You yourself told me that tdsdata uses some downconversion from 32k to 16k!
So, how does the downconversion appears in the measurement?
How does the difference of the sampling rate appears in the measurement?
If you like to understand the delay, you have to dig into the downconversion
issue until you get the EXACT mechanism including the filter coefficients.
AND, is the transfer function the matter now?
As far as the LSC and OMC have some firm relationship, whichever this is phase delay or advance or any kind of filering,
this will not introduce any noise. If so, this is just OK.
In my understanding, the additional noise caused by the clock jitter is the essential problem.
So, did you observe any noise from the data?
The measured data are shown in attached fig.1 and 2.
In the fig.1 it looks like they are the same signal.
However in fig.2 which is just magnified plot of fig.1, it shows a time-delay apparently between them.
The delay time is roughly ~50 micro sec.
The surprising is that the LSC signal is going beyond the OMC signal, although the OMC signal drives the LSC !!
We can say it is "negative delay"...
Anyway we can guess that the time stamp or something is wrong.
Tomorrow I'm going to measure the transfer-function between them to see the delay more clearly.
( And I would like to fix the delay. )