Author Topic: TEKBOX TBJT02 + Analog Discovery 2 Network Analyzer  (Read 744 times)

0 Members and 1 Guest are viewing this topic.

Offline griffinadamsTopic starter

  • Newbie
  • Posts: 9
  • Country: us
  • I would have routed cardiac nerves differently.
    • Engenious Design
TEKBOX TBJT02 + Analog Discovery 2 Network Analyzer
« on: July 13, 2023, 07:56:45 pm »
Has anyone ever used an Analog Discovery 2 and an injection transformer for power supply loop stability measurements?
I am using the TEKBOX TBJT02, 1Hz - 5MHz, 5Ω terminated, 1:1 injection transformer.
I'm using this setup to measure the frequency response of a LM5157EVM-BST Evaluation Module and comparing it to TI's posted Bode plots in the datasheet for the EVM.
I'm am able to come up with something somewhat close, so I think it is working, but I had to turn up the waveform gen on the Disco to 1V. My colleague has concerns that injection voltage would end up changing the operating point of some supplies.

Has anyone else run into the Analog Disco 2 not outputting the correct voltage or maybe the injection transformer is attenuating the signal?  :-//

Any insight would be very much appreciated.
Griffin Adams | Senior Electrical Engineer
www.engenio.us
 

Offline thm_w

  • Super Contributor
  • ***
  • Posts: 7527
  • Country: ca
  • Non-expert
Profile -> Modify profile -> Look and Layout ->  Don't show users' signatures
 

Offline Weston

  • Regular Contributor
  • *
  • Posts: 221
  • Country: us
Re: TEKBOX TBJT02 + Analog Discovery 2 Network Analyzer
« Reply #2 on: July 13, 2023, 09:19:16 pm »
I created a set of student labs using the AD2 to measure the loop gain of a SMPS. It worked pretty well. As the injection signal was lowered the plot got more noisy, but I was able to get good results with only 100mV injected signal. For lower injected signal amplitude the plot will be noisy but it will still have the same general shape and inflection points. With this method the low frequency gain measurement will be pretty noisy due to the high gain, but you should get good resolution near crossover, even with low injected signal level.

Are you sure its all connected properly? Like shown below?

 

Offline griffinadamsTopic starter

  • Newbie
  • Posts: 9
  • Country: us
  • I would have routed cardiac nerves differently.
    • Engenious Design
Re: TEKBOX TBJT02 + Analog Discovery 2 Network Analyzer
« Reply #3 on: July 14, 2023, 03:05:49 pm »
Quote
Did you set the output for low impedance if you are using a BNC board?

I have it set to 50 ohms because I am using a 50 ohm coax BNC-to-BNC cable. I assumed that would be best case. The transformer has a BNC connection on the primary side. It seems like they match impedance in the transformer design. Here is the datasheet: https://www.tekbox.com/product/TBJT02_Manual.pdf

Quote
Are you sure its all connected properly? Like shown below?
I am like 99% sure I have it all connected up like you have shown. I have several other sources showing the same setup that I was working from.


Since my first post, I found this app note that seems to make a lot of sense for my situation. Using it, I was able to get a super clean bode with 100mV near the crossover, like you suggested, but still had to use 1V on the extremes.
https://scdn.rohde-schwarz.com/ur/pws/dl_downloads/dl_application/application_notes/gfm321/GFM321_1e_ControlLoopStability.pdf

Latest measurement:
« Last Edit: July 14, 2023, 03:12:25 pm by griffinadams »
Griffin Adams | Senior Electrical Engineer
www.engenio.us
 

Offline Weston

  • Regular Contributor
  • *
  • Posts: 221
  • Country: us
Re: TEKBOX TBJT02 + Analog Discovery 2 Network Analyzer
« Reply #4 on: July 14, 2023, 06:11:59 pm »
From that plot, it looks like you are getting pretty good results. For characterizing phase margin you really only need good data near the crossover frequency.

When loop gain is high the AC voltage at the lower side of the injection resistor is very low. With 40dB of loop gain, as shown in your plot, its going to be 100x smaller. There may be attenuation due the output impedance of the function generator (if you have it in high Z mode), and I am not sure if you are using 10x probes or not. Either way, its a small signal that is superimposed on any SMPS ripple.

One possible issue not yet discussed: Have you put the jumpers to AC coupling on the BNC breakout board? A DC offset on the signals can cause issues.
 

Offline griffinadamsTopic starter

  • Newbie
  • Posts: 9
  • Country: us
  • I would have routed cardiac nerves differently.
    • Engenious Design
Re: TEKBOX TBJT02 + Analog Discovery 2 Network Analyzer
« Reply #5 on: July 14, 2023, 06:23:56 pm »
From that plot, it looks like you are getting pretty good results. For characterizing phase margin you really only need good data near the crossover frequency.

When loop gain is high the AC voltage at the lower side of the injection resistor is very low. With 40dB of loop gain, as shown in your plot, its going to be 100x smaller. There may be attenuation due the output impedance of the function generator (if you have it in high Z mode), and I am not sure if you are using 10x probes or not. Either way, its a small signal that is superimposed on any SMPS ripple.

One possible issue not yet discussed: Have you put the jumpers to AC coupling on the BNC breakout board? A DC offset on the signals can cause issues.

Thanks for the feedback.

I do have them on AC coupled and 1x on the probes.

This particular power supply eval board has Bode plots in the datasheet. I'm using the Vin and load current for Figure 5-16.
https://www.ti.com/lit/ug/snvu739a/snvu739a.pdf?ts=1689319865318

Comparing my result to theirs, I think I can be reasonably confident in my measurement setup. All this for the goal of measuring the margins on my custom board with a completely different supply and be somewhat confident with those measurements.

Thanks again.
Griffin Adams | Senior Electrical Engineer
www.engenio.us
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf