so ive been chasing this one for a couple weeks now and its driving me nuts. my ic-7300 will start out fine, full 100w on HF, sounds great, but somewhere around 5-10 minutes into a session it starts dropping output. like it'll be down to maybe 40-50w and the ALC is doing weird things, kinda spiking where it shouldnt be. no error codes on the display, fan seems to be running, nothing obviously wrong externally.

first thing i did was check the SWR on the antenna side thinking maybe a connector was thermally cycling loose or something but thats all clean. reseated all the connectors i could get to on the PA board without fully stripping the rig down. didnt help. the dropout seems to happen faster if im doing digital modes vs SSB which makes me think its heat related but the fan ramps up like it should and the heatsink doesnt feel crazy hot when i pull the cover.

i havent scoped anything yet, mostly because im trying to figure out where to even start. has anyone seen this specific symptom on a 7300 or even a similar icom? wondering if its the finals starting to go or maybe a bias issue that drifts with temp. i do have a service manual PDF but honestly some of the alignment sections are a bit over my head without some guidance on what to actually look for first.

that pattern — full power cold, dropping after warmup, worse on digital — that screams bias drift to me. the 7300 uses a temperature compensated bias circuit for the finals but if one of those compensation components is getting flaky it wont track properly and the transistors end up running out of their linear region as things heat up. ALC going weird is consistent with that too because the radio is trying to compensate for something upstream.

before you go digging into the PA board id check if icom has any service bulletins on this, i vaguely remember something about a batch of units having issues around the bias network but dont quote me on that. if you do get into it, the thing to measure is quiescent current on the finals when cold vs after it starts acting up. if theres a big swing thats your smoking gun. the service manual should have a table for what those values should be, the alignment procedure walks you through it even if the language is a bit dense. its not as scary as it looks once you have a DMM in there.

had something similar on an older 746 pro, turned out to be a cold solder joint on one of the driver stage connectors that only opened up when the board expanded. not saying thats your problem but worth doing a close visual inspection with a loupe under good lighting before assuming its the finals themselves. i reflowed a couple joints that looked fine to the naked eye and it cleared right up. digital modes push the duty cycle way harder so any marginal connection is gonna show up there first.