Low inductance full bridge layout. / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

Low inductance full bridge layout.

1 2 next

Move Thread

LAN_403

Jorrit

Wed Sept 06 2006, 03:00PM

Registered Member #140 Joined: Sat Feb 11 2006, 11:43AM
Location: The Netherlands
Posts: 14

Hello,

I'm building my third drsstc. I will be running with a Fres of 40 Khz.
The coil is finished already. The MMC is also almost ready.
But i'm still working on the electronics.
The driver will be Steve Connor's pll driver, so that shouldn't be a problem hopefully.

But I still have some questions about the full bridge. I got a 1200 Volt 3000 A 3phase igbt module which I want to use for the coil. I haven't found a datasheet yet so I don't know the switching times but I will give it a try. If it doesn't work I will switch to a fullbridge with 300 A 1200 V igbts.

The igbt I want to use is a pp30012hs. And contains 3 half bridges in one module.
The layout of the full bridge should have very low inductance because I don't think it's a very fast igbt

So does anybody have some suggestions for a very low inductance layout ?
I have some ideas but I need to draw it before I can explain it. My english ain't that good so.

I have attached a picture of the igbt module on the heatsink.

Another picture.

Sulaiman

Wed Sept 06 2006, 03:57PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

Operating at 40 kHz means (usually) moderate dI/dT
so unwanted voltages across stray inductances (E=L.dI/dT) are moderate also
hence you need not go to extremes to achieve low stray inductance
though you should still keep it low.
A good quality invertor-grade capacitor
with short thick wires to the bridge should be ok
anything better would be excellent!

In your mind, draw out the path that the main current takes
then minimise the area enclosed by the current path.

others may differ!

Steve Ward

Wed Sept 06 2006, 05:36PM

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

What would be ideal is a laminate bus assembly. What i like to use is 2 sided copper clad board as the base, then solder thick copper sheeting onto it to give it the conductor size required. Anyway, you devote one side of the board to be the + and the other the -. Drill holes and remove copper where connections are needed. Connect your large electrolytic capacitors to this same double sided copper sheet. When using large sheets as conductors, you get far less inductance than if you used wires. Making the sheets wider lowers the inductance (as its similar to paralleling many small inductors). Here are some large H-bridges i have designed:

Check out the pictures in each of those links. Your job is a little easier since they put the +/- terminals for each bridge in a convenient place.

Also, these big IGBTs can be switched a bit faster than spec sheets may suggest because we arent hard switching quite so much current.

Do you have any pictures of your previous 2 DRSSTCs?

Jorrit

Wed Sept 06 2006, 05:57PM

Registered Member #140 Joined: Sat Feb 11 2006, 11:43AM
Location: The Netherlands
Posts: 14

Thanks for the ideas

I was also thinking about a laminate bus assembly like you mentoined Steve. And got some good ideas from your pictures.

The only problem is I got to put 2 caps in serie (sorry don't know the english word for it but it's the opposite from putting the caps parralel)
I want to power the coil with 230 vac rectiefied and doubled. I got 4 12000 uF 200 vdc caps who I want to put in serie so I get 3000 uF at 800 vdc. And connect these caps to the bridge with short and thick wires (25mmÂ²).
And on the copper plate I want to put 2 1900 uF 400 vdc caps in serie. I don't have any snubber caps so I want to keep the inductance as low as possible.

Attached is a picture of the coil.
I don't have any pictures of my old drsstc's on this computer but wil post them when I find them

Reaching

Fri Sept 08 2006, 10:47AM

Registered Member #76 Joined: Thu Feb 09 2006, 10:04AM
Location: Hemer, Germany
Posts: 458

you really need those snubber capacitors in your bridge design to prevent the whole thing from ringing up. when this ringing occurs the voltage increases to maybe twice the voltage rating of your igbt module, its very dangerous. you dont need so many yF like steve uses only 1yF decoupling can help you a lot if you use good caps. for a bridge this size is suggest using 10yf or more very close to your igbt brick.

Jorrit

Fri Sept 08 2006, 02:33PM

Registered Member #140 Joined: Sat Feb 11 2006, 11:43AM
Location: The Netherlands
Posts: 14

Well I don't have any snubber caps. And I don't have a source to buy them cheap

And I believe Steve Connor also doesn't use snubber caps on the full bridge of his new drsstc (odin the all fragger)
I'm trying to keep the layout as low inductance as possible with the caps very close to the igbt.
So I hope it will survive without snubber caps.
I`m aware of the potentional problem but I will scope the whole thing beforce turning it to 11.

Steve Conner

Fri Sept 08 2006, 02:37PM

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

Odin hasn't been tested yet (and probably won't be tested for a while due to me being real busy with work) so YMMV.

I did find on my 220kHz DRSSTC, that certain values of snubber caps could make the ringing worse than it was with the electrolytic alone. But I also found other values that made it a good deal better. I never tried running it at full power with no snubber caps.

Jorrit

Fri Sept 08 2006, 02:56PM

Registered Member #140 Joined: Sat Feb 11 2006, 11:43AM
Location: The Netherlands
Posts: 14

Well I could try to find some snubber on ebay or something. Just to be shure.
Any ideas on what amount of snubbber caps would be good ?
Or should I try different values and measure which gives the best performance ?

I'm probally going to buy the copper clad tommorow and finish the full bridge.

Steve Ward

Fri Sept 08 2006, 05:43PM

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

For the record, my DRSSTC-2 system has had its highest power , and longest runs with no snubber capacitors installed, only the main electrolytics. Unless i find some reason to change it, this is how it will remain. I found that even adding little 2uF low inductance caps on the buss increased the ringing of the buss voltage, probably because it increased the Q of the buss. I use large inverter grade capacitors on that coil, so they have lower ESR than computer grade caps, so as Conner says, "your mileage may vary".

Jorrit, I also have to series lytics on my coil. Make a third layer to your laminate, use thick wide sheets to make the series connection. This will keep the inductance down. Hope this idea is clear enough. Keep in mind that all of the capacitors in series form a "loop" that is in series with the IGBTs, so you must keep *all* of these inductances low.

Thats a serious looking coil setup, what are the dimensions of the secondary?

HV Enthusiast

Fri Sept 08 2006, 06:53PM

Registered Member #15 Joined: Thu Feb 02 2006, 01:11PM
Location:
Posts: 3068

wrote ...

I found that even adding little 2uF low inductance caps on the buss increased the ringing of the buss voltage, probably because it increased the Q of the buss.

With a snubber you need a resistor in series with it (usually) to dissipate the power and to properly damp the snubber.

Actually, if you really want to design an effective snubber, you basically take a measurement of the ringing of the buss without any snubber capacitors, and then with an arbitrary snubber capacitor, say your 2uF. From the frequency of the ringing and with the capacitor value you added (vs. without the capacitor value), you can effectively experimentally determine the inductance on the buss / bridge and properly design a snubber network (capacitor, resistor)

1 2 next

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