I wonder if anybody had the chance to witness a factory test for accelerometers and/or was exposed to the steps to produce the calibration curves in the FAT.

What type of mounting is used? If it is a standard stud size (screw) what torque and surface considerations are taken? (ISO 5348 mentioned 1.8 N.m and recommended applying an oil film between the two surfaces –though this standard is not for FAT).

What vibration amplitude is used to excite the sensor? Is it constant throughout the frequency range?

Are there any precautions/limitations when a shaker table is used to check general purpose accelerometers?

I had the privelage to tour a facility and go through the production area, and the calibration station. This was many years ago and I did not question how they did things.

What are you attempting to accomplish? Field cal of an existing permanent install?

Hi Mick,

Part of it is to test the new sensors and installed sensors on site, using the shaker table that we have in house.

In addition to a specific case where we had a vibration switch giving strange alarm and trip signals. We tested the loop under the input of the shaker table and found that the contacts responds under different input values if we change the frequency of the signal.

Any ideas?

We follow a specific procedure for our accelerometers, but do not deal with vibration switches. Are these on Recip compressors?

No, actually they are mounted on a few old vertical motors.

Can you share theme of your procedure? If it is copyright or company owned information, ignore my request.

I'm most interested in pitfalls, limitations, common mistakes, tricks, shortcuts etc.

Standard procedure is to stud mount on the shaker with oil, and then sweep across the entire frequency range past the mounted resonance of the transduer taking a sample say every 500 Hz.

typically you use a calibrated transducer for reference, mounted to the table as well, set the table to each frequency, set the amplitude to 1G on the reference transducer, and then take a sample on the transducer to be calibrated.

Reference transducer and sample must be bolted back to back as tight as possible to avoid mounting resonance interference there are reference accels that have holes for bolts both ways. We use white noise from the 2-channel FFT that is synced to the sampling and do a transfer function or correlation plot btw. reference and sample. Since we are the way we are we do that in velocity. Cross channel will compensate for shaker output variation across frequency range if that is not a special controlled device. Next limitation is the hi freq. cut off of the shaker system. Producing the nice plots in the accel datasheet are normally done in another way. If you are checking velocity probes they may not be possible to check w/o the electronics as there may be linearization in the electronics or you need a good identical reference transducer. Good luck. Olov

One method that has been used with seismic transducers is to suddenly drop them vertically, a step function. This can be achieved by suspending the transducer (probably in a jig of some type) from a springy cord (bungy) and impacting the cord. One may correct for the gravity pull at one's particular location on earth as needed.

One needs the transfer function between the transducer signal and a step function to generate the frequency response curve. Pre-triggering seems to be a good idea for this test, so as not to loose the start.

Thanks for your replies.

A draft form is attached for your comments.

Seismic_Vibration_Sensor_Check.pdf (11 Kb, 18 downloads)

Good? Bad? No comments to improve!

Note:

The sheet is originally in an excel file for automatic plotting and to perform conditional formatting.

My only comment was that it would take a lot of time if you have 300 transducers at the end of a papermachine... So it depends on what you use it for if it´s for regular check of the transducers used for route collecting I think it´s a good idea. Brutal way is to just have 1 point like 159.16 Hz, that is as far people have come here, some extend to check both like 5 and 10mm/s for linearity test but rarely anything more, except the nuke plants at initial FAT they require documents similar like this and more. Olov