I am building a half bridge SMPS with an output of 14V. It works ok, but with a lot of ringing in the output, for about 4V pk-pk! Though it came down a bit with tweaking and adding the snubbers, it is still about 600mV pk-pk, with a frequency of 20 to 25MHz.,
While looking at the waveforms, I had noticed that this is found in the ground as well! I am measuring at the secondary ground, with the scope probe and scope ground connected to the same point. I was ruling out the scope (low cost) as when I switched off the SMPS, there is no ringing. Is it the radiated noise or something that is picked up by the probe? This ringing is happening during switch off and switch on of the MOSFETs. I am attaching the waveforms.

I am really confused, and kindly request someone to explain what this is, and also help me solve this.
Thanks for your time.

Ground Waveform (!) during MOSFET Turn On 600mV pk-pk 20-25MHz
Uploaded with ImageShack.com

Ground Waveform (!) during MOSFET Turn Off 300mV pk-pk 20-25MHz
Uploaded with ImageShack.com

 

A properly functioning SMPS requires careful layout on a printed circuit board. This is probably a breadboard with conductor impedances that are too high. It would be helpful to observe the relationship between the "ground bounces" and the switching edges.

 

It could be common mode noise through the scope.

 

A properly functioning SMPS requires careful layout on a printed circuit board. This is probably a breadboard with conductor impedances that are too high. It would be helpful to observe the relationship between the "ground bounces" and the switching edges.

I have attached the switching waveforms alongwith the noise in Ground and Output. Surprisingly, the Noise in Ground and Output is almost identical to the looks leading to CM noise as suggested by mik3.

All the yellow traces are rectifier output 10V/div, and the red ones are Noise 100mV/div in either Gnd or Output Voltage as mentioned in the caption of the image.

Output Noise with diode output
Uploaded with ImageShack.com
Output Noise during TurnOn
Uploaded with ImageShack.com
Output Noise during Turnoff
Uploaded with ImageShack.com

Gnd Noise with diode Output
Uploaded with ImageShack.com
Gnd Noise during TurnOn
Uploaded with ImageShack.com
Gnd Noise during TurnOff
Uploaded with ImageShack.com

The power section is a pcb with wide traces almost 8-10mm wide, on a 70micron copper thickness! I started off ambitiously for about 30A! My bad layout(?). The control section is wired in a bread board.

Is it possible to have a common mode filter at the DC output? Will it be possible to identify the source and eliminate this at the source?

 

Are you using low noise probing techniques for the measurements?

If not use them because EMI can induce those signals due to the big loop of the probe GND lead.

Prepare your low noise probing and then connect the probe tip to its GND. Move the probe around close to the PCB without touching it to see if there is any pick up.

If not, touch the probe tip to various points on the GND track while the probe tip is connected to the probe GND. If you see those signals it means is common mode through the oscilloscope.

 

It sounds like Mik3 has it, but a soft recovery diode might also help.

 

I am not using any low noise probing techniques! Lol! Learning it! I am reading up on this. Without any low noise probing, I moved the probe tip and ground connected together near the pcb without touching. Near the diodes, there is 100mV pk-pk, and when I moved farther it reduced.
On the ground, with probe tip and ground connected it is 200mV pk-pk, and it is there all over the ground! Bang On!

Originally I had put a 10Ohm/10000pF snubber across the secondary of the transformer, which I changed to 18Ohm/1000pF snubbers across the diodes. The noise came down to 300mV pk-pk from the earlier posted 600mV.

I am trying to get a cable ferrite (not available here locally) to put it on the probe. In the meantime, I connected a 10K resistor to the probe tip and measured the noise to be 40mV pk-pk!

Are these methods of putting a cable ferrite near the probe end, or connecting a 10K resistor to the probe are alright for low noise probing? Is the 10K resistor with 10pF offering attn to the noise?

I have used P3717-AL combination line filter choke (got a couple of samples!) to the line input. This gives less than 5db CM Attn at 10MHz (maximum in the graph). All the noise here is about 20MHz., and will the noise improve with a different choke?

I am using MBR20200 diodes which was easily available. Can you please suggest some soft recovery diode, if I need to change this?

 

Some of the noise probably is being picked up in the scope's ground leqad, but the best way to solve the problem is with properly designed/placed R-C snubbers. You need a snubber directly across the switching device and also the rectifier(s), and in some cases across the primary and secondary of the transformer windings. The lead length on the snubbers must be as close to zero as possible, even 1/2" of lead length will render them ineffective.

 

I have a RC snubber across the primary side of the transformer, nothing on the switches.

I had a RC across the secondary, which I removed and added an RC across the rectifiers. This one saved a quite a bit of power, with slightly better results. The snubber measurement calcs. yielded 680pF and 20 Ohms, and I am using 1000pF and 18 Ohms. Now with this snubber across the diodes, the ringing has reduced to a level, I am unable to measure the frequencies for an additional snubber across the secondary!
I would check and try to reduce the lengths of the snubbers. Of course, my layout is questionable; atleast I can see some distance to rectifier to transformer, and the LC and ground loop is big, which I am trying to reduce - cutting tracks, and putting short jumpers!