I am trying to get some more chips/cables. The problem is, with an FTDI chip, you expect the project to be through in a week. We do have a USB solution with SILABS F340, and it works fine, but at a higher cost. Prolific cable works fine, but it is a general serial-USB cable where I haven't explored the possibility of having my own PID/VID/Serial No, direct driver instead of a VCP driver etc.

With F232R, I can have VID/PIDs configured without any external chips, and that is why I went for that chip.

"There could be more than one failure mode, and any failure that has data-content dependence, is not likely to be physical. Send test patterns, that progressively get more baud-rigorous, and even do a Baud-scan, and plot the results."

Can definitely do that. But again, the idea is to get a ready made solution that can manage reasonable baud-errors and normal noise issues similar to a controller. I am sure that the manufacturers must have done all the tests and if they are kind enough, can help a worried user instead of expecting the user to do R&D on the chip. They can tell us 'If the baud error is within x percentage, chip handles it, else this happens and you need to do that' or 'If you have noise issues connect his cap here and that cap there and make sure to do this thing'

"The fault descriptions here seem to be vague, and sometimes conflicting. "

I have tried my best to elucidate the query, but the problem is that, the issue itself is a bit vague. The internals, how the pipe works etc are all in a black-box encapsulation, and when the chip hangs, it doesn't respond to any query from the driver. It is simply dead. You don't even have to have the application software running, to make the chip hang. But again, the chip works great if the communication is between two PCs (may be the baud error is within limits or vcc/gnd noise levels are decent). As I mentioned, I can plug-in or plug out the connections at any end any number of times, and it works. There are no waveforms that can be checked, and there are no error signals/messages that can be analyzed.

I can change the chip, and that is a bit depressing. Not that I cannot handle that, but wish it was avoidable.

Regards, Raj