Count rising edges between two pulses

[neilsontr]

I'm looking for either code or hardware that will help me with a gate circuit.  We will have 2 pulses on the same channel, and I need to count pulses from a separate line during the time between the 2 pulses.  Our existing systems utilize some external hardware (essentially a flip flop) to provide this, but I'm looking for a way to get rid of this external hardware.

[alyssa.h]

Dear neilsontr,

What you have described here is a little confusing. Can you please sketch out or draw a diagram of what you are trying to accomplish? Also, what is your current hardware setup?

Regards,

Alyssa H.

National Instruments

[neilsontr]

here is a quick hand sketch. 

[Kevin_Price]

This can be a (mostly) straightforward Counter task.

A method that's pretty universal for just about any NI board with counters would be to configure a Period measurement task, setting it up with Units="Ticks".   You'll need to specify the Gate signal in your sketch as the signal whose period you want to measure.  Then the sorta tricky part is that you need to identify the Counter Input signal in your sketch as the counter's timebase source.

To get just 1 measurement, don't make a call to DAQmx Timing.vi before starting the task.  To get a buffer full of continuous measurements, *do* call DAQmx Timing to set up a buffer and configure it as "Implicit (Counter)" .  Note: different era boards have different counter behavior for buffered period measurement.  Newer ones return the 1st sample at the end of the 1st *full* interval when the 2nd Gate edge arrives.   Older ones return the 1st sample at the end of an intial *partial* interval when the 1st Gate edge arrives.

There are other methods that might be simpler conceptually but they only work on newer boards such as the X-series MIO boards.

-Kevin P

[BowenM]

This is also fairly trivial to do with just code and two analog input channels.  Just set some voltage threshold.  If you go above that threshold you are "in pulse", dropping below it becomes "out of pulse".  Increment your counter every time you detect from in to out.

I'm not at a computer with LV right now, but in pseudo code

For (each element in data array)

  {

   If (element < Threshold && InPeak == true)

      PulseCount = PulseCount + 1

  else 

  if (element > Threshold && InPeak == False)

     InPeak == True

  }

Just store "InPeak" and PulseCount in shift registers and have a for loop every time you read from your analog input.  Only process for loop in between your two trigger pulses (that you'll detect the same way!).

[Kevin_Price]

I like the suggestion from BowenM.  In my experience on the forums, it's a better match for the way most users find most natural to think about these problems.  I just happen to like counter/timers kinda a lot.  

I'm pretty sure his pseudo code will need to be tweaked some, but the basic concept can work and most people will probably find it easier to understand.

-Kevin P

[neilsontr]

We currently do not have any hardware specified, I'm working on that now.  Our existing systems have used M series DAQ (or whatever the standard multifunction daqs are/were).  Those systems had external circuitry that we developed, I'm trying to move away from custom hardware if I can.  I can move to any sort of DAQ available now, but would prefer to keep it out of RIO so that I don't have to start over with our programs (and I have no experience with RIO).

[neilsontr]

We can certainly do this, and have done this (essentially a software trigger turned on and off by that line).  For this particular situation though, our preference is to make the gate external and not software.  There is some concern of accuracy if it is essentially turned on/off via software.

[BowenM · Accepted solution]

Kevin is correct - I have nothing in my psudo code to set InPeak == False, but I'm sure you can figure that part out 🙂

---

@neilsontr wrote:

There is some concern of accuracy if it is essentially turned on/off via software.

---

The software should be easy enough to validate.  I see no reason to have any concern at all with this method.  To put this into perspective I wrote a quadrature encoder demodulator using (almost) the exact method I described above.  I was able to count all 4 edges, direction, and frequency extremely accurately up to ~20 kHz pulse trains.  The only limiting factor was the maximum rate of the AI card I used.