Load cell voltage spike NI-9215

[ciccio29]

Hi

Im having trouble with a NI 9215(screw terminal) when using a cDAQ-9178. I have a Load cell powered with a 12V battery that provides a +-5V signal in output. The picture below show an example of measured signal with NI MAX(samples to read=100, rate=100Hz) without load applied on the load cell.

One can note that the mean value of the signal is correct(i've checked it with a multimeter) but there are several voltage spikes.

The layout of the connections is:

NI-9215 AI0+=Load cell signal+

NI-9215 AI0-=Load cell signal- (which is a cable connected to the negative terminal of the 12 V battery, as recommended by the load cell manufacturer)

I've also tried to connect the NI-9215 AI0- with the NI-9215 COM with a 1 MΩ resistor, but it doesn't have effect.

this signal behavior is more noticeable when a determined load is applied, for example see the picture below.

The graph show the draft signal(kgf) acquired with the load cell, the mean value is correct, but these spikes are really annoying.

Could someone give me some advice on what I am doing wrong thanks.

[Henrik_Volkers]

Just some assumtions:

1.  the sensor has an internal switched power supply (needed if only a single ground reverenced DC supply is provides but the output is +-5V 😉 )

2. this switched powersupply introduce current spikes on the supply lines

3. supply lines are resistive ! (incl. the ground line in this case, and see my signature :D) 

4. the current spikes on the ground supply line are also measured since you shared this line 🙂

(Or you have a  powersupply nearby that couple these spikes into your cable/sensor)

Possible solution: seperate the signal and power supply lines. two twisted pair should do the job. connect the A0- as close as possible to the load cell output. try to avoid a shared power supply and signal line)

If you can't seperate them because of a fixed cable (hard wired to the sensor and no seperate ground provided) claim/hit the sensor maunufactor (BTW a link to the sensor datasheet would have helped 😉 )  and try software filter.

Software filter:

Lowpass filter always help , thats what the DMM do 😉 , for small spikes and higher bandwidth median filter migth do the job.

As more as you (and we!)  know about the signal provided and the information needed (How fast is the signal changing, how fast do you need that information... also called needed bandwidth 😉 ), as better the software filter can be matched.

Sample the signal (with and without load) with a higher samplerate (seems like you have a 60Hz powerline and the spikes are every 1/60, so it could also be a poor designed classical powersupply involved)  10kSPS (or 100kSPS) for 120ms are fine. and post a png and a vi including a graph with save data if you want to motivate others to provide a filter.