oscilloscope triggering for long chain of events (segmented sample memory)

[zac]

I am trying to measure the frequency of and capture a waveform that goes through a long sequence.  A simplified version might look like this:

(1) square wave, 500 hz, 30 second duration
(2) square wave, 125 hz, 60 second duration
(3) square wave, 700 hz, 300 second duration
....

The amplitude of each part may be different or the same as the previous part.

I want to measure the frequency and duration of each part of the sequence and haven't figured out a good way to trigger the scope to do it.  I have a TDS754D with 8M segmented memory.  I'm considering getting one of the new rigol scopes with deeper memory (56M for DS2000 or 140M for DS4000) , but don't know if their triggering would be any easier than on the TDS754D. 

[h0nk]

Hello zac,

if this is a single task: use a USB-based logic analyzer.
A Saleae-(or-a-clone)-LA has plenty of (disk-)space to capture the whole thing and
analyze frequency and duration.
If necessary use a amplifier and a schmitt-trigger to deal with the changing amplitude.

My scope has a pulse trigger which can adjusted to trigger if the pulse width is equal, unequal,
greater or lesser than a defined value. This will help to distinguish the frequency to trigger the scope but will not give any duration.

If You must doing this measurment on a daily basis, i would recommend to write a program
for a PIC, ARM or an AVR. Any controller with some Timers should do the job.


Best Regards

[Gallymimu]

I'd consider labview with a moderate cost DAQ board.  Your frequencies are low enough for the cheaper USB DAQs.  Then you can save all the data and do any programmatic analysis you might need.

[zac]

This is just a one time use for a project.  I'll expect to only make a few dozen sets of measurements.  I'm hoping to avoid building custom hardware/code to do it.  The waveform in question is also not a square wave so I would ideally like to capture the first cycle or two on the scope and also the duration (or number of cycles).  

I haven't used labview in many years and don't have a working setup so that would probably take too much time.  Perhaps the easiest way is to build a circuit that will trigger the scope at the beginning and end of each sequence.  I was hoping one of the new rigol scopes might have some capability to do this without external trigger.  

Posted on: November 21, 2013, 08:04:19 08:04 - Automerged

Hello zac,

if this is a single task: use a USB-based logic analyzer.
A Saleae-(or-a-clone)-LA has plenty of (disk-)space to capture the whole thing and
analyze frequency and duration.
If necessary use a amplifier and a schmitt-trigger to deal with the changing amplitude.

I may be able to convert the signal into a pulse or square wave of the same frequency.  Does their software do automatic analysis of the capture to measure frequency and duration?   Something like (frequency, duration) or time index showing when the frequency changes?  If not, would it be practical/possible to scroll through up to a half hour worth of captures (even at very low sample rates like the lowest of 25 khz) to look for the time indexes when the frequencies change?  (At 25 khz sampling rate, that would be 45 megasamples in 30 minutes)   

45 megasamples...the Rigol DS2102 has a 56 megasample memory.  I wonder if the sample rate can be set low enough for the 56 msa to cover 30 minutes...I only need to measure frequencies up to about 1 khz.  My TDS754D can be, but it only has 8M memory depth.  Have any of you tried to scroll through a huge capture buffer like that on the rigol scopes (or with their software on the PC)? 
 

[h0nk]

- Does their software do automatic analysis of the capture to measure frequency and duration?
Saleae: frequency yes, using the cursors for duration: yes.

- Something like (frequency, duration) or time index showing when the frequency changes?
No.

- If not, would it be practical/possible to scroll through up to a half hour worth of captures (even at very low sample rates like the lowest of 25 khz) to look for the time indexes when the frequencies change?
The software so far operates very fast, but im afraid it would be slow when it needs to operate on disk data.
You may export the data to CSV and analyze them with some self written software.

I think there is no such simple thingie ready were You can simple drop Your signal in and get the expected results.


Some easy suggestions:

Use a PLL (4046) and record the voltage of the VCO. This will give You a direct graph of frequency vs. time.

Use Retriggerable Monoflops to discriminate the frequency and use the Logic Analyzer to record their output.

Use a Acoustic measurment package. The one i use (SpectraLAB) is unable to log the maxima of the FFT on disk.

Use a Frequency Counter with some GPIB/RS-232-output and log the readout with a terminal program.
Perhaps this can be done with Your scope to.


Best Regards

[wild]

why not using a PC soundcard input to sample the signal?
There are lots of free software packages suited to the analysis of the signal!