how can I tell if signals from two devices are truly synchronised?

[ClaireJ]

Hi there,

How does one check that signals from two devices (two separate devices in a single X-series chassis) that should be synchronised actually are?   I am using a PXIe6361 and PXIe4331 on a PXIe-1073, with Labview 2001 SP1 64-bit. All devices are using the Sample Clock from the 4331 device, and an AI Start Trigger, so they should be synchronised.

I thought that writing the signal data to file and checking the time stamp for each column of data would be the most accurate, but I have been told that timestamps are software created and therefore don't reflect the actual time that the signals were acquired by the hardware.  When I do this, the timestamps vary by up to 150ms which is larger than I expected.

If I set the x-axis of the waveform graphs (on the GUI) to "Time" then it appears that the first data sample is taken at different times for the two plots (one plot per device).

If I set the x-axis of the waveform graphs (on the GUI) to "Ignore time stamp" (so that the x-axis just starts from 0 rather than a date-time) then the first data point occurs at "0" for both graphs. However, I'm not sure that this reflects the actual alignment of the signal.

What is the best way to check if signals collected on different devices in the same chassis are actually synchronised?

Thanks,

Claire.

[johnsold]

Claire,

Feed the same signal to both devices. Choose a signal with a well defined shape so that you can uniquely identify certain features of the waveform.  For example the output of a counter which varies in a binary or BCD pattern has sharp transitions and a pattern which allows easy identification of several transitions. The signal can be attenuated with resistive voltage dividers if you are using a low voltage input range.

Lynn

[ClaireJ]

Hi Lynn,

Thanks for your message.

I've connected an analog trigger to both the 4331 (just using the AI+/- terminals as its a strain gauge device) and the 6361.  However, I'm still not sure which output to look at to compare timings - the waveform graph with timestamp, the waveform graph that ignores timestamps, or the output .csv file. 

Thanks,

Claire.

[johnsold]

Claire,

That will trigger both at the same time and should synchronize things.

What I was suggesting is to measure some distinctive signal on both devices and look at the data collected.  When some feature shows up on both channels, that is where they are synchronized.  The attached VI is a variant on a demo I put together recently for a dfferent purpose.  Run it and change the offset value.  When the offset is zero, the sample points coincide and everything is synchronized. With other offsets the sample points do not line up.  This is not a perfect example, but it may help.

Lynn

[ClaireJ]

Hi Lynn,

Thanks for your help and for sending the demo.

I understand the concept of how the signals will look if they're not synchronised and your demo shows that nicely. I guess I have been perplexed by someone else telling me that the timestamps in the output file (and following from that I assume timestamps on a waveform graph) do not give an indication of whether signals are synchronised. The reason they gave for this was that the timestamps are manufactured by the software, not the DAQ hardware.  They suggested that I put the setting "ignore waveform timestamps" on my waveform graphs, and then check that both signals come in at the same time (i.e. both start at zero), but I'm not convinced about this. 

When I use an analog trigger, neither the timestamps in my output file or on the two waveform graphs are synchronised. If I don't use the trigger, then there is far less disparity in the timestamps in the output file. I've attached two output files here, and my VI.

This is my first attempt to synchronise a voltage module and a strain gauge module on an X-series chassis, so I want to make sure that I'm achieving the best synchronisation that I can, and the difference in behaviour with and without the trigger worries me.

Thanks,

Claire.

[johnsold]

Claire,

I cannot help much with that because I do not have access to any DAQ hardware or DAQmx to see what is going on.

Lynn