Listen to Latest SolderSmoke Podcast

Saturday, February 13, 2016

ZL2CTM's Teensy SDR SSB Superhet -- Very Cool



Hi Bill

I thought I would drop you a quick line to show you something I have been playing around with for the past couple of weeks. It's a SSB superhet using a PJRC Teensy 3.1 microcontroller (YouTube link below), and a great audio library by Paul Stroffregen. Suffice to say I'm really happy with it.

There is a direct conversion front end, albeit using a VFO 15kHz down from the incoming RF. The 15kHz IF is then fed into the Teensy audio line in which can handle up to 22kHz. From there everything is in software. First is a 2.4kHz BPF tuned to the LSB, followed by mixer with a 15kHz BFO. Finally, there is a 2.4kHz LPF. Next step is to add both CW-wide and CW-narrow filters which can be selected during run time.

As you can tell from the video, my antenna is not the best and I have quite a bit of QRM in the shack. I really need to think about a better antenna, but that's another story.


I have been following your R2 endeavors with interest, and as I said to Pete I will attempt to replicate the R2 on a Teensy. I will try and use the divide by 2 arrangement you tried as the logic looks right.

I will also be adding in a transmitter. For that I'll use the microphone input and then take the audio from the line out straight to the RF pre/power amp. 

I must admit that I really enjoy homebrewing hardware/software hybrids as you get the best of both worlds. As i say that, I wonder how much ADCs cost these days to directly digitize RF...  Now that would be interesting.

I'll say again that I really enjoy following the podcast. I have loved electronics since I was a small boy when my parents bought me a battery, switch and light bulb. This age of cheap DDSs, microcontrollers and the like is amazing, and I get so much enjoyment putting them all together to make functioning ham radios. I hope more get into the homebrew field as the entry barrier is dropping fast.  

Keep up the good work.
73s
Charlie
ZL2CTM



13 comments:

  1. Another project I want to build. Must stop looking at these things!
    Seriously though FB! keep up ideas flowing.

    ReplyDelete
  2. I think the block diagram in the video is incorrect. I believe the second conversion is followed by a DSP phasing IF, not a simple BPF, otherwise there would no unwanted sideband rejection. Also, recognize that this is a superhet architecture and there's nothing to eliminate the 30 kHz image product from the 1st conversion. Interesting, but not very practical.

    Joe
    W3JDR

    ReplyDelete
    Replies
    1. I think Joe is right. If you are trying to receive an LSB signal at 7.200 MHz with this RX you'd tune the VFO to 7.185 MHz. The LSB signal would then be in the passband of the SDR's 2.4 kHz filter. So far so good. But look what happens if a signal appears at 7.172 MHz. That signal will also be in the filter passband. This is the classic problem of a low frequency IF -- it is hard to eliminate the image. That's why we have IFs up at 5 or 10 MHz or higher -- this allows us to build bandpass filters for the front end that will take care of the image. Charlie is able to tune to the other side of zero beat and NOT hear the kind of image you get in a DC receiver, but if he continued to tune up 30 kHz he would hear the signal again. I guess this is why we go through all the trouble to generate I and Q signals in front ends of SDR receivers. To take car of the image. Still, I really like Charlie's RX and have learned from it.

      Delete
  3. I think the block diagram in the video is incorrect. I believe the second conversion is followed by a DSP phasing IF, not a simple BPF, otherwise there would no unwanted sideband rejection. Also, recognize that this is a superhet architecture and there's nothing to eliminate the 30 kHz image product from the 1st conversion. Interesting, but not very practical.

    Joe
    W3JDR

    ReplyDelete
  4. The Teensy audio port has a BW of 22kHz, which filters the 30kHz image.

    ReplyDelete
  5. I don't think so. That image is at the RF input frequency and can't be filtered. The second image, the opposite sideband audio image frequency, can only be eliminated with a VERY sharp BPF at 15 khz or a phasing detector of some sort.

    Joe

    ReplyDelete
    Replies
    1. You are correct Joe. My error there. Apologies.

      I think I'll see if I can implement a phasing receiver using I and Q inputs.

      Thanks again Joe.

      Delete
  6. This is awesome, I am trying some Teensy DSP code now- inspired by this work by Charlie. Could you/Charlie post the Audio System Design Graphic and the code so we can reproduce these results? I am interested in your filter parameters. I am using the biquad and not seeing very sharp cutoffs in my test (but am guessing). Great work Charlie!

    ReplyDelete
  7. This is awesome, I am trying some Teensy DSP code now- inspired by this work by Charlie. Could you/Charlie post the Audio System Design Graphic and the code so we can reproduce these results? I am interested in your filter parameters. I am using the biquad and not seeing very sharp cutoffs in my test (but am guessing). Great work Charlie!

    ReplyDelete
  8. If Charlie is using the sketch I think he's using, there are two modes of operation, the filter mode that his diagram shows and an I-Q mode. I'm not sure why the designers moved away from the I-Q mode, but something about low frequency noise was mentioned in the thread. The hardware is simpler, but I don't think an image 30Khz away is a good tradeoff.

    Joe

    ReplyDelete
    Replies
    1. Agreed. It was certainly an interesting exercise and I learned a lot. Thanks again for the input.

      Charlie

      Delete
  9. If Charlie is using the sketch I think he's using, there are two modes of operation, the filter mode that his diagram shows and an I-Q mode. I'm not sure why the designers moved away from the I-Q mode, but something about low frequency noise was mentioned in the thread. The hardware is simpler, but I don't think an image 30Khz away is a good tradeoff.

    Joe

    ReplyDelete
  10. Rather than converting to I-Q mode, how about a simple crystal filter as used in the Minima, and then downconvert to baseband, I think that is easier to build than I+Q mixers. The crystal filter can be very simple, 2 or 3 crystals would be OK, since the Teensy makes a 15kHz '2nd IF'.
    Mike G1ZVN

    ReplyDelete

Designer: Douglas Bowman | Dimodifikasi oleh Abdul Munir Original Posting Rounders 3 Column