accelerometers -- heads or tails

[longano]

Greets all.
I'm new here, looking for a few pointers.
am getting into some basic vibration measurement
and purchased an accelerometer (10mV/g) and a
DAQ board (100,000 samples/s).

my accelerometer is mounted on a electric vibrator
whose frequency i control via an inverter (0-150Hz;
0-9000 rpm)

questions:
1. when my motor is OFF, shouldnt i be getting
a flat line from my accelerometer? ie, no acceleration?

2. at any given freq/rpm.. should my accelerometer
output look like a beautiful sine wave??

3. is raw accelermeter data what i should be looking
at? or is there conditioning that needs to be done
to see what i'm after?


with the accelerometer at a standstill (on my desk!)
i get a very ..."noisey" waveform. i cant make heads
or tails of it.

when running, say 20Hz/1200rpm i get a very UGLY
looking waveform.. i can SORT of see sinewave information
in there.. but i cant seem to get my software to extract
any information.

accelerometer mount is lubed and torqued per manuf. specs.

any thoughts?
thanks,
-tony

 

[GregLocock]

1. when my motor is OFF, shouldnt i be getting
a flat line from my accelerometer? ie, no acceleration?

yes, definitely

2. at any given freq/rpm.. should my accelerometer
output look like a beautiful sine wave??

depends, but if the shaker is running gently, yes.

3. is raw accelermeter data what i should be looking
at? or is there conditioning that needs to be done
to see what i'm after?

Don't filter anything until you sort the problems out.

Your accelerometer has very low sensitivity for this sort of work. I'm guessing, but doubt that many shakers could manage 50g with a bare table, 10g seems more likely.

Having said all that I can't see how to resolve your problem, without looking at the setup.





Cheers

Greg Locock

 

[MikeyP]

You don't mention if you have any signal conditioning for your accelerometer. Is it an ICP device? If so you may require a special amplifier. In my experience it its usually the accelerometer cable rather than the transducer itself that is at fault when you have noisy signals.

Have you tried connecting the accelerometer to an oscilloscope to rule out the possibility of a dodgy ADC?

Do you have a manufacturer and model number for it?

M

 

[longano]

hello again. thanks for the quick response!
to clarify:

Omega hardware (

http://www.omega.com)

on Labview software (v6).
Omega Accelerometer ACC103 "laboratory accelerometer"
15grams, 10mV/g output up to 500g's.

the acc. is co-ax cabled to an ACC-PS1 powersupply,
which in turn is wired to my DAQ board. the batteries
in the powersupply are fresh and new. (is this a
signal conditioner? or just an amplifier? terminology?)

my DAQ board has selectable gains of 1,10,100&1000
(if this helps the matter)

i will not have access to an oscilloscope until next
week.. but even when i get it, i am unsure what the
signal should look like (from the accelerometer).
-- which makes troubleshooting a tad difficult.

layout:
Accelerometer > Acc Power Supply > DAQBOARD > Labview
Labview > DAQBOARD > Inverter > Motor.

the vibrator/shaker is fairly large, capable of
generating 1000kgF. it is isolated from the floor
via springs. when the vibrator hits the spring's
natural frequency, the setup moves +/- 5mm easy.
at that point the accelerometer is a blurr.

but then again, so is the signal i'm getting.

all of the mechanical connections are tight and sound.

if someone would be kind enough to explain how, i'd
be happy to post a screenshot of my signal.

 

[IRstuff]

You didn't indicate how many g's you attempted to program.

A typical DAQ board might have 1V or 2.5 V max input range. If you've only got 0.1g, we're talking about 1mV of raw signal coming from the acc, which means that you'll need to have nice clean cables and good shields and grounds.

The fact that you getting signal with no acceleration suggests that you've got some sort of noise problem. You might try hanging a resistor in place of the acc to see if it's a cable problem.

TTFN

 

[longano]

interesting development in the past few hours:

(quick note: my accelerometer is screw-mounted to a small stainless steel block.. about 10mm x 10mm x 50mm. the block is then attached to my 'structure' with a C-clamp .. this way i can move the accelerometer anywhere i can get the clamp around)

1. removing the clamp makes my signal "cleaner" (not great, but cleaner) --why?

2. lightly touching the accelerometer with my finger cleans the signal up even more.

this suggests a grounding problem to me. if so, where would a grounding problem possibly occur? everything (outside of my computer) seems fit as a fiddle.

thanks!
-tony

ps.. you fellas are great. i'm doing this research on my own (ie no $$) and have very few resources to turn to. the supplier, Omega, has been a let-down.

 

[IRstuff]

The webpage shows the accelerometer cabling and photo as having the shield connected to the case of the accelerometer.

There may be some coupling of voltage from the vibration table, particularly if you've bolted the accel to the table. A way of minimizing this is to superglue the accel to the table. While epoxy would provide more insulation, it's harder to remove. Superglue can be removed with acetone.

If you can otherwise mechanically insulate the accel from the table, you could still try to bolt the accel on the table.

Once you're electrically isolated, you might try grounding the table to the PC chassis to divert the coupled energy somewhere else.

Otherwise, you might need to look at some sort of isolation amplifier to decouple the vibration table from the DAQ

TTFN

 

[tpet]

You may just have a bad accelerometer also. When you're calibrating the accelerometer, is it holding the signal or floating all around? You can do a physical calibration or shunt calibration, by shunting a known resistance across one leg of the bridge. If either looks "squirrely" the accel is probably bad.

tpet

 

[longano]

re: faulty accelerometer

i've taken some steps to isolate the signal (better grounding).. i measured 38VAC between the shaker and the metal table its attached to. ie.. between motor ground and (any of) the 3phase (380V) feed lines. i'm not sure if this is normal or not, but i put a piece of paper between the acc and the test piece and my signal cleared up ALOT. still not perfect, but better. i'm going to run an earth ground from outside to all of my equipment.

the acc signal does have a DC offset, however. at a standstill its around -2VDC and slowly climbs to ~1VDC when in operation. touching the acc also cause the DC offset to fluctuate. at standstill i now measure ~1mV (~1g) so i still have some cleaning to do.

i am assuming that an accelerometer will not
measure 1g unless in a freefall, correct?

re: calibration. please explain how to do this with the resistor. the acc has a coax going to a chargeamplifier. where does the resistor go? and what should my measurements be with a known resistance?

thanks a 10^6,
-tony

 

[EnglishMuffin]

An accelerometer in free fall in a gravitational field should measure ZERO g (ignoring slight field gradient effects across the accelerometer which are immeasurable).
Try PCB accelerometers and signal conditioners next time.