Simple H-Bridge construction with low parasitic inductances (for SSTC or ...)

[davekni]

A basic half-bridge SSTC makes a great first electronic Tesla coil project.  For example, this SSTC-1 design from KaizerPowerElectronics:
http://kaizerpowerelectronics.dk/tesla-coils/kaizer-sstc-i/

There are two sections to the electronics.  The driver:
http://kaizerpowerelectronics.dk/wp-content/gallery/2009_01_22_-_kaizer_SSTC_I/KaizerSSTCIdriverschematic.gif

And the half-bridge:
http://kaizerpowerelectronics.dk/wp-content/gallery/2009_01_22_-_kaizer_SSTC_I/KaizerSSTCIbridgeschematic.gif

Many first-time builds have problems due to half-bridge construction.  Wires and even device leads have inductance, often enough to disrupt operation.  A wired bridge certlainly can work, as shown in the above links.  However, lower parasitic inductance reduces voltage spikes, making for a more robust build.  Below is a simple technique (no ECBs) for building a half-bridge with low parasitic inductance.  Part labels are from the above KaizerPowerElectronics SSTC-1 bridge schematic.

The key to low inductance is copper planes.  Ideally the planes are on opposite sides of an insulating layer.  Four rectangles of copper foil tape is all it takes for a bridge, two on each side of a flat insulating layer.  Fiberglass or other phenolyc would be ideal for soldering temperatures, but may not be readily available.  Here I use 2mm-thick polycarbonate.  Even though the melting temperature is below soldering temperature, it works well.  Perhaps counter-intuitive, but it works best to use a hot soldering iron.  A hot iron can make joints quickly before the polycarbonate has time to melt.

Apply two rectangles of copper tape to one side of the insulating sheet horizontally, with a small gap between rectangles.  Apply the other two copper rectangles to the other side vertically, again with a small gap.  The gaps form a cross shape.

Simplest construction is to solder parts along the edges.  It's not space-efficient, but works well:





For better space efficiency, parts can be mounted on top of the foil/insulator sandwich.  This requires drilling a couple holes for one lead each of C8 and C9, and cutting a larger clearance hole in the top copper foil.  One lead each of C8 and C9 is bent out horizontally to solder to top copper.  The other lead goes through the hole, then bent horizontally and soldered.







Size of these copper planes and insulator is not at all critical.  Make them fit the pars being used.  The key is just overlapping planes instead of wires, and minimal lead length from part bodies to copper planes.

This can alternatively be constructed from double-sided copper-clad board.  Just cut a horizontal gap in top-side copper and a vertical gap in bottom-side copper.  I find that a dremel-tool cut-off wheel works well.  A hobby knife or round file can work too.  For clearance around through-holes, a few turns by hand of a large drill bit works nicely.  (Not for foil, which tends to tear.)

Notice that the gate-drive wires solder to the FETs (IGBTs in my sample) close to the device bodies, not to the foil planes.  This minimizes source (emitter) inductance that is shared between gate-drive and output current paths.

It is best to use copper foil tape that is wide enough, then cut it down as needed.  Soldering multiple strips of copper tape together is difficult.  Solder resists bridging between sections, as it doesn't wet the sticky layer.  Here's two example EBay listings for 50mm tape (no indorsement - just search result):
https://www.ebay.com/itm/Copper-Foil-Tape-EMI-shielding-for-Guitar-Slug-and-snail-barrier-6x50mm-Folded/273719723048
https://www.ebay.com/itm/3m-50mm-Guitar-Copper-Foil-EMI-Shielding-Tape-for-Electric-Guitar/133531273698

BTW, the top and bottom copper layers can be swapped.  There's nothing magic about which signals are on top.  Also, this works for a full-bridge, by removing C8 and C9 and adding two more IGBTs/FETs along the edge opposite the existing two.

Here's a very-well-constructed half-bridge example by ZakW.  See the two pictures at the end of his post:
https://highvoltageforum.net/index.php?topic=1910.msg14500#msg14500

[davekni]

I've since rebuilt this as a full H-Bridge with two parallel TO247 IGBTs for each switch (8 total):




The gap in top-side foil between VBus+ and VBus- isn't visible in these latest images.  Please refer back to the original post.  It's also hard to see, but I've added some 10nF 630V 1206 caps across the VBus+ to VBus- gap adjacent and under the three 3.9uF snubber caps.

The gate-drive diodes and resistors are now on cut copper-clad board using SMD parts.  Had to change values anyway, so went with my typical build technique.  This gate-drive change isn't critical.  (I also added small 3-turn inductors in series with the turn-on gate drive resistors to add a bit more dead-time.)

This H-Bridge build is used for all my indoor low-frequency QCW experiments as posted here:
https://highvoltageforum.net/index.php?topic=1268.msg10067#msg10067

[Da_Stier]

Very intersting, I like the additive construction with copper tape instead of milling / etching copper clad.
What ferrite cores do you use for your ethernet cable GDT?
They look like the typ you would get out of VGA cables

Greetings,
Michael

[davekni]

Michael,

The GDT cores are from a surplus store.  Likely were for cables originally.  Only spec's I have are what I measured.  They are 28mm long, 18mm OD, 9mm ID, 2.8uH/turn^2, ~25uVs/turn.  So this 3-turn GDT on two cores is good for ~70kHz and up at room temperature and +-20V gate drive.  To be safe, I don't plan to run it much under 100kHz.

[davekni]

Since I'm a fanatic for minimizing parasitic inductance, here's an over-kill low inductance H-Bridge using two CM600DY-13T bricks.  This would be more practical using an ECB rather than copper foil.  What I built makes any future repair difficult.

This is the H-Bridge before VBus planes.  VBus+ leads are all towards the center.


VBus+ plane added, soldered to capacitor + terminals and temporarily bolted to brick + terminals:


VBus+ plane covered with polyimide insulating layer:


VBus- plane added on top and soldered to capacitor - terminals:


Cables added, including current transformer on one H-Bridge output.  Note that parasitic inductance on the H-Bridge outputs is not critical.  No need for planes on the output.  Cables are fine.  Cables are fine for power input too as long as there is enough local capacitance on the parallel planes to handle high-frequency currents.


Used this bridge for my low-frequency QCW experiment:
https://highvoltageforum.net/index.php?topic=1268.msg11784#msg11784

Edit:  A clarification about parasitic inductance on H-Bridge outputs:  The two IGBTs within each half-bridge DO need low inductance connection between them.  For half-bridge bricks such as these, that low-inductance connection is internal.  When connecting individual IGBTs such as the TO247 packages at the beginning of this thread, it is important to keep inductance low for the low-side collector to high-side emitter connection.  The rest of the output cabling from that connection to MMC or primary is less critical.

[shrad]

I can give some food for thought...

For some laser experiments I discussed with a guy who has several patents and is really into the discharge stuff...

If you want the best possible setup you have to go coaxial, and you use a capacitor formed from two concentric tubes of metal with the inter tubes area filled with deionized water and the extremities closed by machined PTFE caps + orings. This itself is a self healing spark gap which relies in breakdown of deionized water and intertube distance for discharge voltage. If your primary is really low inductance like circular foil, you can reach insane energy density...

It would be fun to experiment with this

[John123]

Since I'm a fanatic for minimizing parasitic inductance, here's an over-kill low inductance H-Bridge using two CM600DY-13T bricks.  This would be more practical using an ECB rather than copper foil.  What I built makes any future repair difficult.

This is the H-Bridge before VBus planes.  VBus+ leads are all towards the center.


VBus+ plane added, soldered to capacitor + terminals and temporarily bolted to brick + terminals:


VBus+ plane covered with polyimide insulating layer:


VBus- plane added on top and soldered to capacitor - terminals:


Cables added, including current transformer on one H-Bridge output.  Note that parasitic inductance on the H-Bridge outputs is not critical.  No need for planes on the output.  Cables are fine.  Cables are fine for power input too as long as there is enough local capacitance on the parallel planes to handle high-frequency currents.


Used this bridge for my low-frequency QCW experiment:
https://highvoltageforum.net/index.php?topic=1268.msg11784#msg11784

Wow that's really cool Dave! Do you need special tools to make it using copper foil or will scissors work?

[davekni]

Yes, the copper foil is scissor-cut, with the exception that I made some round holes with a hammer and cheap punch (from Harbor Freight Tools).  It is 100um foil, a bit thicker than typical foil tape, but still easy to cut.  The punch isn't necessary, just convenient since I had a set already.

Thank you for the compliment!

[rikkitikkitavi]

Excellent postings davekni.

Very informative and educational, a step-for-step instruction how to do this the right way..

Low inductance layouting is difficult.

"insert smiley with hands clapping".


 

[costas_p]

Davekni in your ebay search listing , the foil tape is 65um thickness,

I guess , designing a Full or Half bridge on a  pcb, 2oz (70um) copper weight will be sufficient for small to medium SSTCs/DRSSTCs, am I right ?

Going over 2oz copper on China pcb factories is a bit expensive.

[davekni]

Yes.  I used 2oz copper for a prototype 800A peak H-Bridge at work.  Production version moved to 3oz for margin.  My 2600A peak (3500A eventually) DRSSTC is built from 40 TO247 IGBTs.  Each half-bridge is constructed using a 2oz ECB.  The ECBs are augmented with copper foil.