Author Topic: Using a generic programmable scope with wavegen to make automated bode plots?  (Read 1658 times)

0 Members and 1 Guest are viewing this topic.

Offline JPortici

  • Super Contributor
  • ***
  • Posts: 2656
  • Country: it
- Of course one could use a separate programmable wavegen. Let's keep it contained for now shall we?

So, it seems like the bode plot function on the keysight dsox-1000g is (for now) a gimmick because it steps the frequency over a limited number of points. (10 apparently)

Question: If someone had a scope with generator included, this one could make its own (computer) application by sending commands and reading the measurements?

-> Set AWG wave: Sine
-> Set AWG amplitude
-> Set AWG frequency at F start
Repeat n times (for an average)
-> read back measurement: amplitude A
-> read back measurement: amplitude B
-> read back measurement: phase B - phase A
draw point.
Increase AWG frequency by delta (delta being the number of points)
Adjust vertical/orizontal scale.
repeat m times, until F stop is reached.

seems pretty straightforward... not as handy as having it already inside the scope though, but probably more powerful
 

Online pigrew

  • Frequent Contributor
  • **
  • Posts: 567
  • Country: us
- Of course one could use a separate programmable wavegen. Let's keep it contained for now shall we?

So, it seems like the bode plot function on the keysight dsox-1000g is (for now) a gimmick because it steps the frequency over a limited number of points. (10 apparently)

Question: If someone had a scope with generator included, this one could make its own (computer) application by sending commands and reading the measurements?

seems pretty straightforward... not as handy as having it already inside the scope though, but probably more powerful

Yes, very possible. I actually have put together some Matlab code to do this for audio frequencies for the 3000 series of oscilloscopes.

I use the high-res mode of the scope for acquisition (while keeping the sample rate high enough to prevent bandwidth limiting of the signal harmonics). This lets me also make distortion and noise measurements. FFTs are used with the Welch method to calculate the PSD.
« Last Edit: March 07, 2017, 02:41:55 pm by pigrew »
 

Online rstofer

  • Super Contributor
  • ***
  • Posts: 7407
  • Country: us
That's pretty much the way we did it in lab class.  Just set a bunch of frequencies and measure the output amplitude and phase.  The data could be stuffed into something like wxMaxima (or Excel, or ...) to get a decent looking Bode Plot.

The quality of the plot will be determined by the number of sample points.  I suppose the data could be fitted to a curve (least squares?) to improve the appearance.
 

Offline voltsandjolts

  • Supporter
  • ****
  • Posts: 900
  • Country: gb
... a gimmick because it steps the frequency over a limited number of points. (10 apparently)

No, its more like 50 points, enough to be useful I reckon.
 

Offline MrW0lf

  • Frequent Contributor
  • **
  • Posts: 921
  • Country: ee
    • lab!fyi
https://bitbucket.org/hexamer/fra4picoscope/wiki/Home

Standalone open-source bode plotter, maybe source code can be used as basis for some other scope interface also:


I will include it in my Pico 2408B review/test, however my AWG is only 1MHz with standard signal. So this app probably more useful on 5000/6000 models with 20MHz AWG (if works).

On the other hand my relatively expensive handheld Agilent U1733C LCR meter is only up to 100kHz, works in steps and even with these limitations has proved very useful in small electric motor R&D.
« Last Edit: March 07, 2017, 04:56:45 pm by MrW0lf »
 

Online rstofer

  • Super Contributor
  • ***
  • Posts: 7407
  • Country: us
The Analog Discovery will run the Bode Plot out to 10 MHz.
 

Offline djnz

  • Regular Contributor
  • *
  • Posts: 174
  • Country: 00
Ever since I read of the Bode plot feature on the new 1000X scopes, this has been on my to-do list. It seems quite doable, and I am pretty sure it could be done a little more intelligently. For example, by seeing the rate of change of the magnitude (derivative / double derivative) and taking data points more densely when that rate is high or changing fast so that you get better resolution around interesting frequencies.
 

Offline JPortici

  • Super Contributor
  • ***
  • Posts: 2656
  • Country: it
... a gimmick because it steps the frequency over a limited number of points. (10 apparently)

No, its more like 50 points, enough to be useful I reckon.
the 1000x seems to be doing 10 points at the moment

see: https://www.eevblog.com/forum/testgear/new-keysight-scope-1st-march-2017/msg1153743/#msg1153743
 

Offline voltsandjolts

  • Supporter
  • ****
  • Posts: 900
  • Country: gb
... a gimmick because it steps the frequency over a limited number of points. (10 apparently)

No, its more like 50 points, enough to be useful I reckon.
the 1000x seems to be doing 10 points at the moment

see: https://www.eevblog.com/forum/testgear/new-keysight-scope-1st-march-2017/msg1153743/#msg1153743

See for yourself, at 6mins in:


Edit: ...and the answer is: it does 10 points per decade

http://literature.cdn.keysight.com/litweb/pdf/5992-1965EN.pdf?id=2832547

In the 3000X series the points per decade is adjustable, allows at least 30 points per decade.
http://literature.cdn.keysight.com/litweb/pdf/5992-0594EN.pdf
« Last Edit: March 07, 2017, 07:01:02 pm by voltsandjolts »
 

Offline JPortici

  • Super Contributor
  • ***
  • Posts: 2656
  • Country: it
you're right!
i have a feeling that the delta f in the 1000 is fixed... or reduced according to range. i've seen another video where the guy was plotting an audio filter responce, way less points
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf