flyback efficiency problem

[xaccto]

I have a simple flyback circuit as



why CD4069? because I had 25 of them.
input is 10-12V, output 250V.
not exactly isolated, as I connect the grounds for 12V and 250V thru a resistor.
The feedback is using a voltage divider thru a zener diode, when the zener conducts,
a schmitt trigger switches and switches OFF the CD4069 oscillator.
The core is a bit of an unknown, looks like a ETD29, recycled off a PC board of unknown origin.

I started with an airgap by placing spacers at the legs of the core

measuring efficiency by multimeter measuring current thru a load resistor, and another the volts. Power supply has volt/amp meter - questionable accuracy but lets be generous as at least 10%.

checkout the huge spike in the primary when the switch turns off

25% efficient so far.
mosfet running hottish, so is snubber diode

Ok, so no air gap......


efficiency ~32%

so just checking the capability of 3xCD4069 buffers......substitute a TC427


maybe slight improvement to ~35%, certainly no big improvement

C back in the snubber...


mosfet not so hot, neither is the snubber diode, but still very warmish.

the cicuit is still very protyepish, as in stray inductances etc...


i'm disappointed, I thought should be easy to get a switcher to at least 60% efficient, but i'm stuck at half that.......

i need some ideas please

thanks

 

[xaccto]

how would efficiency be as a function of frequency of pulse width ? should it matter ?

in the above pic, the back emf spike, the width is almost half of the on-time pulse width, is that normal ?
with air-gap this rose to ~60V thats ~ the breakdown voltage of the mosfet being used, IRF044,
with-out air-gap this came to about 40V, so much better, but the width of the spike is still large, so still alot of energy in that....
how in principal does one reduce that ?

Thanks

 

[FvM]

The good point with your circuit is, that the MOSFET is basically protecting itself by absorbing the leakage inductance energy, unfortunately
at the expense of efficiency.

Leakage inductance is the main problem of your circuit. Normally, it's mainly a matter of transformer design, but I fear,
most of the leakage inductance is formed by your breadboard wiring.

Consider, that primary and secondary current loop are the place, where leakage inductance gets effective. The area of both
loops marked in the schematic bust be kept physical minimal. Please also consider, that the current flows from MOSFET source
back to the VIN bypass capacitor, so this is the loop of interest.

 

[xaccto]

hmm, thanks FvM!

seems I should then proceed with a PCB and see what I get then?


would you or anyone have a oscilloscope diagram of a proper designed flyback showing the Vgs and Vdrain similar to above so I can see what I should aspire to ?
thanks again

 

[marthinus]

You need an airgap in the transformer.The primary voltage should be clamped by
the reflected secondary voltage(the voltage on the capacitor).
The snubber's job is to clamp the voltage due to the leakage inductance only.

 

[FvM]

seems I should then proceed with a PCB

Not necessarily. Reducing leakage inductance is also possible with a breadboard, if you understand the basic problem. In a lab, prototypes are sometimes assembled upon a tin sheet acting a s low inductance ground plane. There are many options. Even twisting of specific wires can considerably reduce leakage inductance of your "circuit".

You need an airgap in the transformer.

Most likely yes, but not necessarily. It's clearly required to achieve power density with a flyback transformer, otherwise you get saturation at low power levels. It's also a matter of understanding power converter basics, particularly the relation of main inductance, leakage inductance and saturation current.

would you or anyone have a oscilloscope diagram of a proper designed flyback

There are many manufacturer application notes showing the same.

 

[xaccto]

just completing this thread with publish of final circuit and report that PCB of circuit
worked much better than the prototype.

I've been experimenting with this circuit as a battery rejuvenator for nearly a year now with various sealed lead-acid batteries.
Got some temporary response from one previously dead battery, but on the whole
no success, could be the batteries I tried were already too far gone.

This circuit really hits the battery though, you can hear a decent click in the battery
when it pulses.





[