DC PD calculations for OMC
The full package for the analysis of OMC DCPD amplifier with various transimpedance (Made by Koji Arai)
Method
The circuit model in LISO was made. The transimpedance resistor (rT) was varied from 10 Ohm to 3000 Ohm with a half-decade log-spacing (i.e., 10, 30, 100, ...). The transfer function (total transimpedance gain) in Ohm (=V/A), the input-referred current noise in A/rtHz, and the max input current range to have no saturation, in A, were calculated in each case.
How to interpret the plots
- transimpedance.pdf: This is a basic check to see how the input current is converted to the output. At DC, the transimpedance gain is rT*2 because of the differential output. Between 100-10K, the gain is boosted to rT*(11^2).
- input_current_noise.pdf: The input noise is 20pA/rtHz (= the shot noise of iDC~1mA) when rT is 100Ohm. When rT is smaller than r8 (100Ohm), r8 cause the dominant thermal noise. When rT is larger, it dominates the noise. However, above rT=1kOhm, the input current noise of the 1st opamp dominates the noise. That's why the improvement of the input referred noise for rT=1k, 3k is suppressed.
- input_range.pdf: This indicates the input current amplitude which makes the circuit saturated when a sinusoidal wave at the amplitude and frequency. For example, 120mA at DC makes the circuit saturated. This is because the first (and second) stage opamp only has the ±12V output range in LISO. This limit is hit when 120mA is given to the transimpedance resistance of 1K.
Notes on the model
- In the LISO model, LT1028 was used instead of LT1128. However, this is not a problem in our application.
- The output of the circuit is to be differential. This requires a proper noise estimation, strictly to say. i.e., the noises from the output stage are uncorrelated; however, the noise from the previous stages are coherently added. To deal with this issue, I have added a low noise buffer+differential amp using ideal opamps (op00) to the output stage. So notice that this is not a part of the real circuit.
How to run the LISO model
- The primary model is "D060572.fil" .
- The liso directory contains "fil" which is a Mac version of LISO. Replace this with an appropriate version if you are not on Mac.
- Use a Perl script "mfil" to run different model runs in a batch.
- mfil produces "runAa" files. "A" ranges from 1 (rt=10Ohm) to 6 (rt=3000Ohm).
- "a" ranges from 1 to 3. 1 is the transimpedance gain calculation. 2 is the input noise analysis. 3 is the max range calculation.
- A Matlab script "D060572.m" summarizes the results. It uses my class file "freq_data_tools.m" , which needs to be placed in the path.