Wednesday, July 29, 2009

short update on the WSPR/QRSS trx

This project has been resting for a a while. Yesterday, some progress was made. The local oscillator now has got a buffer/amplifier producing much more power than the mixing stage would ever need. This stage uses a 2N2222A in an emitter follower like configuration.

Even though the signal is pretty clean, in order to sure, I added a crystal as filter. Output level is adjustable to the needs of the mixer by a trimmer.

Here's a photograph of what there is sofar.

Presently I assume that I would need something like 2V peak to peak to properly drive the Polyakov-mixer, thus the buffer will be run with stabilized 5V.
Oscilloscope shows 1V/div, no further surprise here...

Let's see how far I come today....


  1. Hi there,

    Searching for information on required peak-peak LO voltage of Polyakov mixers, I found (again) your blog. Just wondering, do you have any conclusions on this?

    I'm thinking about a general-coverage (3-30 MHz) phasing DC receiver using Polyakov mixers. This would require wideband 45 degree LO phase shift, which can be done as you did in your 30m subharmonic grabber with an adjustable RC network. However the LO output level will change as the RC network is adjusted for different LO frequencies, so it will need some buffering and amplification, and I was wondering how much LO level to aim for.

  2. Hi there too!

    What level you use depends on the diodes you're intending to employ. In order to provide clean doubling with as few harmonics as possible, you would like to distribute your LO's (sine) signal such that the diodes for the positive half-wave open exactly between pi/4 and 3pi/4; the ones for the negative between 5pi/4 and 7pi/4. At those phases you signal must have Vd (forward voltage drop), which would be 0.7V for Si.

    The RC phase-shifting network is rather frequency sensitive. For wideband applications, I would probably go for some logic (digital) circuitry, e.g. having the LO running at 4 times of the subharmonic LO-frequency, do a 180 degrees phase-shift and divide the individual signals by 4, which would result in a 45 degrees phase shift. I could imagine that one could do some digital manipulation of the pulse width to bring as to reduce the same to the actual width you need the diodes open. Maybe something that runs at 8 times the LO to derive a gate signal from... The digital way, the level would not matter...