PRA pop-up Data Acquisition Stopped Early

[Rich Feldman]

Hello, experts.
I've recently become a PRA user. Posted a glowing report on another forum, after the software helped me to demonstrate detection of neutrons with a BF3 tube, charge sensitive preamplifier, and a radioisotope source whose total activity is on the order of 100 n/s.

In a few sessions I accumulated an hour or two of run time with no problems, on the Windows 11 Pro laptop where this message is being composed. But when we tried same program on the computer of someone who's borrowed my detector, we saw the Data Acquisition Stopped Early pop-up. A week later, reverting to this PC, the message usually appeared within a few seconds of starting the acquisition.

Troubleshooting hints would be welcome. Even confirmation that the stoppage isn't related to the shape of pulses in the sampled waveform. Pulse height histograms seem to behave properly even when audio input comes from built-in microphone. Yesterday I got a complete 40-minute acquisition using internal microphone, in unsuccessful effort to get a D.A.S.E. popup to snip for this story. Started acquisition again, and got DASE within a few seconds. Today I got the pop-up within seconds after arranging several windows for a nice screen capture, and whistling at the microphone.

As mentioned before, I'd welcome guidance about looking at computer issues (sound card driver, competing processes, etc.) vs looking at hardware issues (funny pulses or no signal from radiation detector due to a bad connection, etc.) Thank you!

[Sesselmann]

Rich,

Thanks for posting.

How do you manage to complete pulse shape acquisition with such low count numbers?

I have not seen this message in PRA before, that said I'm not using it much these days, because ImpulseQT does the same thing, it uses the same method as PRA for sound cards.

Neutron detection with a sound card means you need to amplify and stretch the pulse through a low pass filter otherwise you won't be able to catch it. You should also set the sample rate as high as it goes (384 kHz) and use all 60 sample points.

After completing the pulse shape acquisition for a good average number of pulses, you can click the "Copy" button and paste the values into a spreadsheet to see what the pulse looks like when you plot it. ImpulseQt has this feature built in.

Something else you can try is to delete the PRA.ini file in case it has been corrupted.

Other than that I am not sure what to suggest..

[Rich Feldman]

Thank you. We're going to try ImpulseQT as an alternative to PRA, and see how that goes.

PRA served well before "Data Acquisition Stopped Early" started rearing its head on a regular basis, on 3 different PC's.
That was without using any pulse shape training (and I never did find location of any PRA.ini file). Neutron detecting tube gets bias from a NIM module, and feeds a charge amplifier that delivers long-tailed positive pulses (decay RC = 140 us) shaped just like the red line in Gamma Spectacular logo. Followed by a low pass filter, in a box with BNC connectors, that helps us make an oscilloscope trigger on pulses and not on low frequency excursions.
The last successful session was reported in this post on another forum: https://fusor.net/board/viewtopic.php?p=106480#p106480
That includes a snip of audio input according to PRA.

[Sesselmann]

Richard,

Thanks for posting a picture of the pulse..

This work a little bit different with sound card spectrometers, we don't actually want a perfect pulse like the GS logo 😱.

The reason being the sample rate is too slow, so when that fast pulse rises to a sharp point it often falls between two samples and consequently gets missed, resulting in "under-sampling".

You need to pass the pulse through a low pass RC filter causing the top of the pulse to be nicely rounded.

If you look at the three dots forming the tip of my pulse you can see that there is not much difference in the height, this means I get a more accurate sample.

Both PRA and ImpulseQt use the "exact" same method for calculating pulse height:

1) First normalise the pulse to zero (this eliminates baseline drift)

2) Calculate the distortion factor (this is used to filter out pulse pile up)

3) Calculate: pulse height = max(sample) - min(sample)

We are essentially measuring how much energy that went into deforming the tension in a capacitor.

Therefore gaussian pulses work better with a sound card MCA, and as always, higher sample rates are better, even if this means oversampling.

Steven


PS: We put a lot of R&D nto the preamplifier circuit in my latest GS-PRO-V5 (actually v5.6), the pulses are now perfecly gaussian and have virtually no undershoot.

Shape and distortion from ImpulseQt.

NaI detector pulse GS-PRO-V5

[Rich Feldman]

Thank you for the explanation about why sharp-pointed pulses are not ideal for sound card sampling.
I'll experiment with low pass filtering.
Will consider buying a GS-PRO, after looking for a manual to clarify the HV and Signal connectors. One application would be with the existing BF3 tube. Today that has an adapter to BNC connector, and has been happy with 1800 V bias.

When we run PRA software, Data Acquisition Stopped Early continues to be a show-stopping nuisance. I bet it's connected to the Windows environment. When it pops up, it's usually in the first few seconds of a run. But it never reared its head in first couple of sessions here, a few weeks ago. In preparation for experiments with ImpulseQT, receiving signals presented at 3.5mm TRRS audio jack (as with PRA), would it be silly to dust off a benchtop Waveform Generator and try to approximate your unipolar Gaussian pulses?

[ColoRad-o]

I have found that an old Windows laptop doing PRA acquisition gives the Data Acquisition Stopped Early if I attach to the laptop by Remote Desktop (replaced now by "Windows App" on macOS). However, I found that if I logged in "on the console", that is, directly onto the laptop after this, I could tell PRA to resume the count, and nothing was lost. Is it possible some other attachments to your Windows machine were being made?

Hope this helps.

[Sesselmann]

A common reason for PRA or Impulse to stop recording early is that either max counts or max time has been reached. These are stop conditions intended for repeating experiments accurately.

A brief note on using Windows remote desktop from a Mac, I found that neither of my Windows computers allowed sound card access when connecting using RDC, however when I switched to VNC everything worked just fine (almost) , VNC protocol does not exchange information on screen resolution so is defaults to the lowest resolution (bring back the 1980's).

I was able to fix the screen resolution with $10 dummy display port adapter plugged into the PC., apparently it tricks the PC into thinking it has a larger screen attached, ahh...... Windows don't get me started 🙄.