my own Manhattan pads, I saw W1REX’s MeSQUARES on the QRPMe site and was curious to try them. They arrive as one sheet of 300 square pads. The material is scored so that you can easily break the pads off with a pair of pliers:

The material these pads are made from is fairly thin (about .03″) and I would normally be concerned about the possibility of shorts to ground with thin pads like this, but Rex has designed them in such a way as to minimize the possiblity of that happening – the square pad of tinned copper on top of the pad doesn’t extend to the very edge – there is a small margin around the tinned area.

The pads are easy to use.  I wanted to put something simple together to try them out, so I built a crystal oscillator in an Altoids tin:

If I were building a more complex circuit in a tin, I wouldn’t build it directly onto the base, as it can get tricky holding the soldering iron at a low enough angle to make the joints. It’s definitely not the way to go if you need a fairly decent component density. I like the MeSQUARES, but wish they were available in smaller sizes too for circuits requiring greater component density. A few days later, I ordered some MePADS which is the same thing, but for mounting IC’s. I’d show you those, but they haven’t arrived yet.

OK, onwards and forwards (or however the saying goes).  I built NT7S’ VRX-1 Direct Conversion Receiver for 40M a while back. It worked well and could hear plenty of stations, but I must admit there was a fair amount of hum.  This is a common problem with simple DC receivers and I’m not sure if it has something to do with my layout, or perhaps whether something I did with the wiring caused a ground loop. (Perhaps feeding the final audio amp with a balanced input wold help eliminate some of it?) After completing it, I put it aside and made a mental note to come back to it one day. Incidentally,  KE7GKM has been making lots of QSO’s with his homebrew station using a VRX-1 on the receive side, so it must be something to do with the way I’ve laid out or wired my circuit. The other thing about simple DC receivers like this that my inexperienced CW ears would benefit from, is some audio filtering.  It seemed like a good excuse to build the active peaked lowpass filter that was designed by W7ZOI and featured on VE7BPO’s site. I built it on a scrap of PCB from Dan’s Small Parts and Kits. While the pieces of board from abcfab on eBay are cut nice and square, the small pieces from Dan’s are not. This is not necessarily a problem if you’re using them to build circuits on, but will require a bit more work if you’re using them to build enclosures. Here’s the completed filter:

A week or two ago, I made a couple of hundred 3/16″  (about 5mm) pads and tinned them all. Now I’m finding that they are rather large, and I’ll most likely only be using them for connection points that have a lot of connections. The trimmer pot nearest you is mounted on these 3/16″ pads. They looked small when I made them, but once I started to use them, they were huge! As the MeSQUARES hadn’t turned up yet, I mounted the NE5532 chip on pads punched from a 3/32″ die (just under 2.5mm).  By bending out the chip leads and trimming them to different lengths, and staggering the pads, managed to do it without any of the pads touching. Pins 6 and 7 were connected together, so I soldered them to one bigger pad. You can get a slightly better view of the way I mounted the IC from this picture:

The two trimmer pots are for adjusting the resonant frequency and Q of the filter (if you take that adjustment too far it starts to ring).

Here’s the board installed on the back wall of my VRX-1:

It is certainly filtering the audio, but of course I still have the hum. Although the filter peaks gently at the frequency set by the trimmer, it is still basically a low-pass filter. I could use something to notch out the 50c/s hum. However, there is so much of it that I think what I really need is to figure out the way I’ve laid out and wired the receiver.

EDIT: I only get hum pickup when an antenna is connected. I just walked out onto my balcony, connected an antenna and the hum was much lower in level than before. I don’t know why I didn’t think of this before, but I live in an old house with substandard wiring and no electrical ground (just the 2 wires at every socket.) I wouldn’t mind betting that if I were to walk out onto the street, the hum would disappear almost completely.

The box that I put the VRX-1 in is one of a really useful line of enclosures from LMB Heeger. Look at this shot of the receiver all cased up:

See the way that 4 little lugs are formed from the top of the case? They stick down just a little and engage with the front and back panel, giving extra rigidity to the enclosure. The aluminum is 16 gauge (0.0508″). The larger sizes aren’t quite as rigid, but the smaller ones like the one above, are, and would be ideal for a VFO or VFO-controlled rig. The above case is 2″ high, the front panel is 4″ wide and the case is 4″ deep. It is LMB Heeger’s Model 143 in their interlocking series, available in crinkly black, painted grey (like the one you see above) and plain aluminum (for that homebrew look). I think Digi-Key carry them, but you can order them directly from LMB Heeger. I’m thinking that a modular station built in these little cases (plain aluminum finish) would look neat. VFO in one, freq counter in another, mixer in another etc etc. Hmmm……

That’s it for now.

Blog at WordPress.com.