Odd behavior from analog output on PLC5 system

[wildswing 0]

Hey folks, this is my first posted topic and it's a spooky one. Got lotsa people scratching their heads over this one. Here's my "in a nutshell" version... I'm using a PLC5 with an analog output card in a RIO rack (due to location). Occassionaly the card does not respond as expected. The 4-20 goes to a drive for speed reference. Sometimes the drive does not slow down as instructed by the logic. Histograms have confirmed that the integer in the BTW is changing correctly. At the same time we have a chart recorder watching the 4-20 at the drive. Once in a while the 4-20 does not change. We see no BTW error, BTR errors, rack faults, drive faults, RIO Que Full bits, RIO Channel Full bits. Tried moving the integer to another channel of the same card. We see the same behaviour there. The changes in analog value last at least 1 second in the logic. Value is programmed to go from from 85% for 5 sec, then 60% and 25% for 1 second each then zero. Most of the time we see the corresponding change in analog output and all works fine. Once in a while, there's a delayed change where it's 85% for 7 seconds or more then zero. Sometime we see a rapid change i.e 85% for 6.5 seconds, 60% and %25 each for only 1/4 sec. Scan time of the plc is 29 mSec. The output card has been set up with the standard BTR/BTW pair logic set for non continuous, either get's enabled when itself or the other isn't. Ideas? Bad output? But we replicated the behaviour on another channel. Problem with something in the card that's common to all channels? Any ideas would be appreciated. Sorry if I don't reply for a while. Vacation next week. Internet access is neither available nor wanted on the golf course. Thanks, in advance, for your replies, Mark Edited 9 Jun 2005 by wildswing

[pseudoquas 1]

Hmm... A "sometimes" problem with the hardware, where the software is confirmed to work properly is invariably due to electrical noise or ground loops. Even on 4-20mA loop, I've discovered ground loops that produce a radical deviation in measured mA, but that deviation is nonetheless stable. It would be nice if you discover that the problem loop has cable shielding that is grounded in more than one location....then an easy fix. Otherwise, perhaps cosmic rays...