S- and P- accidentally swapped on Load Cell. Can my testing data be converted/recovered?

Thread Starter

prime_number

Joined Apr 7, 2016
12
Hi community,
First time poster, long time viewer.

Is there a way to convert the data of a load cell that had S- and P- wires swapped to reflect accurate data?

I recently finished with a fatigue test on a steel beam where I was sampling the data of multiple sensors using a Keysight multi-measuring unit.
One of the sensors, load cell in this case, had been installed incorrectly. The singal- and power- wires had been swapped and was not noticed until the fatigue test was finished. There is not a way to recreate the fatigue test, but I did notice that the data values of the load cell appear to be scaled by 0.5 of values I received on a previous similar test and expected values. When the test was performing, someone did notice that the load cell appeared to be half of what was expected/calculated. I scaled the data by a factor of 2 and the data looks pretty good, but I do not know how to confirm that it is accurate. Can someone confirm this or suggest a more accurate way of converting the data?
The load cell is an Interface Standard Precision Load Cell.
 

KL7AJ

Joined Nov 4, 2008
2,225
Hi community,
First time poster, long time viewer.

Is there a way to convert the data of a load cell that had S- and P- wires swapped to reflect accurate data?

I recently finished with a fatigue test on a steel beam where I was sampling the data of multiple sensors using a Keysight multi-measuring unit.
One of the sensors, load cell in this case, had been installed incorrectly. The singal- and power- wires had been swapped and was not noticed until the fatigue test was finished. There is not a way to recreate the fatigue test, but I did notice that the data values of the load cell appear to be scaled by 0.5 of values I received on a previous similar test and expected values. When the test was performing, someone did notice that the load cell appeared to be half of what was expected/calculated. I scaled the data by a factor of 2 and the data looks pretty good, but I do not know how to confirm that it is accurate. Can someone confirm this or suggest a more accurate way of converting the data?
The load cell is an Interface Standard Precision Load Cell.
 

Thread Starter

prime_number

Joined Apr 7, 2016
12
I do have some (very limited) access to the load cell currently. It is still installed in the setup (about 15 ft in the air) between an actuator and a steel column. In order for me to accurately test it, I would need to remove it and test it in universal testing machine in series with a master load cell to see the relationship.
This is all possible in the future, but I am hoping to avoid all of this if there is, in fact, a mathematical relationship like I am seeing (i.e. factor of 2).
I am hoping to find confirmation about whether this will convert the data to be truly accurate or seemingly accurate.
 

Alec_t

Joined Sep 17, 2013
11,505
I am hoping to avoid all of this if there is, in fact, a mathematical relationship like I am seeing (i.e. factor of 2)
There is a mathematical relationship, but without knowing the internal resistances of the components making up the load cell I don't see how you would determine what the relationship is. Below is a sim showing how an approximate 2:1 ratio of the outputs could be obtained by a mistaken '90° rotation' of a 1%-unbalanced fictional cell. That ratio is highly dependent on the R2/R4 (or R7/R5) ratio.
LoadCell.PNG
 

ericgibbs

Joined Jan 29, 2010
10,463
hi prime,
Looking at the d/s It appears to be a 'symmetrical' load cell, ie: the measured resistance of the +Vx and -Vx will be the same as the +Vs and -Vs connections.
Use a Ohmmeter to measure across +Vx and -Vx and also +Vs and -Vs , check if they are the same resistance.

The only 50% signal reduction you are considering could be due the different location of one pair of the strain gauges on the load bearing beam.
Which in my experience would be unusual.

I would recommend that you contact the L/C manufacturer, they should be able to give an answer.

E
 

Thread Starter

prime_number

Joined Apr 7, 2016
12
ericgibbs,
I have tried calling several times, but have not had any luck hearing back from their technical person. I'll try the link you posted and update the thread if I get a response/solution.
 

Thread Starter

prime_number

Joined Apr 7, 2016
12
UPDATE:
I was able to speak with someone and they informed me that the data can not be converted to reflect accurate, calibrated data.

The reason, according to the manufacturer, is that the bridge was not being equally powered and that it would effect linearity. He further described the internal columns, of a load cell, running symmetrically from the center and how if some of the columns were not reading, it would effect linearity.

Does this seem correct to you guys?
 

ericgibbs

Joined Jan 29, 2010
10,463
hi prime,
Their analysis is IMO correct.
Your readings showing an apparent reduction in signal of some 50%, confirms what they say.

How critical is the load test for the project.?

Eric
 

Thread Starter

prime_number

Joined Apr 7, 2016
12
The cost of the fatigue test was several thousand dollars and lots of time, so I am hoping to help salvage a grad student's research project.

The guy that I spoke with, from the manufacturer, asked about what accuracy/error is acceptable before asking their electrical engineer for a solution. I mentioned that we would be interested in data that is as accurate as possible, and I wonder if that demand is what brought the response of "no solution". Is it possible that this data could reflect somewhat accurate data with an assumed error percentage?
 

ericgibbs

Joined Jan 29, 2010
10,463
hi,
Do you still have that particular L/C available for bench load testing.?

If yes, I would suggest that you electrically connect it exactly as it was in the original test.
Apply test loads to the L/C, in steps of say 10% of FSD upto to the full rated loading of the L/C, in other words carry out a reverse connected load cell calibration.
Draw a graph using this data, knowing the manufacturers published data it should be possible to produce a correction table for the original load data.

E
 
Top