Today I attempted to measure the passband of the output filters. In order to do it with minimal intrusion I coupled the tracking generator output to the input of the filter bank with two turns looped through T11. The input to the spectrum analyzer came directly from the antenna connector of the uBitx. I then removed PA power, turned the drive all the way down and keyed the rig to engage the relays. I then offset the nominal reading within the passband to zero so I could get an idea what the baseline attenuation of the filters is after cutoff.
It is a very consistent -20 dB or so for both the 80 and 40 meter filters. What this means is the filter bank only attenuates high order harmonics by that amount. This is clearly the source of the harmonic issue on my unit. If we are getting only 20 dB of rejection from the filter then that means any spur at -23 dBc or higher is going to cause compliance problems.
Screenshot of the 80 meter filter is attached with markers at the 3rd, 5th, and 7th harmonic frequencies.
As an aside but worthy of mention: In response to my similar post about this issue on the uBitx Facebook page one owner claims to have tested his unit, packaged in a metal box, and found no spurious outputs higher than -52 dBc. Testing methodology was not disclosed. The point is, and I want to reinforce it, we need more spectrum analyzer tests of different units in order to be sure this is not an issue isolated to one or a few units. And we need to do this before sending everyone on multiple schemes to fix what is not yet a confirmed general problem in my opinion.