Snubber for IGBT / General Science and Electronics / Forums

Forums

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

« Previous topic | Next topic »

Snubber for IGBT

1 2 next

Move Thread

LAN_403

Linas

Sun Nov 23 2008, 07:02PM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

Hi all

I found interesting snubber for IGBT overvoltage protection. i need to know your opinion about that. In my DRSSTC i will use two snubber. One will be simple C snubber (cap, between +C and -E),

other CDR (capacitor diode resistor), and here is circuit:

Steve Ward

Fri Nov 28 2008, 11:37PM

Registered Member #146 Joined: Sun Feb 12 2006, 04:21AM
Location: Austin Tx
Posts: 1055

The best approach is to first minimize bus inductance, which reduces the need for snubbers. Once that fails, using decoupling capacitors on the bus right at the IGBT terminals is the next simple step to cleaning things up. When that fails, id suggest slowing down your IGBT turn ON time (assuming this is for SSTC/DRSSTC where voltage lags current) because that will help give a slower diode reverse recovery i believe. Often times, there is not really a significant impact on the actual delay time when increasing Rg... measure for yourself by looking at the Vce and current waveforms simultaneously.

For "soft switching" you probably wont see a big benefit of using RCD snubbers. I found that older IGBT modules simply have too much internal inductance for which you can do nothing about except slow down the transitions. You really are limited by the device internal inductance in many cases.

My CM300 bridges ring quite a bit, but the nice thing is, my "by-hand" math has predicted both the relative magnitude and frequency of the ringing. From that point i have confidence that when operating at full bus voltage, my transient ringing is within spec. I just use conservation of energy and basic L-C relationships to solve the node voltages across the IGBT.

Those black/blue snubber caps are pretty sweet, highly recommended for module decoupling.

Linas

Sat Nov 29 2008, 08:18AM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

I buy from ebay some wima MKP10 0,47uF 630Vdc capacitors for RCD snubber, also get some ultrafast diodes (35ns d2-pak)

But i don't know, how lot ohms use for resistor ?? ( i think 50R)
here how now look bridge, quite sweet

cap is 5000uF 400Vdc

MRacerxdl

Sat Nov 29 2008, 03:43PM

Registered Member #989 Joined: Sat Sept 08 2007, 02:15AM
Location: SÃ£o Paulo, Brazil
Posts: 476

Hmm, how it works that Snubber caps? Putting it direct with Collector and emitter wont make the IGBT shorts the charged cap each time it gets gate emitter on?

Linas

Sat Nov 29 2008, 08:31PM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

Mestre Racerxdl wrote ...

Hmm, how it works that Snubber caps? Putting it direct with Collector and emitter wont make the IGBT shorts the charged cap each time it gets gate emitter on?

if you was asking about C type snubber, it's just stabilizing input voltage, and don't let resonate bridge. And if you talk about RCD type snubber, if turned on IGBT voltage is bigger than DC input voltage, it will comeback deltaV into DC input...

MRacerxdl

Sat Nov 29 2008, 09:37PM

Registered Member #989 Joined: Sat Sept 08 2007, 02:15AM
Location: SÃ£o Paulo, Brazil
Posts: 476

But it wont increases the current across the IGBT?

Linas

Sun Nov 30 2008, 08:28AM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

RCD type snubber no, but C type yes. i was tested yesterday RCD and C types snubber, and then i put C type snubber, IGBT voltage immediately become very nice.(dropped rise and fall times ) it can give very large Ipeak (like 1000A), because they have low L.

Here is idea. of course resistors will be 50â„¦, and 50W, mounted on heatsink. Capacitors will be wima MKP10 0,47ÂµF 630V.Also i think add on snubber Vsat protection, and fall signal will go to over current protection input (into 555 timer, directly after LM311 output)

Steve Ward

Wed Dec 03 2008, 12:10AM

Registered Member #146 Joined: Sun Feb 12 2006, 04:21AM
Location: Austin Tx
Posts: 1055

Be sure to maintain bus-bar to IGBT tab direct contact. You do NOT want current to flow through the steel (or even brass) screws to the IGBT, its much more resistance and it wasnt designed for that. So get your bus bar under the PCB.

The C snubber should not increase the peak current seen by the IGBTs really, unless you have a shoot-through problem.

GeordieBoy

Wed Dec 03 2008, 02:11PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

I would not recommend the connection of a pure capacitance directly across IGBTs in the DRSSTC arrangement...

The load on the inverter in the typical DRSSTC arrangement appears capacitive and the load current changes direction BEFORE the IGBT's switch. (It has to happen this way, because the current has to pass through zero to be detected by the CT, and propagate through the drive circuitry to eventually switch the IGBTs some time later.)

In such an H-bridge driving a capacitive load each switch turns on into an already conductive free-wheel diode on the opposite side of the bridge leg. This has two profound consequences:

1. Firstly the device turning on does so with full DC bus voltage applied across it. (There is no ZVS like you get for MOSFETs driving an inductive load.)

2. Secondly, the device turning on has to handle the reverse recovery current of the free-wheel diode that is presently conducting in addition to whatever load current is currently flowing.

Since the full DC bus voltage exists across the device at turn-on, adding any capacitance at this point will cause excessive turn-on losses as the stored energy must be dissipated in the device. This adds to the burden on the IGBT of recovered charge that must be swept out of the recovering free-wheel diode in order for it to turn off.

Additional capacitance bolted across switches is beneficial in some topologies like ZVS/ZVT MOSFET bridges, but not in this instance. A more common use of snubbers with IGBTs is the RCD snubber to reduce turn-off dv/dt with inductive loads.

-Richie,

Steve Conner

Wed Dec 03 2008, 02:56PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

I assumed the cap was bolted across the DC bus terminals, not one or other of the switches. (In which case "snubber" is the wrong word for it.)

1 2 next

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