Author Topic: Reducing Noise in Gain & Phase Measurement of DC-DC Converter's Control Loop  (Read 2179 times)

0 Members and 1 Guest are viewing this topic.

Offline bendrasTopic starter

  • Contributor
  • Posts: 17
  • Country: de
I am trying to measure gain and phase of the control loop of a DC-DC buck converter (see below for images depicting the setup).
The problem is that DC-DC converter's switching frequency (around 500kHz) gets coupled into the input and output sine waves which I am trying to measure (see below for images of the affected signals). 
I have tried moving measurement points away from the main inductor and using differential probes but that does not seem to help much if at all.

Does anybody know of a way to eliminate or at least reduce pickup of such noise?


Sketch of the Setup:



The Actual Setup:







Noisy Signal:


Noisy Signal Zoomed In:
« Last Edit: October 22, 2018, 10:04:29 pm by bendras »
 

Offline boB

  • Frequent Contributor
  • **
  • Posts: 307
  • Country: us
    • my work www
I think that a real gain-phase analyzer will have a swept filter as well as tracking generator so that would I think help a lot.

That's most likely a good reason that an oscilloscope is not used for this purpose.  Not saying that it can't be done though.

boB

Edit:   For example.....

http://www.ridleyengineering.com/hardware/ap310-analyzer/intro-ap-analyzer.html

The AP310 is a frequency response analyzer designed specifically to make sensitive gain-phase measurements in high noise environments, and is used to measure the following:

Transfer Functions
Loop Gain and Phase
Output Impedance
Audiosusceptibility or PSRR
Input Impedance
Measure large and small signals with its advanced analog and digital signal processing. Receive rugged and reliable results in all environments, including challenging power supply applications.

Measure individual passive power components to accelerate your design process, including:

Power capacitors
Power Inductors
Power Transformers
Leakage Inductance
Winding capacitance

HOW IT WORKS
The Frequency Response Analyzer is a sophisticated instrument that injects a sinusoidal test signal and measures the response of a system to that frequency on two return signals. It makes swept frequency response measurements that give magnitude and phase data plotted verses frequency. The frequency sweeps can be either logarithmic (for Bode-type plots) or linear.

A rugged and easy-to-use Windows interface allows the plotting of relative or absolute signal amplitudes. The analyzer uses the latest in analog and digital signal processing technology that provides an affordable, high-performance, high-productivity measurement instrument.

« Last Edit: October 22, 2018, 10:17:34 pm by boB »
K7IQ
 

Offline KIKi

  • Newbie
  • Posts: 8
  • Country: hr
Did you try soldering GND wire on the bottom layer (put GND wire on other side of the PCB)? You should move GND wire as far as you can from the SW pin (inductor).
 

Offline capt bullshot

  • Super Contributor
  • ***
  • Posts: 3033
  • Country: de
    • Mostly useless stuff, but nice to have: wunderkis.de
The probing technique looks quite inappropriate.
You should use the usual grounding method for the scope probes: Remove the ground clip, connect the probe GND directly to the circuit boards GND plane and the probe tip as short as possible to the measurement points.
Often done by winding a short piece of stiff wire around the GND sleeve and soldering that to the GND plane while pointing the tip to the test point. That should reduce the coupled switching noise significantly.
Safety devices hinder evolution
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 16545
  • Country: us
  • DavidH
The switching frequency should be asynchronous to the swept frequency measurement so trigger the DSO on the swept frequency and use averaging to remove the switching noise.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21606
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Classic appearance of common mode noise.  Connect the ground near where the signals come from, then make a common mode transformer by carrying one extra ground wire with the bundle of wires, through a suitable core, back to the main ground pin where the power wires connect.  The ground loop voltage through the extra wire (which would ordinarily be a shorted turn, but that ignores the fact that ground isn't, in fact, ground, everywhere on the board) induces a similar voltage in all the other signals, keeping them quiet(er).

Maybe.

Use a few turns on a large ferrite, usually a clip-on cable clamp style, but a toroid (ferrite, not powdered iron -- if it's not dark gray, and you don't know what it came from, assume it's not) will do fine, just a little inconvenient to loop your wires through.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline bendrasTopic starter

  • Contributor
  • Posts: 17
  • Country: de
Did you try soldering GND wire on the bottom layer (put GND wire on other side of the PCB)? You should move GND wire as far as you can from the SW pin (inductor).

I  have tried this out (please see pictures below) and it seems to have reduced noise quite a bit  :-+.
 

Offline bendrasTopic starter

  • Contributor
  • Posts: 17
  • Country: de
The probing technique looks quite inappropriate.
You should use the usual grounding method for the scope probes: Remove the ground clip, connect the probe GND directly to the circuit boards GND plane and the probe tip as short as possible to the measurement points.
Often done by winding a short piece of stiff wire around the GND sleeve and soldering that to the GND plane while pointing the tip to the test point. That should reduce the coupled switching noise significantly.

I have tried this out as well (please see pictures below) and it seems to have also reduced the noise quite a bit  :-+.
 

Offline bendrasTopic starter

  • Contributor
  • Posts: 17
  • Country: de
Classic appearance of common mode noise.  Connect the ground near where the signals come from, then make a common mode transformer by carrying one extra ground wire with the bundle of wires, through a suitable core, back to the main ground pin where the power wires connect.  The ground loop voltage through the extra wire (which would ordinarily be a shorted turn, but that ignores the fact that ground isn't, in fact, ground, everywhere on the board) induces a similar voltage in all the other signals, keeping them quiet(er).

Maybe.

Use a few turns on a large ferrite, usually a clip-on cable clamp style, but a toroid (ferrite, not powdered iron -- if it's not dark gray, and you don't know what it came from, assume it's not) will do fine, just a little inconvenient to loop your wires through.

Tim

At the moment I do not have a ferrite core handy, but I will get one and report the findings  :)
 

Offline bendrasTopic starter

  • Contributor
  • Posts: 17
  • Country: de
Classic appearance of common mode noise.  Connect the ground near where the signals come from, then make a common mode transformer by carrying one extra ground wire with the bundle of wires, through a suitable core, back to the main ground pin where the power wires connect.  The ground loop voltage through the extra wire (which would ordinarily be a shorted turn, but that ignores the fact that ground isn't, in fact, ground, everywhere on the board) induces a similar voltage in all the other signals, keeping them quiet(er).

Maybe.

Use a few turns on a large ferrite, usually a clip-on cable clamp style, but a toroid (ferrite, not powdered iron -- if it's not dark gray, and you don't know what it came from, assume it's not) will do fine, just a little inconvenient to loop your wires through.

Tim

At the moment I do not have a ferrite core handy, but I will get one and report the findings  :)

I have tried out the common mode transformer suggestion and it seems that this has reduced the noise a tad in comparison with the other methods  :). See the images attached below.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf