# Ringing/Noise on smps output while BJT turns on

I am trying to build a 4watt SMPS with 5v output.The design uses BJT switching in quasi-resonant mode. Though I could get the 5V output, I am observing ringing in the output which is increasing the Vpp. I am pretty new to SMPS design and couldn't figure out what is causing this ringing. All that I could sense is, this ringing is caused by the leakage inductance of the secondary winding & the capacitance of the rectifier diode. Please correct me If i am wrong.

The blue waveform is between the transformer's secondary winding pins and the yellow is between the output bulk capacitor pins.

I tried with 20MHz bandwidth as well, nothing much changed.

The ringing is observed every time the BJT is turned on(that's my understanding from the waveform, apologies if I am wrong). This ringing is increasing the Vpp and I want to reduce it. I tried adjusting the AC side RC snubber across the primary winding, but it did not help.

Then, I tried to add a RC snubber across the rectifier diode thinking that this might be because of the secondary leakage and rectifier capacitance, but I did not observe any change. I want to understand what causes this kind of ringing on the output.

Operation; DCM BJT: NXP PHD13003C Output rectifier diode: 1N5819 Bulk cap: Low esr, rubycon 470uF/10V

Thanks everyone for your responses, Here is the layout

Links to components I used, Output bulk: Rubycon 470/10V Diode: 1N5819

Schematic:

Added one ceramic cap & increased the output capacitance. Now, I dont see any high frequency ringing the the peaks have also reduced to 50mV. Thanks for teaching me about ceramic capacitors

• OK. First check if the signal you're seeing is real. Are you probing with the long alligator ground clip attached to the scope probe which makes a very nice antenna, or using the proper ground spring? Like this one: electronics.stackexchange.com/questions/40420/… – peufeu Aug 3 '17 at 17:31
• Thanks for your quick response. I am using a ground spring and getting about 150mV of ripple @ 400mA. How can we verify if the signal is real. unfortunately dont have differential probe. – Raju Aug 3 '17 at 17:40
• If you are using the proper ground spring, then it's most likely real. Can you post a pic of the board, layout, etc? – peufeu Aug 3 '17 at 17:50
• I note that the falling edge, a $\mu\textrm{s}$ earlier, doesn't show the ringing despite enough time for some. You are likely right about BJT parasitics (and wiring) being part of the issue here. But tossing snubbers around randomly isn't how to go. When you turn a BJT on, you also pull on the BC capacitance, too, for example. We really do need board pics, schematic, and layout stuff. I +1 your question to give a few pts in the hope that helps in posting more pics. – jonk Aug 3 '17 at 18:36
• Could the output caps ringing, too. Do you by any chance have a low-ESR aluminium cap in parallel with a ceramic? Need specs here, of all the output caps (+layout) – peufeu Aug 3 '17 at 18:45

You didn't provide the all-important schematic, so we first have to guess at your circuit, then guess at the cause.

1. Bad scope probe grounding. Make sure the ground lead of the scope probe is directly tied to the negative side of the output cap, with the tip directly on the positive side.
2. Bad grounding and overall layout. These things are important in switching power supplies. You have to carefully visualize the two primary loop currents and make sure they are properly contained. I can't get into specifics without specifics of your circuit and layout.
3. Too high ESR output cap. You only say the output cap is "low ESR", which of course means nothing. At 470 µF and 10 V, it sounds like a electrolytic. Maybe the ESR is low for a electrolytic, but that's would still be too high. You should put as much ceramic capacitance physically close to the inductor and diode as you can manage. There should be at least a few 10s of µF. Keep the electrolytic, but it can be placed a bit further away.
4. No, the ringing is happening when the switch turns off, not on. From the waveform and your mention of "transformer", this is apparently a flyback design. The low-going pulse is when the primary is being driven, and current builds up. The sudden rising edge is where the primary is shut off and the secondary is now dumping the stored energy onto the output thru the output diode.
5. Make sure the diode is a Schottky. You didn't provide a link, so I didn't look it up. Not only will a Schottky significantly reduce loss at this low voltage, but the almost instant reverse recovery time is very useful.

## Response to schematic

Ah, so it is a flyback design, and the input is line power.

Things look generally reasonable, although there are a few issues:

1. The snubber on the line side could be wasting power. You want to let as much of the energy stored in the magnetic core be delivered out the secondary as possible. You don't really want a snubber on the primary, just something that clips the voltage to what Q2 can handle. Other than that, you want the voltage at pin 2 to go as high as it want when the switch is turned off.

2. Q2 should be rated for substantially higher C-E voltage than what it is switching. That minimizes the need to waste energy each pulse clipping the kickback of the primary. With high enough rating, you can often leave off any clipping.

3. I can't read what the D2 Zener voltage is, but possibly 5 V from the net label. This is a wasteful way to regulate the output voltage. Feedback thru a opto to shut off the oscillations when the rectified DC gets to a certain level will waste less power.

Then you can also regulate to just above what U3 needs to keep its output regulated. With 5 V in and 3.3 V out, you're down to 66% efficiency due to U3 alone. You should be able to get 80% end to end with some care. Below 70% is pretty crappy.

4. The transients are probably due to the direct conduction of the secondary thru D3 and D2 once C2 gets charged up enough. Again, a more sensible output regulation strategy would get around this completely.

My usual trick for low power flyback converters is to put a PNP transistor around a LDO to detect when the LDO input is one B-E drop above its output. There are plenty of LDOs that can do a few 100 mV headroom, and 700 mV or so is a good tradeoff that allows for dips but doesn't cause too much dissipation.

Here is a snippet of a schematic where I used this trick to make a linearly regulated 5 V on the output of a flyback converter:

The raw rectified DC on the output is created by the secondary of the transformer, D13, C40, and C41. Note the use of the electrolytic C40 for the bulk storage and the ceramic C41 for low impedance at high frequencies.

The main trick shown here is Q9 around IC15. When the input of the LDO (IC15) goes one junction drop above the output, Q9 is turned on thru R54. That turns on the LED in the opto-isolator IC16. The output of that going low kills the oscillations driving the primary side switch, Q8.

The roughly regulated 5.7 is also used elsewhere. This can be useful for lighting LEDs and the like where some ripple and voltage slop is acceptable. By using that when possible, it keeps the current requirements for the nicely regulated 5 V lower, allowing the use of a smaller LDO.

A 6N137 opto-isolator is overkill here. It was used in this design because it was needed in a few other places, and it wasn't worth saving a few pennies on this one but then require stocking another part.

• I have updated the schematic & the layout. From point 4, you say that fixing the RC snubber on the AC side should fix the ringing? – Raju Aug 4 '17 at 7:17
• Regarding point 4, the waveform in blue is across the secondary winding. You are saying that the voltage on the secondary starts building up only after the BJT is turned off? I was under the impression that the secondary voltage ramps up simultaneously with the current in the primary. Please correct me if I am wrong. – Raju Aug 4 '17 at 7:36
• thanks for your detailed explanation. point 4 really cleared my misconception. Regarding the removal of snubber, i will try that. BJT is about 450 on CE. Zener is 6.2v. I would like to understand what can help reduce that ringing. As you suggested , I tried the ceramic 22uf cap across the bulk and the Vpp dropped to 88mV without any ringing. I have some strict cost and size constraints so cannot change design much. – Raju Aug 4 '17 at 14:54
• Just woke up and enjoyed this. Learned something too. +1 Thanks! – jonk Aug 4 '17 at 18:23
• @olin, made few changes to schematic. Added a ceramic in parallel to the output bulk. This eliminated the high frequency ringing. Increased the output bulk from 470uF to 1000uF. Thanks for your suggestions. – Raju Aug 8 '17 at 7:51