Can we average out channel error by using multiple channels to measure the same voltage?

[Sean.C]

I don't know how correlated the error terms are between measurement channels, but it occurred to me that, if they were relatively independent, I might be able to sample the same signal with multiple channels and increase accuracy.

For example, instead of measuring a voltage with a single AI at 100kHz, I might connect up 10 different AI lines to the signal, sample at 10kHz on each line.  This should allow me to average out the some of the error associated with each channel(, or so I would think).

Can anyone definitively speak to this?

Thanks,

Sean

[Seth_B. · Accepted solution]

Well, if you are using a multiplexed board (anything that isn't specced for simultaneous sampling) then each channel is connected to the same ADC one after the other, so the best you'd be able to do is account for the extremely, extremely small variations in the pathes of each channel of the multiplexer.  This would still be overwhelmingly dwarfed by the inherent noise of the system and the accuracy of the device.

Your best bets for reducing measurement error is to oversample on a single and average to reduce the effects of noise and to self-calibrate the device before beginning each test to account for temperature variations.  In addition to that, make sure you keep your board calibrated (most boards have a 1 year calibration cycle).

For certain applications, you should also consult the Field Wiring and Noise Considerations for Analog Signals, How Do I Eliminate Ghosting From My Measurements?, and Troubleshooting Unexpected Voltages, Floating, or Crosstalk on Analog Input Channels to help account for ghosting and settling time issues.

http://digital.ni.com/public.nsf/allkb/73CB0FB296814E2286256FFD00028DDF