voltage change in NI PCI 6110 causes jumps

[alex_mikh]

Hello!

We are using NI PCI 6110 along with  BNC 2110 to measure signals from 2 identical detectors (signals are slightly different; in a range 0.2-5V). We get signals from the detectors, subtract offsets and take the ratio. Depending on the signal voltages, we have to control the card input voltage range. When we change the voltage range, we always see a jump in the ratio. The jumps are quite small (<1%), but  they are very consistent and reproducible (i attached some example of the ratio as function of different incident power / voltage on the detectors). Can anyone suggest some ideas what causes those jumps and how to get rid of them? Thanks

[johnsold]

Most likely this is due to slight differences in the calibrations between channels and between ranges. What you can do is to wire the same known signal to both channels and measure it carefully on all the ranges. Record the results. Then for each range and the appropriate signal values calculate correction factors for each channel. Save those values in a file and use them to correct the readings.  This process is called calibration.

If you still have errors, then it is likely due to differences in the sensors. Calibrating the sensors is the solution here as well. The difficulty may be in determining how to generate or connect a known calibrations signal to both sensors.

Lynn

[alex_mikh]

Dear Lynn,

thanks for the feedback!

Some artificial correction (calibration as you called using more scientific language) is what i'm trying to use now to compensate those jumps. It helps, but i was wondering maybe there is a more clever way to compensate it or even to get rid of the jumps at all ..

Alex

[johnsold]

Unless you can keep everything on just one voltage range you will very likely see jumps. 

One option is to use a device with a wider dynamic range.  The PCI 6110 has a 12-bit A/D converter. That has a resolution of 2.4 mV on the +/-5 V range and ~0.1 mV on the 0.2 V range.  If you used a device with a 17-bit A/D converter, it would have a resoution of ~0.08 mV on a +/-5 V range. Using such a device would allow you to make all the measurements on one range, avoiding the jumps.

Lynn