D. Barker, T. Sanchez, D.Bhattacharjee, R. Savage, F. Llamas
Characterized the latency of a Keithley digital volt-meter (KDVM or DVM) by making measurements of a Martel voltage source using kvm_read.py
.
The objectives:
The Martel was injecting 1 V to the KDVM. The python script took 1000 *immediate* samples from the DVM.
The pdf "kvm_ConfigComparison" illustrates the results from our measuremnt. The discretization of the data comes from rounding the time to 3 significant figures and the voltage to 5 significant figures (my bad). The top plot shows a voltage timeseries for both tests. Notice that the measurement labeled as "Once" finished acquiring the thousand samples in ~ 3 seconds before the measurement labeled as "On". The bottom plot shows the time difference between the samples. It is clear however that when the DVM is periodically using the auto-gain and auto-zero functions, the sampling will have inconsistent latencies.
A third measurement was done to characterize, with a better resolution, the latency of the DVM sampling function when the auto-zero and auto-gain are set to "Once". The results are shown on the pdf "kvm_LatencyTest1". A mean of the time difference between samples yield ~ 18.5 ms, which is interesting since 1 PLC takes 1/60Hz ~ 16.6 ms. The histogram shows that, from the Keithley itself, there is a variation of 0.012%.
In summary:
It is probably helpful to follow up with a technical note on how to (will be) configure(d) the Keithley before making a measurement and to document the jargon of the useful configurations and functions used for pcal responsivity ratio measurements.