[bmitc]

I'd just be happy to find a USB to RS-232 adapter using either a Prolific or FTDI chip that just worked. Even non-fake chips are famously flaky. It sort of blows my mind that the supposedly "better" serial evolution in USB can't even get RS-232 communication right. In my experience, it's best if the RS-232 communication is done on hardware as much as possible and then communicated to a computer with Ethernet or another device's actually usable USB driver than to rely on a Prolific or FTDI chip. Relying on these USB to RS-232 converters for anything other than desktop prototyping is a recipe for something not working.

[josephcsible]

Maybe you should go the path that the Arduino UNO and Mega use, and use an ATmega16U2 programmed to do just that. The code they use lives here: https://github.com/arduino/ArduinoCore-avr/tree/master/firmw...

[bmitc]

Most of my industrial career, I've used things like this for actual systems beyond prototyping:

https://www.ni.com/en-us/support/model.ni-9870.html

They go through an FPGA and then exposed via drivers over Ethernet or USB and are rock solid. You can access the serial data on the FPGA or you can use a default driver/personality for them that makes the module plug and play. There are other products I've used that expose RS-232 over Ethernet or fiber that work decently well.

There's also the Pmod interface:

https://digilent.com/reference/pmod/pmodrs232/start

I'll take a look at what you linked. I have often thought about doing something in this area, because it's amazing how reliable serial communication over RS-232, RS-422, and RS-485 can be, within their operation constraints.