Strain Gauges and Heat

[juhop]

I have a problem with my strain gauge (load cell) measurements being affected by heat. I contacted the manufacturer of my strain gauge and they claimed that it is only affected 0.01% per degrees C. However, this is significantly less than what I am experiencing. Has anyone had similar problems before? Could the change in temperature (about+-2 degC) affect the wiring, which could lead to the errors?

 

[Compositepro]

Perhaps thermal expansion of some mechanical component in your system is actually changing the load on your load cell. A strain gage is not a load cell. A load cell contains strain gages.

 

[carnage1]

Thermal coefficient of resistance of copper per deg. C is .0039 However your wiring should be a negligible portion of the circuits total resistance. If you have another load cell available of the same type I would suggest a double blind test to verify your test setup.

Luck is a difficult thing to verify and therefore should be tested often. - Me

 

[IRstuff]

What exactly are you doing with your gauge, and how are you determining that it's in error with heat?

Is your +/- 2ºC value correct? That should be negligible in the face of your other system errors. And why are you calling that "heat?" Can you even confirm that the calibration was done with +/- 1ºC?

TTFN

FAQ731-376

 

[TheTick]

Strain gages are often wired in a "Wheatstone bridge" configuration. To cancel heat effects, add a strain gage to the opposite leg that the measuring gage is on, but don't attach the second gage to anything with echanical strain.

 

[IceStationZebra]

First, a wheatstone bridge will cancel heat affects as long as all of the gages are at the same temperature. If you do not have a full bridge, then YES wiring and temperature make a big difference.

Second, make sure the heat is not causing a non-axial loading on load cell.

Third, check your wiring. You might be reading the resistance of a large chunk of solder in a connector. Or you could have a loose connection or water.

Fourth, if this is a purchased load cell what exactly was the specification? You have to be very careful about the wording. There should be a specification for temperature affect on span and on zero. There should also be a specified operating temperature range that these values are valid.

If you made your own load cell then be aware that the 0.01%/°C value is only valid for a small temperature range. For Vishay gages they have a good paper from Vishay.

http://www.vishay.com/docs/11054/tn504.pdf

I would think that other manufacturers would be similar. (FYI - Omage gages are made by Kyowa)

ISZ

 

[rob768]

The strain gauges may be ok, but are you sure your adhesive has completely cured? Perhaps the heat is affecting the bonding.

 

[juhop]

Thank you for all of your answers.
The strain gauge is used to measure torque. One of my problems is that the strain gauge was attached to the component by a third party company. I asked them about the possible effects of heat to the assembly and their reply was: You can expect a zero shift of up to 0,01% of the full scale per C°, and a span (gain) shift of up to 0,02% per C°. The cell is rated at 50Nm, which still doesn't explain the deviation of about 0.15Nm due to a change of only about 2 degC.
The part is cooled by water;however, as mentioned before, the water temperature varies about 1-2 degC. The part is made from aluminum, which should not be affected very much by thermal expansion.
I will double check connections, since I have not soldered them myself.

 

[rb1957]

one or two degrees really isn't going to matter significantly, 100s would.

it sounds like you're checking your load-cell, and calculating a difference (via your strain gauges) of 0.3%. personally i'd be happy with that correlation. how arrurately were the s/gauges located on the part (a small difference in arm could account for the difference. personally, i'd use the result to re-calc my moment arm (assume the loadcell is good, and delivering 50Nm and your calc needs to be adjusted down by 0.997 to match the loadcell).

 

[IceStationZebra]

I got thinking about Rob768's answer. If one of the gages is not bonded fully this may show up as a temperature drift because 3.5 of the gages will stretch with the temperature expansion, but the last 1/2 will not. This would be evident by the CW and CCW output not being the same for a given applied torque. I would do a quick check with a pipe wrench to see if the output is about right and symmetrical.

And aluminum is not affected very much by thermal expansion compared to what??? Aluminum expands 70% more than steel.

ISZ