I did check the BPF board before I started this thread and there was indeed a leaky diode on the 500khz filter. I
cut the lead on the corresponding diode and reassembled the radio. There was an improvement, but the problem
remains. I suppose it's possible that the 20 meter filter diode is leaking... I also sent a request to join the 440 yahoo group but never got a response.
I want to thank everyone for their help.
Do you realize when a radio is "tested" with another improperly coupled radio, especially a test receiver with AGC, you get can have all sorts of weird effects that make signals sound bad?

?
This is the very same reason we should never set a spectrum analyzer near a radio and use an open "probe" antenna to pick up signal for spectral purity tests. I sat through an entire Agilent analyzer demonstration where the Agilent rep coupled the analyzer wrong! Instead of using a wideband sample tap off the transmitter antenna port, he stuck a short whip antenna in the analyzer input connector!!
An improperly coupled test receiver will pick up stray radiation from RF leaks, many of which are modulated by noises, and that garbage sample that comes from power cords and cabinet leakage is not representative of what is on the air.
Making it worse, if the signal is strong and you have AGC on, you will hear "crap" that is -80dBc when mic gain is down.
I'm not saying the KW is OK, but we should all be aware improper test setups can create significant misleading artifacts.