Powerflex 40 comm slow when motors are on

[mrhilamma]

We have 5 powerflex40 communicating with an Allen-Bradly 820 via rs485.
When the motors are off, communication is almost Instantaneous (write/read). However, when the motors are on, most MSG_Modbus commands fail.
Anyone else experiencing this?

 

[mrhilamma]

Sorry for the brevity of my initial post, I was still at work and had a friend type that out on their phone for me... I shall explain further


We are running a conveyor system using 5 Powerflex 40 VFD's. The drives are being controlled by a Micro820 PLC connected with a RS485 cable. I am using Modbus RTU protocol to communicate with all of the drives.


Currently, while the drives are not commanded to run, I can successfully communicate with each of them. I can use MSG_MODBUS in Connected Components Workbench to send Function6 writes and Function3 reads and get or assign anything needed with (seemingly) no issues.



Sending Function6 write to make the start bit on a drive TRUE will immediately start the drive and it will run at any speed I set the drive to with a Function6 write to the refspeed register. This is when the problem beings....


Once the drive is running, communicating with it again becomes painfully slow, sometimes taking almost a whole minute to successfully send a Function3 read command or several seconds to send a Function6 stop command. Most MSG_MODBUS calls will simply time out or return error 55. If I can manage to successfully get a Function6 Stop command through, the drives will immediately go back to communicating properly and within just a few milliseconds of response time.



I have no idea what/if I've done wrong setting this up. Please feel free to share any ideas you may have! I am new to this forum and look forward to your responses.


Thank you!

 

[drbitboy]

I would for look (sniff) for traffic on the RS-485 network. Maybe the drives are configured to send status at X Hz and the issue is bandwidth, although if they are Modbus slaves, then they should not be initiating any Modbus traffic.

 

[Steve Bailey]

My first thought is electrical noise when the drive is running. Put a scope on the network cable to look for any difference when the drive starts.
If you don't have a scope available, try rerouting the network cable to keep it as far away from the power wires as possible and see if that improves things.

 

[mrhilamma]

Thanks Steve and DrBitboy.


I'll put the scope on the line tomorrow and check for both (activity and noise while on and off).



As for EMI, that is my thought as well... I have all of the 480-3ph power on the right side of the cabinet and all of my low volt DC on the left. My rs-485 is currently daisy-chained between drives such that the hanging loops are 90degrees to the running of the HV wires. Obviously the rs-485 on a Powerflex 40 leaves little wiggle room in where I can get my wires relative to the HV lines.


The PF40 manual recommends ground the rs-485 cable shield in only one place. The PF40P manual recommends grounding the shield at each node (one side only). I followed the first suggestion... Is there any advantage to one over the other? I've been seeing much of both methods on the internet, but little info on it.



Thanks again for your suggestions!

 

[Ken Roach]

The RS-485 port built into the Micro 820 is non-isolated, so whatever noise is on your DC common will be on your network too.

RA recommends the 2080-SERIALISOL module for anything that could have a long network or be noisy, and the application profile for using Modbus/DSI with the Micro 800 function blocks presumes you're using the isolated module.

Do you have terminating resistors in place at the extreme physical ends of the network ?

You're correct to follow the DSI wiring guidelines to ground the shield in only one place.

I'm not sure if there are good ways to measure retries with these components; the drive is going to just know about timeouts, and I'm not sure if the UDFB has a retry counter in it, or just an "the drive has timed out" status bit.

 

[mrhilamma]

Thanks Ken,


Yes, 120ohm resistors are installed across d+ and d- at both extremes of the cable. Total cable length only measures about 1m from master to the last of the 5 slave drives. (All being installed in the same cabinet.)



There are a lot of LVDC IO device wires run near the drives and HV wires... so what you're saying about the noise being on the LV supply makes a lot of sense to me. I'll find out when I scope the signal.



Would something like this solve the isolation issue?



https://www.amazon.com/Industrial-P...uPWNsaWNrUmVkaXJlY3QmZG9Ob3RMb2dDbGljaz10cnVls

 

[Ken Roach]

I prefer more robust ones, including the 2080 isolator, but for a quick test that can't hurt.

 

[mrhilamma]

Thank you guys for all of the help!


After scoping the network common, I found that there were 5-9V peeks of noise all over that line while the drives were running, and only fractions of a volt of noise while they were off. Clearly this was the issue.



I decided to rewire the network following, what I know to be good practice, vs following what is described in the RA Powerflex40 manual. The manual does not have the network common terminals (terminal 04) connected to network common, but instead tied directly to ground for all nodes. Instead the rs-485 standard for using a single twisted pair in shielded wire, is to connect the shield to network common at all terminals - then ground the shield at the furthest node.



Removing the PE connection from terminal 04 and replacing it with network common on the shield fixed the issue entirely. There is still noise on the line, but less than 1V at it's largest peaks, instead of 5 - 9V peaks!! I will also be adding an Isolated repeater from the PLC (master node). But for now, adding an 480uF electrolytic cap and small inductor to the PLC power terminals has cleaned up most of the noise all together. Peaks only around 100mV while the drives are running.



Thanks again for your help and your suggestions! I hope this solution helps anyone else who runs into similar issues in the future.