external clock for AI trad & mx

[GabbyG]

We are using a 6110 to read a train of bi-polar analog pulses. The pulses consist of a negative and positive half-puls with 250 ns spacing. These are sampled with an external clock on PFI0 of the same high speed.

Up to now we were using traditional NIDAQ 7.0 currently with LV 7.0 on a Win2K PC.

Now I ran into a limitation of 16mio samples for a finite scan. While modifying to help this, I decided to move up to mx interface.
This is not giving the same results as with the old interface. The first sample (the negativ one) of the twin puls is read correctly. The second puls is obviously wrong. I suspect it is taken with a delay.
When run with the maximum internal clock of 5MHz you can verify both halfwaves are correct.
The hardware h
as not changed. Is there a different setup time for the sample puls between traditional and mx interface? Well, I doubt it.
What else could I have missed.
Thanks
Gabi

[peterguy]

Hi Gabi,

if you are not having the newest NI-DAQ driver(7.2) you should first update it. You can download the driver for free on our homepage. Perhaps the delay is only a small bug which is fixed in a newer driver-version.
If this doesn't help, can you post your VIs? I would like to look at your code.
Is it also possible to take a Screenshot of the Signal?

best regards,

Peter Weber
NI Germany

[GabbyG]

Thanks Peter for offering to look into my problem.
To start with DAQ-mx I downloaded 7.2. Previously I used DAQ 7.0 together with LV 7.0, but using only traditional DAQ. I have one problem to start with DAQ 7.2 as the DAQ Assistent is not starting properly. It will start and then hang for ever. So I am working with the programmed task setup.
I modified the example "DAQmx Create Channel (AI-Voltage-Basic)" for the demonstation purposes. It shows the same problems as my own program.
I attached a screenshot from a scope and VI FPs for onboard clock and external clock.
The signal are several bipolar pulses, generated with a frequency of 4MHz. The pulses are repeated twice and at the second time sampled with two consecutive clock pulses.

When sampled with the internal clock @4MHz it is obvious that the levels of the scope image are reached. When using the external clock it looks like the reading is taken on the slope or at least not at the top.
What is also very strange is that the setting of rising or falling edge makes no difference.

Again thanks for your help.
Gabi

[peterguy]

Hi Gabi,

the problem is the samplerate of your 6110-Card, i suppose. Actually the samplerate should be more than two times higher than the frequency of the Signal. So its only luck, if your measure usable data. In the worst case, your card measures at every zero-crossing, the result would be a flat line.
Probably, the DAQmx is a bit slower than the traditional DAQ, maybe around 50 ns. So it could be, that you are sometimes measuring at the edge of your signal. I have modified one of your Pictures, just look at the attachement.

If you want, I would sent you a PCI 6115-Card for testing, because it has a Samplerate of 10 MS/s. I think this card should better fit your issue.

Best regards,

Peter Weber
NI Germany

[GabbyG]

I do not expect to see a square wave. The external clock is synchronous to the data pulses so I should always see the maximum (flat top) of the negativ and positiv half-wave. That results in a triangle in the VI display, that's understood. I thought, that using external sample pulses is a purely hardware issue. That makes me wonder how a new driver can affect that.

The traditional was working fine. But now I ran into limitations with the 16 Mio odd maximum number of samples. The data is coming in bursts of some 20000 a time, so I was collecting the whole set in one read. If there is a way to do this in traditional I might do that. But first I'd like to know what is really going on.

The 6115 is probably not a solution. We are a
lmost ready for an alpha test of our machine. It's supposed to to become a low-cost machine. We'll switch to the 2-channel version anyway. So I guess the 6115 would blow our budget.

Since you are located in Germany I might try to contact you directly.
Gabi

[Stuart_G]

Gabriela

One thing you might check is your convert rates. I know there was a change between Trad DAQ and DAQmx on the convert rates. Since you are using an external clock I am not sure if it will matter or not, but make sure these are the same in your applications.

Stuart Gillen
National Instruments

[GabbyG]

To verify that last sentence of rising/falling edge ignorance I made an experiment. I use a programmable function generator. On channel 1 a ramp (100kHz) on 2 a pulse with a width so that the rising and falling edges are at distinctly different point of the ramp. When I switch between sample point at rising or falling edge I should get the level of the ramp at these instants.

This is not true.

"DAQmx Timing.vi / Sample Clock" obviously ignores the entry and always uses falling edge.
Gabi