Self-oscillating halfbridge flyback driver / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

Self-oscillating halfbridge flyback driver

1 2 3 next

Move Thread

LAN_403

Marko

Sat Mar 10 2007, 08:03PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Hi everyone

This friday I was too tred to do anythyng but randomly play with some components at hand.
I wanted to see if a flyback could be driven with a small simple blocking oscillator as used in halogen lamp 'electronic transformers'. Off 230V mains, ofcourse.

I built one out of some BUT11A transistors and FR309 diodes. It worked beautifully when I hooked it up to an old PC power supply transformer.

I loaded it with a bunch of 24V 5W lamps for some 30..40W (as much as my isolation transformers could give) and transistors were only slightly over room temperature after 10 minutes, without any heatsinks.

Circuit stops oscillating if not loaded enough, but restarts as soon as load comes back.

I hooked it to a flyback with 50 turns for primary and pulled off some 5cm arcs, but transistors were overheating in seconds, even when no arcs were drawn and load was minimal.

I wasn't very sure what causes this, the only difference is that flyback is rectified only by a single diode, not being exactly a forward converter, but it has no filter caps, coupling is lower and shouldn't make such problems... I'l try it on AC transformer when I find one.

Output waveform from ATX transformer is much more beautful than I expected.. it's a bit shaky due to diac startup circuit wich kicks in every cycle.. no filter cap on input.

I only blew one transistor, after I tried to run without base resistors. It just shorted out, and I don't have a clue why... currents available in entire circuit are 20mA or so and really shouldn't harm the transistor. So that's a mystery number 2...

Dr. Dark Current

Sat Mar 10 2007, 09:09PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

That's nice. I tried that circuit too , in the past but had some problems with the base drive transformer (it didnt oscillate). when I finally get it to oscillate the driver run relatively cool with low (<50w) powers but was not really good for anything above 100w.

Download

Sat Mar 10 2007, 10:45PM

Registered Member #561 Joined: Sat Mar 03 2007, 02:46AM
Location: Adelaide Australia
Posts: 230

i have seen that designe before using a halogen lamp electronic transformers seems to work well

Marko

Sun Mar 11 2007, 12:58PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

That's nice. I tried that circuit too , in the past but had some problems with the base drive transformer (it didnt oscillate). when I finally get it to oscillate the driver run relatively cool with low (<50w) powers but was not really good for anything above 100w.

The circuit is easy to get into oscillation. I used a small toroid from CFL bulb and wound two turns for base windings and one feedback turn. It made a nice small low voltage supply with ATX transformer.

I popped two more transistors trying to get it working on a flyback. Transistors just overheat and die in seconds, even on AC transformer, wjile waveform and frequency still look good.

IIRC, solid state NST's use the same circuit, and I'm sure it should work up to several hundred watts without problems. At mains voltage, half an amp per transistor should be enough to melt the flyback anyway with little dissipated heat.

So I have some basic 'mysteries' at this point, I hope some you guys can help:

First, I managed to kill a transistor just because I didn't use btase resistors. Since feedback is from current transformer base current should be proportional to primary current and hardly able to blow the transistor. (BUT11A stands 2 amps of base current).

And second is much more basic; with increase of a load, or increasing the number of primary turns on the current transformer, the working frequency decreases.
This is counter-intuitive, as one would expect the little transformer to saturate faster in both cases and frequency to increase!

I'l see if I can get a commercial halogen lamp transformer and gut it for component investigation and fix mine according to it. Guys on the internet usually don't give out component values...

uzzors2k

Mon Mar 12 2007, 08:40PM

Registered Member #95 Joined: Thu Feb 09 2006, 04:57PM
Location: Norway
Posts: 1308

Sorry I can't give you any real advice, but I'm no engineer. However this is pretty good reading, and should help you get it working with flyback transformers. I'm looking forward to when you get it working, something small, simple and offline like this is pretty handy, quick and dirty power supplies!

Marko

Mon Mar 12 2007, 08:59PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Oh, regarding CFL lamps, there are about a ton of schematics.. some people have done really nice work!

As you can see there are different versions of startup circuits, some don't use diacs but just resistors and some kind of DC block to start up the oscillation. AT supplies have same approach as well!

And I mostly done it around these ST appnotes

I although didn't run itno philips appnote you posted, and it indeed looks much more detailed.

Anyway, this seems to require a dissection of a real electronic transformer, as I probably missed out something important. This was more of friday fun than something serious, but I may just get it working better at once.

uzzors2k

Thu Mar 15 2007, 03:43PM

Registered Member #95 Joined: Thu Feb 09 2006, 04:57PM
Location: Norway
Posts: 1308

It may be the lack of deadtime that is catching up to the transistors as power increases. Have you scoped the bases to see if they are overlapping while running the flyback? When I made a smps for my audio amplifier I had a deadtime problem, and the mosfets would die in 30 seconds even with no load. Putting something like the IGBT drive here might help. I would play with this circuit myself, but I only have power mosfets, and no diacs.

Marko

Thu Mar 15 2007, 06:47PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

There actually is a lot of ''deadtime'' on base signals when circcuit is under load wich is probably due to transistor storage times. Circuit as such is protected from shoottrough pretty well because of positive feedback between transistors!

I checked the lamp supply and it uses 2,7 ohm resistors, small one has 1:5:5 and big one has 1:3:3 current transformer.

PC power supplyes from other side seem to use something like 1:8:8, and are fine, I guess it's just a matter of startup 'sensitivity'!

I'm now trying to get a DRSSTC working but I'l play with this circuit more after I get some time and a new bag of transistors! I hoped to get people interested and learn something in process

For mosfets, I just feel they aren't worth of it since it's just so easy and reliable to directly drive them. All schemes with bipolar transistors I seen seem to use this approach regardless If they are controlled or not.

If there is a PWM IC, like in a PC power supply, bipolar transistors with their current transformer are work horses... and controller just reins them with very little power!
Two little transistors slightly bias the core (trough a 1,5k resistor!) to VCC when drive transistor needs to be turned on, and short the transformer out if it needs to turn off.
In AT supplies, there are pull up resistors wich start the blocking oscillation on it's own (wich is required to power the controller) ala those circuits in CFL's. Diac is just sort of option.

robert

Fri Mar 16 2007, 01:49PM

Registered Member #188 Joined: Thu Feb 16 2006, 05:18PM
Location:
Posts: 67

The flyback has a gapped core and therefore has very little primary inductance (compared to the ungapped transformers this is supposed to run with) so primary current will rise very high even with no load.

Marko

Fri Mar 16 2007, 02:00PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Oh... I ungapped the core but failed to mention it

Running a gapped transformer with a halfbridge definitely wouldn't be fun for transistors.

Still, sthe leakage inductance is surely higher than of a transformer with overlapped windings, but I don't think it's a big deal. I took care to make it appear as similar as possible to ATX transformer I used first.

Circuits like these run large inductances alone in electronic ballasts and are fine about it.

1 2 3 next

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