PDA

View Full Version : whaddafuh?


TimothyMoulder
June 6th, 2006, 02:11 PM
Two weeks ago, we had a problem with a particular servo losing position. We replaced the encoder cable. No more problem.

Two weeks later, same problem. Okay, must not have been the cable. Observe symptoms, decide drive is likely culprit. Order new drive.

New drive has a two-day wait, so, just to rule it out, change motor and encoder. Same symptoms. Wait for drive.

Install drive, start up - same symptoms.

At this point we've replaced the entire servo apparatus. Then one of the maintenance men has an inspiration - put the old encoder cable back on.

No more problem. kaaaaaaay...

This makes no sense. The new cable ran for two weeks before problem resurfaced. And after replacing everything else in the system, the problem is corrected - by using the old cable.

I'm a technician - I don't believe in superstition, and I gave up sacrificing live chickens on the worksite years ago.

But I have, on many occasions over the years, been through an ordeal like this one, and after putting in enough hours on the machine, something just gels and the damn thing starts working like nothing was ever wrong. Like it just wanted attention.

Yeesh,

TM

BTalbot
June 6th, 2006, 02:30 PM
Intermittent connection @ drive on encoder connector, old cable was good but had come loose, replaced with new cable (tight connection, worked for 2 weeks, connector came loose, re-installed old cable tight connection, all is well. ( this is just a W.A.G.)

Leadfoot
June 6th, 2006, 02:45 PM
We had a customer constantly "losing" encoders on a shear. During one service call on another problem, one tech found the plug loose, and the +v pin burnt. New connector, new encoder, tighten the encoder plug, keep the plug tight - no more encoder problems.

Vibration sure seems to make connectors get loose.

Thomas Sullens
June 6th, 2006, 03:02 PM
Know the feeling! We had a servo fault on one of the torche tables, You could wiggle the encorder cable and it might work for a while. So we changed the cable going to the encoder, it worked a while same fault, so we changed the gear on the encoder and new it wasn`t the problem because we had this problem with a gear completely gone and it fixed it, but it went back to work about a week same fault, took that gear off put an other new gear on fixed the problem again??? In about a week same problem this time they checked the oil in the gearbox and it was low. So they took the gearbox off cleaned it and changed the oil guess what it has run for about 6 months with no faults our fingers are crossed. Did i mention we also swapped encloders twice and this also each time fixed the problem for a day are so! What was funny you could make a dry run of the part to be cut and it wouldn`t fault only when actaully in the cut would it fault??? OH WELL!That`s what makes it fun???

TWControls
June 6th, 2006, 03:50 PM
We had a customer constantly "losing" encoders on a shear. During one service call on another problem, one tech found the plug loose, and the +v pin burnt. New connector, new encoder, tighten the encoder plug, keep the plug tight - no more encoder problems.

Vibration sure seems to make connectors get loose.
Wrap just a little bit of teflon tape around the threads. Don't let the teflon tape get over the end of the connector when the cable plugs in (for grounding purposes on some).

Problem solved

ward
June 6th, 2006, 04:50 PM
I've had a similar situation occur where everything on an axis had been changed and the problem kept coming back; in that case it was a noise issue where in the process of changing one component or other the resolver wires were moved away from the source of the noise and the problem went away (must have been the replaced component). Then when everything was back running again they would wind up back in the cable tray and subjected to noise again and the problem would be back.

plchacker
June 6th, 2006, 05:41 PM
Encoder/Canon connectors:

I made it a practice that on any hydraulic powered equipment, or any equipment with lots of vibration to use a sleeve of heat shrink over the connector once tight. (heat gun, not torch please) This has several advantages.

Hydraulic servos/heat:

Hydraulic servo valves do strange things if the fluid reaches high temperatures. Symptems that will make you believe you have encoder problems. I have seen countless encoders/cables replaced due to heat trouble. After spending the time to replace the encoder, the machine cools down and runs fine on start up. Bubba says, "It has to be the encoder,'cause thats what fixed tha trouble."

After months and many encoders, I finaly convinced people to put a shell/tube heat exchanger to keep the fluid at the correct temp.
Years later, same encoder/cable. No trouble.

keithkyll
June 6th, 2006, 08:03 PM
I'm surprised that encoders are so troublesome. The approach in the Industrial Control world is to troubleshoot to the modular level. Swap out suspected units until it's fixed. In this case, that appears to be very expensive and unreliable. Wait 2 weeks to see if it's fixed now? A pile of parts that are now in unknown condition. Send them off for testing and repair.
Even if you know how to troubleshoot electronics to the component level, they don't give you any schematics or information, so fixing it is difficult if not impossible.

What I see here is a problem with connectors. All metals form an oxide when exposed to Oxygen. Moisture (water) is a corrosive that also causes problems. Highly resistive surfaces appear everywhere, except at the mating point of the contacts. Wiggle it a little, and now you're on a new, oxidized surface. Remove and replace the connector, and the 'wiping action' cleans the surface. Well, a little anyway. 2 weeks later, it's bad again.
First, gold plated pins are superior to tin/nickel. Next, keep the metals parts away from Oxygen or water. One way is to grease the connectors. Grease forms a protective coating. This method is used on cars for battery posts and lamp sockets. The most common approach in Industrial is to seal the connector.

For troubleshooting, you need an oscilloscope. Encoder signals are slow enough where a low cost or older 'scope will do fine.
The card looking at the Encoder signal is tolerant to some noise. The fault occurs when the noise gets to be too much for it to handle. With a 'scope, you can see bad signals long before they get bad enough to cause a fault.
You don't need to get deep into timing. The quality of the signal is the important part. Wiggle the cable and watch for spikes. High resistance (oxidized) connectors will show up as a low voltage signal. Moving the cable will cause it to vary in amplitude.
You can do this on optical encoders with the motor stopped as long as you're looking at it while it's a 'high'. Turn the motor a bit if it's low. For magnetic encoders, you'll need to do this with the motor running.

Peter Nachtwey
June 6th, 2006, 09:20 PM
I tell my customers to use differential signals. Also, they should use low capacitance cable. I would have to ask my hardware guys exactly how low is low enough but the cable that Temposonics recommends for its MDTs is the place where I would start.

So, does the encoder use differential signals? Is the cable low capacitance cable.

One more thing. The decoder part should be able to phase errors. The decoder should be able to detect if both the A and B phase change at the same time.

I hate these kinds of problems.

TimothyMoulder
June 6th, 2006, 10:15 PM
Peter, if they vex you, I know I'm in expert company...

Since the consensus is connectors - one thing I'd not thought to check - I'll do that tomorrow!

Thanks!

TM

rsdoran
June 6th, 2006, 11:47 PM
When it comes to many situations that are addressed on this site I am not the most proficient person to answer but will enter some questionable discussions.

When it comes to troubleshooting.....

The method described in the original, and most preceeding posts, is called "PART SWAPPING." It is the least most sucessful method of discovering a problem.

In the past I have had problems with keithkyll's answers...I have determined that is because I am jealous, he is very very smart and experienced.

In this situation, actually many, keithkyll has offered the method most appropriate. You should have used an oscope, or whatever "tools" are needed to isolate the problem...NOT assume and just swap parts.

In this area I do not care if you are an Engineer with what you percieve as "KNOWLEDGE" or not. It is not about understanding theory, math, etc, it is about knowing what should happen and how to determine what is happening with the ability and capability of using the correct tool.

ADDENDUM: In many cases where encoders and tachs are concerned the wiring used is very small i.e. 18 gauge or smaller therefore if the encoders moves it is very easy for the wire to develop breaks and work "sometimes". It should be simple to know what signals should be there and what is not there in a failed condition.

IN many, if not most situations, if a drive is involved it can offered the information relevant to the problem.

keithkyll
June 7th, 2006, 02:38 AM
Extra info to my previous post:
If you don't have a scope, a meter with a bargraph will get you by. The bargraph is designed to respond quickly.
When you're checking cables, remember that zero tolerance is the rule. You should be able to wiggle cables and connectors with absolutely no deviation in the scope image or bargraph.

Ron, my initial reaction was similar to yours. In the Semiconductor Industry, assemblies that cost thousands are replaced like light bulbs. Downtime is more costly than the parts. That's the approach in the PLC world too. In most cases, it's the rule. But, what happens when swapping doesn't fix it? Do PLC schools teach the use of an O'scope? With the problems others have described, it seems they didn't know that a scope would give a 'picture' of what's happening.
TM didn't consider connectors, so, for him, this thread is probably answered now. He might be able to solve the problem with visual inspection alone. I take the stand that an intermittent isn't fixed until you can wiggle the cable with no indication on a scope or meter.

Regarding knowledge, everyone has their 'gift'. We all contribute together as team and learn from each other. I've learned from you too.
A team of experienced contributors from various disciplines is what makes this forum so powerful.

Andybr
June 7th, 2006, 03:14 AM
For most general troubleshooting tasks such as the one described above I find a basic PC scope package such as those made by Picoscope is more than adequate and costs very little. I would agree with the other subscribers who say that the practice of swapping parts until the fault goes away is almost the worst possible approach to fault finding. You cannot possibly say that a fault has been rectified if you do not know what the fault was in the first place.
Andybr

Leadfoot
June 7th, 2006, 08:35 AM
I always break out the scope when trouble shooting encoders, right after I check all of the connections. Unfortunately many customers do not have scopes, or the knowledge or willingness to use them. I like the shrink wrap over the connector idea. I have taped them upon occasion. When I was in the NAVY, the cannon connectors had holes for the safety wire to keep them tight.

Ron,

Swapping parts until it is fixed is called the "SHOTGUN" method at my company. It is a great method with older PCB's to "shotgun" the electrolytic caps and is a requirement on bench repairs.

Sliver
June 7th, 2006, 11:14 AM
There is always pressure at our plant to replace something, anything.
Here's how the conversation goes.

So what was the problem?

Not quite sure. Checked x,y,z, Nothing definate.

So what am I supposed to tell production, we didn't do anything?

No, tell them we did some diagnostics and set up a trend etc...

Well just change out the most likely component so I can get them off our back.

You're the boss.



Brian.

russrmartin
June 7th, 2006, 11:55 AM
I'm pretty limited with encoder troubleshooting. We have them all over the place here, but I've never seen a problem with one. Can someone elaborate exactly how you would use the oscilloscope to troubleshoot this? Particularly, what leads out of the encoder would you check, and shouldn't the encoder be turned to watch the pulse? I guess my main question is, which leads do you look at, and what do you expect to see when the encoder is not moving, further, if you are looking at a steady signal, why will a good VOM not give you the same information? Thanks.

Russ

TimothyMoulder
June 7th, 2006, 12:03 PM
Depending on how much functionality you have from the device, you can check:

A & B - Pulses present, phasing (when going one way, A pulse starts before B Pulse, going the other, B before A)

Z (index) pulse - occurrs once per revolution. May be used - may not be used. Depends on set up.

voltages at correct levels, square waves, clean waveforms (low noise).

All of this is only relevant if you can get the thing to move.

I recently learned about a command available on some controllers, in Trio it's the DAC command. It allows you to output a voltage to the drive's command signal, irrespective of the feedback. This is a spiffy way of testing the above, since it is immune to errors relating to the encoder which stop normal positioning cold.

TM

Peter Nachtwey
June 7th, 2006, 04:54 PM
Depending on how much functionality you have from the device, you can check:

A & B - Pulses present, phasing (when going one way, A pulse starts before B Pulse, going the other, B before A)

Z (index) pulse - occurrs once per revolution. May be used - may not be used. Depends on set up.

voltages at correct levels, square waves, clean waveforms (low noise).

All of this is only relevant if you can get the thing to move.

I often disconnect things and test the devices separately. In a recent tech support call our controller was blamed because a valve didn't work. The mechanical guys would not even trust the open loop or DAC command you suggest below. Fortunately the electricians had a mulimeter that was also a current source. They could connect this directly to the valve without the motion controller and test the valve. The multimeter could also check the output of the motion controller when different open loop or DAC commands are issued.


I recently learned about a command available on some controllers, in Trio it's the DAC command. It allows you to output a voltage to the drive's command signal, irrespective of the feedback. This is a spiffy way of testing the above, since it is immune to errors relating to the encoder which stop normal positioning cold.
TM
Yes, in fact we have two forms of DAC or open loop commands. One that cares about errors and another that doesn't. The one that doesn't is used for diagnostic work. The version that cares about error is used in operational mode. Believe it or not there is still a lot of open loop control out there.

Another thing we have is graphs that can plot what is happening like a AB trend but ours is faster with higher resolution. We also have an event or error log that can record events or error as the millisecond level.

TM, You have too many threads going.

TimothyMoulder
June 7th, 2006, 05:00 PM
I wanted to avoid hijacking myself again :P

TM

plchacker
June 7th, 2006, 07:52 PM
At the last mill I worked in the first thing I did was get a scope. Others couldn't understand my insistance on having a good O-scope. :(

I teach my students how to use them in a troubleshooting class.

I remember the holes for safety wires too. Uncle does do a few things right.

keithkyll
June 7th, 2006, 09:29 PM
... if you are looking at a steady signal, why will a good VOM not give you the same information? Thanks.
Russ The main issue here is intermittent connections. Sensors are running at 24DC or 120AC, with a pretty broad threshold. A quick dropout of a few milliseconds might go unnoticed.
Encoder signals are lower voltage and the slightest intermittent will show up as a missed pulse, so poor connections in connectors will cause problems here first.

What's the best way to diagnose an intermittent, and prove you fixed it? First, you need to 'see' it. Wiggling a cable with a bad connection will produce 'glitches'. An osciliscope is the best tool to see a glitch. A meter with a Bargraph option will also show it.
A digital voltmeter uses an A/D converter. The processing takes time. It's hard to see a glitch with a digital. More expensive voltmeters have a Bargraph to overcome this limitation. There's no need for accuracy with a bargraph, so a fast A/D circuit is used for this part of the display.

One poster said his problem occurred when the cable was put back into the cable tray. Extra signals were coupled into the cable. A scope would clearly show the extra signal riding on the main one.
Todays scopes have a storage feature. You can save a trace to floppy or USB. Read and store your signals when they're good, so you can compare when a problem is suspected.

There's a large variety of Encoders, and I'm not familiar with what's typical. I can't tell you which pins to connect to.
This part isn't that critical anyway. Trying to look at phase or timing isn't necessary. The controller will tell you hard faults and phase errors.
The point of this discussion is to use a scope to make sure the quality of the signals aren't marginal.