Noisy USB-9215

[Bob_Y.]

Does anyone have any experience using the NI USB-9215 module? A customer recently purchased one to use to capture peak force measurements, but is getting more noise than was expected. Upon investigation, the input noise specification is not good, 7 LSB for peak-to-peak. The expected signal range for the sensor being used is +/- 100mV. Our NI field engineer suggested oversampling and then averaging to remove some of the noise, which helped, but still does not remove enough of the noise. Hence, we are considering abandoning this module and going a different direction (a more costly direction). It appears that the 9215 is not the right DAQ with Signal Conditioning device for this application now, but I am wondering if anyone has tried to make it work for measuring low voltages at a 100 samples/second rate.

Has anyone else used this module? Have you had noise levels like this?

Thanks for your help,
Bob Young

[AnalogKid2DigitalMan]

Is any noise present when verifying against an oscope? Proper input configuration used (NRSE, differential with 'bias' resistors, etc)?

[Bob_Y.]

After we had some troubles with our sensor, a coworker tried this experiment. He measured a single AA battery voltage using the USB 9215 module defined as being in differential mode (in MAX). Here is what was found:

Setup: I used channel 1 without anything connected to between the negative terminal of channel 1 and the common.
Lowest voltage (low peak) was 1.6070V
Highest voltage (high peak) was 1.6145V

Noise amplitude: 0.0075V (7.5 mV)


Setup: I used channel 1 with the 1Meg Resistor between the negative terminal of channel 1 and the common
Lowest voltage (low peak) was 1.6070V
Highest voltage (high peak) was 1.6135V

Noise amplitude: 0.0065V (6.5 mV)


Setup: I used channel with the 1 Meg Resistor and a jumper wire (hence should be a short) between the negative terminal of channel 1 and the common
Lowest voltage (low peak) was 1.6095V
Highest voltage (high peak) was 1.6110V

Noise amplitude: 0.0015V (1.5 mV)


Also tried was modifing the min & max values for the voltage range that the channel was set up for. Unluckily, we did not see any change in the noise amplitude due to a narrower voltage window.

Let me know if you have any questions on what we did.

[AnalogKid2DigitalMan]

Was there a resistor between the + input and analog common for the battery test? Was noise observed or just battery discharge giving the millivolts difference between highest and lowest.

I'm not familiar with the USB DAQ's (yet), last DAQ I used was a 6025 in full differential with high value resistors on the + and - inputs both going to analog common. No noise issues seen, signals were in 100's of mV.

What happens with a fully shorted input?

For the actual application are wire lengths excessive? Using shielded twisted pair with proper grounding at one end?

Just throwing out some general tips without detailed knowlege.

[e-Cash-Money]

Hello Bob,

The system noise you measured (shorting the input channel) seems to be within the specified noise range of the USB 9215 but there might be some things we can do to reduce the noise that you are seeing. I think AnalogKid2DigitalMan gave some good tips in terms of reducing noise in any system. In terms of software analysis, you might want to do a spectrum analysis on the signal to see if there is a dominant frequency component in the noise that could be coming from an exernal source. If this is the case, we might be able to filter it out of the signal in software.

Just out of curiosity, how did your coworker perform these measurements in MAX? The USB 9215 is supported by the DAQmx base driver. Generally, devices that are supported by the DAQmx base driver do not appear in MAX for configuration as there is a separate DAQmx base configuration utility. However, I am not aware of any sort of test panel or measurement you can conduct in the configuration utility itself. Thanks for the info!

Take care,
E.Lee

[woutert]

Hi,

The new usb-9215A for windows is a usb20 device and is directly supported by max.

So it can be used with all max compatible software, also by vi-logger.

When using two usb-9215A modules and sourcing the same signals to the inputs, the returned signals doesn't seem to be synchronised. Clock shift.

The in my opinion best and simpliest filter for 100Hz data is the median filter.

[woutert]

I tested and have to change my opinion.

The "sample compression" express vi shows that the "mean" function is the best way to get the most precision out of this module.

With the mean function a signal range by short-circuit input is 0,000251754 volt.

With the median function a signal range by short-circuit input is 0,000318676 volt.

Using 10KHz sample rate and a reduction faktor of 100.

Cpu charge is 40%

[Dan_at_Ocean_Systems]

I believe that the noise problem is because the input range setting of the 9215 is alway +/-10V, regardless of the software setting.  Beacuse of this, when you are tring to measure low level signals the ADC is not being optimized.  In effect the noise floor of the device is -80dB because it is always measured from 10V. I'm not sure if NI realizes that the input range can't be changed. I have a call into them to find out.

To over come the lack of the instrument not having a Programmable amp you may want to add a low noise preamp so that you are using more of the 16bits.  If you added 20dB to your signal you should appreciate about a 20db increase in your Signal to noise ratio (i.e. lowing the noise floor.

Dan