Parker AC10 via mb2hal: unreliable communication

I'm connecting to a Parker AC10 VFD via USB/Modbus from linuxcnc 2.8.2.
But the connection is unreliable. There are sucessfull commands but a lot of errors too:
CRC
Connection timed out: select
Request for slave 128 ignored (not 1)

Looks like a physical i.e. wiring problem, but there is a nicely terminated twisted pair wiring between VFD and USBtoRS485.
Even more the same wiring worked well with the Parkers antecessor a huangyang VFD.

Any ideas?



 

andypugh wrote:

Not really a solution, but I see a lot of failed updates with my HY VFD under modbus control when I look at the output.
My solution is not to look at the output
The spindle seems to work just fine.

I'ld love to not look, but dropout rate is to high.
Optical isolated USB/RS485 and new wiring will be the next in poking around in the dark.

How much different is that VFD from a Huanyang? If it's close, you are welcome to try out my custom made VFD HAL component. It was made specifically for a GT Series but if you can program a bit, it should be adaptable. It can go for hours without a single communication error.
Jim

Try using an RS485 adapter with A/B and GND connection instead of those simple USB with A/B only. I had similar issues before, if I remember correctly.

I had a problem with a bunch of Lenze ACTech drives, where if the commanded speed was less than the minimum setting set in the drive the communication would show a Modbus comm error in Linuxcnc. So when ever the spindle was stopped and the speed was set to zero I saw a constant string of comm errors. The slolution was to disable the minimum speed limit in the drive and do the limiting in Linuxcnc. No more comm errors.

Meanwhile I could solve the case by using another USB2RS485 adaptor..

The old one showed some weird behavior on the scope. With only 250 mV dU between A+ and B- it was inside standard but obviously outside the parker AC10s expectations.  .

The new one (with GND connection by the way) showed of better