An unhappy situation.
If you could run a separate screened cable for the RS485, that would help considerably.
Is this an umbilical to go subsea for your modified motor? In which case I guess you are stuck with what you have got.
Yes it's the subsea umbilical. No option for a separate cable.
What is your signalling speed?
If it is not too high, you could consider putting low pass filters at each end of the cable to attenuate the noise from the PWM and leave the RS485 intact.
JimB
I will try that as well as the 3-resistor biasing circuit in the app note posted by MikeML, if other options don't pan out. I don't need any more than 9kbps.
So far I've tried RS485, CAN bus and DSL. I expected RS485 to work the best, followed by CAN bus, and I figured I would get a free take-home for the DSL modem pair. But much to my surprise, RS485 and CAN bus both fell on their faces while the DSL modem pair was almost a great success.
The DSL modem pair performs flawlessly, like it doesn't even know it's transmitting through the worst EMI environment conceivable, 95% of the time. Zero data corruption, zero difference in latency between VFD running or not running. But There are occasional hang-ups where 4-5 seconds elapse with no throughput, then back to normal. Or not; a couple of times I've had to reset the modems.
I really expected more out of RS485; In light of the proof of concept given by the DSL modems, I suspect I'm doing it wrong, hence this thread. I want to make sure all the nails are in the coffin before I bury it.
In the morning I'm going to test out a homebrew opto-isolated digital current loop serial interface. I've read about old teletype machines using these types of communication. 20mA or 60mA ON, 0mA OFF. I tested 4, 20, 60, and 100mA current levels today, and all of them showed equally nasty spikes (>30V) on the scope when the VFD was running. I was surprised that even with 100mA flowing through the wire, the EMI was enough to cause these spikes. But even with the spikes, there was no change in the current level as read by my DMM. I think that current-based (high current,60mA maybe) digital comms has a fighting chance.