D0-More with GS-Edrv100 timeout

[davidbgtx]

I have a do-more with 3 Durapulse GS3 drives set up to communicate through GS-EDRV100's via a stride ethernet switch (all in same cabinet). Getting random trips (maybe not random, seems to be related to speed reference changes)of comm time-out on the drives. Have taken timeout for the GS-EDRV100's up to 1 sec and 8 retries, left poll at 0, in the Do-More. Ran for a couple of hrs with no trips after that, not really sure if the problem is completely solved, no production to run on it for several days to be sure. Should I really have to take time-out and retries that high for only communicating with 3 GS-EDRV's? 

[ADC Product Engineer]

It depends on how you have the wires ran.  This sort of thing is almost always a wiring or grounding issue.

The wiring to the Serial ports should be kept as short as possible.  Also this wiring should be kept away from the high voltage VFD wiring as far as you can get it.

Motors need be grounded to the VFD bonding point on the bottom of the drive, NOT to the main panel grounding bus.

If all of this still results in comm issues, you may need to start looking for a ground loop elsewhere in the panel.  Eliminate any daisy-chained grounds.  They are a big no-no in VFD installations. 

This should help you get started in finding the issue.  If not, please give ADC tech support a call at 1(800) 633-0405 and they can help you figure out where the noise is getting into your system at.

[dcsinc]

I found using the short .5 meter cable that comes with the GS-EDRV100, I had no problems testing "on the bench". But I was going to have to use a longer cable (2 meters) the GS-EDRV-CBL-2 that has the same pin-out to mount the GS-EDRV100 farther away from the drive. I started to have all kinds of communication errors. Note: I already had SW1 switch set to "Source" as I wanted to use the external fault input from an E-Stop. I found that by tying both the "DCM" and "ACM" to common which is grounded at my separate 24VDC panel power supply cured the problem. I have never seen another fault in over a month of testing. It appears their are com faults if these two DCM and ACM are left "floating". 

[davidbgtx]

I'm not using any in/out terminals on the drive, only the short cable that came with the GS-EDRV100. BUT, I'll try grounding the ACM and DCM terminals anyway, can't hurt. I read on AD that I should be using shielded ethernet cables (which I'm not, never needed to) so am ordering some to try that. I have some weird thing with the analog inputs (2 inputs on the same F2-8ad4da-2) also. Both analog inputs were bouncing all over the place and I know this is wrong, but the ONLY way I could settle them down and get good signals, was to ground the shield to the analog proxes AT BOTH ENDS. I tried grounding at either end, didn't work. I'm really scratching my head with this system as to where, I assume my noise/ground loop/ whatever is causing my problems. I hate trying to work on the symptoms rather than finding the cause.

[ADC Product Engineer]

Google 'Finding Ground Loops' for some help on finding the loop.  Most of the results are for AV systems, but the theory is the same.

[davidbgtx]

Ground loops from analog inputs removed, still no better. Tied ACM and DCM on drive to ground, not sure yet if it has solved comm fault issue (no trips so far), however I was having trouble controlling the speed when ramping up and down, and tieing those points to ground completely solved that. I mean night and day difference. Thanks dcsinc.

4/7/2016  - 3 days running not 1 communication timeout.

[dcsinc]

Attached are two files that I wrote a simple Do-More program with the GS-EDRV100 that I sent a command to increment and decrement the $GS3_100.RS485RefSpeed by 0.1 HZ when it was = to the received GS3_100.OutputFrequency output from the GS-EDRV100 between 4.0 HZ and 60.0 HZ. You can see in the first example without grounding ACM and DCM there is significant lag between 20 HZ and 40 HZ. When ACM and DCM are grounded in the second example the drive responds as you would expect. This was caused by slow communication from whatever?  I sent this to Host and AD in an email and neither sent a response as to why this may be happening. I would hope AD would test this and give an answer.
The YouTube video by AD https://www.youtube.com/watch?v=xaa2T3hm0hw&nohtml5=False  controlling a GS drive does not mention that ACM and DCM must be grounded. Without grounding ACM an DCM  I only got 6 complete cycles over 30 minutes. When grounding ACM and DCM I got 20+ cycles over 30 minutes. Obviously AD needs to address this problem when telling customers they can just connect the GS-EDRV100 to a drive without grounding the ACM and DCM

[ADC Product Engineer]

Looking at the waveforms, it is pretty easy to see that the comms are timing out, most likely from induced noise.
 
Tying the two lines to ground is fixing a symptom, not creating a "cure" so to speak.  What it is doing is giving the noise on the comm line a path to ground, which shouldn't be needed in most cases.

In other words; the original noise issue is still there, you have just masked it by creating a ground path for it to drain off on.

[BobO]

I haven't seen your email, but the guy responsible for the support box at Host was out all of last week.

[Greg]

I haven't seen an email on GS-EDRV100. My email: support@hosteng.com

[dcsinc]

Greg, I just resent the email to you. It was back in February.

[Greg]

dcsinc,

I answered the email saying that the GS-EDRV100 was probably not the problem and referred you to the AD drives experts. I think the ADC Product Engineer on here has given some good answers that you should consider if you haven't talked to the drives experts at AD yet.

[dcsinc]

Greg

I spoke to AD about possible solutions. It appears it is the GS-EDRV-CBL-2 cable. It is a 2 meter long flat cable without twisted pairs or shielding. I did another test with a 4 meter piece of CAT5 with RJ12 connectors and it works perfect without grounding ACM. I found a 6 foot RJ12 straight through cable online with twisted pairs and shielding. I will update this when it arrives and I test it.