Looking through the schematics to see if yours is missing a cap in the keying circuit... there could be a single point where ICOM has a wrong value loaded or yours has a wrong value part loaded. Seeing this trouble out of another 7300 would be interesting. Anyone here have a scope and can take a picture of their 7300's release characteristics? 4ms is an odd delay. Maybe they are firing the release from the wrong variable in software (i.e. calculating the amp's key release delay from the PTT release instead of the "decay envelope finished" or some such jazz). If it's a hardware fault, shorted diode in a D/R/C setup near the driver transistor's lair... Seeing exactly the same timing out of another 7300 with different software would answer a few questions. It seems like we have to answer them ourselves anyway as ICOM doesn't seem too interested in fixing this. As for my radio's weird Key On character around 15ms, I didn't even look until reading this post a few days ago because I thought that in 2017, this kind of problem shouldn't happen, especially with a computer running the whole operation. Heck, you can delay the modulation channel a few ms and have the "perfect" compressor in the audio chain like we've been doing in the studio for a couple of decades now. Looking at the attack trace on my scope now, I'm pretty sure what's happening is that the power is set with a constant "in the middle somewhere" voltage and then switched to the actual detected voltage around 15-18ms. During that switching time, the value is 'up for grabs' and the output power is indeterminate until switching is complete and the ALC detector and timing have settled in. They did it this way to speed up the Key On timing. Since I'm not operating QSK, I'll use a 20ms turn on delay... some of the spikes are truly amazing.