I have a PCI-6255 board that I'm planning to use for some sensitive measurements. I'm trying to put together an end-to-end error budget and I have a couple questions. My model is in terms of analog-to-digital converter counts (also known as LSB in your data sheet). In the Absolute Accuracy table on page 5 of the NI 615x Specifications I see a column labeled Random Noise, (uVrms). I think I understand the residual gain error, residual offset error, and associated Tempco, and the INL error but the Random Noise is puzzling me. Here is why: on the +/- .2V range, the random noise is 16 uV. On this range, the ADC resolution is 6.1 uV so the random noise is about 2.6LSB If I look at the +/-5V range I see a random noise of 140uV. On this range, the ADC resolution is 305uV so the random noise is well under one LSB. Therefore, in my error budget I can use "1 LSB" for the random noise when on the 5V range but I have to use "2.6" or "3" LSB when on the .2V range. I expect a signal of aproximately 150mV riding on a 2.5V DC level. I was concerned I might need to offset the signal (remove the 2.5V dc) in order to use the +/-.2 volt range to get better resolution. I also thought I would get better noise performance too. In looking at the numbers here though it looks like I would be better off using the +/-5V range (assuming it gives good enough resolution) because the readings have less noise counts (even though it is more noise volts). I can account for the Gain Error and Offset Error by injecting known signals and doing a "software correction". Not sure, but I believe I can also correct for the INL if I use a polynomial fit of 2nd or 3rd order. It's the random noise I'm having problems understanding. Here's the question: what is the Random Noise shown in the AI Accuracy table for the 625x boards? Can you give an example of how NI might have measured it? Secondly, note 2 says "Sensitivity is the smallest voltage change that canbe detected. It is a fucntion of noise." Was this measured with a signal-to-noise ratio of "1" or "2" or some other standard, or is this a calculated value? Thanks MG | 
