Puzzling Bode Plot Artifacts

Thread Starter

SamR

Joined Mar 19, 2019
5,031
@Audioguru again @tautech et al. AG convinced me to prototype instead of breadboard, so I've been doing that but still getting some puzzling artifacts in the Bode Plot sometimes. That is the really strange part, it doesn't happen all the time. It seems that if I start cold the first pass is clean and then the successive passes pick up 2 artifacts. For example...
1645811806650.png
First pass @ low bandwidth
1645812259176.png

First pass @ higher bandwidth
1645812439792.png
Subsequent passes
1645812500596.png

I've been seeing this on other Bode Plots and assumed it was noise (still do) but not sure why I don't see it at all times. One artifact from 10-40Hz and another ~110-140Hz (60Hz harmonic? but nothing @ 60Hz). There is a very small blip @ 60Hz and would expect any harmonic to be less? Adding 1nF caps between V and GND and placing them on the op amp power pins makes no difference. I even twisted the power and GND cables from the PSUs with no difference noted. What is causing this artifact, and can it be controlled or eliminated? Would this somehow have to do with the capacitors charging minutely over time?
 

Audioguru again

Joined Oct 21, 2019
6,671
Your schematic is missing a supply bypass capacitor, a supply voltage, an input signal voltage and an opamp part number.
Is the 5.6V output the peak or the peak-to-peak?
Your Bode plot shows a cutoff at about 80Hz instead of the calculated 16Hz. 5.6V x 0.707= 4V.

A 1nF capacitor at the opamp power pins helps with radio frequencies, use 100uF for this low frequency circuit.
The input should use a shielded audio cable.
Maybe C1 is a polarized capacitor with its polarity backwards.
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
There is a 1nF supply bypass capacitor both at the PSU feed terminal block for the 3x7cm protoboard and from Pin 7 to GND of the TL071CP. The input signal voltage is a 0.1Vrms Sinus wave and the supply voltage on Pin 7 is 10V with Pin 4 grounded. C1 is ceramic. The input signal is fed from the scopes AWG with 50Ω BNC coax to a minigrabber. I did not cherry pick the parts so there would be some deviation from calculated especially with the capacitors used.

What really puzzles me is why I get a good plot and then noisy plots. As to using 100uF bypass caps for low frequency instead of 1nF, I will try that but was trying to avoid using electrolytics except for the 100uF C2 cap. Please explain your rationale for doing so.

Thanks, Sam
 

Audioguru again

Joined Oct 21, 2019
6,671
A 1nF supply bypass capacitor is good for 20MHz radio frequencies and higher.
A 100uF or more electrolytic supply bypass capacitor is almost always used for audio.
Frequently a 0.1uF ceramic supply bypass capacitor is parallel to the electrolytic 100uF one to prevent audio high frequency oscillation.

I don't know why your plots have low frequency level jumps (not noise). Maybe a 100uF supply bypass capacitor will prevent it.
I wonder if your scope cuts low frequencies which causes the low cutoff to be at 80Hz instead of 16Hz?

An electrolytic audio coupling (not bypass) capacitor causes small low frequency distortion.
A ceramic audio coupling capacitor also causes small low frequency distortion and is microphonic which might cause feedback squealing.
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
I added a single 100uF cap at the test modules PSU terminal block.
IMG_0956[1].JPGIMG_0955[1].JPG
Set the Bode Plot config for 0.1Vpp
1645891760557.png
The first pass was clean.
1645891843441.png
The artifacts again appear on subsequent passes.
1645891905426.png
It still rolls off ~110Hz instead of 16Hz... That is not my main concern though, the artifacts are what have puzzled me.
 

Audioguru again

Joined Oct 21, 2019
6,671
Your bode plot shows a load of 50 ohms. The minimum load on most opamps datasheets is 2000 ohms but a few can drive 600 ohms.
Also, you do not have an output capacitor so the 50 ohms load to ground is killing one of the output transistors with high DC current all the time.

I think the opamp is limiting the output current which messes up the frequency response.
You should always look at the output sinewave to see if it is messed up.
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
You should always look at the output sinewave to see if it is messed up.
It is clean.

Your bode plot shows a load of 50 ohms. The minimum load on most opamps datasheets is 2000 ohms but a few can drive 600 ohms.
So instead of the BNC coax minigrabber @50Ω I should use a HiZ scope probe for the input signal?
Edit: Also, the output is floating. Should it be loaded to GND? When I do, typically I use a 10MΩ resistor to GND. In this case I did not.
 
Last edited:

Audioguru again

Joined Oct 21, 2019
6,671
Maybe your scope has an input resistance of 50 ohms, not the mini-grabber. Can you switch the scope to have a high input resistance?
I always use a shielded audio cable to feed the 1M ohms input of my scope.
Sometimes my scope is looking at the signal across an 8 ohm load of a power amplifier.
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
Using a 1X probe with the config set for HiZ I get this. Note I increased the sweep to 1MHz as the span increased.
1645897597692.png
1645897658292.png
1645897712451.png
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
That screen is for the Bode Plot configuration using the scope's internal AWG. That is the setting for the AWG output which is a BNC connection on the scope completely separate from the inputs. The scope inputs are HiZ 10X inputs. C1 on the same Ei circuit input pin as the AWG output, and C2 on the TL071 pin 6 output. The AWG signal is connected using a 1X HiZ probe now.
IMG_0959[1].JPG
 
Last edited:

Thread Starter

SamR

Joined Mar 19, 2019
5,031
OK, it seems the internal AWG config controls do not have a sweep function. Apparently, that lies within the Bode Config. I can do an external AWG connection but will have to do a bit of manual reading and experimentation to proceed. Need to anyway so now is as good a time as any. Then I can compare the internal vs. eternal AWG controlled by the Bode Plot function outputs.
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
Just to update... Spent a few hours perusing the manuals, hours spent finding and installing USB cables, and a few more hours downloading, installing, and configuring the recommended National Instruments software only to find out that it costs over $1k. A bit above my budget... Then a few more hours uninstalling the software (quite a bit of it). There is another possible approach (without having to write my own custom C application) of installing LAN and utilizing the scopes built in Web Server. Even then the capabilities are a bit in the dark so I won't really know what it can do until after I get it up and running. Lots still to do so, maybe a few more days of working on it. I did find out by trial and error that the scope's Bode Plot function cannot connect directly to the external AWG by USB although it has the option of addressing a LAN address for an external AWG. We shall see after I pull the USB cables and install LAN cables, switch, and address everything to get connectivity. So, lots still to do.
 
Last edited:

Ya’akov

Joined Jan 27, 2019
9,069
Just to update... Spent a few hours perusing the manuals, hours spent finding and installing USB cables, and a few more hours downloading, installing, and configuring the recommended National Instruments software only to find out that it costs over $1k. A bit above my budget... Then a few more hours uninstalling the software (quite a bit of it). There is another possible approach (without having to write my own custom C application) of installing LAN and utilizing the scopes built in Web Server. Even then the capabilities are a bit in the dark so I won't really know what it can do until after I get it up and running. Lots still to do so, maybe a few more days of working on it. I did find out by trial and error that the scope's Bode Plot function cannot connect directly to the external AWG by USB although it has the option of addressing a LAN address for an external AWG. We shall see after I pull the USB cables and install LAN cables, switch, and address everything to get connectivity. So, lots still to do.
If you install just the LabView VISA connectivity software and the Siglent applications, it will provide enough information about control to do simple automation in the form of scripting (external to the programs, but over the interface). The web server allows viewing and control but no automation, just an SCPI dialog capability.

It is clumsy and poorly written. I use it for remote viewing but the control screen is a mess. You have to make a huge window for it to fit it doesn't resize.

1646046869060.png
If you don't open the controls it's pretty sane, and I find the ability to view the live display helpful. The live screen is sent over VNC and you can use a VNC client instead of the web page.

1646047052536.png
web page

1646047244703.png
VNC viewer
And finally, the SCPI screen, which is of limited value but can be quite handy for learning and testing the SCPI command set.

1646048663709.png
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
I received my ethernet hub and cabled my scope and AWG to it. Put the AWG address into the scope's Bode Plot config, cabled the module up, ran the plot function, and same thing. First sweep fine but in continuous mode subsequent sweeps have the same artifacts @ low freq. This time however it is a single area of artifact below 100Hz. Time to shoot an email off to Siglent.
Sweeping TL071CP with 20V single supply and 1V signal.
Single sweep
1646337532899.png

Continuous sweep
1646337486616.png
 

Attachments

Audioguru again

Joined Oct 21, 2019
6,671
Is the scope set to cut low and high frequencies?
The -3dB low frequency cutoff should be at 16Hz but is shown at 55Hz.
The -3dB high frequency cutoff should be at 300kHz but is shown at 35kHz.
Also, the phase shift should not suddenly reverse at 2kHz.
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
I have both Ch1 & Ch2 with the bandwidth limited to 20MHz to eliminate HI Hz noise so there is no scope cutoff involved here. The low cutoff is a bit high but then the resistors and capacitors are not precise values so I'm fairly happy with 55Hz instead of the precise calculated value of 16Hz. Yes, I was a bit puzzled at the HI freq rolloff being a bit low. The phase shift is over 50Hz so yes, it is rather steep. My big concern was the presence of artifacts in the plot. By allowing the Bode Plot function to utilize the external LAN connected AWG instead of the scopes internal AWG tells me it is not an AWG problem but rather a problem with the Bode Plot function itself. I have sent an email explaining the problem the Siglent NA distributor and am waiting for a reply. In the meantime, I have finished out the TI Handbook and am moving on to Floyd's Electronic Devices to firm up my transistor knowledge.
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
@Yaakov How the heck do you get the scope server up? I set the IP Config to manual and then put the IP Address in the Edge browser and nada... Til it times out... Ran windows diagnostic and it recognizes the IP Address as connected but not responding.
 

Ya’akov

Joined Jan 27, 2019
9,069
Do you have this Web Server menu on page 4 of the Utility menu? (It's just for setting a password, but f you don't maybe your firmware doesn't have it.)

1646358924899.png
 
Top