Sounds like exactly the same problem ICOM has in some radios.
> Probably easier than modifying internal circuits, one might introduce
> a make delay between KEY OUT and KEY IN. The simplest delay would be a
> relay. A reed relay should proved a few milliseconds delay and an
> armature relay should introduce about 10 milliseconds delay. A relay
> will introduce audible keying noise in the shack and the delay on
> break may keep the signal up when the processor is starting to shift
> the oscillators back to their receive frequencies. That's just moving
> the chirp from start to end of each code element, not necessarily
> progress. It would be preferable to use something like a TLC555 timer
> and a bit of logic to delay the MAKE without adding delay to BREAK.
> Likely a millisecond of delay will be very sufficient. I've not
> detected the microchirp so I can't be sure it will have been cured.
> I'd prove the microchirp was cured using the simple relay, then create
> the necessary quiet circuitry to accomplish the make only delay.
If you do a make-only delay, you also will have to stretch the signal
from the key input to restore the weight. That shouldn't be difficult.
The receive window on QSK will shorten, but anything is better
than clicks or transmitting while the VCO is moving!!
73, Tom W8JI