AD2 Averaging Trigger Peak Frequency Phase Alignment

[brandonhill7]

I am using the AD2 to record a small signal that needs to be averaged to be brought out of noise.

I have a very clear 10+V trigger pulse about 50 ns wide on ch1, and the small signal is a 3Mhz burst about 70 us later on ch2. This waveform is repeated at about 10kHz.

This works well, and the averages are filled quickly when "Acquisition #X/X" is displayed, but stalls completely when "Acquisition #X/X @ X kHz" is displayed as shown in the attached image. I am trying to understand what is going on, and how to fix it. Right now I just have to keep pushing start and stop repeatedly until "Acquisition #X/X" is displayed.

I have tried to figure this out on my own, but I can barely find anything discussing what is going on, except for in this manual where under 3.1. Time Group the following is stated:

Quote

Average and Oversampling require repetitive signal and stable trigger with Single or Repeated acquisition. The individual captures are aligned and combined based on peak frequency phase of trigger source channel or Channel 1 when other trigger source is used, or trigger crossing of source channel when no peak or too low frequency is detected. For peak frequency alignment Acquisition # X/X @ MHz is displayed and for trigger crossing alignment Acquisition # X/X.

I think I'm savvy enough to understand what this is saying, and how and why you might gain accuracy by aligning two separate acquisitions to match the phase of the least noisy component. Except in this case I have a large and sharp edge way above any noise floor to use for alignment, and more than that, the phase alignment doesn't seem to work, slowing acquisitions to a crawl.

How do I disable peak frequency alignment? Is the only way to write a script replicating the averaging? I do need to automate this in waveforms SDK, which I have previously worked with, but I need to understand if I am going to run into this problem there too?

Thanks for any help or understanding!

 

Note: It seems like the phase alignment works when the time position shift is set to zero, but not when the position shift is set to the required amount to record the burst at 70 us from the trigger pulse. It may be that waveforms is trying to find the peak frequency of the window that is being recorded, and not of the trigger pulse, in which case I can see why this doesnt work, since there are no longer any recognizable features in either ch1 or ch2 at this point, just noise. That doesn't change the fact that triggering only based on the edge without phase alignment works fine for me. If I sample at 100MHz I have plenty of resolution of the 3MHz burst, and I don't need alignment to fractions of the sampling period.

Edited April 27 by brandonhill7

[attila]

Hi @brandonhill7

The average and oversampling needs stable and repetitive trigger. The acquisition will stop if there are no more trigger events and the trigger is set to normal.

To prevent the FFT phase alignment to be used, reduce the number of samples to have less than 8 periods, like for 158kHz signal ~4ki samples, or set the hidden option Scope.Trigger.Phase.text = "Xing" for alignment based on trigger crossing interpolation.

[brandonhill7]

Perfect! I was hoping there was a hidden flag like that, and it works perfectly for me!

Thanks very much for the quick answer Attila! I have found this forum and your other posts in particular very helpful as well.

Edited April 29 by brandonhill7