Signal Drift on SCC-SG24

[ntaINC]

We have NI hardware to take measurements from 3 load cells, it consists of a NI-6220, SC-2345, (2) SCC-SG24, and (2) SCC-LP02. During a recent system calibration I noticed that all the channels exhibited considerable drift when the load cells were in a no load condition (and probably while under load as well but I didn't test to verify). For example, a 25,000 pound capacity load cell had a no load measurement of 142 pounds and an hour and forty minutes later it measured 56 pounds. The degree of drift seems to be related to the how long the computer containing the hardware has been turned on but continues well past the longest warm-up time specified in the hardware's datasheets; in the previous example, after an additional hour and a half the load cell measured 25 pounds. I performed a test using a load cell simulator and saw the same drift, this should ensure that it is the hardware and not the load cells causing the drift. I also connected one of load cells to different NI Hardware (cDAQ-9178 and NI-9237) and didn't see the same degree of drift.

Basically, after warming the system up for 4 hours the measurements steadied out enough to render any continuing drift inconsequential during the duration of the test. This amount of warm up seems excessive compared to the other NI hardware we have. Should I be concerned? Have others experienced this?

Thanks for contributing any knowledge or information on the issue I am experiencing.
Tessa

[WillPwr]

Hello! 

Longer warm-up times can sometimes occur if the load cell isn't able to properly dispel heat  - often if the load source is warm or cannot act as a sufficient heat sink. This does not sound like the case in your situation, since you don't see the warm-up time with the 9237, but what are you measuring load from/on?

It sounds like this could also be a grounding issue (see this Forum Post), particularly since you see this on all your channels (as opposed to only on a single module). Be sure that your signals are properly grounded when using the SC-2345. 

I'm also curious - how long has this system been in use? I know you said you only recently noticed the issue, but was the system ever working as expected? That is, without the warm-up period you see now?

Please let us know once you've had a chance to check the wiring/grounding for the 2345, and let us know if you have any questions. If that does not fix the issue, we'll be happy to continue troubleshooting. 

Thanks!

[ntaINC]

Will,

Thanks for replying and sorry about the length of time until I responded. I am not sure if the system has always exhibited these charateristics but I suspect so. Here is a post I made not too long after we got it (https://forums.ni.com/t5/Signal-Conditioning/Voltage-shift-in-Load-Cell-output/m-p/2141398#M6875) and I am wonder now if the 2 aren't related.

The reason we may not have noticed it is because the system typically is turned on long before we use it and stays on until that cycle of testing is completed. As I said in my original post, the longer the system is on, the less noticable the drift. 

As far as our application of the load cell, we test building products so the load cell is being pushed against a wall section using a hydaruilc cyclinder with everything typically at ambient temperatures in a temperatrure controlled space. My test with the 9237 was done in the same space.

Based on the article you posted, grounding could explain the effect observed but given the complexity of the SC-2345 with the SCC-SG24 and SCC-LP02 modules we're using, I'm not sure how to even go about solving it. According to this KB (http://digital.ni.com/public.nsf/allkb/4804BB9C978629B986256DCC0065C7C8) all the grounds on the SC-2345 are the same and all tied back to the PCI card. But the SCC-SG24 doesn't have a ground input, though the EX(-) is connected to analog ground, and then the LP02 grounds the AIGND and AISENSE together. When I setup the TASK in MAX my only option for connection type is NRSE (which I think is due to the SCC-SG24 conditioning the signal into a -10 to 10 volts) instead of differential (like I think I do on the 9237). How else should the system be grounded?

Thanks again for any help or insight you can provide,

Tessa

[WillPwr]

Hello, Tessa!

First, performing a nulling operation is often a good way to eliminate some offsets in hardware. However, the behaviors you're seeing aren't usually symptoms of offset, since the values change after longer runs. You can certainly perform this nulling as a part of your code (it's generally a good idea), but there is possibly something else going on here.

Regarding the grounding - they key point is that your strain gauges are wired according to page 17 of this document:

SCC-SG Series Strain Gauge Modules User Guide

Based on your last post, it sounds like it is so far. The grounding will be taken care of if your EX(-) is connected properly (as you mentioned, it runs to AO Ground), and if the condition listed in Table 1 from the below document: 

Field Wiring and Noise Considerations

Note that you'll be taking a floating measurement from your strain gauges.

Please let me know if offset nulling has any effect, and we can continue troubleshooting if need be. Thank you, Tessa!

[ntaINC · Accepted solution]

I was finally able to remove the drift by removing the Lowpass filter from the setup. Appearently the SG24 and LP02 don't play nice together.

[WillPwr]

Thanks for following up with your solution on this one! We appreciate it.

Best,