kili wrote:Nope, it is already desoldered. The conditions are the exact same, only the program that is accessing the ch341 changes. I really don't have the knowledge to make any educated guesses, but reading your answer and the AN1116 makes it seem that there might be some problem reading the status register since it is used also in the identification process, as you already said.
I'm amazed, I can't see anything in the code that explains this behavior with this unique component. The other eeproms of the 95xx family have no problem. Apart from mixing the A8 and A9 addresses with the write opcode all the rest of the code is common.
I asked my usual correspondents if they had this component for testing with a logic analyser, but no one has it, even me. I ordered one from Aliexpress, I may have it in January, or in three months or never.