Bode-Plot Test Filter 101

[blackdog]

 Hi,

In a number of topics I see people making Bode Plots.
They do this in some cases with a simple bandpass filter with the frequencies of 1KHz and 30KHz.

The original schematic and explanation.
https://www.electronics-tutorials.ws/filter/filter_4.html

When I saw the filter I thought, that could be a bit better and still keep it simple.
What I did is visible in the diagram below.
I find it a handy little tester for learning to use the Bode Plot function of a scope.
I have used this test filter on my two Siglent Scoop, this along with a Siglent Function Generator linked through my network.

First I added a 50 Ohm termination resistor by using two 100 Ohm 1% resistors in parallel.
By the way, to give you as good a picture as possible, I selected the components to smaller than 1% deviation.
However, this is not necessary if you just want to do some Bode-Plot testing.

The High Pass filter consisting of C1 of 0.15uF and R3 of 1K, I lowered 10x in impedance of this section, originally these were 15nF and 10K
I do not find this a problem, this because almost all Function Generators have a 50 Ohm output impedance.

The output of the Function Generator, due to this lower impedance in the middle of the bandpass filter, will be about 0.21dB lower due to the chosen component values.
This is not a problem, because the Scope used to test the filter measures both the input and the output, so the extra attenuation has little/no influence.

The second section of the filter has also been modified in terms of impedances, but the first filter is now loaded with a nearly 4x higher impedance,
this has reduced the attenuation in the midrange (about 5KHz).
Also, due to the chosen lower component values, the input impedance of the scope has become of less influence.

Ok, below is the modified schematic of the Bandpass filter.


.
I built the filter in a TEKbox case I still had lying around.


.
Here I measure the attenuation of the filter when connected to the function generator at a frequency of 5KHz.
0.000dB would be the perfect value, this was measured with a very precise 50 Ohm termination resistor.
The value indicated here -0.2093dB at 5KHz is so low that it has little effect on the Bode Plot measurement.


.
Here I do a phase measurement on the scope, in doing this I first let the scope warm up and do an autocal.
Then set the two channels to "gnd" first, and put the trace exactly horizontal in the middle.
Now change the frequency of the generator so that the Sines perfectly overlap in the center of the screen.
You can see it better, if you make the time base time shorter.

On this scope picture are more than 1 waveform on the image, this is necessary for the phase measurement at the bottom of the image to work properly.
It looks like that the two traces are slightly offset, but as I indicated they are set to zero to the pixel.
Keep in mind the accuracy of the input attenuator and that the sine is not always a perfect sine as it passes through filters.
Oh by the way, these measurements were taken with a 12Bit Siglent scope.


.
And now finally a Bode Plot 
Settings
Sweep: 100Hz to 200KHz
Point Decade: 40
Log sweep
Sweep type: simple
Level: 1V RMS
Input impedance: High Level (1-Meg)
DUT Input: Channel-1
DUT Output: Channel-3

It is also easy to see in this picture that pase "0" takes place around 5.4KHz, as could be seen on the multimeters above.
The graph is nice and tight for both phase and also frequency.


.
Now I have adjusted the scale to 0.5dB/Div.
Here you can now clearly see how big the attenuation of this filter is just under 0.5dB.


.
Now I did a measurement with 150 points per decade, the graph does not improve, the small deviations are now in other places, I think also has to do with the resolution of the diagram.
No problem for me, making Bode Plots with the Siglent scoops is many times more comfortable, than doing it with some homemade equipment.


.
And last picture to show what the possibilities are when you adjust the scales.
Here I have shifted the phase scale to show at which frequency point it passes through the "0", this is visible in the orange frame.


.
I hope this topic will help those who are going to work with the Bode Plot function a bit to set up the scope properly.

Tips
Do not take more measurement points per decade than necessary.
Do not choose the frequency range larger than necessary.
Especially at frequencies below 1KHz the measuring time can increase.
Check if it is necessary to dynamically adjust the output voltage of the generator, in my measurements I did not use this feature.
If you are going to measure phase margin of amplifier circuits or power supplies, then this dynamic adjustment is often needed.
For each DUT to be measured, you will first have to find out what signal level is needed, at what frequency, so that the DUT can be measured optimally.

SHOOT!

Kind regards,
Bram

Translated with www.DeepL.com/Translator (free version)

[tautech]

Capabilities, tips and guidance here:
https://www.eevblog.com/forum/testgear/siglent-sds1x04x-e-bodeplot-ii-(sfra)-features-and-testing-(coming)/

[mawyatt]

Nice Bram 

Like you, we've found the Bode Plotting function quite useful 

BTW the passive Twin "T" Filter makes an interesting Bode Plot candidate. See starting at #3486

https://www.eevblog.com/forum/testgear/siglent-sds2000x-plus-coming/msg4275217/#msg4275217

Best,

[rf-loop]

My opinion is that testing different oscilloscopes FRA (BodePlot) need bit more than just simple RC filter. It is too easy. There need also be some more challenging tests for drop out toys from real tools.

Dynamic range and resolution... and if go further also need test so that there is harmonics and non harmonics interference signals for look freq. selectivity. (or some other way to test FRA:s (BodePlot) receiver frequency selectivity what is also very important in many real cases exept if just playing with some simple passive RLC circuits. Example some DUT's may have high EMI etc or some other kind of active nonlinear circuits may produce heavy amount of spurs.  These below are not at all enough for testing, just random pick-ups from huge amount of separate tests.  Also not at all "best ones".

Perhaps one can make some universal test board for do enough set of tests for FRA (BodePlt) so that results are also comparable.




SDS2504X HD


Some "random" 455kHz IF (AM) filter. FRA Freg span is here 50kHz




SDS2104X Plus
Just cheap surplus XTAL picked out of a miscellaneous box. Everyone have some kind of these.
Frequency span here 20kHz (2kHz/div) ~2.46MHz center
(look also here resolution and dynamic range)





SDS2104X Plus
Just one simple passive RC band pass (separate channels from separate points in DUT, C2 from final OUT)



SDS1104X-E
Just same simple passive RC for compare




SDS2504X HD
One random sample from dynamic range tests
C3 is just for display noise level.
Signal is 0dBm and before and after peak attenuated using 100dB (step)attenuator.
This is dynamic range down from 0dBm just because some other manufacturer also tell dynamic range from 0dBm.
Also dynamic range depends frequency. Noise level with very low frequencies and higher frequencies are more.
As can see phase data is here very noisy. (Naturally)
Also signal data is noisy but still far away from FRA base noise level.

It is also good to note that -100dBm signal is 2.24uVrms  (6.3uVpp).   

As can see in this last image.


SDS2000X HD FRA noise level from 10Hz to 60MHz (using input channels 200MHz BW).




It would be good to do tests that can discriminate between toys and tools.
What would be the value of such tests that everyone passes almost cleanly.
It somehow seems that some manufacturers have added the BodePlot function just to get one more feature in the glossy and gorgeous sales brochure. This should be found out with tests.  Also it need note that with high dynamic range DUT there need take care about channels cross talk  (DUT In is also FRA reference channel what can leak  to DUT Out channel inside oscilloscope (and also external cross talk) and may then rotten result specially when dynamic range in test is more than just some tens of dB)

ETA: Added (below) also some example done with SDS1204X HD
Some around 84kHz resonator.  In this image span is 1kHz and frequency steps 2Hz. (501points,  corresponds 96607 points/decade)
Also in this case there is lack of enough resolution. 2Hz frequency hopping is way too rough to reliable detect this down peak minimum level.
but just for fun.
But unfortunately I have never seen these kind of things made using example Rigol's BodePlot - why?

[jonpaul]

Just seeing this interesting post.

Been doing Bode plots since 1960s, and seems a few misconceptions  A few moments please,  to clarify:

1. Any electronic system can be characterized by the frequency and phase response over the band of interest. Audio, RF, UHF, microwave, etc.

Generally the generator is at a Zo to match the system, eg 50 or 75 Ohm as is the analyzer. This is network analysis and often done with log (db) Y axis and lin or log X axis (freq)



2. A bode plot is specific to characterize the stability ofd a 2ond order system as in servos or control systems.

The location of the network poles and zeros determines the stability, and transient response, eg underdampled, critically damped or overdamped. 

Bode Plot is always log/log DB vd log F.

http://web.mit.edu/2.14/www/Handouts/PoleZero.pdf

Asymptotic Slope of the Bode plot   is +/- 6 db per pole.

3. The freq plots are less useful than an S-Plane  pole/zero location diagram.

4. Rather than the filters shown a better Bode plot demo is  multipole low pass like a 4-6 pole elliptic or butterworth, and  a low Q RLC circuit.
https://www.analog.com/en/technical-articles/a-filter-primer.html
https://www.slideserve.com/lynley/feedback-control-systems-fcs
Enjoy,

Jon

[blackdog]

Hi,

Thanks for all the input!

I started this topic to give users of modern scoops a tool to learn the basic functions of Bode Plot and that it is similar to others who use the same tool.
So I made a filter that is reproducible and show that it is built neat and stable, the breadboard way is not a good way for stable measurements.
And yes that is important to me, there is nothing as annoying as measurements that constantly vary in value where they should not.

So this topic is Bode Plot 101
I show how to make neat measurements when using the suggested filter, so beginners have something to compare their own measurements/settings with.

It shows a few examples of how to adjust the scale of your measurements to get a better picture.
A well built filter and well set Bode Plot function on the scope will give pictures without any noise or interference in the measurement setup I have shown here.
In my experience, you have to invest quite a bit of time in your measurement instruments to make certain measurements without making too large measurement errors.

There is little wrong with RF-Loop's photos, but this is clearly not a 101 Bode Plot. 
The measurements shown by RF-Loop only really make sense to others if you know the properties of the DUT and the measurement conditions.
This is no different from all kinds of other measurement setups where you explain something...

Kind regards,
Bram

[rf-loop]

Hi,

Thanks for all the input!

I started this topic to give users of modern scoops a tool to learn the basic functions of Bode Plot and that it is similar to others who use the same tool.
So I made a filter that is reproducible and show that it is built neat and stable, the breadboard way is not a good way for stable measurements.
And yes that is important to me, there is nothing as annoying as measurements that constantly vary in value where they should not.

So this topic is Bode Plot 101
I show how to make neat measurements when using the suggested filter, so beginners have something to compare their own measurements/settings with.

It shows a few examples of how to adjust the scale of your measurements to get a better picture.
A well built filter and well set Bode Plot function on the scope will give pictures without any noise or interference in the measurement setup I have shown here.
In my experience, you have to invest quite a bit of time in your measurement instruments to make certain measurements without making too large measurement errors.

There is little wrong with RF-Loop's photos, but this is clearly not a 101 Bode Plot. 
The measurements shown by RF-Loop only really make sense to others if you know the properties of the DUT and the measurement conditions.
This is no different from all kinds of other measurement setups where you explain something...

Kind regards,
Bram

Yes, this your original post and circuit is well ok for its purpose as you have already told. Also I will not criticize it at all for its said purpose. 101...

It was such a good idea that I wanted to expand on the idea a bit.
Let's hope that someone would make a general purpose and versatile enough circuit(s) on the one board what can use for simple evaluate and compare BodePlot function from different manufacturers (oscilloscope) and their performance and functions. Something like oscilloscopes BodePlot evaluation board. My images was just random examples and without enough data about setups and circuits. Just only for give some idea...
When different manufacturer and models tests are done with the same DUTs, the results are also comparable. A simple RC circuit(s) alone is not sufficient for this purpose. You can't use it to evaluate real performance differences and limits, etc. They are quite good for learning and experimenting with the basic use and basic settings of BodePlot. Especially if the users are at the very basic level, right from the start.

[iMo]

Perhaps you have to define a couple of parameters people do consider important when using the Bode plots.
Then you may design the test circuit..