Torque sensor drift during continuous measurement and data logging 1

[marktheyeti]

Hello,

I'm dealing with what appears to be a torque sensor malfunction that I have not seen before. Using a rotary torque sensor to measure applied load to a rotating shaft (180 rpm), I have recorded a very consistent, gradual increase in the measured value over a period of several days. I know, due to other measurements in the system, that this is not a changing condition of the applied torque, but rather a malfunction in the rotary torque sensor. I'm hoping that someone else here has experienced this type of sensor failure or knows what might be causing this consistent signal drift. The measured equipment is known to be in steady-state condition, and measurements at other points in the system indicate that the actual applied loading cannot be changing in the way the torque sensor indicates.
This is a wheatstone bridge-type rotary torque sensor with four brushes contacting slip rings to transmit the signal from the rotating shaft to the non-rotating housing and signal wire connector. The slip rings and brushes appear to be in good shape, with no visible damage and a significant amount of brush life left.
Any help of advice is appreciated.

Regards,
Mark

 

[IRstuff]

Have you asked the manufacturer?

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[marktheyeti]

The original manufacturer, BLH Electronics, appears to be out of business. My next step is to contact an alternative manufacturer and ask for an evaluation.

 

[Compositepro]

I would suspect the slip rings. They can easily change resistance slightly with age or operating conditions. Mercury wetted slip rings ought to be used.

http://www.mercotac.com/

However you need to be systematic in troubleshooting. Measure voltage at various points with another instrument to find which component is drifting.

 

[MintJulep]

Has the sensor been unused for a long time?

 

[3DDave]

I'd look at the usual causes for strain gauge drift; uncompensated temperature or self-induced temperature effects showed up first in Google search. If there's more to the history and operating conditions, that might make diagnosis more precise.

 

[marktheyeti]

CompositePro, I will try to evaluate individual components as best I can. I will review your information on slip-rings, thank you.

MintJulep, yes the sensor was unused for an extended period.

3DDave, I did first consider temperature effects, but that seems unlikely because the sensor almost certainly would have achieved steady-state conditions after several days operating under the same conditions.

 

[3DDave]

Thanks - I was looking for feedback on how little my advice was worth. It helps to know that you already considered the case, and then didn't mention that it was considered.

 

[marktheyeti]

3DDave, Your suggestion was appreciated and well-received. My response was not meant to dismiss it but to continue the conversation by providing more information relevant to your suggestion. If I've missed the point here, please explain. Thanks.

 

[MintJulep]

Potentially the slip rings oxidized during the period of non-use and the oxide film is now being worn away, slowly decreasing the resistance of the measuring leg.

 

[Dougt115]

How can you be sure the torque sensor is the issue and not something else? The equipment is known to be in steady state based on another monitoring system. Could that system be in error? There are many more components that could be giving you the incorrect data than the torque sensor.

If possible change out the sensor for another sensor and see if the results repeat. If they do I would suspect the motor or monitoring system. It may just be a characteristic of your system.

 

[marktheyeti]

MintJulep, Thank you, that makes a great deal of sense. I will see what I can do to test this theory, and then I'll post my results.

Doug, The output sensor is measuring torque applied to the output shaft of a worm gear speed reducer. I am also monitoring the input shaft torque (and motor power consumption) and have seen no change over a period of days. Temperature of the reducer is also being monitored and there has been no change over the same time period. If there was a change in the applied load or the efficiency of the reducer, this would have been reflected in the other system measurements, not the output torque sensor only.
I did review all of the wiring connections from sensor, to amplifier, to PCI board, to PC, just to be sure a gradually loosening wire wasn't at play. It is possible that one of these other components could be causing the issue. I will do what I can to examine their performance. Thanks for the suggestion.
I do not have another sensor with the required load rating to try in place of the current sensor.

 

[MikeHalloran]

Strain gages drift when used near the limit of their rating.
I.e., the ratings are a little optimistic.

Note also that when an individual gage is overstrained, the Wheatstone bridge will tend to hide that fact. Have you rechecked the alignment between the output shaft and whatever is providing the braking torque?




Mike Halloran
Pembroke Pines, FL, USA

 

[BrianE22]

Many strain gauge transducers are sensitive to the temperature gradient across (from one mounting face to the opposite mounting face) the device. Has anything changed that would affect the gradient?

 

[marktheyeti]

Mike, The sensor is not near it's limit. It's a 30,000 lb-in sensor measuring a load of less than 5,000 lb-in. The alignment between the shaft being measured and the shaft of the sensor is not perfect, but I've employed a gear coupling between the two to transmit the load. The magnitude of misalignment falls within the capability of the coupling.

Brian, It's possible (likely) that there is a temperature gradient from one end of the sensor to the other, but this condition would have been brought to steady state early in the test. The drifting phenomenon was observed continuously throughout the test, even when steady-state conditions surely would have been reached (days, rather than hours).

I sent the sensor for evaluation last week. Their technician noted that this type of drift is observed when the strain gage reaches the end of its useful life. They're running a bench diagnostic to confirm this. I'll post feedback when I have it.

 

[Dougt115]

It's great that you have two data sources.

You could try testing a second unit to see if the readings drift in the same manner. If they do then I would focus on the torque sensor.

My second area of investigation is the wheatstone bridge itself and you are already having that checked. My off handed thought was if one of the resistors is drifting for some reason.

Good luck.

 

[Compositepro]

"End of life of a strain gage" generally means that the adhesive bonding the strain gage to the load cell is degrading.

 

[marktheyeti]

Feedback from analysis:
Static bench testing did not show this drift.
Electrical evaluation did not clearly identify a source of the problem.
It "sounded rough" to the technician - it made noise while rotating. This I knew.

Not all that helpful. I've started soliciting quotes for replacement.