Tim replaced the 1.842 GHz SAW filters on the GPS low noise amplifier (LNA) with 1.57542 GHz SAW filters which were UPS'd from Digikey overnight Saturday delivery. He checked the installation using the 2m receiver's LO which broadcasts at a 1.56 GHz (for whatever reason).
We tried the SAW filters using the GPS cylindrical patch antenna (CPA), LNA w/SAW filters, and the Jupiter board. We got 4 - 6 satellites, which is comparable to the 6 or so satellites from the active roof top antenna.
We set up the avionics system on a ladder outside Andrew's house. We put a couple of laptops on a table, along with a lamp and a bunch of other junk, and used cantelope (the CAN bus monitor we built) to turn on and off the ATV system. We saw absolutely no difference in the number of satellites the receiver saw, which is great. We might have seen some kind of startup transient kill a satellite, but it didn't happen more than twice during a test of several dozen on/offs.
We verified that the ATV was indeed broadcasting by walking back to Andrew's parent's cabin and turnin on the ATV receiver with the little rubber ducky antenna... which picked up the 3W transmission just fine :)
So: ATV/GPS issues is resolved.
The 2m receiver's local oscillator is ~ 153 MHz. Yet we see components of the LO all the way up to 1.56 GHz!
Using Paul's signal analzyer, we saw that the LO's 3rd harmonic is larger than the 2nd harmonic, so it's more of an odd than even function. This means the LO is probably like a square wave - the rise time is too high (the bandwidth of the circuit is certainly too high), it's being overdriven (it might be being clipped into a square wave).
You'll note in the figure the weirdly shaped magenta trace. That's the output of Q2551 which pulls up just fine, but can't pull down at all. It's pull down is a 680 ohm resistor, so it sucks... causing the clipping you can see. The bounce we think may be the parasitic capacitance coupling the emitter to the base.