Snubbing the Flyback Converter

Thread Starter

pini200

Joined Feb 15, 2016
14
hello
to diode output i add a Snubber (c and r )
c=23nF,R=50ohm
when i turn off and turn on the adapter is unstable
the voltage all the time is up and down
if i remove the capacitor on the system become stable
the schematic is attached(if needs more details i add)
the transformer 1600uH 1:10,
vin 220, vout 12 (up to 500mA)
note :this is NOT the primary Snubber (for mosfet transistor),
 

Attachments

ian field

Joined Oct 27, 2012
6,539
hello
to diode output i add a Snubber (c and r )
c=23nF,R=50ohm
when i turn off and turn on the adapter is unstable
the voltage all the time is up and down
if i remove the capacitor on the system become stable
the schematic is attached(if needs more details i add)
the transformer 1600uH 1:10,
vin 220, vout 12 (up to 500mA)
note :this is NOT the primary Snubber (for mosfet transistor),
Its more usual to find the snubber across the primary. Generally a diode charges a capacitor with the flyback peaks and the capacitor has a bleed resistor in parallel with it.

You can use the same type of snubber across the secondary rectifier to damp the peak reverse voltage. Its a handy trick to get away with the lower voltage ratings of Shottky barrier rectifiers. SB rectifiers usually have higher junction capacitance - sometimes that helps calm things down a bit.
 

Roderick Young

Joined Feb 22, 2015
408
I'm no expert on snubbing, but could you describe in more detail by what you mean by "unstable?" Is it possible to post scope traces? Of particular interest would be the ringing seen on the flyback transition with, and without, your RC snubber.

Does the output stabilize after the supply reaches its final output voltage?
 

Thread Starter

pini200

Joined Feb 15, 2016
14
I'm no expert on snubbing, but could you describe in more detail by what you mean by "unstable?" Is it possible to post scope traces? Of particular interest would be the ringing seen on the flyback transition with, and without, your RC snubber.

Does the output stabilize after the supply reaches its final output voltage?
i add pictuce
(not the output never stabilize with snubber)
 

Roderick Young

Joined Feb 22, 2015
408
Interesting! I have no good answer at the moment. The supply output is swinging at a very low frequency, possibly just to replace charge that has leaked off the main filter capacitor shown in your circuit. It could be that your feedback scheme lets just one flyback pulse through at this low frequency, and that this makes the output voltage zoom to over 12 volts (only 11.x is seen on the scope at this slow speed). Then, perhaps due to a long time constant in the feedback circuit, it takes a long time to generate the next pulse. We'd have to see the complete circuit, including feedback, to decide on that.

Possibly your snubber capacitor burned out, and has a parallel resistive element to it. Might want to check that capacitor on the highest scale of your ohmmeter to see if there is any measurable conductivity. A capacitor intended for a snubber must not only withstand the peak voltage applied, it must also withstand the surge current. I made the mistake of using a tiny 1206 ceramic capacitor in a recent circuit - that soon began to display resistance, and glowed yellow-hot when it finally burned out. I may have made the problem worse by applying too much heat during soldering. They say that dipped silver mica is the best snubbing capacitor, but those are expensive, so I'm now using a physically large foil-mylar capacitor of about 16 nF.

Also, how did you calculate the 23 nF value for your snubbing capacitor? If memory serves, I remember capacitors on the order of 100 pF across a rectifying diode such as you have there. The only reason I have such a large capacitor is that I have a large equivalent inductance, low frequency, and high power.

Another experiment you could try is to put more of a load on the circuit (maybe a low-wattage car bulb, or computer fan), and see what effect that has. And yet another experiment would be to increase the value of your main filter capacitor. If doing either of these changes the output you see with your snubber in place, post new traces here, and people can comment.
 

MrAl

Joined Jun 17, 2014
6,979
Hi,

I have to agree that it looks like the main output cap is too small as compared to the snubber cap.
For an extreme example, if the snub cap was 1uf and the output cap was 1uf, we'd see about 1/2 the total AC voltage appear across the output cap. This should make it obvious that the snub cap has to be smaller than the main output cap. For an AC only analysis again, if we used a 1uf snub cap and 10uf output cap then we would only see 1/10 th of the voltage we saw with 1uf and 1uf. Going even higher should reduce those spikes even more.
This also assumes that the output cap ESR is low enough too, but it probably will be ok assuming the snub resistor is 50 ohms.
Note that the AC part of the voltage has a lot to do with it too though. If the AC part jumps up by 100v even if we have 1uf and 10uf we'll still see about 10v AC on the output, so we would need to go higher with the output cap value.
Doubling the output cap value while measuring the output should tell the story of what is needed.
 

ian field

Joined Oct 27, 2012
6,539
Hi,

I have to agree that it looks like the main output cap is too small as compared to the snubber cap.
For an extreme example, if the snub cap was 1uf and the output cap was 1uf, we'd see about 1/2 the total AC voltage appear across the output cap. This should make it obvious that the snub cap has to be smaller than the main output cap. For an AC only analysis again, if we used a 1uf snub cap and 10uf output cap then we would only see 1/10 th of the voltage we saw with 1uf and 1uf. Going even higher should reduce those spikes even more.
This also assumes that the output cap ESR is low enough too, but it probably will be ok assuming the snub resistor is 50 ohms.
Note that the AC part of the voltage has a lot to do with it too though. If the AC part jumps up by 100v even if we have 1uf and 10uf we'll still see about 10v AC on the output, so we would need to go higher with the output cap value.
Doubling the output cap value while measuring the output should tell the story of what is needed.
If its "hunting" the output reservoir/filter may be too big.

Books on servomech will contain some pertinent information.
 

Roderick Young

Joined Feb 22, 2015
408
If its "hunting" the output reservoir/filter may be too big.
....
I see the output jumping to 11.something volts, then decaying slowly. I would think that if there is a capacitor that is too big, it would be in the control loop, but not at the output. Also, if I'm reading the trace correctly, the "hunting" is happening over a 1-second timeframe. It really would help to see more details of the circuit, including how the output voltage is sensed.
 

Roderick Young

Joined Feb 22, 2015
408
Ah! If that schematic had been in the first post, I wouldn't have put down all the ramblings and speculation, which you can now ignore.

Try connecting the bottom of \(R_{BIAS}\) to \(O_{FB}\) instead of where you have it.

As a tip, if you have a schematic where signals are near to each other, it's better to connect them with a visual wire rather than a label (or in addition to a label). Makes it easier for a human like me to read.
 
Last edited:

Thread Starter

pini200

Joined Feb 15, 2016
14
Ah! If that schematic had been in the first post, I wouldn't have put down all the ramblings and speculation, which you can now ignore.

Try connecting the bottom of \(R_{BIAS}\) to \(O_{FB}\) instead of where you have it.

As a tip, if you have a schematic where signals are near to each other, it's better to connect them with a visual wire rather than a label (or in addition to a label). Makes it easier for a human like me to read.

i didn't understand why connect FB to Rabais you can explain me?
 
Top