Thursday, May 21, 2009

WSPR trx using DigiKey-crystals

The idea here is: using a superhet design to create a full-blown SSB transceiver, with proper USB generation. The transceiver should be able to cover the WSPR and the QRSS band of the 30m band. That's what I found, and propably will build.

Thinking crystals and searching the DigiKey page at the same time, that's what I found:

- standard crystals for 7680 kHz
- SMD crystals for 5223.438 kHz

Doing the maths reveals:
2 x 7680 kHz = 15360 kHz
15360 kHz - 5223.438 kHz = 10136.562 kHz
BUT: 10138.7 kHz - 10136.562 kHz = 2.138 kHz
=> we are off by about 2 kHz.
Remember, this is subharmonic, thus the 2 kHz offset would need a 1 kHz correction at the LO.

The concept would be the following:
  1. The SSB-filter will be built from the SMD crystals, this allows for a small filter, that even could have it's one casing.
  2. Regular XO on 7680kHz, pulled up by 1.069 kHz (which should be manageable) as local oscillator.
  3. A standard J310, or similar front-end, no AGC but some switchable attenuation (RX train).
  4. Here comes the trick, a subharmonic mixer รก la RA3AAE, i.e. two antiparallel diodes (RX train).
  5. The subharmonic mixer propably also works for TX... If not, a regular doubler is required, in combination with a regular mixer (TX train).
  6. Some sort of final (TX train).
So many idea, and so little time...

BTW: I am looking for a combination of standard crystals that would get me to 10148kHz. Any idea? Please let me know!


  1. Do not forget that the centre frequency of the WSPR/QRSS band is 10.140, so the total pull-up needed is aboyt 1.8 kHz on the osc frequency. possibly doable given the experience with pulling up the 5.0688 MHz xtals.

  2. And the history repeated with the PSKmail-TRX-LO, in which a 6144kHz crystal produced a frequency 10kHz above its' nominal frequency. Thus, a safe bet seems to be to use 74HC86 for the oscillator. BTW, I could not observe this effect, when I build the 22m test osci with a 74HCT00.