Frequency response measurements

Frequency response measurements

Postby wireless » Thu Aug 29, 2019 3:24 am

As a new user of the 2820E system, I am still working through the Multi Instrument manual (phew!).

Can you help me understand how to make frequency response measurements of - say - a CR filter set at 1kHz.
- I have established how to set up the signal generator to provide a swept frequency, but am not sure how to convert this into a frequency plot where I wish to establish the corner frequency and rate of roll-off.
- Para 3.2.6 looked like a promising section before I discovered that this apparently needs another $100 to convert to PRO.
- The section on LCR measurements at para 9.6 might also be relevant to what I am doing but this apparently needs a separate (and further?) $100.

I'm hoping that my investment ($370 so far) contains a way to produce a simple frequency plot??

Many thanks
Bob
wireless
 
Posts: 9
Joined: Wed Aug 28, 2019 4:00 am

Re: Frequency response measurements

Postby VirtinsTech » Thu Aug 29, 2019 9:51 pm

There are a number of ways to do frquency response measurement using Multi-Instrument, even with a Lite or Standard version, such as:

1. Single-Channel Methods

1) FFT analysis + white noise excitation
very easy to setup, no synchronization between the signal generator and oscilloscope is required, average required and thus a bit slow. Rectangle window should be used in the Spectrum Analyzer.

2) FFT analysis + linear frequency sweep
oscilloscope frame width must be equal to or a little longer than signal generator frequency sweep duration. The oscilloscope trigger parameters need to be set properly to ensure that the entire sweep is right captured inside one oscilloscope frame, and the FFT Size must be equal to or greater than the Record Length of the oscilloscope. No average required. Rectangle window should be used in the Spectrum Analyzer.

3) FFT analysis + MLS
MLS length must be greater than FFT Size, the rest is the same as 1). Rectangle window should be used in the Spectrum Analyzer.

4) FFT analysis + Impulse
Oscilloscope trigger parameters needs to be set properly to capture the single impulse. Low measurement SNR. Rectangle window should be used in the Spectrum Analyzer.

5) FFT analysis + Multitones that aligns to FFT bands
easy to setup, no synchronization between the signal generator and oscilloscope is required, no average is required. Please refer to Section 4.3.1.8.2 of the software manual. Rectangle window should be used in the Spectrum Analyzer.

6) FFT analysis with peak hold + linear frequency sweep
Signal generator frequency sweep duration must be much greater than oscilloscope frame width to ensure no notch in the frequency response plot during the frequency sweep and peak holding process. Rectangle window should be used in the Spectrum Analyzer.

7) Octave analysis + pink noise
very easy to setup, no synchronization between the signal generator and oscilloscope is required, average required and thus a bit slow. Rectangle window should be used in the Spectrum Analyzer.

8) Octave analysis + logarithmic frequency sweep
oscilloscope frame width must be equal to or a little longer than signal generator frequency sweep duration, the oscilloscope trigger parameters need to be set properly to ensure that the entire sweep is right captured inside one oscilloscope frame, and the FFT Size must be equal to or greater than the Record Length of the oscilloscope. No average required. Rectangle window should be used in the Spectrum Analyzer.

9) Octave analysis + Multitones that aligns to octave bands
easy to setup, no synchronization between the signal generator and oscilloscope is required, no average is required. Please refer to Section 4.3.1.8.2 of the software manual. Rectangle window should be used in the Spectrum Analyzer.

10) FFT analysis + frequency stepped sinewave (Device Test Plan required)
highest accuracy, need to configure a device test plan, very slow. A low spectral leakge window such as Kaiser 6 is recommended in the Spectrum Analyzer.
...

2. Two-Channel Methods (Multi-Instrument Pro required)
Bode Plot with any wide-band stimulus shown in 1.

Multi-Instrument comes with some pre-configured frequency response measurement Panel Setting Files. Some of them can be found in the third toolbar from the top (called Hot Panel Setting Toolbar): FRwhite, Frswph, Frswlog, Bode Plot. You may need to change the setting to suit your specific need (e.g. frequency range of interest, input & output voltage range, etc.)

As for impedance measurement, it is possible to do it without LCR meter add-on. For signal-channel method, please refer to Section 3.2.1.10 Impedance Measurement Mode. For two-channel method, please refer to Section 3.6.1.10 Impedance Analyzer Mode.
VirtinsTech
Site Admin
 
Posts: 276
Joined: Tue Oct 01, 2013 3:06 pm

Re: Frequency response measurements

Postby wireless » Thu Aug 29, 2019 10:22 pm

Wow!
That was a very well detailed response and I am very grateful for your time to answer this. I am delighted to see that there are so many possibilities in the package I bought.

Can you please answer me a follow-up, which is probably a generic question anyway.

Can you help me to understand the relationship between the time sweep and the spectrum resolution? With respect to the attached image snapshot, I have a simple setup of a sawtooth stimulus (red scope display) feeding a filter (output on blue scope display). The spectrum is now showing all the harmonics of both input and output.
In order to get a reasonable time display, the time sweep is set to 4mS.
However, this seems to automatically set the Spectrum display to 500kHz (which is a bit wide for my needs) but more importantly it sets the FFT resolution to 30.51Hz which for the frequency of interest of the test is too wide and results in a low resolution spectrum display.
If I change parameters to get me a lower FFT resolution (say 1Hz) then it looks as if the time sweep has to change so that the cycle waveform is no longer visible.

I hope I have explained my issue. I guess that there are interactions between the parameters so that not everything will be possible, but an explanation would be helpful please.

Inter-relation of sweep and resolution.PNG
Inter-relation of sweep and resolution.PNG (74.07 KiB) Viewed 26473 times


Thanks
Bob
wireless
 
Posts: 9
Joined: Wed Aug 28, 2019 4:00 am

Re: Frequency response measurements

Postby wireless » Thu Aug 29, 2019 10:43 pm

SOLVED.

I have answered my own question - the solution was simply to use the small magnifying glass in the oscilloscope window - this allows the expansion of the time domain display whilst leaving the FFT resolution where it was already set.

Thank you
Bob
wireless
 
Posts: 9
Joined: Wed Aug 28, 2019 4:00 am

Re: Frequency response measurements

Postby VirtinsTech » Thu Aug 29, 2019 11:26 pm

Great! You have figured it out by yourself. Just to add, it looks like that the settings in your screenshot are as follows:

[Oscilloscope Sampling Rate] = 1 MHz, thus the frequency measuring range in the Spectrum Analyzer is from 0Hz to 500kHz (according to Nyquist Sampling Theorem).

[FFT Size]= 32768, thus [FFT frequency resolution] = [Oscilloscope Sampling Rate] / [FFT Size] = 1000000 / 32768 = 30.517578 Hz

[Oscilloscope sweep time] = 4 ms, [Number of Sampling Channels] = 2 and [Sampling Bit Resolution] = 16 Bits, thus the DSO memory buffer size consumed is 1000000 x 0.004 x 2 x 16 / 8 = 16000 bytes. Note that this is less than the max. DSO memory buffer size (Memory Depth) 40000 Bytes. In case this memory depth is exceeded, the software will automatically adjust the oscilloscope sampling rate or sweep time (depending on which one you are adjusting) such that the memory size required is within its upper limit. The software also set a lower limit (2000 samples at least per channel) as well, below which these parameters will also be automatically adjusted.
VirtinsTech
Site Admin
 
Posts: 276
Joined: Tue Oct 01, 2013 3:06 pm


Return to Questions

Who is online

Users browsing this forum: Google [Bot] and 18 guests

cron