Conducting an experiment with rxondelay, I set my node at 1000ms and connected to another node on a different server that I'm able to monitor its transmitter.  I applied a signal to the receiver with rxondelay.  The transmitter on the connected node keys up and as I count to 10, No audio was present so I lowered the rxondelay by half, 500ms. Conducting the same test, I found that around the 6 count audio came out the transmitter.  Lowering the rxondelay to 100ms, and got a count of 2 before hearing audio.  I determined that rxondelay is working. In my case it stops the ping pong effect. I find a 15ms delay is ideal for a gm300 receiver.  It would interesting if someone out there could conduct this same test....  I also found using rxsquelchdelay=XXms is another useful setting to eliminate any squelch tail noise, but it must be used with dsp or flat audio on ACID.