Strange RS-232 scanner to PLC noise problem

So I had a strange problem today that defied what I've learned about data communication.

basic setup is: Scanner -> PLC via RS-232 using only the Tx, Rx, and GND at 9600/8/None/1

Basically what I saw was that whenever any of the VFDs for the motors where running, my scanner read rates went to hell. But the RS-232 connection continued to work fine. The PLC read the bad-scan data with no problem.

I finally resolved the problem by connecting the shield for the RS-232 line to earth ground on both ends. I was grasping at straws by the point I tried this and just happened to get lucky.

This makes no sense to me. Doesn't this create a ground loop? I've always heard to only connect the shield to earth ground on one side to properly eliminate noise and never on both ends.

I almost have the feeling that the scanner which is on a seperate 120vac circuit isn't grounded properly and by hooking up the ground on both ends I was providing the scanner a proper ground but I really have no idea. I have a scope meter, but am pretty clueless with what I'm looking for as far as detecting line noise.

Any info you kind fellers have on the subject would be greatly appreciated.

monkeyhead said:

I almost have the feeling that the scanner which is on a seperate 120vac circuit isn't grounded properly and by hooking up the ground on both ends I was providing the scanner a proper ground but I really have no idea.

Click to expand...

That sounds like a reasonable conclusion. Now, assuming that this IS the reason, you'll want to find a way to 'properly' ground the culprit. IOW, I wouldn't rely on the shield ground as the ONLY path to ground...



-Eric

I once looked into grounding, and whether or not both ends of a shield should be connected depends on the frequency of the noise and of the voltages being used in the vicinity. For frequencies in the 60 Hertz range, calculations (and experience) show that one end is better, and for 300 Hertz frequencies (as I remember), both ends grounded is better, as far as noise elimination is concerned.

The ground-loop problem is another issue that should be handled with proper grounding on all power circuits before the control circuit noise issue is addressed.

I'm almost hoping that's not the solution, because I checked the ground for the 120vac circuit until it headed up to the ceiling and everything looked good. But who knows how many other circuits are sharing that ground... the distribution in our building is a mess and completely undocumented. i have a plug with the ground tied to the hot that I use as my circuit 'tracer'.

Lancie1 said:

The ground-loop problem is another issue that should be handled with proper grounding on all power circuits before the control circuit noise issue is addressed.

Click to expand...

you just cleared up a lot of confusion for me with this statement. now i need to go slap my data com instructor for not understanding the material he was teaching.

Although every situation is unique, (and no matter what I say, there will be 427 other opinions)... It seems that in GENERAL, manufacturers recommend grounding 'signal' leads, be they pulse train (encoder), or analog, at one point and one point only, while they recommend grounding 'Data Transmission' leads everywhere.

There are NOTABLE exceptions to the guidelines though... Controlnet, for instanct, should NEVER have it's outer shield conductor grounded. Ethernet twisted pair cable should be unshielded if both ends can't be grounded (UTP), or shielded (STP) if both ends can be bonded to ground at an appropriate connector. One grounded and One ungrounded end on STP cable will work, but will cause major data errors (invisible except for network slowdown) at 100MBps and over. Go figure.

Best advice (even though it sucks), is follow whatever the MANUFACTURER of your device recommends. Failing to do that will often (cough Rockwell, cough Siemens) result in a "No Support" policy.

holy wow... i've just spent the last couple hours researching all this ground loop / inductance / noise stuff. and although more knowledgeable on the subject, i'm still just as confused.

but what i've come to realize, is the problem is the noise is disrupting the internal communication of the scanner and not the RS-232 connection between the PLC and Scanner. so right now i'm thinking that by connecting the two frame grounds via the shield i'm actually re-directing the noise to the cabinet's ground instead of through the scanner's electronics. but the noise is not great enough to distort the rs-232 signal.

kinda makes sense to me since EIA-232 states that 1's range from +3v to +15v and 0's range from -3v to -15v and are fairly discreet, where in contrast the scanner decodes the analog signal from the optical reciever and that circuit is probably much more prone to noise disruption.

who knows... i'm gonna run a nice fat conductor between the scanner's frame ground and the PLC cabinet's frame ground and disconnect one side of the shield and see if it has the same effect. if it does i'll test accross the two grounds and see if i've got potential and then call it a ground loop problem.

my brain hurts.

This does not sound like a comm's problem at all to me. It looks more like a problem with the trigger input on the scanner. Most of our scanners are triggered by a PEC which is connected to an input on the scanner. This input can usually be turned on by even a very low voltage and it would certainly be possible for noise to trigger false scans. It may be that the additional earthing you are providing is helping to prevent this but since the scanner is still transmitting bad read messages the RS232 part of the system must be working correctly.