I have a sensor that produces tiny currents which are converted to volts for conversion via a delta-sigma converter. Actually, there is no hardware yet, but I have done this before. The problem in this case is that everything is close to the limits imposed by Johnson noise rather than by what the converter ca acheive. I'm trying to et a good estimate of the noise based on bandwidth, resistor value in the I-to-V and converter rate. For now I'd settle for the analog bit befor the converter since I'm confident that this is the limiting factor.
Whata I understand is the Johnson noise is sqrt(4kTRB). Where k is the boltzman constant, T is the temperature, R is the resistance and B is the bandwidth. So far so good. Up until now I've alsways assumed that the bandwidth of an RC low-pass filter is simply 1/(2 pi R C). However, when I try to calculate this from first principles I get a different answer, namely
1/(4RC). I started to do this from first principles because I wanted to work out what the bandwidth of a sinc(x) filter is. Sinc(x) appears when integrating the measurement over a time interval.So two questions: What is the Johnson noise of a resistor R with parallel capacitor C. I'd expect sqrt(2kT/pi/C), where the resistance cancels. What is the Johnson noise of a resistor R when the voltage is averaged over time t?
..and no I'm not a student. I wish I were that young. My brain burned out with age. Just some pointers required.
Peter