A Passive Tunable HF Preselector

When I built the VE7BPO DC Receiver Mainframe recently, it wasn’t intended to end up as a final finished project. The intention was more to have it as part of an experimental platform. The little box that contains the DBM, diplexer, and AF amplifier that make up the mainframe will most likely stay largely the same, now that they are built and boxed up. However, the outboard functions of local oscillator and antenna filtering can swapped around and changed at will. The mainframe includes a spot for an onboard plug-in bandpass filter. It was constructed so that the bandpass filters from QRP Labs could be plugged in, but this circuit section could be constructed from scratch, if desired. The first BPF I constructed was for 40M, and it did a fine job of removing many of the spurious responses I was experiencing with no antenna filtering in circuit. I purchased a 10-pack of these filter kits from QRP Labs, intending to, at some point, assemble most of them for listening to the amateur bands on this little receiver. That may happen, but I also wanted to listen in between the amateur bands. AM reception is not great on a direct conversion receiver, but there are quite a few non-ham SSB and CW signals to listen to outside the ham bands, and it would be good to be able to do that on this receiver. A tunable passive HF preselector seemed like a good way to get this particular show on the road.

The circuit is a very straightforward and standard double-tuned bandpass filter. Although my preferred variable capacitor of choice would have been an air-spaced component, I wanted to fit this into the same LMB Heeger 143 enclosure that I used for the other two receiver modules, the DC receiver mainframe, and the Si5351 “VFO”. I think it would be possible to find a suitable air-spaced part that would fit into this space but, for the sake of timeliness, I plumped for a polyvaricon.

GQRP Club member sales can supply the polyvaricons (to members) with a mounting kit that consists of 2 different lengths of mounting screw, to allow for different thicknesses of front panel. They come with a bolt and plastic spindle, for attaching a knob. Also supplied, are the inductors with adjustable ferrite cores. The inductors are Spectrum 5u3L types. The nominal inductance is 5.3µH, though it is adjustable over a fairly wide range. The L denotes that the secondary is a low impedance winding, suitable for matching to 50 ohm systems. If you are not a GQRP Club member, these coils are available from Spectrum Communications in the UK, who also sell on eBay. Some of the coils in this series are direct replacements for the Toko KANK series, which were popular with UK homebrewers in the past. It is possible to wind a similar coil on a toroid. I wanted these coils though, for the ability to easily adjust the inductance.

The 1pF capacitor that couples the two tuned circuits might seem rather low. In fact, it surprised me too. I began with higher values, of 47pF and then 39pF, but found that the coupling was too tight, and I ended up with 2 distinct peaks in the response, spaced far apart, with a large dip in between them. The final value of 1pF was far lower than I had expected. The only reason I can think of for this, is that the separate gangs in the polyvaricon are not as well isolated as they would be in a larger air-spaced part. I’m thinking that some coupling is happening inside the polyvaricon perhaps? A value of 1pF gave an acceptable response curve above about 8MHz. For the lower frequencies, an extra 10pF capacitor is switched in. Without it, the insertion loss at the peak of the response is a whopping 32dB. Switching in the extra 10pF reduced the insertion loss to 10.5dB which, although still a little high, is a lot better. The switch needs to be flipped to the high position over about 7 or 8MHz, otherwise the response is far too broad. With the switch in the “Lo” position on my preselector, at the maximum frequency setting (lowest capacitance), the peaks were spaced 5MHz apart, and the difference between the dip and the two peaks was 14dB – far too much. With the switch in the “Hi” position, at the highest frequency setting, which is 15.6MHz in my unit, the insertion loss at the peak is 4dB, the passband ripple 1.75dB, and the bandwidth at the -3dB points is 1.66MHz. At 3.5MHz, in the “Lo” position, the insertion loss at the peak is 10.27dB, the passband ripple a mere 0.83dB, and the bandwidth at the -3dB points 180KHz. Your first thought might be that 180KHz is not enough to cover the 80/75M band, but remember that this preselctor is tunable, so you can put the peak wherever you want it.

Some photos of this simple accessory –

The stack of modules that make up the complete receiver. From top to bottom, the preselector, the mainframe and, at the bottom, the VFO –

The stack as seen from the rear, showing the interconnections. It was starting to rain, and you can see some raindrops –

I recently acquired a NanoVNA, which was very useful for adjusting the trimmer capacitors and the inductors on this preselector, as well as for adjusting the fixed bandpass filters. Adjusting the trimcaps and inductors on the preselector is an exercIse in compromise, so it is very helpful to be able to see the effects of your adjustments almost in real time, as you make them. If you don’t have a NanoVNA, I imagine you have heard all about them. If not, there is a lot of information out there about them. Alan Wolke W2AEW has a fantastic YouTube channel, with several instructional videos on how to use a NanoVNA. A search of his channel for “NanoVNA” will yield many helpful videos. I have not yet watched it, but just found his introductory presentation to Fairlawn ARC on the subject of NanoVNA’s. Without going into too much detail, a NanoVNA can be used as a small and very portable antenna analyzer, and network analyzer. It can display the SWR curve for an antenna, over any frequency range you desire. Need to look at complex impedances? The NanoVNA has a Smith chart display too. It can also plot, in graphical form, the response curve of a filter. It is so useful to be able to build a lowpass or bandpass filter, and see the response curve, making adjustments easy. The majority of us regular hobbyists who couldn’t justify the purchase of a more advanced, and much more expensive network analyzer can now purchase a NanoVNA for somewhere between $50 and $150, depending what features you want. The original version has a 2.8″ screen, and is around $50-$60. The newer versions have a 4″ screen, which is much easier to read. Mine, the NanoVNA H-4 (the 4 denoting the screen size) was $93 on Amazon, delivered the next day. There is a more expensive version still, which has a metal case. I am doing fine with the plastic case so far. The very original version didn’t even have a fully enclosed case. I am happy to pay $30 or $40 more for a bigger screen and fully enclosed case. It’s much smaller and lighter than my old MFJ-259B, and does far more. Does anyone want my 259?

This NanoVNA will be fantastic for tuning up antennas in the field. Here it is, connected to a little test rig I built up, for testing the QRP Labs BPF’s. I didn’t switch it on, as the display often doesn’t show too well in the daylight –

The unit can be configured to display up to 4 traces simultaneously, each one showing different characteristics of the circuit under test. Here it is, with two traces activated. One is showing the response curve of the 80M BPF. The other, which was left on accidentally, shows the SWR, which isn’t of interest here. The unit was set to sweep from 1.7MHz, the top of the AM broadcast band, up to 10MHz –

Just for fun, here’s a closer view of that BPF –

While on the subject of this BPF, I used different values of capacitance from the ones Hans Summers supplied, for the coupling capacitor. His filters for the lower HF bands are not designed to cover the entire band. The intended usage is for receivers for digital modes, for which a narrower bandwidth is perfectly acceptable. I used a higher value of coupling capacitor to get the bandwidth I wanted. The bandwidth of this filter, with a 113pF coupling capacitance (47p + 56p), is about 885KHz. It’s a little wider than I wanted, so the next step may be to try a slightly smaller value of coupling capacitance. Insertion loss at the peak of the response curve is 5dB. By contrast, the insertion loss of the 40M BPF is only 1.12dB at the peak of the response curve – a very acceptable figure. In the assembly instructions for the QRP Labs BPF’s, Hans quotes an insertion loss of just 1.27dB for a b/w of 465KHz with his filter. The figure of 5dB for my BPF seems a bit high. The insertion loss of 10.27dB for the preselector when tuned to 80M, seems way too big.

When using the receiver with the preselector, I jumper across the socket for BPF that is inside the receiver mainframe enclosure. Breadboard jumper leads work well for this. Interestingly, reception on 80M is much better using the preselector than the BPF. Although the insertion loss is greater, and I have to turn up the volume to compensate, the SNR is much better, making reception of stations when the band is noisy, much easier. With the internal BPF plugged in, the SNR is higher. It is the same when listening to 80M at night with no filter inline at all – a higher SNR. Reception on 40M is about the same with the BPF as it is with the preselector.

I have not yet used this little direct conversion receiver very extensively for general HF listening, but a few observations, based on my experience so far –

  • The unfiltered output from the simple Si5351 is not perfect (surprise, surprise) and contains some spurious components, as well as the expected harmonics. The 40M band is largely clear. There is one fairly prominent one that is audible in the very bottom 300Hz of the band. There are a few others, at much lower levels, at a few points throughout the band, but they are masked by band noise when an antenna is connected. Outside the amateur bands, there are other spurii dotted throughout the HF spectrum. Annoyingly, there is a rather loud one at 10MHz, which makes reception of WWV troublesome. Future experiments could focus on reducing and/or eliminating these spurii, or looking at a different method of generating an LO signal.
  • The preselector (or fixed bandpass filters, if used) is very effective at eliminating unwanted modulation products from AM BC band stations, as well as from spurii caused by harmonics of the LO mixing with RF signals from the antenna
  • I’m happy with the mainframe circuitry. It is a good module for future DC receiver experiments. As it isn’t a single signal receiver, there is an automatic 3dB SNR disadvantage compared to a superhet or SDR. This is par for the course, however, and expected.

5 thoughts on “A Passive Tunable HF Preselector

  1. Thank you dave for this information, I will have a indepth look at this, you have sent it just at the right time, just received my order of prototype boards, so I will crack on with project. Best regards.

  2. I was looking for info. on preselectors and found this.
    I have sharp filters for 9 h.f. bands but wanted to add 60m. Yet another filter! It would use up all my switching with nothing left as I want to try and get vhf as well so decided on a preselector using perhaps 3 or 4 pairs of switched coils, this is like mine but I have two 4172 coils for testing, good for about 8-15 MHz. This explains my double hump response. I’ll buy and try 3333 coils with 3335 for higher freqs. Tx/rx based on G4CLF board.
    A preselector is also useful if we ever get any more new frequencies. Who ever thought we’d get 30m to start with? Then 12m, 17m, and 60m?
    73, Bill, G4GHB.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s