DO(16x) + AO(2x) ELVIS II sampling help DAQmx 200621

[Cuprum]

Hello,

I need to achieve continuous generation of signals about frequency from 1 to 200 Hz. The DO will goes to DAC and on the output became as analog signal on the same frequency. For convert to analog I use resistor-ladder. DAC working good as separated Vi. I tested AO single and double on one Vi. The problem is following:

I use DAQmx palette. But after generating I gets error 200621 (buffer underflow).

Is there method to achieve number of samples which cant give me signal with suitable smooth signal?

In the other sight isn´t that synchronization related problem?

For synchronization of 16 DO I use SW-timed synchronization.

For synchronization of 2 AO I use Continuous non-regeneration synchronization.

Any tips?

Thank you.

- Cuprum

[astro_naut]

Hi, 

Have you tried running basic example Vis on your system? 

Could you please share the code you are trying to run?

[Cuprum]

Hi,

yes I tried examples from Help > Find examples >HW input and output > DAQmx.

Did not find any solution.

This error still occurs by one time from ten random deploy of code.

My specs of laptop: processor i7-7700HQ, 16 GB DDR4 ram, nvme 256 GB Samsung storage

The code is in attachment.

Kind Regards, Cuprum

[Kevin_Price]

Remove the 1000 msec wait from the loop.  That should make the buffer underflow error go away.   Once the loop iterates rapidly a few times at startup to fill the buffers, DAQmx will do the waiting for you.  Since you're writing 1 sec worth of data per loop, you'll eventually settle into a 1 sec per iteration loop rate.

You're writing a multi-sample digital waveform to an unbuffered, software-timed DO task.  I don't think your device supports hardware-timing for DO, so you should be writing just 1 DO sample (that can contain values for all 16 lines).  I don't know what you'll get when you try to write a multi-sample waveform to a 1-sample-at-a-time task.

Then you have the problem of sync.  Or rather, the lack of sync.  And the basic impossibility of sync with that device.  You're writing AO data to a buffer.  There will be some latency delay as the data works through the task buffer and the device's own hardware buffer before it becomes a real world signal.  Meanwhile, you'll have to write to DO with software timing where there will be no buffer and much less latency.  And you'll only generate 1 DO state for every 1000 AO states.  And that's just an average -- you won't actually have enough timing control for that 1 DO state to occur *exactly* every 1000 AO states.

In an earlier thread you were using a 6259 device.  *That* one at least is capable of buffered DO and hardware-driven sync.  It's a little indirect though as you'll need to understand and use "correlated DIO".   Basically, the DO task will need to borrow its sample clock from the AO task.

-Kevin P

[Cuprum]

Dear Kevin Advice,

thank you for your reply which opened my eyes.

Now I got seriously problems with my digital output timing.

The project is cardiac pacemaker impulse tester. Those impulses will be reference for waveform located on the inside of heart muscles.

I need three analog outputs, based on three locations inside of heart.

Elvis II its not capable to gives three analog outputs, so I decided to bring third one by using 16x DO for

resistor ladder digital to analog convertor.

I used to generate that impulses by triangle pattern, with timing for analog outputs it works very good. I can set delay, amplitude, frequency & number of samples.

But for digital outputs, it doesnt works very well. It does not sync the impulses itself. Minimum DO I can use is 8x DO.

I sending here output on oscilloscope. The offset is not at zero (I am using differential amplifier for correcting this).

I tried to set DAQmx write (Digital > Single Channel > Multiple samples > 1D U32), there is difference that I am not using Analog to Digital vi.

Is there solution for this? Thank you.

[Kevin_Price]

I've never used any NI ELVIS hardware, but over in this thread, you seem to have confirmed that DIO can only be software-timed by marking that response as your solution.

That being the case, I don't think this particular device is capable of letting you accomplish what you want.

As I stated before, there are 2 buffers worth of latency in the AO task where sample timing is governed by a hardware clock, largely independent of your loop iteration rate.  The DO task will have to be based on software timing where you can generate 1 sample per call to DAQmx Write, and sample timing depends almost *entirely* on loop iteration rate.

You're just not going to be able to get reliable sync this way.  You need a more capable DAQ device.  You certainly should not be expected to run tests on something as important as a pacemaker without also being equipped with a more suitable DAQ device.

There isn't a lot of point trying to give you small incremental help b/c it's all heading toward a dead end.

-Kevin P