really simple smps design?

[oldroborg]

I'm planning to build a fairly high output smps (A toroidal linear power supply replacement), ideally a resonant design. I'm looking for the 'most likely to work/least smoke output first power-up'
I'd like it to be able to sustain 2Kw continuous output at 60-70v from a 240v 50hz supply. The cleanlines of the transformer output is not too important as its likely to be filling up some big capacitors.

cheers,
Rob.

[croccodillo]

I developed a similar beast at work: a 2KW output SMPS, but with an high voltage output (several Kilovolts).

I used a TL494 running at 50Khz; such an IC drives an IR2113 mosfet driver that in turn drives an half bridge made up with two IKW75N60T IGBTs.

I used IGBTs instead of Mosfet because the quasi-resonant circuit I have implemented works in the so-called Zero Current Switching, in which condition IGBTs work better than Mosfets.

The resonator tank is made up with 4 470nF capacitors in series, for an equivalent capacitance of 117.5 nanofarad; I had to use 4 capacitor in series for the too high voltages you will find onto the resonator capacitor itself; you can have up to 4500Vpp on the resonator capacitor (and for money: 4 of these capacitors cost a lot less than a single high voltage, high power, SMPS PP Epcos capacitor).

The tank inductor is a toroid with a 84uH inductance.

These values gives a resonating frequency of about 51Khz.

You can use an ETD70/N87 core for the inductance: place a gap of 0.127" on each leg of the core (the exact thickness of a PCB, for a total gap of 0.254") and wind 13 turns of Linz conductor (or a multifilar winding); remember you're playing with about 10 Amps at 50Khz, here.

You can use the Epcos Ferrite Software to calculate and verify the inductance.

The transformer is an ETD70/N87 core too, with 17 turns on the prymary; you have to calculate your secondary winding on your own (mine is 400 turns, for about 7500Vpp).

No gap on the transformer, the topology implemented does not need it.

With such a core you should be able to reach 4KW of output power without any problem.

Actually I pushed the SMPS up to almost 3KW, without any problem at all.

I cannot give you too much details (actually this is my work), so the POWER STAGE SCHEMATIC is the only thing I can show you.

I have removed the whole output circuit, because it is not useful for you (you don't have to output 5-10Kv at few hundreds of milliamps) and because it would be too detailed...

I think the schematic and the base idea is more than enough to begin with; it took me several months to me to make everything work fine (and stable).

Remember you are playing with 2KW of output power; in the best case you'll have an efficiency of 90%, meaning you'll have to dissipate about 200W of heating.

You'll need a quite big heatsink and a fan. No way to build this beast fanless.

At 2KW/70V output you'll have about 30A; you'll have to be careful with the output stage.

Ciao,
Giovanni

[oldroborg]

cheers, thanks for the reply. I'll check out your power stage later I was actually thinking about using a SMPS controller with alot of fault detection built in, At the moment i'm nowhere up to speed on what's currently available. Zero crossing switching does seem to be a very desirable feature, especially at these high power levels.
I suspect it would be hard to resist the temptation to venture off the yellow brick road and build an ampliverter if i had a pwm modulator and some dedicated high voltage mosfet drivers in the mains side of things

cheers,
Rob.

[richie00boy]

I can host a scanned PDF from a very old GE data handbook I've got, with I think 3 basic designs that could be scaled up. They use excitation windings, no chips. Really simple. If it would be of interest let me know and I'll dig it out.

[Pafi]

Zero crossing switching does seem to be a very desirable feature, especially at these high power levels.

Why? You said you want resonant converter! This is it!

[luka]

Hi

I made smps just like croccodillo wrote, but used mosfets, and it is most basic thing I think

PS:Can this be moved to Power Supply Design, thanks

[oldroborg]

Originally posted by richie00boy
I can host a scanned PDF from a very old GE data handbook I've got, with I think 3 basic designs that could be scaled up. They use excitation windings, no chips. Really simple. If it would be of interest let me know and I'll dig it out.

I'd be interested in looking at these, i've built similar discrete creations in the past by rewinding TV flyback transformers, but kindof the opposite of what i want now! If it's not too much effort to host , i'd love to see these, cheers.

Originally posted by Pafi


Why? You said you want resonant converter! This is it!

Definitely, it looks really interesting, now i need to understand more about it. Several KV has me more worried than potential smoke output during development I suppose if i don't inhale the smoke it'll be OK

On a similar tack, I discovered the power supply design forum yesterday (unconvieniently, very well hidden, inside other stuff!) & have spent the last day (literally) browsing through the vast wealth of smps goodies there, and probably going to spend at least the next two weeks continuing to do so, i'm not sure exactly what i want to contemplate building...

cheers,
Rob.

[croccodillo]

Series quasi-resonant converter work in the so-called ZCS (Zero Current Switching) mode, so your goal is reached.

I have some interesting documents for you, you can download the entire list from HERE.

Please check first of all the Texas application note AN658/1194 (AN658_1194.pdf).

I made smps just like croccodillo wrote, but used mosfets, and it is most basic thing I think

Well, I tried Mosfets and abandoned them.

For this kind of SMPS IGBTs are much better.
This is because when you switch a Mosfet on, you short-circuit its drain-source capacitance, and thus, if you have high voltage across them, you'll have high dissipation (In resonant converters, you switch them on at zero current but at maximum voltage).

Instead, IGBTs do not have so high dissipation, due to their low C-E capacitance; but they have poor switching-off characteristics, so switching them off on zero current is better.

Hope this is clear.

Ciao,
Giovanni

[Pafi]

Croccodillo!

This is because when you switch a Mosfet on, you short-circuit its drain-source capacitance, and thus, if you have high voltage across them, you'll have high dissipation

To avoid this, there are two ways: operate it on a little higher freq then resonance! This way the load current will discharge Cds (or Cce)! Other: use magnetising inductance for the same purpose!

It's possible to combine ZVS and ZCS with their benefits. Also it's possible to unite trafo and resonant choke.

[oldroborg]

Originally posted by croccodillo
Series quasi-resonant converter work in the so-called ZCS (Zero Current Switching) mode, so your goal is reached.

I have some interesting documents for you, you can download the entire list from HERE.
Giovanni

There's some really good info here, very good of you to upload it all, thanks for that, it'll take me a while to digest it all! Hopefully report back with progress

cheers,
Rob.