Author Topic: Long Period Measurement (~10s with DS1054Z)  (Read 1501 times)

0 Members and 1 Guest are viewing this topic.

Offline metrologist

  • Frequent Contributor
  • **
  • Posts: 881
  • Country: 00
Long Period Measurement (~10s with DS1054Z)
« on: January 24, 2017, 05:45:40 AM »
I am using a hearing amplifier to pick up a vibration/sound from a simple machine. The machine makes a tick sound about once every 5 seconds (full period is two ticks) and I would like to accurately measure the period and record samples to characterize the timing over a day or more.

I thought I would use my scope to start with capturing one period:

First I set my scope to single sweep, 2s/div, 12Mpts memory depth (1 microsecond resolution). The rest of the settings are default. I get a trace with three distinct events, so I save this as CSV and take it to my PC.

I have to use Access to read all 12M points. I filter points that are above a voltage threshold and copy that data set into Excel for plotting (~50k points now).

An XY scatter plot shows the full 12s sweep, but there are like 6 or 7 events along the sweep that hit the peak value instead of just three, so it makes no sense. The X value is the point number (1 to 12M) and Y is the voltage. I'm not seeing where this approach went wrong.
 

Offline rstofer

  • Super Contributor
  • ***
  • Posts: 3361
  • Country: us
Re: Long Period Measurement (~10s with DS1054Z)
« Reply #1 on: January 24, 2017, 06:33:16 AM »
How about a screen capture of the scope image and a graph from Excel (converted to PDF?)?
 

Offline metrologist

  • Frequent Contributor
  • **
  • Posts: 881
  • Country: 00
Re: Long Period Measurement (~10s with DS1054Z)
« Reply #2 on: January 24, 2017, 07:40:47 AM »
I can recapture a measurement tonight, but does that mean this seems like it should work, or is there a better way to do it?
 

Online tautech

  • Super Contributor
  • ***
  • Posts: 9990
  • Country: nz
    • Taupaki Technologies Ltd.
Re: Long Period Measurement (~10s with DS1054Z)
« Reply #3 on: January 24, 2017, 08:07:26 AM »
Use some loooong Holdoff, a different Trigger type and Single to record a # of the events at a slow timebase setting.
Once captured, use Horizontal position and a faster Timebase to see the event in detail.

See this post and particularly the Gif as to what to expect.
http://www.eevblog.com/forum/testgear/siglent-sds1000x-series-oscilloscopes/msg933997/#msg933997
Avid Rabid Hobbyist & NZ Siglent Distributor
 

Offline metrologist

  • Frequent Contributor
  • **
  • Posts: 881
  • Country: 00
Re: Long Period Measurement (~10s with DS1054Z)
« Reply #4 on: January 24, 2017, 08:46:34 AM »
I need enough resolution, but it sounds like that is what I did. I think my setting gives me 1 ppm resolution.

Using zoom mode, I could not figure out how to see the resolution. The automatic measurements shows the period to hundredth of a second.

I was looking at the delay setting and wondering if that number is showing the period when I am zoomed in on the third pulse.
 

Online tautech

  • Super Contributor
  • ***
  • Posts: 9990
  • Country: nz
    • Taupaki Technologies Ltd.
Re: Long Period Measurement (~10s with DS1054Z)
« Reply #5 on: January 24, 2017, 08:54:38 AM »
As you adjust the Horizontal position to scroll through the captured record it should indicate where you are relative to the trigger and even though you can't see the capture record in full it will give indication to where the other events are in relation to the capture Trigger.
Avid Rabid Hobbyist & NZ Siglent Distributor
 

Offline metrologist

  • Frequent Contributor
  • **
  • Posts: 881
  • Country: 00
Re: Long Period Measurement (~10s with DS1054Z)
« Reply #6 on: January 24, 2017, 09:38:55 AM »
It did not seem like I was able to figure out how to get a measurement that way, but I'll look at it again tonight. I recall trying to manually place cursors or use tracking cursors to no avail.

 

Online newbrain

  • Frequent Contributor
  • **
  • Posts: 385
  • Country: se
Re: Long Period Measurement (~10s with DS1054Z)
« Reply #7 on: January 24, 2017, 10:15:56 AM »
First I set my scope to single sweep, 2s/div, 12Mpts memory depth (1 microsecond resolution). The rest of the settings are default. I get a trace with three distinct events, so I save this as CSV and take it to my PC.
[...]
An XY scatter plot shows the full 12s sweep, but there are like 6 or 7 events along the sweep that hit the peak value instead of just three, so it makes no sense. The X value is the point number (1 to 12M) and Y is the voltage. I'm not seeing where this approach went wrong.

Something is inconsistent in your description: if you use 2s/div the whole shown trace will be 24 seconds, matching the 6 events you see: the screen width is 12 divisions, independent of the memory and sampling rate settings.

Moreover, my 1054Z will set a 500ksample/s rate for 12Mpts memory, not 1Msample/s (giving 24s).
If you set 1s/div, you'll have a 12s trace shown on the screen and 1Msample/s.

Are you sure of the parameters you reported?
Could you please attach a screen printout (and maybe a zip of the csv)?
 

Offline metrologist

  • Frequent Contributor
  • **
  • Posts: 881
  • Country: 00
Re: Long Period Measurement (~10s with DS1054Z)
« Reply #8 on: January 24, 2017, 11:25:13 AM »
It probably was 1s/div. The scope should turn on with the same settings and I'll verify when I get home.

I had a CSV file with 12M records (lines of data), and I think it will be too large to post? It took the scope like half an hour to write that file.

Really, all I need is a precise timing of the clicks. That can be relative to previous or in absolute time. I'm hoping to collect all this data and see patterns in the timing.
 

Offline FrankBuss

  • Supporter
  • ****
  • Posts: 1809
  • Country: de
    • Frank Buss
Re: Long Period Measurement (~10s with DS1054Z)
« Reply #9 on: January 24, 2017, 01:31:33 PM »
It probably was 1s/div. The scope should turn on with the same settings and I'll verify when I get home.

I had a CSV file with 12M records (lines of data), and I think it will be too large to post? It took the scope like half an hour to write that file.

Really, all I need is a precise timing of the clicks. That can be relative to previous or in absolute time. I'm hoping to collect all this data and see patterns in the timing.

You can store the file on Google drives, 15 GB is free. Reading it with Access and then filtering it for Excel sounds really complicated. 5 lines or so of Python code should do it. When you post the file, I can try it.

Offline metrologist

  • Frequent Contributor
  • **
  • Posts: 881
  • Country: 00
Re: Long Period Measurement (~10s with DS1054Z)
« Reply #10 on: January 24, 2017, 02:17:07 PM »
For some reason my scope will not trigger on the output of my audio amplifier - I see no signal from it at all. I could not figure out the problem. It works with earbuds and is loud, but when I connect my probe to one of the channels there is no signal. It worked fine yesterday.

I played with settings a bit anyway and single trigger works only with one event, and to capture additional events is cumbersome to get a zoom view after.

Normal trigger does not allow enough delay to see the second signal. When you try to offset, the delay jumps from 10s to 5s, and then you have to zoom out.

I have not played much with zoom mode, but that does let you offset the signal enough, but then zoom is limited. This might work.

The period is actually closer to 13.5s, so I had sweep set to 2s/div and yes it does show 500kS/s, so resolution is half what I thought.

Well, I'm kind of back to the higher level problem of detecting the signal I want, a quiet click sound.
 

Offline JPortici

  • Super Contributor
  • ***
  • Posts: 1852
  • Country: it
Re: Long Period Measurement (~10s with DS1054Z)
« Reply #11 on: January 24, 2017, 05:05:22 PM »
Quote
but there are like 6 or 7 events along the sweep that hit the peak value instead of just three?/quote]

in such situations you should use peak detect, otherwise all the data between two points will be discarded/averaged, same for on screen trace so when you zoom in.. surprise surprise! maybe those peaks were always there and you couldn't see them without zooming
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf