Monday, January 4, 2010

30m Subharmonic I/Q-SDR Receiver

As promised, some detail about the subharmonic I/Q-SDR receiver. The RX has got a preliminary front end, just for testing, therefore not worth showing. Anyway, some pre-amplification is not that hard to design.

The trick with this design is, to shift the phase by exactly 45 degrees. This is accomplished by the RC combination. Theoretically, a phase shift of 45 degrees is reached when the condition R=XC is fulfilled. I chose a 1k resistor, thus, the capacitor should be 15.6994...pF  32pF (see comments below). Ah well, let's add a trimmer...
For the I/Q-stuff functioning, the levels of both LOs need to be exactly the same, therefore, I added a voltage dividing trimmer to the In-phase path. This trimmer is 2k in order to balance the RC phase shifter.
The total level is adjusted by the trimmer between the oscillator and the filter crystal. This trimmer seems to have a tiny influence on the phase shift...

So, here you got it, the first sneak preview of the 30m I/Q-SDR stuff. See my previous posting for other possible frequencies, most will not have the convenience of a canned oscillator however.


  1. Hi Joachim,

    I am sure you mean 90 degrees phase shift between I and Q and not 45 degrees. BTW, what is the idea of having to xtals? Are they in sync?

    Andreas (dl8oam)

  2. Hi Andreas,

    no, I do mean 45 degrees phase shift at the generated frequency for the following reason:
    I/Q-SDR requires 90 degrees phase-shift for the quadrature, but on the beat-frequency; that's what your comment is about.
    The trick with sub-harmonic mixers is however, that those require signal at half (!) the beat-frequency. The mixer itself doubles the frequency. Doubling frequency also means doubling the phase-difference, hence, the phase difference at half the beat-frequency needs to be 45 degrees.
    Therefore, the generated 5.0688MHz signal is split, one path phase shifted by 45 degrees, which at the mixers results in 10.1376MHz signals at the mixers, which are phase shift by 90 degrees.

    73, Joachim (pa1gsj/dl1gsj)

  3. PS: Forgot about your other question. This circuit employs one oscillator and one crystal. The oscillator generates a square-wave-ish signal, lots and lots of harmonics. For the sub-harmonic mixers to work best, a pure sine is required. And this is the job of the crystal, it is used as a crystal filter to get rid of all the oscillator's harmonics and form a signal close to a sine wave. One could do that with capacitors and coils.... many more parts would be the result.

  4. I posted a sub-harmonic mixer circuit here:

  5. Sean,
    interesting document. It will certainly give nice results. As I see it, it is a push pull configuration at the RF end, i.e. 180 degrees of phase shift. Note, it may look similar but will function in a totally different way.
    I figure, your design could improve my 30m grabber receiver (

  6. Nice receiver. A similar circuit is TinySDR (, which also uses subharmonic Polyakov mixers and phase shifting, though I must admit the phasing arrangement of TinySDR has me scratching my head (90 degrees at the LO plus an additional phase shift at the RF - perhaps 45 degrees? - with a final software-based AF phase shift).

    I'm thinking a fixed-frequency, crystal-controlled DC rx like this might be a good IF stage for a simple single-signal superhet.

  7. Hi qrp-gaijin,
    I know the TinySDR circuit... I must admit that I don't understand the phase shifting either. I believe, you are right, it seems that the all important quadrature shifting takes place in the RF section.

    Concerning you superhet idea, I was actually thinking along the same lines, e.g. for a traditional 9MHz i.f., 4.5MHz crystal are cheaply available. In this way rigs like the Drake TR4 can be equipped with a panoramic display...

  8. I am not clear about the termination impedance your mixers are seeing. As I understand it proper termination of the subharonic mixer with a diplexer is important for good performance (see e.g. and Do you have any thoughts on this?

  9. The design is derived from my QRSS-grabber receiver which itself is based on PA2OHH's receiver design

  10. Your calculation for R=Xc, with R=1K and F=5.0688 MHz should result the value of capacitor 31.4 pF. Did I miss something? Thanks.

  11. You're absolutely right. For whatever reason I missed that factor 2.
    My RX, the one with the trimmer, works. I hope your's will work too.

  12. I have not built one, yet. I will try to get one built soon. Another suggestion: Would it be more stable to generate 45 degree shift using a fix NPO capacitor at 32 pF and use a trimpot (1K value)with the center grounded? Using a VOM we can roughly adjust the value of the resistor 494 ohms to match the capactor's Xc. Just a thought.

  13. I made a wrong calculation :-). For F=5.0688 MHz, and fixed NPO cap 32 pF. The Xc is 981.2 Ohm. And the trimpot should be 2K, adjusted with VOM to get to 981 ohm.

  14. Regarding the comment from 2 years ago "I know the TinySDR circuit... I must admit that I don't understand the phase shifting either.": I've been thinking about TinySDR ( again lately, and decided to follow up here, since nobody else seems to have presented a proper explanation of how TinySDR works.

    I recently saw a document (, pages 51-52) about TinySDR that again glossed over the role of the HF phase shift components C3 and R1. And I know that C3 and R1 are indeed used for HF phase shifting; LY1GP told me so himself in private correspondence.

    So here are my thoughts about the TinySDR phase shift scheme. I haven't built or simulated this circuit yet, but I think I am starting to understand what is going on with TinySDR.

    The 90 degree phase shift used on the VFO is unusual. It is very confusing because the schematic indicates a "90 degree" phase shift at the VFO, and most people think, "OK, 90 degrees phase shift, it's a normal I/Q scheme". But we are dealing with Polyakov mixers here, so the 90 degree phase shift at the VFO frequency f becomes a 180 degree phase shift at the effective mixing frequency of 2f! So the purpose of this phase shift is not the normal 90 degree delay for the quadrature channel, but instead this 90 degree phase shift is, I believe, used to ensure that only one of the Polyakov mixers is active at any time (where a Polyakov mixer is defined as a pair of antiparallel diodes; TinySDR has two such pairs). When one mixer is conducting (either top diode or bottom diode of the pair), the other mixer is not conducting. This, I think, ensures that the HF signal coming in from the tank reaches only one of the mixers. I can only assume that if both mixers were conducting at the same time (as would happen if no VFO phase shift were used), some problems would happen, maybe something like the HF phase shift network no longer providing a proper 90 degree phase shift.

    I need to simulate and/or build this circuit one day to verify the above hypothesis, but it does seem there is more going on in the TinySDR circuit than a casual glance would imply.

    1. Dear qrp-gaijin,

      I know the TinySDR-design. You hit the exact same question that I had about it. Personally, I still believe that it wont work, for the exact reason you are mentioning. A Polyakov mixer, which the TinySDR employs, requires an l.o.-oscillator operating at half the frequency of the effective l.o.-frequency required. This however means, that the phase shift, in order to achieve 90 degrees of effective phase shift, will be required to be 45 degrees, hence, my rather complicated phase shifter.

      However, I believe that the TinySDR still will work to some extend. Why is that? The 90 degrees phase shift are required to cancel out the other side-band using I/Q. Any direct-conversion receiver will work just fine with any random SDR-software product for receiving signals. Given that the unwanted sideband is sufficiently far, e.g. some frequency without any traffic, the results will be just fine.
      Here's an example: figure that you want to receive the 20m CW band. You would be running an 13.999MHz effective LO (Polyakov or not, it does not matter). With a sampling rate of 96k, you would cover the range from 13.951MHz to 14.047MHz. You will be very happily listening to signals in the 20m CW band. Since there is nothing between 13.951MHz to 14.000MHz, there will be no side-band interference at all. In this case, there would not be a need to suppress the side-band. And, in a design that does not suppress the unwanted side-band, you would not notice any difference.

      I hope it helped a bit...


  15. I still think that TinySDR could cancel the opposite sideband if a 90 degree phase shift in the RF section (using C3 and R1) combines with the 180 degree phase shift from the VFO for an effective total of 90 degrees phase shift at the signal frequency.

    I tried to simulate the circuit in QUCS but unfortunately my results are not conclusive yet.

    Another thing that might be possible with TinySDR is just to eliminate the bottom diode in each Polyakov mixer, meaning that each mixer then becomes a single diode mixer. The effective VFO frequency then would be the same as the VFO frequency (not doubled), so the 90 degree phase shift at the VFO should work properly to give the I/Q signals. No RF phase shift should be needed in this case, so C3/R1 could be omitted from the TinySDR circuit.