High-Speed DAQmx Acquisition

[eoinbutler]

I'm working on a project that reads the voltages on several channels and needs resolution of the order of tens of microseconds.

What I'm doing is using DAQmx in On-Demand, Continuous timing and using a loop that reads the samples one by one and stores them in a buffer array (there will come a time when I want to do range checking on each point as it comes in, so that's why it's being done this way). However, when the time comes to read the stored data, the time stamps are about 15ms apart, and there are multiple (5-10) readings with the same stamp. The time stamps are read from the t0 of the waveform output by the DAQmx Read VI.

Leaving alone the fact that I'm not reading fast enough, is there a way to make it so the timestamps are more accurate?

[TCPlomp]

You should use timed data acquisition (continous) and not point by point

A range check can be done on multiple items:


Ton

Message Edited by TonP on 09-03-2007 05:01 PM

[eoinbutler]

That's not possible, the reason is, if it goes out of range, I need to change the gain setting on hardware for the next point. That means that the range check needs to be done point-by-point.

[TCPlomp]

What is the purpose of this measurement?
If your interested in the evenly spaced data you need a multipoint reading.
If your interested in protecting your hardware (as you mention) you can use single point and not worry about the time between samples.

Ton

[eoinbutler]

The project is to monitor hardware over a long period of time for events which take place on the microsecond timescale. The events involve a change in voltage of several orders of magnitude, so in order to properly follow it, the gain (controlled by a signal conditioner) has to be changed dynamically during the event.

Message Edited by eoinbutler on 09-03-2007 10:49 AM

[eoinbutler]

Someone IRL pointed this out to me.
The timer doesn't need to give absolute/system time to the microsecond accuracy. All is needed is the time after some fixed reference point (VI start or something).

[Dippi]

Hi eoinbutler,

I think that the most people have a problem by understanding what you exacly need.
Please write down at first only your exact requirements.

You can also post this in German if it is easier for you to explain it in German (but please both Englisch and German).

Best regards
Dippi

[eoinbutler]

Okay, I'll try to be as clear as I can. I'm afraid that I don't speak German.

I have a piece of hardware on which I want to measure the voltage drops at at high temporal resolution (typically on the order of milliseconds). The voltages from from the hardware are fed to a signal conditioner, which can apply one of four different gain settings. This is in turn applies the signals to the DAQ card. The signal conditioner gain is controlled by Digital out lines from the same DAQ.

Now, every so often in a long period of time (hours, days), an 'event' (a rapid change in voltage levels)  may happen very quickly (over a period measured in tens of milliseconds). In order to record the event accurately, the gain on the signal conditioner must be changed to keep the signal on the DAQ card in the range -10 - 10V, and fill that range as much as possible. Which means that LABview must sample the data AS IT COMES in, decide if the gain needs to be changed, and act as appropriate.

The difficulty I was having was that I couldn't make the read/check loop fast enough. I think I've solved that now, and I might post how once I'm sure it's working.