Tclock synchronization - how often to initialize?

[blakney]

Hello All,

I am afraid there is no category for digitizer <and> AWG <and> CVI so today digitizer group gets stuck with this.

I am using a 5421/5122 or 5442/5122 in PXI/PXIe chassis under CVI with Tclock synchronization. In this application, it is important that the delay between signal generation and signal acquisition be very reproducible (at least under identical experimental conditions). I add both devices to a Tclock session, use a marker from the AWG to trigger the digitizer, and set the trigger delay on the digitizer for any additiional required offset.

My problem is that I am also looking for speed. As such, I only reset the Tclock when the AWG or digitizer settings are redefined by the user. Currently, I reset on "ANY" change to either device.

As we automate the changes to respond to "pseudo" real time response the Tclock reset begins to require a more significant portion of the duty cycle. I was wondering what parameter changes "require" a new synch on the Tclock.

Obviously, changing the sampling rate on either device would necessitate a change. (The sampling rate on the AWG is fixed at the beginning of each session so that will not change - but  the digitizer sampling rate will).  Note : there is only one record on the digitizer for each pass of the instrument - one acquisition / one transfer / one reset. generally only single channel as well

What are the other events I should flag to reset the Tclock?

Record size (both), number of waveforms (AWG), trigger delay (digitizer),  marker location (AWG) and  amplitude (both) are subject to change throughout the experiment.

One final question, does anyone know the approximate time difference in Tclock init for a PXI 5122 versus PXIe 5122 in conjunction with a PXIe 5442?

Thanks in advance ,

Greg

[Ana P]

Hi Greg,

I understand that you want to speed up your program by saving time in setting the TClk. At this point I’m a little confused on what you mean by reset the TClk, do you mean that you initiate a TClk session? Or that you synchronize the TClk signals of the sessions? Are you aborting the generation every time before changing the events in the AWG? Or do you create a new session for the individual devices? It would be helpful if you can give us a little bit more information on how your program is configured.

The TClk is a signal internally generated by each device from the sample clock. The TClk software calculates the TClk frequency based on the sample clock, but I believe that the TClk initiate function is not deterministic; the time it takes to perform will depend on your operating system. I hope this is helpful,

[blakney]

Sorry about that Ana,

I see that I was mixing terms.

After each experiment, when nothing changes on the AWG or digitizer, I simply initiate the Tclock. This does not take that long and  is not such a problem.

Whenever the user changes any parameter on either the AWG or the digitizer, I currently set a flag and this forces a new Tclock syncronization. I am asking what parameters require a new Tclock synchronization and when can I opt instead for a simple Tclock initiation (clockrate changes on either device would obviously require this.)

I only create new instrument sessions at program start (load) and I abort the AWG output generation between each experiment when I transfer the data from the digitizer. At this time, I reset two 6733 and a Spincore NMR timing card as well.

Note : there are many experiments per session and / or program execution.

I understand that Tclock initiation is not deterministic (nor is anything else under WinOS) but the Tclock initate is chump-change compared to the Tclock synchronization. I can live with some variation in delay time as we timestamp the data received (tenth second or so) but this at least keeps us from propagating time errors over hundreds or more experiments.

Greg

[Ana P]

Hi Greg,

Thanks for sharing more information. You are right in that clock rate changes will require synchronization of the TClk. Actually you will only need to synchronize the TClk when you are changing parameters that will affect the sample rate. There is a list of these attributes in the NI-TClk Synchronization Help, under Performance Optimization:

I hope this will be helpful in speeding up your application,

Message Edited by Ana P on 04-01-2008 12:03 AM