Would be nice if this were a "soft" failure that could be fixed by resetting the code stored in the CPU.
Every time we see this, a check of the input pin on the CPU that connects to the mike socket shows a short to ground.
Inside the CPU. That's a hard failure. Can't be fixed by re-flashing firmware. It's just broken. Permanently. Like a shorted protection diode, you need a new one.
Bought my first Saturn Turbo 25 years ago cheap because it had this problem. The previous owner or tech had trouble with the relay inside the amplifier. Rather than just replace the bad relay, there were unshielded wires snaking out from under the amplifier's shield cover, across the inside of the radio and connected at the far end to an "ice-cube" size relay from RatShack. The coil on this relay was powered from the main 13.8 Volt supply, and was keyed with a wire that snaked over to pin 3 of the mike socket. This put the full 13.8-Volt supply voltage onto the input pin of the CPU.
Showed the exact same symptom as your new radio. Ended up buying a 2950 mobile radio with a damaged main circuit board, but with a good CPU. Made one radio from the two.
Over the years we have fixed this problem the same way, with mixed results. One customer wanted to fix his Galaxy DX2527 base after lightning hit his landline phone, which jumped to the D104 next to it on the desk. Same symptom. Stuck on transmit, could not change frequencies. He obtained a box with four salvaged CPU boards from the same 1995-version junked 2950 mobile radios. All of them were dead as a post. Leads me to wonder if the "horse-trader" who sold him the boards knew that none of them were any good. No way to tell.
Finally found him a good CPU salvaged from a mobile. After trying a half-dozen that were bad.
From 1992 or so to the present, this type radio has had between four and six different revisions of the CPU and display. For the most part they will not interchange. A used replacement must come from a radio of the same production type to be compatible.
Led me to wonder what a person would pay for an aftermarket design. A drop-in replacement would take some programming effort. I'd favor a OLED display, rather than the LCD the factory uses. But I estimate the development cost to be several thousand bucks more than I have on hand to spend on such a project.
There is an opportunity here for someone with deep pockets who can let the money ride while enough of them sell to break even.
But that ain't me.
Not this week.
73