Small flyback SMPS's / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Small flyback SMPS's

1 2 3 next

Move Thread

LAN_403

Marko

Sun Apr 22 2007, 08:07PM

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

Hi all

Last days I was prodding around small flyback power supply design. I can buy UC3842 current mode controllers but I had an idea to make something 100% ghetto entirely from junk laying around.

I wanted to use TL494 as a controller for general reliability. Discrete circuits don't please me for much.
TL494 is a bit difficult because it lacks UVLO and has a wierd output stage, but I've overcome that.
As a switch I used IRFU410 mosfet from CFL ballast.
In order to make it work I needed to put a 100ohm pulldown resistor on gate in order to turn off the gate in few tens of ns wich I considered fast enough; since duty cycle is low for most of time theresistor doesn't dissipate a lot of power.
Additional transistor would be much better solution, but that's for some other time.

I made an UVLO circuit with hysteresis using a diac and additional capacitor.
When first cap reaches about 30V diac fires and dumps it into another cap wich ends charged to about 15V, starting the controller up.

SG3525 looked like it owns UVLO with hysteresis; I tried it out, but it proved useless because it needed about 6 miliamps off current even in UVLO; this would require huge and very hot resistor, and isn't much of good.

With such a primitive voltage mode control, it is important to set the minimum deadtime right; otherwise transformer will saturate under load and blow the mosfet.
Calculation is a bit difficult in my case, but I found 3us of ON time to work well at frequency of about 100kHz.

The good thing is that almost any of generic flyback-mode transformers (from PC power supplies, chargers.adaptors) will work well here.

Circuit I built has pretty cool line and load regulation and is very reliable when set properly. It even stands off shorted output as long as the diode and transformer don't overheat! Mosfet is without heatsink and only gets up to about 50 degrees C.

EDIT: zener in the schematic is wrongfully dimensioned for 600V device.
Use something like 150V max for the particular mosfet.

1177272421 89 FT0 Tl494 Flyback Schematic Shrink

This is how tl494 circuit looks, with optocoupler isolation adn working diac-based UVLO circuit. I loaded it up to 20W; at much more transformer windings get quite hot.

SG 3525 without isolation. Small IC on other side of board is UC3842 awaiting a run.
These motor's bearings scrabble and it draws lot of power without any load.
The output power is about 9,5W; input is about 11W apparent wich looks like pretty good power factor and efficiency. Maybe even as much as 90% efficiency.

Offline flyback converters absolutely require some kind of regulation to work efficiently.
When circuit is lightly loaded the controller is actually off most of time (since duty cycle can't be kept small enough for continuous operation) producing characteristic waveform-ghosting as in the picture.
At even lower loads (only the LED) it works in burst mode with few cycles and long off periods, in lower audio range.

uzzors2k

Sun Apr 22 2007, 09:26PM

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

That is pretty neat Firkragg! No more lame iron core transformers.

I always thought the flyback pulse from any offline flyback converter would be astronomic, so I've never bothered. Also 200V zeners are uncommon in my junk box, could a snubber be used instead?

Marko

Sun Apr 22 2007, 09:37PM

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

Well I just blocked that big motor to see how far could it go, it drew some 40W at substainal current, the TVS across the mosfet melted solder after some 20 seconds and soon after that everything went to pieces.

I blew a dozen mosfets as ritual sacrafices, but I'm not too much sorry for them.

I always thought the flyback pulse from any offline flyback converter would be astronomic, so I've never bothered. Also 200V zeners are uncommon in my junk box, could a snubber be used instead?

You just pointed out a mistake: the zener is for 600V mosfet, for 500V 150V zener should be used! Snubber is although just a secondary protection and doesn't do much in overall.

For fun I tried to run SG3525 at lowest duty cycle (about 800ns) without regulation, and voltage spike was still phenomenal and instantly blew the mosfet when I removed 100 ohm resistor load.

In order to work the circuit needs to be able to go into ghost and burst modes.

Still, if you can get UC3842 you should really do it, the IC would spare you a lot of trouble.

I think I'l also make some boards for the 3842 in time.

Dr. Dark Current

Mon Apr 23 2007, 07:41AM

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

That looks very nice. I have one question for you, with standard flyback drivers the usual flyback spike goes ~10-20 times the supply voltage, that would equate ~3-6kV with off-line operation. How it comes that your zener doesn't melt instantly?
I wanted to make an off-line flyback driver using just one transistor in flyback mode, I have some 1,5kV transistors, and I always thought they would get killed by the flyback spike, so I never even tried it...

Dr. Shark

Mon Apr 23 2007, 12:52PM

Registered Member #75 Joined: Thu Feb 09 2006, 09:30AM
Location: Montana, USA
Posts: 711

Very nice work, I know you have tried doing something like that for a long time (just as I have...), and I am glad you finally found something that works. I will copy your circuit and see if I can get it to work, maybe using a MOV instead of that exotic 200V Zener.

Ken M.

Mon Apr 23 2007, 07:58PM

Registered Member #618 Joined: Sat Mar 31 2007, 04:15AM
Location: Us-Great Lakes
Posts: 628

I also think this is an awsome set up, But Joe, I think he said that the 200v zener was amistake and that it is actually a 160v, 500v, or a 600v not sure which one it is...

Marko

Mon Apr 23 2007, 08:17PM

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

Guys, I think you should read these

Never mind the controllers used, but I think explanations are quite nice.

I would still recommend you all to try the UC3842 first,

it is absolutely pretty controller and far better suited than TL494. It is widely available and as cheap as tl494. It's not ghetto only for reason because it's somewhat harder to find in junk.

In offline application like this, we can't have the switch that will withstand significant voltage spike on the primary.
Most controllers and mosfets are practically limited to ON time of few hundred ns and usually have very fast turn off time. This still traps significant amount of energy in the primary winding and will likely kill the snubber and the switch.

Circuit needs to be constantly loaded in order not to roast the snubber and the transistor.

When it is very lightly or not loaded, the circuit can't reduce the duty cycle any more and in that part feedback is cruical.

As the output capacitor charges up to a point the feedback will turn the controller *off*, and it will start again when it falls for a degree.
At light loads the controller will start 'skipping' pulses creating the wierfd ghosty waveform I posted a pic of. (note that there is just few tens of volts of 'spike' under load!)

At even lighter loads, the controller will tend to operate in bursts of few pulses with very long off periods; repetition frequency cyn be s low as few hundred Hz.

When a large constant load is applied the primary inductance will dramatically decrease and controller will be able to safely increase duty cycle. (until the transformer saturates and everything blows up.)

It may look wierd to you, but most commercial offline flyback converters seem to operate that way.

Transformer itself is not very critical, but it tends to be ungapped since storing a lot of energy isn't important, and especially if there are multiple outputs and cross-regulation is important.

Regarding snubber, there are countless possibilities. See the appnotes!
With clamps, I usually try set them to 90..95% of mosfet's breakdown voltage.
In case of 500V device the primary zener should be about 475-325 = around 150VV.
For 600V device 200V zener is fine. Seriesed are also good.
Clamp can also be used across the mosfet itself but then it needs to have higher voltage rating.
Clamp sees very little stress except at high overloads. My circuit worked well without it, but I just didn't want to avalanche the mosfet at any point.

uzzors2k

Mon Apr 23 2007, 08:47PM

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

Actually I have a UC3842 in my junkbox.

Complete with the flyback converter it came with, which I found it in a monitor. I'll make it into a power supply when I finnish some other projects.

So just to make sure I understand things properly; when lightly loaded the energy stored in the transformer isn't used, creating a large inductive spike on turn off, which is burnt in the snubber. While at medium to high loading the energy is used up, smaller spike, so no energy goes to the snubber. The minimun load also keeps the regulation working, which I am familiar with in other regulated smps' topologies. I haven't gone through the app notes yet, but they look like good reading.

Steve Ward

Mon Apr 23 2007, 09:03PM

Registered Member #146 Joined: Sun Feb 12 2006, 04:21AM
Location: Austin Tx
Posts: 1055

I have one question for you, with standard flyback drivers the usual flyback spike goes ~10-20 times the supply voltage, that would equate ~3-6kV with off-line operation.

Standard? You design the converter to do whatever voltage you want.

Just trying to clear up any misconceptions.

Firkragg

I thought the PC power supply transformers were actually forward converters? But im actually not sure of this.

Marko

Mon Apr 23 2007, 09:14PM

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

Hi steve, uzzors

I thought the PC power supply transformers were actually forward converters? But im actually not sure of this.

Yes, the main transformers are predominantly halfbridge forward converters.

I was talking about the small auxiliary power supply transformer wich operates in flyback mode.

I do have a small 80W ATX supply that uses flyback converter and UC3842 on high voltage side, but that's some other story.

when lightly loaded the energy stored in the transformer isn't used, creating a large inductive spike on turn off, which is burnt in the snubber. While at medium to high loading the energy is used up

Actually, with no load, the converter will charge up the output capacitor and turn itself off.
With very light loads it operates in burst mode and output capacitor creates enough load at that point; the snubber will actually start to get hot at very high loads where primary current becomes significant. (leakage inductance always remains).

1 2 3 next

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