Author Topic: Looking for a Low Cost Way of Measuring DC-DC Converter Control Loop Response  (Read 2599 times)

0 Members and 1 Guest are viewing this topic.

Online nctnico

  • Super Contributor
  • ***
  • Posts: 17970
  • Country: nl
    • NCT Developments
Any 4-channel oscilloscope should be able to do it.
Put a current shunt in series with the load and use two channels to look at the difference across it (i.e. the current reaching the load).
That is not going to work because you'll be looking at milli-Volts on top of several Volts (or more). The resolution of an oscilloscope won't be enough. A better way is to put the current shunt at the ground point with one end and use 2 channels: one for voltage and one for current.
Not if you use the math function.
No, because you are still limited to the ADC resolution and dynamic range of your oscilloscope. If you need a total range of 5V then each bit of the ADC is 19mV. If you use a 0.1 Ohm shunt then each bit of the ADC is equal to 190mA.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline exe

  • Supporter
  • ****
  • Posts: 1329
  • Country: nl
  • self-educated hobbyist
Connecting one end of the shunt to ground only works moderately well if you are careful to ensure the ground point for the scope is right on the shunt. The ground bounces around too much for millivolt level sensing to work unless the ground for the measurement is well placed.

That's why I'm gonna use analog discovery with differential inputs :).
 

Offline Kevin.D

  • Frequent Contributor
  • **
  • Posts: 278
  • Country: england
It's possible to get a bode plot of the feedback loop using a scope and function gen if you have the time.
search app note AN-1889.
Also I remembered  Robert Bolanos has a video also showing this .


Regards.
 

Offline Mechatrommer

  • Super Contributor
  • ***
  • Posts: 9204
  • Country: my
  • reassessing directives...
Any 4-channel oscilloscope should be able to do it.
Put a current shunt in series with the load and use two channels to look at the difference across it (i.e. the current reaching the load).
That is not going to work because you'll be looking at milli-Volts on top of several Volts (or more). The resolution of an oscilloscope won't be enough. A better way is to put the current shunt at the ground point with one end and use 2 channels: one for voltage and one for current.
Not if you use the math function.
No, because you are still limited to the ADC resolution and dynamic range of your oscilloscope. If you need a total range of 5V then each bit of the ADC is 19mV. If you use a 0.1 Ohm shunt then each bit of the ADC is equal to 190mA.
what about AC coupled?
if something can select, how cant it be intelligent? if something is intelligent, how cant it exist?
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 17970
  • Country: nl
    • NCT Developments
Any 4-channel oscilloscope should be able to do it.
Put a current shunt in series with the load and use two channels to look at the difference across it (i.e. the current reaching the load).
That is not going to work because you'll be looking at milli-Volts on top of several Volts (or more). The resolution of an oscilloscope won't be enough. A better way is to put the current shunt at the ground point with one end and use 2 channels: one for voltage and one for current.
Not if you use the math function.
No, because you are still limited to the ADC resolution and dynamic range of your oscilloscope. If you need a total range of 5V then each bit of the ADC is 19mV. If you use a 0.1 Ohm shunt then each bit of the ADC is equal to 190mA.
what about AC coupled?
But then you introduce a high-pass filter from the AC coupling which makes the measurement harder to interpret. AND if the load step causes a significant amount of ringing and/or voltage drop you'll overdrive the oscilloscope input anyway causing all kinds of weird effects which aren't there.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Online David Hess

  • Super Contributor
  • ***
  • Posts: 10010
  • Country: us
  • DavidH
Yes, the load step response works and is also the easiest but you need an oscilloscope which can differentiate the response and then return magnitude *and* phase response from the FFT which the mentioned Keysight and Siglent oscilloscopes won't do.

I'll try to export raw data from the scope and do analysis in Python.

This Tektronix article gives an idea of what is required.  I have considered buying one of their old TDS DSOs which support magnitude and phase FFTs just to do this:

http://www.testunlimited.com/pdf/an/55W_8815_2_2009.01.07.10.50.04_2765_EN.pdf
« Last Edit: August 10, 2018, 12:06:13 am by David Hess »
 

Offline rx8pilot

  • Super Contributor
  • ***
  • Posts: 3513
  • Country: us
  • If you want more money, be more valuable.
It's possible that a InfiniiVision 1000 X-Series scope could work for you, it does gain and phase up to 20 MHz if you get a "G" model.

This can be a challenge on a scope because of the low level signal that is typically buried deep in the noise of the switching elements. The latest software appears to be filtering/averaging, but I cannot see what it is doing in the background. The dynamic range is considerably better than the original application.

I have been able to rough it manually using averaging and filter math to get what I needed. I manually step through the frequencies and measure the gain and phase with the scopes measurement tools. In general, you are looking for two distinct points - the phase margin and the gain margin which means you don't have to analyze in 1hz steps. This is rather slow and somewhat limiting - but cheap. Forum member Robert Bolanos demonstrates this in his YouTube video posted a few replies up.

The next step I took was to use a differential amplifier Preamble/LeCroy DA1855 to better see and filter the low level signal. That requires some time compensation to get an accurate phase measurement but the benefit is that the low-level signal is gained to a point where the scope is not on the edge of its front end performance. This is not critical, but helpful. The main use I have for my 1855A differential amplifiers is to measure current shunts.

Next for me was getting a VNA. I found an HP 3577A that was originally used in the Paleolithic period by cavemen trying to analyze the energy transfer at the contact point of two sticks being rubbed together. I searched for quite a while and ended up with a mildly broken one for $200 plus about $75 to fix it. I added a Jensen video isolation xfrmr VB-1BB that was another $20 off of eBay. I am still learning how to get this setup and deliver trustworthy measurements. So far, it seems rather fantastic relative to my previous efforts. It goes down to 5hz and has 1hz RBW, averaging, etc.....lots of relevant features for FRA. The alternative I was looking at was suggested by power integrity guru Steve Sandler was a Keysight E5061B with low-frequency option. This is an amazing tool - but with an appropriately amazing price. The 3577A used as an FRA is a great bang for the buck as long as you have a bench big enough and strong enough to hold it.


Factory400 - the worlds smallest factory. https://www.youtube.com/c/Factory400
 

Offline chris_leyson

  • Super Contributor
  • ***
  • Posts: 1296
  • Country: wales
Quote
That's why I'm gonna use analog discovery with differential inputs :).
I'm hoping to have a stab at measuring the loop response of an offline flyback using an Analog Discovery, time permiting. It's got enough dynamic range to do the job but I don't know how well the "back end" software will deal with the unwanted switching frequency. Will have to try it out and see.
Robert Bolanos has some excellent tutorials of flyback control loops, he uses a cleverscope for the measurements.
 

Online _Wim_

  • Frequent Contributor
  • **
  • Posts: 751
  • Country: be
This guy made a excellent description how to perform gain phase measurement using a low-cost pico scope and the FRA software tool:
http://www.simprojects.nl/images/Gain_Phase_measurements.pdf

He also made a very detailed description how to make your own injection transformer for less than 30€:
http://www.simprojects.nl/images/DIY_signal_injection_transformer.pdf

I have made 2 of these, one using the large core (Coilcraft as suggested in the article), and one using a slightly smaller core (Würth WE-CMBNC XXL) trying to achieve a bit higher frequency response. Attached are the two results.
 
The following users thanked this post: Wolfgang

Offline JohnPi

  • Regular Contributor
  • *
  • Posts: 79
  • Country: us
I'm trying to do this also. My plan is to use an injection transformer in the feedback loop; drive it with a signal generator, and use a scope to capture input & output waveforms. By capturing the scope data for each waveform (over precisely 1 cycle), I can correlate (dot product) with a sine & cosine and get its phase & amplitude. atan2(sig2, sig1) gives me the phase. For a high gain loop, the signal will be small, but I can get within a few degrees with 4 mVpp on a 200 mV/div range.

This method will be slow (sort of like a spectrum analyzer). The Bode 100 https://www.omicron-lab.com/products/vector-network-analysis/bode-100/ can sweep from 100 to 1 MHz in ~10 s. I don't know if it uses discrete frequency steps, or somehow does a smooth sweep and calculates something. They use a 24-bit ADC inside (see teardown), so can get good SNR.
 

Offline Hydron

  • Frequent Contributor
  • **
  • Posts: 404
  • Country: gb
The "scope" based units (e.g. cleverscope, keysight/siglent bode plot mode, pico FRA software) use FFTs to do this, not a direct measurement. This allows for low level signals (which you will get in this case) to be pulled well out of the noise, as you're just looking at the magnitude and phase of the FFT output at that frequency, ignoring all the rest. If you're grabbing the data off the scope to a PC in the first place, just do the FFT on the PC as well (with a decent record length for good FFT resolution, though this will slow the sweep).
 

Offline vealmike

  • Regular Contributor
  • *
  • Posts: 192
  • Country: gb
Sig gen, scope, injection transformer.
http://www.ti.com/lit/an/snva364a/snva364a.pdf

Some of the Keysight scopes with built in sig gens have a bode plot feature, and will do the hard work for you.
If you don't have one of these, just adjust the gen & take the readings manually.
 

Offline mzzj

  • Frequent Contributor
  • **
  • Posts: 874
  • Country: fi
This could be one ready-to-go option if audio frequencies are enough for you:
http://www.holmacoustics.com/holmimpulse.php
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf