Grounding a noisy load cell 2

[Dodger6600]

I have a 25000 pound (15V max) load cell that is supplied by a battery pack/converter which puts out 12.22 VDC. The load cell is read via DAQ board by laptop. The issue I am having is that the noise level (without any load) is approximately 1mV which according to what I have calibrated with compression machines is about 1000 pounds/mV. How can this be? Granted the load cell came with a spec sheet calibrated at 10 VDC the initial reading without any load shouldn't be that high. Do I need to ground the DAQ board or the battery pack? Also any ideas as to where all this noise is coming from with the load cell?

 

[MiketheEngineer]

No authority on this subject - but personal experience always says GROUNDING is good..............

 

[Skogsgurra]

One of the oversimplifications that plauge the engineering world. Grounding is not at all always good. And grounding the load cell's active parts is a disaster.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[analogkid2digitalman]

Inherent power converter noise? Can you run right off of a 12V battery source to retest?

-AK2DM

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"It's the questions that drive us"
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

 

[btrueblood]

Are you sure you have 1000 lb per mV? So the maximum output is only 25 mV? At those voltage levels, you will need good shielding to avoid stray voltage pickup. And a nice, stable, instrument-grade amplifier with a gain of 200x to pull the 25 mV signal up to the 5 volt typical input range for a PC-based A/D board.

 

[Comcokid]

You didn't indicate what kind of DAQ board (data acquisition) you are using. Load cells are usually built using a Wheatstone-bridge type of arrangement consisting of strain gauge resistors. Interfacing to this arrangement requires the right kind of interface and the right kind of A/D converter.

There is a world of difference in what you might be able to achieve between trying make measurements using a 8 bit single-ended A/D converter in a inexpensive USB stick, verses a 24 bit National Instruments computer acquisition A/D with differential channel inputs and front-end programmable gain. What you are calling noise may be the wrong equipment.

Reading a load cell or other strain gauge/piezo resistive sensor where the full scale output is only a few millivolts is not straight forward. Usually a load cell will be connected to a low pass filter, probably a differential amplifier, and then to a sigma-delta type of A/D converter. Additional filter may be performed in the A/D or software.

 

[BrianE22]

How high a frequency response are you looking for? In addition to grounding you might want to try a small capacitor across either the excitation or across the output.

 

[LiteYear]

1mV of noise could come from lots and lots of sources. The sensitivity you state seems realistic - 1mV/1000lbs is 25mV at full scale, which is 2mV/V full scale at 12.5V excitation. 2mV/V is often called a "Gage Factor" (GF) of 2, which is pretty common. Well actually I'd call it Gauge Factor but then I wouldn't measure weight in pounds either ;-)

A GF of 2 means that full scale output is 0.2% of your excitation voltage. So a noise source doesn't have to be particularly large to destroy your signal.

All is not lost though - there are plenty of circuits designed to cope with this kind of sensitivity. There's power supply filtering, Wheatstone bridge arrangements, signal conditioning, shielding, and data acquisition amplification to consider. It is definitely not just a matter of grounding or not - in fact in this case a floating differential measurement is probably best. You need to look at your measuring circuit in detail. Does the load cell spec sheet have any application recommendations? The fact that the calibration is at 10Vdc is not relevant - the cell was just be a little more sensitive at 12.22Vdc.

 

[ScottyUK]

What else is fed from the battery? Anything which draws enough current to be able to modulate the voltage at the battery terminals will be show up in your measurement. Motors would be a prime candidate, as would any variable load. With 12.2V available you have options to regulate the excitation voltage it to 10V and reduce any effect from supply variation to a negligible level.

 

[Dodger6600]

I conditioned the signal to try and rectify the problem and turns out Comcokid called it right. Turns out I was supplied with an inadequate DAQ board that isn't sensitive enough to read such low changes in voltage. Long story short, another is on its way.