Not sure where to start. 1

[willhan]

So, we've installed this fancy new system with a dual motor driven chain, and the damned thing has had problems since day one. It slows down from time to time and interrupts production. We have considered a number of options to cure the problem, for example, we just changed the motors' encoders at the engineer's recommendation, and the problem is still occurring. We are not sure if the problem is one of controller programming, bad wiring, mechanical, or something else.

I have included some shots that may help to illustrate the issue further. There is an 8 hour time series of both of the motor's currents. When the chain is not loaded, you can see an oscillation with period of about 40 minutes, the time it takes for one complete chain cycle. When production starts, the current begins to rise, and peaks after about half an hour or so. Then the chain randomly jerks, slows down, and sometimes even fully stops then returns to it's normal operation.

I appreciate anyone who might be able to point me in the right direction. What other information should I provide to help mitigate the problem?

 

[willhan]

So, I guess I can only submit one file at a time. Here is the motor speed for an 8 hour period in the morning.

http://files.engineering.com/getfil...ccf-89ed-d7af8fbb7573&file=Pump_Speed_8hr.png

Here is a zoomed in version of one of the "spikes" on current:

http://files.engineering.com/getfil...-4a7e-b66a-c7ddfa146168&file=Pump_Current.png

And, here is a zoom in on one of the spikes affecting motor RPM:

http://files.engineering.com/getfil...a1-4aee-ae80-bf8b02686789&file=Pump_Speed.png

 

[willhan]

Oh, the files are called "Pump...", these are not pump motors, they drive a chain.

 

[ScottyUK]

A bit more explanation of how the load behaves and/or what the process does might go a long way to helping understand what the waveforms are indicating. A few sketches, a process flow diagram, a handful of photographs... remember that none of us have ever seen this machine before. We're engineers, not telepaths. ;-)

 

[2dye4]

It's tricky to drive one mechanical system with two motors, especially if the two motors have feedback loops
making their output speed stiff.

One quick though without knowing much about the system is make one motor master of the correct speed of the chain and
the other motor is controlled to provide proportional torque only to aid the master.

 

[willhan]

So, I've drawn a little diagram of the system (which is why i get earn big money). The process is like this: The machine is a miniaturized blast freezer. Hooks hang from the chain every foot. The chain itself is over 800 feet long, and it can hold some 800 hog carcasses at a time. I'll assume that a dressed carcass weighs 150lbs. So, the motors are pulling 60 tons of meat. The motors are configured in a "Virtual Master - Dual Slave" configuration. They are both 25HP, with something like 900:1 reduction.

An indexer just before the entrance to the cooler catches carcasses, and they ride the chain for 40 mins before exiting and being unloaded.

We changed the Encoders last night, bc we were only getting a 3.5V feedback signal, and 5V is the minimum. The number of "spikes" has halved today, but the problem is still significant.

 

[IRstuff]

So, possibly noise in the system. Possibly the signal lines are noisy or the power.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

Of course I can. I can do anything. I can do absolutely anything. I'm an expert!

 

[waross]

This may be chain problems. From the sketch it appears that the chain returns outside the cold room. If a section of the chain became wet during construction, it will freeze and the resistance to bending will increase as more chain enters the cold room and more joints freeze up. Some joints that may be wetter than others may be frozen so hard that they put a heavy load on the motors as the chain wraps around a sprocket and the frost must be broken before the chain can bend around the sprocket. I realize that lubrication is problematic in a food handling area, especially at blast freezer temperatures.
See if you can see the take-ups moving as the motors load up.
Check the chain for damp sections.
I have seen serious problems (shafts bending and teeth spitting out of exposed gears) in a blast freezer. This was a metal mesh conveyor which slid on plastic covered rails. As the blast room was cooled down, it went through a temperature zone where quantities of hoar frost formed on the sliders and was scraped off by the conveyor mesh. It turned out that during the period of hoar frost formation, the co-efficient of friction of the sliding conveyor increased many times normal. Although the open gear drives were over sized for the friction at normal operating temperatures, the gears would self destruct when the hoar frost formed due to the increased drag of scraping many feet of hoar frost off of the support rails.
If the problem is worse as the temperature drops below freezing and frost forms when the blast freezer is first started, it may be well to look for a hoar frost issue.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[Compositepro]

If I understand you sketch correctly, the length of chain at the bottom is outside your chiller and is where loading and unloading occur.
It appears that you are running your chain backwards. Chain tensioners should always be at the lowest tension point in a chain loop. This is on the out-going side of a drive sprocket. I would guess that your problem is caused by the chain jumping a tooth on the drive sprockets. A chain as long as yours will have a lot of stretch in it. Going as slowly as it does the tooth jump is probably fairly slow and quiet.

 

[ScottyUK]

Have you checked the obvious cause for low voltage from the encoders - Ohm's Law? Put a scope on the encoder supply at the encoder terminals. Is it as expected and stable? If that is OK, is the encoder output signal reaching full amplitude at the encoder terminals?

I had a problem exhibiting similar symptoms from a different type of transducer a few years ago and the problem turned out to be power related: the average supply current was stated in the data sheet and we'd designed around that, but the actual current was a series of pulses and the cable resistance was too high to allow those pulses to be drawn without significant volt-drop. A decent-sized capacitor local to the transducer sorted that out, but the problem had us puzzled for a while.

 

[willhan]

Thanks for the responses. I've made some notes and we'll look at these things. @waross, is there any way to mitigate the problem of hoar frost?

 

[LionelHutz]

I'm a little confused on how you have 2 encoders feeding a virtual master with the motor drives as slaves. Can you provide some info on the control system, ie what control loops? Draw a process diagram?

It could be jamming. The red curve jumps right away when the chain catches because that drive is pulling a chain section with little give. The green current curve builds as that drive first pulls the slack out of the spring loaded take-up before it hard jambs.

 

[willhan]

Should we have only 1 encoder feeding the virtual master? I'm not familiar with the virtual master setup.

 

[jraef]

I've never heard of using a "virtual master" in a permanent control loop, that term is used, in my experience, to TEST a master/slave control loop by making either of the drives a master in order to test or prove the relationship and/or scaling functions. Might it be that someone set it up this way for commissioning and forgot to return it to normal?

My only other thought is that there is some sort of higher lever controller that is deciding which is the master and which is the slave, maybe by looking at both encoders to determine error as the chain stretches and expands in different temperature zones. If so, and the chain jumped a tooth, I can see that causing havoc in the control loop. The general slowdown might be the system changing back and forth between masters attempting to correct an error that was not accounted for in the design. Those blips of current dropping off looks to me as well to be evidence of that; as the chain slowly slips over the sprocket there is a momentary loss of load, which drops the motor current briefly.


"You measure the size of the accomplishment by the obstacles you had to overcome to reach your goals" -- Booker T. Washington

 

[willhan]

The current is spiking when the slowdown occurs. My supervisors and mechanics don't think that the sprocket is slipping. We believe that it would be completely catastrophic if that were to happen.

We are getting normal square waves on the A,A*, B, and B* lines on both drives, but the Z and Z* look strange. The documentation says that they should be square waves with frequency 1/3 of A and B, but we are seeing a strange form that looks a lot like noise.

 

[willhan]

Oh, I see the abrupt drop in current after it spikes up. Here is a screen capture from today before lunch. Red is the infeed motor, and green is the outfeed.

 

[willhan]

The drives are configured as Speed Trim Followers. The master sends its torque signal to the followers who develop a "trim" error by comparing the master torque to their own.

 

[LionelHutz]

I'm not understanding the need for the encoders. Where are the encoders connected?

Most newer drives have a very good sensorless vector control algorithm. I can't see this application requiring closed loop vector control for these motors and I can't understand what else they would be doing.

You should be able to program a drive to output a 4-20mA speed signal. It might be telling to have both VFD speed and current on the same plot.

The applications I've seen basically have a master VFD. You tell the master VFD to run at a certain speed. A torque or motor load output is sent from the master to the slave VFD. The slave VFD is set up to run up to a speed where it is producing the same amount of torque or the same motor load.

Can you not involve the manufacturer or integrator?

 

[willhan]

Here's a .pdf with more information.

http://literature.rockwellautomation.com/idc/groups/literature/documents/wp/drives-wp001_-en-p.pdf

 

[waross]

Mitigation of frost issues; Our solution was to re-engineer the drive gears to withstand the load of scraping frost.

The current is spiking when the slowdown occurs. My supervisors and mechanics don't think that the sprocket is slipping. We believe that it would be completely catastrophic if that were to happen.

This may be pointing to a mechanical problem. However it may be a mechanical problem cause by a control issue.

We are getting normal square waves on the A,A*, B, and B* lines on both drives, but the Z and Z* look strange. The documentation says that they should be square waves with frequency 1/3 of A and B, but we are seeing a strange form that looks a lot like noise.

Is this the output from the encoders? If so, have you bench tested the encoders? What type of wiring is used between the drives and the encoders? Length, wire gauge? Shielding? Possible voltage drop in the conductors?
Do the carcass supports run on rollers? are all the rollers turning or are some frozen?
Does the chain tend to bunch up on the output of the drive sprockets? Is there excess movement of the take-up rolls?
Consider: Is the chain speed actually slowing down or is a lost or faulty encoder signal indicating a stoppage or slowdown when the chain is actually running normally? The speed and current spikes may be indicative of an actual slow down, or they may be the drive's reaction to a faulty signal showing that the drive has lost speed abruptly.
Unforeseen occurrences; Is the portion of the chain subject to occasional wash down water which may be causing a problem as the wet chain transits the cold room? This is one example but the list of the unforeseen is endless.
If you have bunching or excess take-up roll movement, you may need more tension on the take-up rolls.
If the encoders give good signals on the bench it may be well to inspect the field wiring to the encoders.
A mechanical problem causing electrical upsets or an electrical problem causing mechanical upsets, that is the question.

Bill
--------------------
"Why not the best?"
Jimmy Carter