Skm flyback destroyer 🤗

[oneKone]

A while ago I started messing around with driving skm bricks to drive my diy flybacks. I had varying success with using a gdt and 40-50% duty cycle. After speaking to "Steve arcs" I decided to try what he used, a3120 gate drive IC and lower duty cycle in favour for longer dead times.

This project is something that's been kicking about for a few months now, recently I thought I'd make it into a proper project vs letting it slip onto the back shelf and letting it be forgotten.

The first thing I stated working on was the gate drive side of things. I had (have?) No real experience with opto gate drives ic so my approach is minimalist. The boards simply have bus capacitors, gate resistor/dt diode and gate protection diodes.



My next plan was to use 24v 1a smps and modify them for the gate drive power supplies. In the end I had two floating supplies with an adjustable voltage range of 15-30v DC. At 17v (gate drive) when pulling an arc the a3120 would hit uvlo, a sign to me that peak current even for a short time was enough to pull the smps voltage down. As a short term fix I adjusted the voltage upto 19v. I've since ended up ordering 18v 2a smps, I'm simply hoping this will be enough (just a pure blind guess).


I'm currently finishing up the bridge side of things. The plan is to make an enclosure this weekend from plywood. Originally I was thinking of making an aluminium enclosure but added complexity was something I wasn't keen to take on.

Cheers

[davekni]

Sounds like a fun project.

A standard SMPS may have trouble with rapid common-mode slew rate of high-side gate supply.  If you had the main bus powered during your test where UVL hit, it could be the supply getting confused rather than over-current.  (To estimate current, just multiply gate charge times frequency.)  I'd suggest a large common-mode choke in series with the high-side gate supply leads.

There are dual isolated gate driver chips that include adjustable dead-time insertion.  I use Si8274 on both my DRSSTC and the isolated 10kW "PFC" that feeds its bus supply.  (BTW, they don't survive shorted outputs.  Fried one two days ago when a strand of wire shorted two pins in a connector in my gate-drive wiring.)

If you want a bit more work, another option to standard SMPS and common-mode filtering is to make your own isolated supply.  Here's a self-oscillating circuit I use on my PFC:


Output voltage is a couple diode-drops below input voltage with no separate regulation.  Adjust input voltage to change output voltage.

I have a higher-power version for my DRSSTC to drive all four banks of IGBTs - same topology with larger FETs and correspondingly lower feedback network impedances.  Can't find that schematic at the moment.

[TMaxElectronics]

My next plan was to use 24v 1a smps and modify them for the gate drive power supplies

Interesting idea, never thought of that before But I agree with dave that there is a good chance the common mode voltage will confuse it (it certainly did in my early experiments with some other isolated 2-output dc-dc).

You might also want to consider (or test) those small isolated DC-DC converters that you can buy from digikey. Something like 102-PQQ6W-Q24-S15-S-ND, though you could prob get by with a smaller one.

And as dave said you could also just make your own supply, it's surprisingly simple. Either dave's layout or just a simple flyback.
You can buy transformer bobbins with matching ferrite cores on aliexpress for less than a buck each. I've used these, They can easily handle the low-ish power required for gate driving: https://de.aliexpress.com/item/32788766492.html
Regulation of the flyback could be done with a simple microcontroller like a pic12f1572 or a dedicated pwm-regulator.

What controller do you use? If it has programmable logic you could just control the deadtime in that and keep the gate driver simple and dumb
I think I'll just use a digital isolator, half bridge driver and a pair of mosfets in an SO-8 package for mine but I need to drive quite a bit more current

[oneKone]

Cheers for the info @davekni, I'll have to look into making my own supplies. It'll be handy for future projects.

@TMaxElectronics I'm just using a tl494, straight up copied from mads website.

I ended up finishing the bridge last night, I'll hopefully go I to detail on how I make my bus bars in a seperate thread. Overall I'm happy with how it turned out. The switch is an skm400gb126d, mdq100 rectifier, two epcos 10,000uf 400v bus caps. I thought about trying to make it more compact but it would have sacrificed the ease of working on the bridge.

[Patrick]

Will you have inrush current problems with those caps ?

[oneKone]

Will you have inrush current problems with those caps ?

For me it won't be a problem as this will only ever be used with a variac. But now you mention it maybe I should add a softstart just incase the bus ever gets powered wrong.

[oneKone]

I made some progress on the build this past weekend, unfortunately I didn't get any build pics ... Juggling house renovations and hobbies is challenging!

For the most part I got everything setup and ready to go. I just have to make an engraved panel for the frequency pot and get some aluminium mesh bent for the sides. Unfortunately my oscilloscope transformer failed so I'm now without a scope, it may be upto a month before I order a new one so I'm trying to decide to run it (having previously run the system) or simply wait. I guess I could just use a multimeter on a ct and take note of the primary current.

Here's some pics is the progress.

[alan sailer]

It looks quite clean. Well done.

I tend to like housings that have round edges because everything in engineering tend towards right angles.

Darn that René Descartes!

Cheers.