Author Topic: Common ground ringing  (Read 2831 times)

0 Members and 1 Guest are viewing this topic.

Offline garethwTopic starter

  • Regular Contributor
  • *
  • Posts: 88
  • Country: gb
Common ground ringing
« on: September 01, 2019, 05:56:51 am »
I have assembled a buck converter using an LM2594 DIP. It uses a 100uH inductor and has a switching frequency of 150kHz. The circuit is assembled on a breadboard and performs well.
I then tried the same thing with an MCP16301 (SOT-23 soldered to DIP breakout board). Assembling the circuit from the data sheet using a 22uH inductor, again on a breadboard. This IC uses a switching frequency of 500kHz.
Both circuits step down 12V DC to 5V. Both drive a small load of 60mA but I have tried higher to see if it helped.

My problem is ringing from the MCP16301 into ground and across the whole board. The  first circuit had no issues at all.
Is this a classic breadboarding limitation? If so, I’m guessing the frequency limit (of the breadboard) must be between the 150kHz and 500kHz of the two circuits. I’m surprised the first circuit had no issues at all and the other is awful. The only major difference is the switching frequency.


Sent from my iPhone using Tapatalk
Father
Husband
MENG Electronic Engineering student
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Common ground ringing
« Reply #1 on: September 01, 2019, 01:53:30 pm »
As in, you're probing anything, including even just ground itself, and you pick up the same ringing?

That's common mode noise, and indicates a voltage drop in ground itself, where the RF current (the ringing) is flowing through ground, inducing the voltage drop.

Shrink your switching loop.  Use closer positions and avoid AC currents in and out of the ground rail.

Reminder that capacitors act as AC short circuits, and inductors act as opens.

A typical case for a buck converter might be:
1. Chip plugged in, GND jumpered to the ground rail.
2. Bypass cap between VIN and GND, right in front of the chip.
3. Diode between GND and SW, beside the chip.  Jumper from SW pin to diode.
4. Inductor from diode to output filter cap.

#1 isn't a big deal in and of itself, because the regulator doesn't draw much ground current, if it is using an external diode.  (Both devices you mentioned require external diodes.)

#2 can be okay in and of itself, but where it fails is #3.  The switching loop includes the VIN bypass cap, regulator and diode.

If the diode were moved to a standing position and placed in front of the regulator itself (if there's room to), then the switching current flows all right in front of the regulator.

You should then use a modest inductance (~0.1uH, or a ferrite bead maybe) between the VIN pin, and the breadboard VIN (where you probably have a bigger like 100-1000uF electrolytic), to avoid switching currents being drawn in parallel with the bypass cap.

The output filter loop (inductor and cap) doesn't matter as much, because the inductor has high impedance at RF.  It can still carry some, in which case the cap should again be returned to the diode anode, or VIN bypass cap ground, directly.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: garethw

Offline garethwTopic starter

  • Regular Contributor
  • *
  • Posts: 88
  • Country: gb
Re: Common ground ringing
« Reply #2 on: September 01, 2019, 02:12:21 pm »
Thank you for your reply. Answering your first question; Yes, I get a similar ringing everywhere, ground, 12v rail, 5v output of the smps, even the outputs of the LDOs. The common mode noise you mention sounds spot on. I will go through your suggestions and see if I can make some improvements.


Sent from my iPhone using Tapatalk
Father
Husband
MENG Electronic Engineering student
 

Offline fourfathom

  • Super Contributor
  • ***
  • Posts: 2005
  • Country: us
Re: Common ground ringing
« Reply #3 on: September 01, 2019, 02:23:12 pm »
Some of this could also be scope probe ground connection noise pick-up.  Clip your probe ground to the probe tip and then probe your circuit ground.  If you see a signal then your scope ground lead is too long and you are seeing inductive coupling.  There are probing methods available to reduce this effect.
We'll search out every place a sick, twisted, solitary misfit might run to! -- I'll start with Radio Shack.
 

Offline noname4me

  • Regular Contributor
  • *
  • Posts: 93
Re: Common ground ringing
« Reply #4 on: September 01, 2019, 09:51:00 pm »
How many probes are you using?

Are you connecting the ground clip on each one? To different ground locations on the board?

Sent from my SM-N960F using Tapatalk

 

Offline garethwTopic starter

  • Regular Contributor
  • *
  • Posts: 88
  • Country: gb
Re: Common ground ringing
« Reply #5 on: September 01, 2019, 10:10:37 pm »
I'm currently using a single probe at a time. I have two probes, one with a standard ground lead and the other with a small spring attachment near the tip. I get lower ringing readings when using the probe with the small ground pin.

Father
Husband
MENG Electronic Engineering student
 

Offline schratterulrich

  • Regular Contributor
  • *
  • Posts: 50
  • Country: at
    • Elektronik & Layout
Re: Common ground ringing
« Reply #6 on: September 02, 2019, 12:02:48 pm »
Common mode noise generates currents in the probe cable shield and probe ground loop. This is a common problem with single ended oscilloscope measurements.
I think common mode noise is not only caused by currents in the gnd system but also by fast voltage signals with short rise times or digital signals which are unavoidable capacitively coupled to the environment and thus shift common mode potential.
Perhaps the difference of the switching frequency of your ICs is only 3 to 1 while the rise time increases tenfold...

Howard Johnson wrote in his book "High-Speed digital design, A Handbook of Black Magic" how to avoid common mode noise in oscilloscope measurements in detail.

The same recommendations from the book to mitigate these effects are reproduced in the following pdf
http://ece-research.unm.edu/jimp/650/slides/measurements1.pdf

My suggestion is:
For measurements on the power supply a simple and effective way is to use a coax cable instead of a probe. This is possible due to the low impedance of the power supply. A short GND connection (shield of coax) is essential. Also the quality of the cable is important (shielding effectiveness, shield impedance)

You will get less noise with this method.

For very low frequency measurements you can use the standard 1 Meg input impedance of the scope. With AC coupling you can optimize the vertical setting to see the ripple in more detail. With AC coupling you will see frequencies down to about 10 Hz I think. Switch on the BW limit on your scope as with this configuration you can't measure high frequencies anyway.

For high frequency measurements you have to use a 50 Ohm termination at the scope input. Be aware not to overload the 50 Ohm termination. If your scope has an internal 50 Ohm termination you can use AC coupling. Then your lower frequency limit is about 200 kHz. (At least with my scope)
 
The following users thanked this post: garethw

Offline garethwTopic starter

  • Regular Contributor
  • *
  • Posts: 88
  • Country: gb
Re: Common ground ringing
« Reply #7 on: September 02, 2019, 01:57:30 pm »
Thank you for that pdf link. It looks very interesting. The book that you mentioned, High Speed Digital Design, is in my Uni library so I will take out a copy. [emoji3]


Sent from my iPhone using Tapatalk
Father
Husband
MENG Electronic Engineering student
 

Offline garethwTopic starter

  • Regular Contributor
  • *
  • Posts: 88
  • Country: gb
Re: Common ground ringing
« Reply #8 on: September 05, 2019, 09:05:53 pm »
After reading the PDF linked by schratterulrich I came across this tip for reducing noise pickup due to probe shield current:

Quote
Use a 1:1 probe instead of a 10:1 probe.
The 10:1 does not attenuate the shield voltage effect.
Rather the attenuation of the logic signals by 10x amplifies makes the
noise appear 10 times larger

Setting the probe and scope to x1 did indeed reveal the noise to be much lower than was first thought.
A very valuable lesson learnt!
Father
Husband
MENG Electronic Engineering student
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Common ground ringing
« Reply #9 on: September 05, 2019, 11:03:24 pm »
Wat.

No, what you're seeing is the probe bandwidth itself is tiny, usually about 7MHz.  Just punch in a low bandwidth setting on the scope!

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

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Common ground ringing
« Reply #10 on: September 05, 2019, 11:14:12 pm »
In fact, referring to page 16, I just recreated the diagram exactly as shown.  Generator: function generator, 50 ohm source, 3m cable, signal pin to scope probe ground ring, probe set to 10x, scope set to 200mV/div.

Note, that's with the probe tip open circuit (this is justified below), clippy tip and ground clip removed.  Just the coaxial pin and ring exposed.

I see about 20mV.  The waveform shows it's approximately flat in bandwidth, probably either residual coupling from the open circuit tip, or actual leakage in the probe shield.

The signal amplitude by the way is 2.5V, so the shielding effectiveness is at least 40dB.

If I touch the probe tip pin, the signal becomes considerably stronger, and may become peaky as well.  This is due to my body capacitance and resistance acting to ground the probe tip, at some frequencies or all, while the shield voltage moves independently of it.  This proves that 1. the probe tip is moving with the shield when it's open-circuited, and 2. there is real, and substantial, voltage dropped across the probe shield, which does couple with the signal line within.

I sincerely hope that PowerPoint isn't based on a real book.  I've heard loopy things about the Johnson book (but haven't reviewed it myself to see just how loopy it is), but this is just flat out, and very easily demonstrably, wrong, information.

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

Offline schratterulrich

  • Regular Contributor
  • *
  • Posts: 50
  • Country: at
    • Elektronik & Layout
Re: Common ground ringing
« Reply #11 on: September 08, 2019, 02:23:14 pm »
I made the following measurement setup:


It shows the equivalent circuit diagram of the coaxial cable - the screen and the inner conductor are inductively coupled (L1, L2). however, it also shows an ohmic resistance (R1). And this is not coupled into the signal.
Simulation Results:

The difference signal of the coax shield and inner conductor is measured with channel 1 at the oscilloscope end.
I measured the signal across the shield with a 1:1 probe with channel 2.
The measurements show the same result as the simulation.



The signal of the coax inner conductor stays is at 1,5 mV indepentent of frequency. (yellow trace)
The signal across the shield is also 1,5 mV at low frequencies but increases at higher frequency (inductance) (blue trace)

And that is exactly what Johnson claims in the book.
« Last Edit: September 08, 2019, 02:40:54 pm by schratterulrich »
 
The following users thanked this post: T3sl4co1l

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Common ground ringing
« Reply #12 on: September 08, 2019, 03:29:24 pm »
Also a good introductory demonstration of creating an EMC simulation. :)

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

Offline schratterulrich

  • Regular Contributor
  • *
  • Posts: 50
  • Country: at
    • Elektronik & Layout
Re: Common ground ringing
« Reply #13 on: September 09, 2019, 08:57:53 pm »
I searched the web for "transfer impedance" and found the following interesting link:
https://interferencetechnology.com/differential-transfer-impedance-of-shielded-twisted-pairs/

It shows how the signal of a coaxial cable can drive the shield into radiation!


I wanted to try this myself and connected the shield of a coaxial cable to port 1 of my VNA.
Then I connected the inner conductor to port 2.
Something like that


My guess was that if the cable length is in the quarter wavelength range, there will be a radiation resistance in addition to the ohmic resistance. This should lead to a larger coupling to the inner conductor...

The result shown as S21 parameter:


At 313 MHz the S21 (represents the shielding effectiveness in this case) peaks at -25 dB !!

My conclusion from this experiment is that the shielding effectiveness of a single braid cable can be very bad.
I haven't thought about this circumstance yet.

« Last Edit: September 09, 2019, 09:00:53 pm by schratterulrich »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf