GearBox Test Bench Calibration


The Quality Kid
Hi there!

I have been put in charge of a project regarding the calibration process improvement of a Test Bench installation for GearBoxes we have in our plant. This installation has different kinds of measuring devices such as (Acelerometers, Temperature Sensors, Microphones, Pressure transducers, Acquisition Boards...)

Currently, the calibration is done individually on each of the devices in our lab or externally. So for example for the pressure transducers we take the devices to our lab, and we use a pressure calibrator with an integrated multimeter to calibrate it).

Now, the company is seeking to do the calibration of the devices on the installation, so for the pressure transducers instead of taking the measures through a multimeter, the measured value is obtained through the PC where the operator checks the values on serial production. I am not sure if this method is acceptable as there are many uncertainty inputs which I think we are not considering (the cable from the transducer to the acquisition board, the acquisition board, the software which transforms the value in mA to bar...). Is there some kind of norm / standard related to this, if I am honest I am completely lost...

Also, the company also wants to create a unique calibration certificate for the whole installation if the previous step is feasible, but I do not know neither if this can be achieved... also, I am afraid that in this scenario any change/repairmen on any of the devices will mean to re-calibrate the whole installation.

Any kind of help/guidance will be highly appreciate it!

Kind regards / Saludos.


Trusted Information Resource
This really isn't that unusual. In fact, I feel it is a better approach, as you are validating the entire system instead of individual pieces and then assuming that they work together perfectly.
We do that here with chambers, measuring multiple temperature and pressure transducers along with gas flow. The tricky part is determining a way to stimulate the transducers on your test bench without disassembling it all.
Calibrating the system will require setting up specifications. The easiest and most conservative method is to just algebraically sum the tolerances of each component (transducer, acquisition board, signal processing, etc.) and use that as a system specification. Some prefer to use an RSS combination to get tighter tolerances, that's not my favorite method.
Top Bottom