Friday, June 8, 2018

A/B testing of receivers in a digital age

Over my many years in this hobby, I've frequently read reports, reviews and opinions on various radio's comparative performance derived by A/B comparisons with another rig.

Typically, the two rigs would be connected to an A/B switch so that each could be alternately connected to an antenna. The op would then listen and look at the S-meters as the receivers were tuned to specific stations. The result, minus QSB and QRM, would then be made obvious(?) and a write-up would inform us readers of which radio handled weak signals or near-by QRM better than the other.

The problem, of course, is the nature of QRM and QSB - they are both intermittent. Also, S-meters and the ops ear are not calibrated nor is the op immune to a subconscious preference for one rig to perform better than the other.

Enter WSPR and FT8 (note to FT8 haters - FT8 is a tool, not just a mode).

Decodes from these modes both show received signal strength. Substituting the A/B switch for a low-loss splitter would allow both receivers-under-test to receive the same signals at the same time. The resulting number of decodes, and their reported SNR ratio, would then give a far more accurate indication of how each receiver performs.

For the past few days, I've been exactly this test, comparing my Flex 6300 with my new SDRduo.

What I've learned so far (about the method, not the results) is that WSPR provides the best method of testing receiver sensitivity while FT8 (with its super crowded, one station on top of another in many cases) provides a very good indication of each receiver's blocking dynamic range.

I had hoped to post the results of all this testing by 10 June but I believe it will be a few days beyond that before I can collate the results. And I want to wait until the weekend's VHF contest before testing the rigs on 6m.


  1. And there is something additional important too: the sound card. Either built in the rig or external. And the sound card used in FT8 can have an tremendous impact on decoding results. Just connect them and check the basic noise floor with no input connected, using 'Spectrum Lab' (freeware) by DL4YHF. The worst are at -60db and the best are down at -125db.
    73's de OE1MWW

    1. Good idea, Wolfgang - I will give that a try, and with the upcoming QSX as well. Yes, it makes sense to baseline each rig/soundcard before applying RF.

      Thanks for that idea and 73,