Noisy Load Cell on U6 | LabJack

Noisy Load Cell on U6

3 posts / 0 new
Last post
ncirincione's picture
Noisy Load Cell on U6

Can anyone determine why I'm getting so much noise on this load cell setup? The load cell is an Omega LCKD-1000 and I'm using a U6-Pro

  • I've tried powering with the 5V VS and monitoring it with another AIN and using that in my equation.
  • I've tried using a stable TO-220 ST 3.3V regulator with 3.3 hard coded into the equation
  • I've tried using another AIN to monitor that 3.3V supply and using that in the equation.
  • I've tried grounding the shielding on the load cell cable to GND and also breaking that connection.

No matter what I do, the scaled signal looks like garbage as shown in the screenshot below.

This is the equation used for scaling (with the -90N being the offset I was seeing before I started the stream)

Attached is a PDF file of recording just raw voltage at 100 Hz and a setting of BIP1V and BIPP1V. I would just barely need to use the BIP1V to get the full range out of this load cell. With the BIP1V I can see 11 levels of bit noise.

In the end, this just seems like a noisy signal. Is there anything I can do in the LJ settings (settling time/ resolution) to clean this up? The larger application has 8 channels with some of them amplified so I have to use the BIP10V setting on those (and to measure the VS for the ratiometric sensors). I've been getting by with sampling faster than filtering, but it just seems really messy.



ncirincione's picture
It doesn't look like the

It doesn't look like the screenshots of the LJStream UD front panel and the equation came though inline, so here they are attached.

LabJack Support
labjack support's picture
We expect values from

We expect values from streaming to be noisier than if you were to sample in command response mode (such as by using LJLogUD). Your sampling rate under command response mode may be lower than 100Hz depending on how many channels you sample and at what resolution.

For a reference of stream mode noise, we expect 4.5 counts of noise at BIP1V and resolution index 1 (default). This corresponds to around 137uV. We would expect around 48uV effective resolution at the same settings when using command response mode. Please see the following pages:

To help reduce noise, we would recommend using the 24-bit sigma-delta converter on the Pro. This is enabled by using command response communications and setting the channel resolution index above 9. Doing this would notably increase your per channel sampling time as described in the specifications of Appendix B linked above.