Dave Richards AA7EE

December 3, 2011

The 3rd and Final Stage Of K2 Assembly

The K2 has been finished for about 3 weeks and it’s time I made the post concerning the final stage of assembly. After completing the second stage and gaining a working receiver on 40M, the majority of the work was done. At this point it felt as if I was about 3/4 of the way through, and I think I probably was. The final basic  K2 consists of 3 boards – the front panel board which you can’t see in this photo, as it’s obscured by the control board which sits behind it, and the RF board. All boards are now fully populated, with the exception of the occasional jumper or connector here and there which are reserved for extra options. The on-board frequency counter is shown plugged into the test point that reads the BFO frequency – useful in regular operation if you want to adjust the filter settings:

Although I know my own toroids don’t look as pretty as the pre-wound ones from Mychael AA3WF would have,  I’m getting better at winding them, and they don’t look too bad at all.  There are many examples of attention to detail  in this kit. One of them is that when you pull the toroid leads tight and then solder them, the toroids are actually straight on the board. Often in kits, the holes for toroid leads are placed such that when you pull the leads through and solder them, they are skewed just a little. It’s a small point, but it makes boards look a bit messy. Hats off to Elecraft for helping to line my toroids up in nice neat rows:

A view from above:

Here’s the view from underneath. Note the nuts securing the 2 PA transistors to the case:

The 2 chassis pieces that I received for the back of the transceiver had already had the green masking tape removed. That was a little disappointing, as I wanted to leave the tape in place over the holes that are currently unused.  I’m sure Elecraft would have replaced them for me but it didn’t seem worth contacting them about. I’ll use some regular masking tape:

What a beaut!  –

Everything about the K2 seems to be working the way it should. Sensitivity and handling of strong signals seem fine.  I plan to measure the minimum discernible signal at some point when I get a calibrated signal source. There are only 2 things that are not quite as perfect as I’d like them (though neither are deal-breakers):

Firstly, due to the way that the transceiver handles the D-A conversion, when stepping through the filters, the BFO is not on precisely the same frequency for each setting of the filter.  My filter settings are (nominally) 1.5K, 1K, 700Hz and 350Hz (though I understand the real bandwidths are narrower in practice). When receiving a signal and stepping through the bandwidths, the signal is at the same pitch except when the 350Hz filter is selected, when it lowers in pitch about 15 – 20 Hz. It’s not a lot, but to someone like me who is sensitive to pitch, it’s a bit annoying.  No matter how hard I try to adjust the BFO frequency, I cannot get it any closer. I understand this is due to the number of bits in the D-A convertor and was one of those compromise decisions that often have to be made during product development.

The other thing is not really an issue. It’s more of an indicator of how used I get to certain receivers and how I often am very picky about how I want my radios to operate. I suspect I’m not the only operator who wants his rig to operate exactly the way he wants it to so that it “fits like a glove”.  On comparing the K2 receiver to the Norcal 2N2 receiver, I notice 2 things. Firstly, the rushing background band noise seems to be pushed to a lower level in the background by the 2N2’s filtering.  It’s difficult to make a direct comparison though, as I don’t know the exact bandwidth of the crystal filters in the 2N2 and the K2  – this would be something worth measuring.  The K2 has more audio output power, so part of it may also be that I simply have the AF gain turned higher on the K2, creating the illusion of more rushing background noise.  On some very weak signals, I find that the 2N2 has a very slight edge.  I don’t think this is a sensitivity issue; I think it has more to do with the audio filtering – the audio on the 2N2 seems to peak more sharply at a fairly specific frequency (I set up all my rigs for a 500Hz sidetone), while the K2 doesn’t have this extra peak. My guess is that the audio circuitry on the 2N2 is tailored more specifically to CW, as it is a CW-only rig. The K2’s audio chain needs to accommodate wider SSB signals so needs to be fairly flat with a passband of a few KHz.  So I think the next step is to figure out some kind of extra audio filtering for the K2 – perhaps a KAF2 or the DSP option?

These 2 small points aside,  I’m happy with my K2 so far. Sure, it’s fun to compare figures and performance characteristics, but there is another very important factor that determines how useful a transceiver will be to the operator, and that is the feature-set and how easily accessible those functions are. No complaints in that regard yet.  Considering the relatively small size of the front panel and the fact that each button controls 2 different functions, Elecraft have made the most often-needed functions the easiest to access. I found it straightforward figuring out how to record and playback the keyer memories, as well as using the Fast-Play function, by which you can playback certain keyer memories with a single button push (great for contest operating).  Use of the dual VFO’s and operating split was intuitive – even for this guy who is fairly new to the world of operating split-frequency on HF.

In the few weeks since having it, I’ve worked T2T, YN7SU, HA3UU, JE4JPQ, JF1RWZ, PA0LEG, CO8WZ, 7N1PRD, BD4FM, CA2LQA, HL2DC, JE2UFF, DK1AX, CO6RD, PV8ADI, ZP6CW, ZP9MCE, GW4EVL, PJ2/W8WTS, C6ATA, JF1SQC, JR3NZ, JA6WFM, ZM1A, UA0ZAM, JF1NSD, JA7FTR, XE1CT, XE2B, JA5FDJ, PW7T, ZS4TX, PV0F, LS1D, C5A, HK1N, LW5EE, XE2AI, HK1R, JR1MQT, LU8YE, LT1F, VK4KW, JA1KGW and PY3ED. Oh – and a bunch of Stateside stations too 🙂

I think this rig is a keeper.  I’m already contemplating either the audio filter or DSP filter, the QRP internal tuner, the 160M option and (shock horror) perhaps also the SSB option 🙂

EXTRA NOTE:  It is now the end of July 2012 and I’ve owned this K2 for 8 months.  I’d like to add to the comments I made above about my comparison of the K2 audio response to that of the Norcal 2N2, which had been the main rig I was using for CW before. I had been used to the more peaked audio response of the 2N2 but after a few months with the K2, realized that the K2’s flatter response within it’s passband was preferable. I can now work a CW station slightly off-frequency without having to engage RIT.  With the 2N2, if I was working a particularly weak signal, adjusting the RIT was sometimes a necessity to maximize copyability of the signal.  My preference for a peaked response was based merely on what I was used to.

December 1, 2011

Second Stage Of K2 Assembly Completed

I actually completed all 3 stages of K2 assembly 5 days after finishing the first stage, and about 9 or 10 days after beginning the assembly, which included a couple of days off in the middle. However, as soon as I’d finished building it I wanted some time to relax, play with my new radio, and generally recharge my batteries, hence the lack of a timely follow-up post.  Sometimes I temporarily lose my urge to communicate.  Sorry about that.

At the end of this post, I mentioned how just one part was missing from the kit – a 20-pin connector that connects the front panel board to the main RF board.  It only took a day to arrive so while waiting, I decided to make the best use of my time by pre-winding the toroids.  I had originally planned to buy the pre-wound toroid kit from Mychael AA3WF, reasoning that if I ever wanted to sell my K2, I’d want really great-looking toroids in it. When push came to shove however, budget started becoming very important, so I decided to wind my own and was glad I did. They don’t look quite as nice as Mychael’s, but I think they’re not half-bad. Here are 3 of them.  The one with the yellow core is the VCO coil and is shown with just one winding (I hadn’t wound the secondary at this point).  You can also see the 2.1mm crochet hook I used to pull the windings through the cores.  It’s a method I learned from Jim K8IQY and I find that it helps in keeping the windings fairly close to the core:

This second stage of assembly,  after which you have a working receiver on 40M,  is lengthy. There are a lot of resistors and capacitors to install and as many builders before me have commented, the only way to deal with it is to just get on with it.  For relatively inexperienced builders, this stage could be a bit of a nail-biter, as you have so many parts to install before getting any feedback on whether the circuits you’re building do indeed work. However, if you have a reasonable amount of experience, the manual is so detailed, and entire assembly so well thought-out and described that you just know you’ll either end up with a working rig, or Elecraft will help you correct any mistakes made during assembly.

An advantage to building such a late model K2 is that all the mods, such as the thermistor board to stabilize the VFO further,  the few extra components to shape the keying waveform, and the 2 diodes to improve the handling in the presence of very strong local signals on nearby frequencies, are included. There’s only one little thing that slightly bugged my detail-obsessed mind, and that was that some of the mods look a little messy to me.  It’s probably unrealistic to expect the board layout to be revised for every single mod after the K2 hit the market, but the board is of such high quality and looks so good, that a few components soldered at odd angles on the underside of the board offend my over-developed aesthetic sensibilities. Luckily, I got over it pretty fast. The components that are either horizontal or vertical are part of the original design while the ones that look patched on afterwards are, well, patched on afterwards.  In the real world, I’m sure that the economics wouldn’t allow having the board layout revised again for these small changes:

In one sign that through-hole components are becoming harder to find, to the left of the above picture in the middle, you can see D36, which is now an SMT part on a small board, modded to fit the main PCB. I feel a little sad at the slow passing of through-hole technology, as not everyone is comfortable with SMT construction and so the building of your own rig will be an experience available to fewer people in the future. Just under D36 is L33 which was originally a choke with axial leads. As I understand it, using a toroidally-wound component for this part and mounting it on the underside of the board gave greater suppression of the opposite sideband in SSB mode. It is wound from fine wire and there’s no need to worry – it comes already wound, though you do have to be very careful with the leads when installing it.

Another view of this part of the underside of the main RF board:

The entire underside of the main board after this second stage of assembly. Near the left side of the board about 2/3 of the way down, you can see the outlines where the 2 PA transistors will be installed in the 3rd and final stage of construction:

At this point, the majority of the circuitry has been built, and that is quite apparent in the following views. The length of mini co-ax connects the built-in frequency counter to one of 3 test points on the board. The K2, as you probably know, has a built-in DVM and frequency counter to assist in building and aligning the radio:

A close-up of the PLL upgrade board that increases the stability of the VFO. It’s the light-brown vertical board just to the left of the crystal:

Here’s the 5 pole variable bandwidth crystal filter. I did a slightly less-neat-than-normal job of soldering the ground leads to the crystal cases. The crystal nearest the front is the messiest-looking. These kind of things really bug me but it’s a perfectly serviceable joint, so I pulled myself together and got over it. In this picture, you can also see the crystals for the second, fixed bandwidth crystal filter and to the left of them (and slightly out of focus), the other SMT part that has been substituted for the original through-hole part. In this case, it’s U12, an MC1350 IC which you can see installed on a small PCB that solders in to the holes for the original DIP part:

This wider shot also shows the 2 BFO crystals. The manual (bottom left side of page 54 in revision H1 of the manual dated April 26, 2011) notes that the leads from one of the BFO crystals (X3) need to be folded over and soldered particularly close to the board in order to prevent fouling a rubber bumper that will be placed over it in a subsequent step.  I seem to remember that one (not both) of the leads from X3 did protrude directly underneath the rubber stem bumper.  I found that by pre-cutting the lead so that it didn’t quite poke out of the PCB underside and applying solder so that it filled the plated-through hole but didn’t spill over onto the pad, I ended up with a nice flat surface on which to lay the rubber bumper for L33 (the BFO inductor). If you’ve built a K2 before, this will make sense to you. If not – it will become clear when you get to that point in the instructions.  You are instructed later on (right-hand side of page 60 in my revision of the manual) to flush-trim all leads under or near L33 but if you were strictly following the manual, you would already have installed X3 at this point and folded over it’s leads which will allow the bumper to “exist” as it were, but will not allow it to sit completely flat.

Aww what the heck, here’s another very similar shot. In the background is the control board, and underneath the processor chip, you can see the crystal and associated trimcap that should be adjusted to exactly 4MHz so that the frequency display reads accurately. The manual tells you how to do it – you don’t need any extra test equipment:

This shot shows the VCO inductor. It’s the toroid wound on a yellow core to the right of the picture:

Oh – and the receiver? Well, it seemed to be working well on 40M.  I’ve read blogs from some builders who after a fairly lengthy period of assembly, were having such a blast listening to 40M on their K2’s that they gave themselves a break before continuing. I don’t remember exactly, but I think I had a short break of maybe a day, while entertaining company.  As I had been several months without any capability on the upper HF bands, I wanted to forge on so that I could listen to the higher bands and catch some sunspot action.

In the next post, I’ll show some pictures of the completed basic K2 and share my impressions so far.

November 7, 2011

First Stage Of K2 Building Completed – More Pictures

The missing part arrived from Elecraft yesterday (the day after I called).  The fact that I live just 50 miles from them helps in getting things delivered swiftly. The part was a 20-pin connector for connecting the main board to the front panel board. Once that was installed, it didn’t take long to partially assemble the case and plug the completed front panel and control boards into the RF board – which at this stage had just the DC power, latching relays and the I/O controller circuits installed. Before performing testing on this stage of the build, I had to install the bail on the base of the case – a procedure which some builders have had trouble with.  Following the procedure in the manual requires you to compress the bail which, if you have a vice, can probably be accomplished without too much bother, but if like me, you’re trying to do it with your hands, could be quite difficult.  It was in my case, at least.

This is what we’re trying to accomplish:

The method I used, which was adapted from one I found described on the Elecraft reflector, was to install one of the oval feet and place one end of the tilt bail in it. Then I rummaged around in the junk box and found a machine screw that fitted through one of the holes in the other foot but was longer than the supplied screws. I installed this screw through one of the holes in the remaining oval foot, but only screwed the nut on a little, allowing me to lift the foot enough to get the end of the tilt bail underneath it (you do have to compress the tilt bail a bit but nowhere near as much as you would if you had followed the procedure in the manual). Then I installed a regular length screw and nut in the other hole and screwed it down fairly tight. The next step was to replace the long screw and nut with the supplied (shorter) screw and nut and screw it all down tightly.

The first IC to be installed on the main board (called the RF board) is U1, the I/O controller. It controls all the latching relays for the micro-controller, as well as other input and output functions.  You can see it here, flanked by some of the latching relays:

A similar view:

The control board plugged into the main (RF) board:

A view from above:

It passed all the tests. The band changing relays work. The display does too, as do the circuits that drive the signal strength bar-graph LED meter. I can twist the tuning knob and the frequency readout counts correctly. The keyer and keyer memory work and sound great. It even looks great from the front. We know, of course, that at this point it is a gutless wonder; it really needs a synthesized VFO, as well as transmitter and receiver circuits. As gratifying as it is to play with it at this stage, the result of the next stage will be to have a working receiver on 40M.  I’ll talk to you next when I’m at that point!

November 5, 2011

Getting Started With Building The K2

I’ve been wanting to build an Elecraft K2 for several years now, but the desire has been getting stronger, until maybe a year or so ago when I started seeing it as the logical endpoint in a progression that has included the Norcal 2N2/40, the Fort Tuthill 80 and the CC-series of transceivers (which is ongoing, as we are still in beta-testing.)  At some point I realized that if I could successfully put all these kits together, there was no reason I couldn’t build a K2 as well.  If you can solder pretty well, can identify parts, and can follow written instructions, you can put a kit together.

A number of people have asked why I would consider a K2, now that the KX3 is about to be released. The answer is that I wanted to build myself a multiband full-featured HF rig from a kit at the component level, which I wouldn’t be able to do with the KX3. The K2 has been around a long time now – something like 13 years. It doesn’t have the cutting edge SDR technology that the KX3 will have, but it’s still a solid performer and very capable. If you want to build a full-featured HF rig from a kit containing individual components (as opposed to modules that you connect together), the K2 is the only choice out there. I had no problem with the fact that the K2 has no direct competition in the kit world, as from everything I’ve read, it seems to be such a great-performing transceiver – especially for the QRP CW enthusiast.

I’d been umming and aahing about ordering the K2 for a while and although the plan was to wait until the new year, the recent increased sunspot activity and excellent propagation on the upper HF bands prompted me to hurry up so that I can get in on a bit of the DX action too (as my current station consists just of 3 monoband QRP rigs on 80, 40 and 20M).

I ordered the basic version of the K2 (no options) online last Sunday evening. They shipped it on Monday, and the $12 Priority Mail option got it to my door the very next day, as I only live about 50 miles from Elecraft.

I know that the K2 has been extensively documented over the years on many blogs and websites, but allow me the obligatory “I just opened the box” shot:

I’m not going to go into great detail about the K2 kit as that has been done on so many other websites over the years, but I will offer a few of my thoughts and share a little of my experience.

Documentation is great. The manual is a lot like I imagine one of the old Heathkit manuals would have been like – very detailed with clear, step-by-step instructions. For the experienced builder, some of the descriptions and suggestions on how to install components will not be needed, but it’s definitely a good thing to have all that information there. In fact, I found that some of Elecraft’s suggested methods for mounting components differed from my preferred practices, in which case I opted for my way. More on that in the next blog-post.

There are 3 boards in the basic K2, the control board, the front panel board, and the RF board. The first board to be assembled was the control board; the brains of the transceiver:

You can see the multi-pin connectors at the bottom of the board that are used for all inter-board connections in the K2. That represents one big difference from the Heathkit days – no complex wiring to route around the inside of the enclosure. Not only does it simplify construction, but it must contribute a great deal to reliability too. On the reverse side of this board, you can see the extra caps that have been added (as recommended in the manual) to improve the keying waveform:

The control board wasn’t particularly exciting to build – just a board that needed filling with components.  However, the next stage – the assembly of the front panel board, felt a lot more engaging, as I got to slowly see the front panel of my new transceiver take shape:

To aid in making sure that all switches are mounted at the same height above the board, Elecraft include a really neat little spacer tool that you place underneath the switches before soldering them to the board. This ensures that all the switch buttons protrude an equal amount from the front panel to give a nice, uniform appearance.  A small PCB is supplied for constructing an RF probe to help with alignment. Attached to this PCB are two small strips of board that are broken off to make the switch spacers. The manual instructs the builder to snap the protrusions at 4 points as indicated in the manual, to make 4 spacers.  As well as being used to set the switch heights, the spacers are used later to set the exact height above the board of the LCD backlight. Although not mentioned in the manual, I found that it would be easier if I initially broke the PCB at only 2 points, to make 2 long spacers for setting the switch heights. This way, each spacer could fit under 2 switches at a time.  On reaching the stage where I installed the LCD backlight, I snapped each spacer in half to get the 4 spacers required for setting the height of the backlight.

I also want to talk about soldering, but first of all, another view of the front panel board just for the heck of it:

The back of the front panel board (with the front panel attached):

When installing the encoder, Elecraft recommend that the 4 wires that attach to the encoder are wrapped around the connection posts before being soldered. I didn’t do this because I figured that if the wires were wrapped around the terminal posts, it would increase the chances of a short between the posts on the encoder. The other reason was that it felt like overkill to me.  While I like to make solid electrical connections, I also like to plan for the possibility that I might need to disassemble parts of the transceiver in the future.  Here’s how I made the connections to the encoder:

If I ever need to de-solder the encoder, all I have to do is hold the iron to each post and move the wire away with a small screwdriver – or wick the solder away with de-soldering braid.  Job done – and personally, I think it looks neater than if the wires were wrapped around the posts.

While I’m on  the subject of soldering, take a look at how I soldered the IC to the left of the encoder. I’m not claiming that it’s the neatest or prettiest soldering job in the world. I’m still trying to find a pair of flush cutters that will cut a wire completely cleanly and horizontally, without leaving a bevelled edge on the wire. Does such a pair exist? If I have to spend a lot of money to get such a pair I’ll do it, as I’d like to have my PCB’s look neater if possible. Anyway, what I wanted to point out is the fact that I have filled the plated-through holes with solder but have not allowed the solder to build up on top of the board. Many people when soldering boards with plated-through holes like an accumulation of solder on top of the board, and it is just not necessary. Depending on your personal taste, I can see that it might possibly make your joints look nicer to have a little build-up of nice shiny solder around the wire on top of the board. Thing is, if you ever have to remove a part from the board in order to replace it, that’s a whole lot more solder you have to suck up or wick away with desoldering braid.

I think one of the reasons folk often put more solder than is necessary on joints is to “make sure” that it’s a good connection, and if they can’t see solder on a joint because it’s in the hole, they perhaps think that it’s not there, so they put a little more on top “just to make sure”. It’s kind of like putting one sugar in your coffee, and then adding an extra one (actually, I’m not sure that it is, but I’m feeling a bit sleepy and am in stream-of-consciousness mode). If you’re dealing with plated-through holes, all you need do to make an excellent connection is make sure both the tinned pad on the board and the component lead are hot so that the solder will melt onto them, then hold the solder close to the top of the hole and experience a wonderful moment of zen as you see the solder wick down by capillary action into the hole. If you use a nice thin solder (I use .02″) then you’ll be able to apply just the right amount to get the job done. If you’re fairly new at soldering, allow me to give you a tip. Once you’ve made sure the tip of your iron is clean (I wipe mine before every joint, unless I’m soldering several in a row one straight after the other), then a great way to ensure maximum heat transfer from the iron to the pad and component lead is to melt a very small amount of solder onto the iron. The solder melts, makes contact with the iron, pad and lead all at once, and you’ll notice the solder suddenly wicking down into the hole and making a perfect joint. Bingo. It’s a beautiful thing!

Incidentally, if you’re soldering on a board that is single-sided without plated-through holes, then you do need a little fillet of solder on top of the board.

EDIT:  I just read a short essay on soldering on the Elecraft site in which “Dr Solder” at Weller says that you should never have a solder joint in which the hole is slightly under-filled, leading to a dimple in the hole. This is what many of my joints in the above photo are like. Hmmm…..now I’m wondering if I should have put just a touch more solder on those joints.  I think I’ll leave them as they are and only resolder them if they are problematic.  I have a sneaky feeling they’ll be fine though.

Here’s what that front panel board with the front panel attached looks like from the front:

It’s really gratifying seeing the transceiver slowly take shape. The whole process of putting this together has given me even more respect for folk who put kits like this together – or who design any product like this.  So far, almost every component has fitted the corresponding holes on the board exactly – and with the rate at which these things change, it’s something of a feat to make a kit available – and have it still available for purchase 13 years later. Every single fastener, spacer, enclosure piece – they are all part of a whole, and it takes a great deal of creativity and engineering experience to fashion a product like this.

The next step was to assemble the DC and control circuits on the main board so that the transceiver case could be assembled and all the boards plugged into each other to ensure the correct operation of the control circuitry, before proceeding with the build of the receiver and transmitter circuits. I got very close to completing this step when I came across my first missing part, and kicked myself for not doing a complete inventory earlier.  I had performed an inventory of the control board and front panel board parts, as well as the bag of miscellaneous parts. On looking at the sheer number of parts in the bags for the RF (main) board, I decided to wing it and hope there was nothing missing, which there was – a 20-pin connector for mating the main RF board to the front panel board, a rather essential part.

At this point, it was late on Sunday evening, so I decided to conduct a complete inventory of all remaining parts so that when I called Elecraft in the morning, I could put in one order for all the missing parts.  As it happened, that was the only part that was missing. Only one missing thing out of many hundreds is pretty good. There was one other part which, although present, I wasn’t completely happy with, and that was the main tuning knob. I’m fine with the weight, feel and look of it, but the machining of the one I received was a little substandard; one of the holes for the set screws had a very ragged edge, and the knob looked like it had bumped up against some hard object or sharp edges, as there were a number of marks on the side. It wasn’t terrible but compared to the high quality of everything else in the kit, it looked a bit shabby. Madeleine at Elecraft was very helpful and suggested that they send me another tuning knob.  I may end up getting a different knob, but would like to start out with the stock one.  I’ve spoken to Madeleine over at Elecraft a number of times now and she’s great. It’s a real pleasure to call a company and have my phone call taken by someone who is articulate, friendly, and communicative.

So I spent part of yesterday taking pictures of the progress so far and writing this blog-post.  Later today when the 20-pin connector arrives, I’ll finish off the DC and control circuits, assemble the enclosure, plug the boards together, and run the first tests. Fingers crossed – hope there’s no blue flash or whiff of smoke 🙂

Blog at WordPress.com.