Self-oscillating halfbridge flyback driver / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

Self-oscillating halfbridge flyback driver

previous 1 2 3 next

Move Thread

LAN_403

Steve Conner

Fri Mar 16 2007, 02:05PM

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

That's neat Firkragg! So an ATX power supply is actually more or less self-oscillating, and the PWM IC just pokes it now and again?

Marko

Fri Mar 16 2007, 03:46PM

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

Sort off. Driver actually uses very little power to drive bipolar transistors just because of large positive feedback from current transformer. If drive was removed in operation halfbridge would continue to work as a blocking oscillator (and probably overvolt and unleash chaos to low voltage components of a supply).

AT supplies also have pullup resistors wich seemingly start the oscillation and power the controller up, as there's no auxiliary supply to do it.

cbfull

Fri Mar 16 2007, 05:57PM

Registered Member #187 Joined: Thu Feb 16 2006, 02:54PM
Location: Central Ohio
Posts: 140

There is a version of that floating around here somewhere that I posted a while back, but I modified it to use resistors to place a bias on the transistors to start the oscillation instead of the diac/capacitor startup circuit.

I was very pleased with the results, and it is the circuit that I will be using from now on. If you like I can try to locate the thread, or you can just search all posts for cbfull, it shouldn't take you very long to find. The nice thing about the circuit is that you don't need any extra gate/base transformers, they are intrinsic to the design.

Marko

Fri Mar 16 2007, 11:41PM

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

Yes, I remember the thread and app note...

The circuit saves on current transformer but still needs feedback windings to be wound, and I actually find external transformer more neat than it. For the rest of the circuit, it seems doesn't seem to have some special advantage over resonant version.

OK.. Can anyone answer; why operating frequency of (blocking) circuit *decreases* with increase of load? IN all occasions I would expect the little transformer to saturate faster under higher current and thus make the frequency increase. It's still boggling me up!

Dr. Dark Current

Sat Mar 17 2007, 06:59AM

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

Firkragg wrote ...

OK.. Can anyone answer; why operating frequency of (blocking) circuit *decreases* with increase of load? IN all occasions I would expect the little transformer to saturate faster under higher current and thus make the frequency increase. It's still boggling me up!

possibly storage times of driver transistors increase with load?

uzzors2k

Sun Mar 18 2007, 08:10PM

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

Happy Birthday Firkragg! Did you get that DRSSTC working in time for your birthday?

Back on topic, what frequency does the driver drop to while running a flyback? I'm sure you've thought of it already, but if the the frequency is dropping too low, saturating the flyback, which would overheating the transistors. I'll try making a mosfet based version later this week some time, so I can see for myself whats happening.

Marko

Sun Mar 18 2007, 09:23PM

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

uzzors wrote ...

Happy Birthday Firkragg! Did you get that DRSSTC working in time for your birthday?

Thanks..

I ordered some new transistors as well as some ribbon connectors an LM311's for DRSSTC I forgot to buy earlier.

I somehow feel that mosfets will be harder to get working because of their narrow operating region. You may easily overvolt the gates with just slight miscalculation of your current transformer and expected max current. You may need some kind of hard clamping and would probably waste a lot of power there.

Most of new CFL bulbs actually use mosfets in their designs, but I'm not sure how is circuit constructed since it has no current transformer and takes feedback from ballast inductor directly. Ic could be just saturating the whole inductor wich may even not be too big problem since mosfets are very fast, or it may be a resonant circuit ala CFbull's scheme.

Anyway I would much rather use SG3525 and normal GDT since it is way more robust adn mosfets are easy to drive.

Marko

Tue Mar 20 2007, 10:49PM

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

possibly storage times of driver transistors increase with load?

Apparently, they do; more precisely, storage and fall times increase as transistor goes deeper and deeper in saturation.

Large amount of charge needs more time to get removed from the base and thus switching times adn losses increase. Anyway, I tought that any such effect should still be maybe an order of magnitude weaker than response of the current transformer.

I fiddled around a lot last days, and it seems that I basically misunderstood how the circuit was supposed to operate.

At first I tought that it's just better to drive a transistor deep into saturation and make sure it stays there, basically with as much current I can allow it's base to take.

Such gross overdriving of bases must have caused very long storage and fall times. As load is purely inductivetransistors were switching off at maximum current with very long fall times, suffering massive switching losses and blowing themselves out.

The trick, or better said all the point of current transformers are used for is to always keep the ratio of collector and base current same. (proportional base drive).

That way transistor is always kept at or just over saturation with smallest possible charge storage and fall times and circuit operates at maximum efficiency. From what I know, delays can be minimized as much as a factor of ten.

Current transformer, seemingly, needs to be designed with a turns ratio *just lower* than minimal expected current gain gain of the transistor, in order to keep it just around saturation and in sweet spot of operation.

BUT11A's are relatively small, high gain transistors (Hfe 10..20 or so) and would now seem to be best with bigger ratio transformer, 1:8:8 from pc power supply or even bigger looked just fine.

Large transistors like MJE13009 have much lower gain, for this example it wanders from 4..8! This explains why these work well just with 1:3:3 or similar CT.

Current gain varies with load put on transistor, but at low powers higher gain and little bigger switching loss just isn't important. So, transformer is designed around lowest expected Hfe.

****

Just this evening (night) I lashed up a new design using ATX supply BDT (1:8:8??). I also ditched the diac startup circuit and switched to DC block cap + diode with pullup resistor, exactly the circuit as in AT power supply.

I was a bit confused with orientation of the transformer, and after I got it to oscillate the power transformer was giving off terrible sound, looking like it's saturating and transistors were getting hot quickly.

PS. as a sort of update; my frustration continues.

Two times I tried the reconfigured circuit, the transistors failed immediately with their bases shorted out. Even after I completely copied the supposedly working circuit from old AT supply, with same components, I again blew the transistor without having any clue what was going on.

So I'm back on the beggining with a pile of 4 dead BUT's. It's hard to believe that such a small and simple-looking oscillator can be so complex and frustrating to actually get running.

uzzors2k

Wed Mar 21 2007, 08:34PM

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

This is proving to be quite the mystery. I haven't gotten to try anything myself yet as my SGTC is keeping me busy. (Not that I could with contribute much, lol.) Aren't there any engineers here?

I saw this a few days ago, a lot of talking in circles but it may help.

Marko

Wed Mar 21 2007, 09:14PM

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

I don't know what was with those PC-power supply BDT's, as I can't see neither number of turns nor orientation, and even while following the original PCB and schematic I could have done something wrong.

I switched back to my little toroid transformer and diac circuit, and it works all fine now up to 30W or so with ATX transformer.

Since I now got and idea how to design a proper proportional-drive transformer, I started increasing secondary turns from 3 to up. At over 6, the transformer started throttling.
5 turns seemed to be a sweet spot for operation, and switching times indeed got much shorter!

I measured both rise and fall times to be closely 400ns compared to some 1,5..2us as of before, when I used something like 2:3:3 transformer.. what was I doing!

Maybe it could go even faster with a bit of twiddling on a base circuit, but I just didn't want to touch anything.

Here are some better pics so you guys can finally see what's all about now. I used a filter cap so I could actually scope something clearly.

This is primary voltage, at some 40kHz with some 20W load.
Frequency tends to drop pretty much with load. I guess I could get no more than 50W this way without saturation... I use bigger ferrite for CT, as this one is really tiny one from CFL lamp.

The PC power supply CT's from those 200W supplies are often about 1:8:8 or so.. they may have just been too big!

This is closeup on a falling and rise edge, 2us div:

Base voltage, 1V div, at about 50kHz.

previous 1 2 3 next

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