Snubber concerns for switching power supply / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Snubber concerns for switching power supply

Move Thread

LAN_403

cjk2

Wed Feb 05 2014, 04:39AM

Registered Member #51 Joined: Thu Feb 09 2006, 04:17AM
Location:
Posts: 263

I have designed a 400V DC power supply. The input is 400V from a power factor correction section. The output is also 400V isolated from the input. The winding ratio of the transformer is 45:56.

My problem is this: There is lots of ringing on the output rectifier diodes (D12, D14, D15, D16). I have read that this is caused by stray capacitance in the diodes and transformer windings plus leakage inductance in the transformer. The secondary of my transformer has about 30 uH of leakage inductance. This value can likely not be reduced as it is needed to enable zero voltage switching of the primary side.

The ringing is at 2.2 MHz with no snubber attached. I have looked at the voltage across all four diodes and they all display the same ringing.

It seems that an RC snubber is of no use here as it will have to dissipate about 10W per diode. This translates to a 4% loss of efficiency in my design which is too high.

Are snubbers really required on these diodes? Can I stick 1kV diodes in the circuit and call it a day?

Dr. Dark Current

Wed Feb 05 2014, 09:56AM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

Hi, you forgot to state the topology of the converter. Also a quick sketch of the power section would help.

Edit: Oops just noticed the schematic. I will have a look at it and then edit this post.

So, as you said, the ringing is caused by the leakage inductance of the transformer and the slowness of the diodes. 1kV diodes are actually the minimum I would use. You could have a look at 1200V diodes, the price of SiC diodes has gone down lately, but there are classic diodes with good parameters. The 1200V version is developed quite a lot as it matches the 1.2kV IGBTs.
And to reduce the ringing, you should use R-D snubbers designed for best damping of the waveform if possible. The C is selected based on allowable dissipation in the resistors, and R must be the "right value" (you should best determine it experimentally, probably something around 100 ohms?)

-Jan

Finn Hammer

Wed Feb 05 2014, 02:13PM

Registered Member #205 Joined: Sat Feb 18 2006, 11:59AM
Location: SkÃ¸rping, Denmark
Posts: 741

Chris,

I like the gate drivers you are using.

I wonder, perhaps you could use Mosfets as Syncronous Rectifiers, and controll their turn on to limit the rate of current rise so that you avoid ringing.
You would burn more power during the first 500nS, but save during the rest of the period, due to lower R=on in the fet.
I realise your circuit is high powered, but the waveform looks exactly like what I had on some CM600 gates, before I slowed down the switching time on the totem pole gate drivers, after which I had a perfect slope up to a rounded knee with no overshoot at all.

Just my 5 cents.

Cheers, Finn Hammer

cjk2

Thu Feb 06 2014, 02:47AM

Registered Member #51 Joined: Thu Feb 09 2006, 04:17AM
Location:
Posts: 263

I made some progress tonight on this issue. I read a paper here that introduced a "novel snubber" that recycles energy into the output capacitor.

I will still have a problem with ringing at lower duty cycles as the paper predicts. Perhaps a TVS diode from the output of the full bridge rectifier to the output capacitor will help to clamp this ringing. I tried different values for C32. I have currently picked 33nF. Smaller may be better though, as it seems that increasing this capacitor value makes the power level required for ZVS higher.

In other news, I achieved 94.4 percent efficiency tonight at 144W output power.

1391654791 51 FT161039 Novel Snubber Sch

1391654791 51 FT161039 Novel Snubber 33nf

cjk2

Thu Feb 06 2014, 04:36AM

Registered Member #51 Joined: Thu Feb 09 2006, 04:17AM
Location:
Posts: 263

Here is the result with a traditional RCD snubber. I used an UF4007, a 1k 5W resistor and a 2.2nF ceramic cap. The power dissipation in the resistor is low (about 1-2W I estimate) and the ZVS of the primary works well. This is probably the way to go for the final design.

This gives 93.8% efficiency with a 147W load

Alex M

Thu Feb 06 2014, 04:12PM

Registered Member #3943 Joined: Sun Jun 12 2011, 05:24PM
Location: The Shire, UK
Posts: 552

cjk2 wrote ...

Here is the result with a traditional RCD snubber. I used an UF4007, a 1k 5W resistor and a 2.2nF ceramic cap. The power dissipation in the resistor is low (about 1-2W I estimate) and the ZVS of the primary works well. This is probably the way to go for the final design.

This gives 93.8% efficiency with a 147W load

Much better results! Have you got any ferrite beads lying around? using one of them on the diode leads might help to reduce the ringing even further.

cjk2

Thu Feb 27 2014, 02:19AM

Registered Member #51 Joined: Thu Feb 09 2006, 04:17AM
Location:
Posts: 263

For the benefit of anyone that encounters a problem like this in the future, allow me to explain how I solved this problem:

I tried several different diodes including 1kV ultra fast rectifiers, 1.2 kV ultra fasts, and finally 1.2kV SiC diodes.

The SiC diodes work great and eliminate nearly all of the ringing. Best of all, they cost less than $2 each.

I also combated this ringing problem by moving the resonant inductor on the primary side out of the leakage inductance of the transformer and into an external inductor made of type 2 material. This inductor has its floating end clamped to the primary rail with another pair of SiC diodes. I should note that the secondary side snubber never conducts but will probably be left in for good luck. I have ferrite beads on all gate and drain leads on the fets, and a bead on each diode lead. Overkill maybe but this seems to work well. I have hit up to 93% conversion efficiency total for this isolation stage and a prior PFC stage not shown.

1393467337 51 FT161039 Sic Everywhere And Tvs Snubber

1393467337 51 FT161039 Sic Everywhere And Tvs Snubber Wave

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob