Maybe OT: High-Speed Headaches!

Hi guys,
I'm having fits with what should be a simple project.

On a rewinding application, I'm trying to count revolutions of a large bedroll using a flag and high-speed proximity sensor.

Equipment:
PLC-5/40e with 1771-VHSC counter card
Dedicated 24VDC Power supply
TURCK Ni 4-M12-VP6X-H1141 12mm Inductive Prox

The flag is mounted laterally just under the circumference of the 20" diameter bedroll, which is travelling at a surface speed of 1300 fpm. The prox is stationary and picks up the lateral approach of the flag. The flag is 1.5" and passes within 2mm of the face of the prox.
So I calculate
1300 * 12/60 = 260 in/sec = .00384 sec/in
.00384 * 1.5 in = .00576 secs of on-time, every 1/4 sec or so.

The spec sheet for the sensor says its switching frequency is 2000hz, and the 1771-VHSC manual says that a channel in counter mode can handle 1MHZ, so picking up this flag consistently should be no problem.

I know that the number of revolutions per machine cycle is 60. I need to trigger an event on 59, and end it at 60.
The problem is that the sensor does not always count 60. It counts anywhere from 60-68 pulses per cycle. I've never seen it count < 60, but I have verified with an array tracking machine cycle encoder positions that the machine occasionally misses a pulse entirely (There should be a pulse every 6 degrees, but occasionally there are 12). Mostly though, I see double counts < 1 degree apart.

The wiring for the prox and the VHSC input channel is in a dedicated conduit, about a 20 foot run to PLC panel. The power is on a dedicated power supply.

Here's the really odd part: I've used three different power supplies, wired into three different AC (120V) power sources. I have a SOLA 2.5a, a SOLA 10a, and an AB 10a power supply. The AC sources are off the main control power circuit, off of a dedicated circuit (fused) from the control transformer, and a dedicated power cord plugged into the panel convenience outlet (fed from same control transformer).
Now, of all the combinations that I have tried, the ONLY thing that worked consistently was using the 10a AB power supply plugged into the convenience port! With this combo, it counts 60 pulses every time, just like it should.

Never seen anything like this, and don't know what to make of it. Obviously, the one working method is not a workable solution. I've double and triple-checked all the power wiring in the cabinet, and even ran it outside of the panduit channels to avoid noise. Haven't tried to scope the 120v signal, but I've never heard of the AC voltage causing these kinds of problems with a device on the 24volt side of a power supply.

Anybody have any ideas what I might be missing here?

Thanks,
Dustin

Devil's advocate here, describe the AC & DC commons and how they are bonded throughout.

jstolaruk said:

Devil's advocate here, describe the AC & DC commons and how they are bonded throughout.

Click to expand...

The DC common wires are only two, one to the prox, the other to the return terminal of the counter card, not tied to other commons for other channels on the card, both wires are a home run back to the power supply.
All the AC commons are tied together on a T-strip, with one lead going back to the grounded transformer terminal. Pretty standard stuff.

So the intent was for the DC common to be ungrounded? Just wondering if thats difference between the Sola & AB suppplies - one brand has the DC common tied to ground internally and the other doesn't.

New installations ground the DC common at the power supply but I sense this may be an older system? (PLC5). Any reason the DC common can't be grounded? The Turck may need it.

Interesting. Well neither the sola nor the ab supply have a terminal for DC ground. I just assumed that they are grounded internally to the AC supply ground. But you know how well assumptions usually work. Definitely an easy experiment to try an external ground to the dc common of the power supply. Hadn't even crossed my mind, thanks!

If this is a new install:


If this is existing that started acting up or all of the above checks out good, check for VFD noise being related and it doesn't always have to be parallel conductors...but that's the first thing to look for.

Filter the AC feeding your sensor PS.

If at all possible, ditch the inductive prox. sensor and put a real encoder on there...you already bought the card for it!

If you ground the DC common, please carefully analyze the implications for all the devices.

It is usually preferable not to ground your sensitive stuff especially near VFDs. As long as in practice your measured voltage level with respect to ground does not rise above the isolation ratings of the devices, you're golden. This may happen with 460vac to 24vdc supplies, but with 120vac feeding an ungrounded supply, I have never seen +24vdc float higher than about 50 volts.

As soon as you bond instrument common to ground, you are inviting VFD noise from near and afar.

If the power supply has a metal case, there will be a ground screw for it.

...the ONLY thing that worked consistently was using the 10a AB power supply plugged into the convenience port! With this combo, it counts 60 pulses every time, just like it should.

Click to expand...

Consider HOW the working solution is different from the non-working trys. Grounding is apparently not the difference. It could be simply that the 10 Amp AB supply has enough reserve capacity (lead in the pencil) to switch the prox at high speed, and the other power supplies don't. It could even be that the convenience outlet has a higher operating voltage (with everything else running), where the other input supply AC voltages are being pulled slightly lower by other loads.

Try the 2.5 Amp supply on the convenience outlet. If it works, then your other circuits are being pulled down somehow, or the convenience outlet is wired to some other source. It it doesn't work, then maybe you need a 15 Amp supply to make sure of 60 good counts.

Another thing to try is to raise the 24 volts supply output voltage to the maximum allowable by the connected equipment, maybe to 26 volts. Higher voltage with same current = more effective power delivered.

I would check the differences in the specifications for the Sola 10 Amp supply and the Allen Bradley 10 Amp supply. Do any of the rating differences jump out? Is there any difference that would allow the AB unit to deliver more power to the Turck proximity switch?

Equipment:
PLC-5/40e with 1771-VHSC counter card
Dedicated 24VDC Power supply
TURCK Ni 4-M12-VP6X-H1141 12mm Inductive Prox

Click to expand...

I really appreciate you listing the equipment details right up front. That is so rare, it stands out! I bet with that good description, someone is going to figure out the solution.

The Turck prox is 200ma typical and with the dedicated supply to that circuit, any of those should work.

Are you using the complementary outputs of the prox to feed the 177a module? Shielded cable? With shield properly connected to the module per the User/Installation Guide? Power Supply noise could be an issue.

The Turck prox is 200ma typical and with the dedicated supply to that circuit, any of those should work.

Click to expand...

At high speeds, it is a different ball game. The thing is: BMacAttack said it worked perfectly with the 10 Amp AB power supply running off the convenience outlet - same Turck proximity switch in all 3 cases, but only one worked. Why?

The prox is seeing a pulse that is the same as a 174hz signal - occuring every 1/4 second. Not a high speed for this prox, capable of 2KHz. And its power ramping up to 50 times its nominal?

From my experience, its pointing to a circuit that is not shielded, or improperly grounded, or noise being picked up by not using the complementary signals (not rejecting common mode noise).

But why does it work with the AB 10 Amp supply? Noise or grounding doesn't explain that, and I think Bmacattack is a truthful fellow, so why?

All are switchers, maybe the Solas have more noise on their outputs?

I think the OP will need to supply some more info if he doesn't get it fixed.

Sorry guys, wasn't able to get on here today, but I appreciate all the suggestions and speculation!

Grounding the DC common on the 2.5A Sola PS helped quite a bit, but I would still see a count of 61 or 62 about every 10th machine cycle-still not acceptable though.

I would have liked to really trace this issue down to a root cause, but I really needed to move on, so to that end I mounted the big 'ol AB 10a PS in a corner of the cabinet, ran my power off the terminals of the convenience outlet, and jumpered ground to DC com.

After that, I was able to work on some other projects while watching the machine run for 4+ hours with nary a missed or extra count.

Try the 2.5 Amp supply on the convenience outlet. If it works, then your other circuits are being pulled down somehow, or the convenience outlet is wired to some other source. It it doesn't work, then maybe you need a 15 Amp supply to make sure of 60 good counts.

Click to expand...

Tried that. An improvement over machine power sola, but still erratic.

Another thing to try is to raise the 24 volts supply output voltage to the maximum allowable by the connected equipment, maybe to 26 volts. Higher voltage with same current = more effective power delivered.

Click to expand...

Was actually able to get 28 vdc out of the Sola's, with no apparent effect.

I would check the differences in the specifications for the Sola 10 Amp supply and the Allen Bradley 10 Amp supply. Do any of the rating differences jump out? Is there any difference that would allow the AB unit to deliver more power to the Turck proximity switch?

Click to expand...

Didn't look TOO closely into this, but of course they both put out 240w. Might look at this tomorrow.

Are you using the complementary outputs of the prox to feed the 177a module? Shielded cable? With shield properly connected to the module per the User/Installation Guide? Power Supply noise could be an issue.

Click to expand...

Complementary outputs? Do you mean the N.O. and N.C. output pins? If so, yes the black wire (N.O. output) is wired into terminal 23 (Channel 3-A) of the module. Not using the white wire.
Not using shielded cable, but did run 14ga wire in it's own conduit to a short cable running from quick disconnect on back of prox. I realize that is overkill, but it is something I tried after difficulties were encountered. Originally, I was just using a 4-wire micro cable (4/22ga). Thought the long run on the small wire couldn't push enough juice to fire the card inputs at high speed. I also realize that this set-up doesn't shield the wires in the cabinet. Of course, even with shielded cable, I would have to strip the cable for a few feet to run from the module to the power supply...

In any case, shift guys have reported "noise" problems with this machine for awhile now. 1st scan bits latching while machine running, funky encoder counts, getting booted off-line, etc. So I know that there are definitely issues in this old panel. Not pretty in there. It's a 30-something year old machine with wireways full to bursting, all the ugly stuff you see inside a panel on a machine that has been evolving for that long.

Anyway, I appreciate you guys more than I can say. Thanks again!

Cheers,
Dustin

Megger yer moterz...

22awg wire would have been fine for a very long run, but I never use anything smaller than 18awg for sensors simply to improve mechanical strength and ease of termination.

In a pipe more than a foot from power conductors, you could benefit from them being twisted, and shielding should not hurt but it should have been okay unless trash is getting on your pipes.

I like to buy 100' long 18awg 4 or 5 pin 12mm DC Micro connectors...home run shots, shields for high speed stuff, shield drains at one point only, and you may even need to add a spike suppressor at the input point to kill induced noise from servo amps or VFDs.

A dual trace storage scope would be the ticket to nail this down...

If moving to an external 120vac source helps, stick a Corcom in the panel...

EDIT: Suppressing high speed inputs:
http://literature.rockwellautomation.com/idc/groups/literature/documents/rm/gmc-rm001_-en-p.pdf
see ch.9

Didn't look TOO closely into this, but of course they both put out 240w. Might look at this tomorrow.

Click to expand...

240 watts from one PS, being switched on/off every 1/4 second, might not be equivalent to 240 watts from another brand of power supply with the same rating. One PS might have switched-mode DC that is closer to a straight horizontal line, where the other supply output looks like a sawtooth or has a ripple riding on it.

I suppose your final result is proof that there WAS a difference in the supply outputs. Regardless of how small, the difference was large enough to not allow the prox to switch fast enough in two cases.