Connecting flyback like an ignition coil

[John123]

What happens if you connect the HV return pin of the flyback transformer to the drain of the MOSFET like this? Could that solve leakage inductance complications and put it in series with the secondary?

Have I invented a masterpiece or a power supply/driver killer?

[klugesmith]

Are you going to try it? No need for MOSFET, you can get the same voltage by just touching wires on the primary. With your resistor, instead of On time, to control the primary current.

Your schematic is incomplete.  Transformer symbol needs dots to indicate the winding polarities.

How would omitting the diode change your expected operation?

[davekni]

I have one old flyback transformer that is wired that way internally, so it can work.

In normal TV use, there's a capacitor across the flyback primary winding to make it a resonant circuit.  That limits the peak voltage when the FET is turned off.  Without that, the FET drain voltage will rise until avalanche breakdown.  Some FETs are rated for reasonable amounts of avalanche energy, so may be OK.

Concerning just touching the wires, I've personally had a hard time achieving the same voltage.  The momentary arc between the wires as they separate slows the current fall rate well below what a FET can achieve.  Try it both ways to see.  Of course, the FET can be controlled to turn on and off at a much higher frequency, so produce much more power out.

[John123]

Are you going to try it? No need for MOSFET, you can get the same voltage by just touching wires on the primary. With your resistor, instead of On time, to control the primary current.

Your schematic is incomplete.  Transformer symbol needs dots to indicate the winding polarities.

How would omitting the diode change your expected operation?

The usual TV transformer configuration for max sparks, like this I believe. The diode is built into the transformer so I can't remove it, I just thought it could work as car ignition coils are like this just without the diode but they still utilize the flyback effect to produce a spark. Maybe I need to connect it the other way around instead.

I tried tapping wires manually but the sparks were not very distinct, looked like a very weak disposable lighter spark and depended how good I tapped it. I do want to try it on a real driver but keep chickening out at the last minute 

I have one old flyback transformer that is wired that way internally, so it can work.

In normal TV use, there's a capacitor across the flyback primary winding to make it a resonant circuit.  That limits the peak voltage when the FET is turned off.  Without that, the FET drain voltage will rise until avalanche breakdown.  Some FETs are rated for reasonable amounts of avalanche energy, so may be OK.

Concerning just touching the wires, I've personally had a hard time achieving the same voltage.  The momentary arc between the wires as they separate slows the current fall rate well below what a FET can achieve.  Try it both ways to see.  Of course, the FET can be controlled to turn on and off at a much higher frequency, so produce much more power out.

Well I was thinking the leakage inductive pulse would be redirected into the secondary coil with this configuration, but you're saying there would still be a big spike across the fet regardless? If so then it kinda renders this idea redundant, I just thought it might help with those super simple un-snubbed drivers on the internet last longer than 5 minutes. Apart from having no primary current sensing it's often primary switch overvoltage that does them in, even avalanche rated fets don't last all that long.

[klugesmith]

Thanks for augmented schematic.  Applause for the current sense note.  Mind the inductance of your 0.1 ohm resistor.  I've found that small wirewound power resistors in the 1-10 ohm range turn the corner & behave like inductors at around 1 MHz.

I think leakage inductance in a LOPT is not a desirable and necessary feature, as it is in MOT or NST.
It's an unavoidable evil that's aggravated by putting primary and secondary coils on different legs of the core.  Dave might know better.
The air-gapped core _is_ essential, but that's for energy storage, not to create or control leakage inductance.

Good luck !  I can't remember what circuit i used to drive a flyback "transformer" for continuous sparking, in about 1976.  Nothing looked up on the Internet, that's for sure.

Transformer in quotes because flyback converters (and ignition coils, which came first) use coupled inductors in a way not even close to an ideal transformer.  In a real transformer, primary and secondary currents flow at the same time, and their ampere-turns almost cancel each other out.  The difference (net ampere turns, just enough to magnetize the core to match volts per turn) is ideally zero, with no energy ever stored in the magnetic field.

Your one-transistor circuit reminds me of a Joule Thief.  Anybody ever try making a Joule Thief (blocking oscillator circuit) from a LOPT?

[John123]

Thanks for augmented schematic.  Applause for the current sense note.  Mind the inductance of your 0.1 ohm resistor.  I've found that small wirewound power resistors in the 1-10 ohm range turn the corner & behave like inductors at around 1 MHz.

I think leakage inductance in a LOPT is not a desirable and necessary feature, as it is in MOT or NST.
It's an unavoidable evil that's aggravated by putting primary and secondary coils on different legs of the core.  Dave might know better.
The air-gapped core _is_ essential, but that's for energy storage, not to create or control leakage inductance.

Good luck !  I can't remember what circuit i used to drive a flyback "transformer" for continuous sparking, in about 1976.  Nothing looked up on the Internet, that's for sure.

Transformer in quotes because flyback converters (and ignition coils, which came first) use coupled inductors in a way not even close to an ideal transformer.  In a real transformer, primary and secondary currents flow at the same time, and their ampere-turns almost cancel each other out.  The difference (net ampere turns, just enough to magnetize the core to match volts per turn) is ideally zero, with no energy ever stored in the magnetic field.

Are wire wound resistors likely to be too inductive around 40khz? As I ended up using 10 ohm versions of these on the gates of my half bridge driver. Look like these but I've never had a failure.

The current sense resistors I used are some special vishay types.

Your one-transistor circuit reminds me of a Joule Thief.  Anybody ever try making a Joule Thief (blocking oscillator circuit) from a LOPT?

I tried one of those really simple slayer exciter circuits once with a small tesla coil-like secondary and it worked really well considering the junkbox parts and zero calculations I did.

So something like these for a slayer exciter with a TV flyback transformer?

[davekni]

All the flyback transformers I have use one ferrite core leg for both windings.  I measured coupling factor on one, but don't recall the exact value.  It was over 90%, I think around 95%.  With a separate hand-added winding on the other leg, coupling factor was 81%.  To my knowledge, flyback circuits always desire high coupling (low leakage inductance).  For low-voltage flyback designs, as is common in small wall-warts, coupling over 99% is reasonable.  With all the insulation and space for HV diodes, coupling can't be that high.

Concerning resistor inductance, yes, that's caused issues for me as well.  Resistors vary over a wide range.  Sounds like you may have 0.1 ohm resistors designed for current sense, so designed for low inductance.  I've seen resistors labeled as "non-inductive", although there's always some inductance.  Can't personally offer any good rule-of-thumb.  For this flyback drive, are you using the resistor for just measuring current with a scope or meter?  Or, are you using it to terminate each pulse when the current ramps up to a given setpoint?  The latter is typical of my designs.  In that case, more inductance can be handled by low-pass filtering the current signal (say 1us time constant) to filter out the initial inductive spike.

A common snubber circuit for flyback drives is to use a diode feeding a capacitor and resistor:


Adjust the resistor value until the voltage across it (plus B+) is just under the FET Vds rating.  Pick a capacitor that forms a time-constant with the resistor of several cycles.  The resistor may need to dissipate significant power depending on leakage inductance.  A higher-voltage FET and correspondingly higher clamp voltage improves efficiency.  If power isn't too high, a TVS diode instead of a resistor avoids the need for adjustment.  I use this clamp circuit at home and at work.