TV flyback transformer mosfet protection

[Carabus]

Hi,
How can I protect my mosfet against voltage spikes when the primary is turned off? I have tried putting a shottky in reverse to source voltage, but that messed up the arcs.
I want a big arc which I can briefly achieve by using 50v primary voltage.
Currently, I am using irfp250n

My primary is just wire wrapped around the flyback ferrite 8 turns.

[Carabus]

My cheap LCR says that the primary inductance is 0.01mH.

[CaptDon]

I would probably start by getting rid of the CMOS 555 and putting a real old school 555 in its place. Any snubber circuit used to kill overshoot on the mosfet will probably kill most of your arc. MAYBE that is why horizontal output transistors are rated 1500vce and your mosfet isn't??

[Carabus]

I use old 555, I put cmos in multisim with no big reason. Maybe using IGBT is better? The Ones rated for kilovolts.

Are you talking about internal rectifying diodes?

[T3sl4co1l]

What is this for, a model of a TV deflection circuit or something?  Or are you using the transformer alone just for HV?  If the latter, why not just a free-running oscillator and driver?

To a certain extent the drain voltage doesn't need to be clamped separately, as the transistor can handle some by itself; it won't last long this way, however (check the datasheet for avalanche ratings; they often give single-pulse figures, rarely repetitive -- and when they do it's only a tiny fraction!).  Good options are: RC snubber, RCD (clamp) snubber, RCD (dV/dt rate) snubber, or TVS.  Some of these can be combined, like say a rate snubber to reduce switching loss, then a clamp snubber or TVS to limit voltage, then an RC snubber to dampen ringdown; etc.  But probably you'll only need one of these.

Tim

[Carabus]

This is for musical arcs. I will try your suggestions

[Carabus]

Can you explain how to select tvs diode for this application? Also where should it be connected? Gate-  source maybe?
mosfet switching frequency is 36-37kHz

[T3sl4co1l]

Source to drain.  Use one with clamping voltage below Vds(max), or a series stack totaling the same.  SMAJ or P6KE series should be fine, unless you find they get too hot in which case choose a larger size (SMCJ or 1.5KE, etc.).

Musical arcs as just generating tones (given from a logic-level input)?  Or should it be a singing arc (general purpose audio input)?  The latter you may find some success modulating the 555's pin 5, or other methods (like using a 555 with better modulation, or better still, using a UC3843 or the like).

Tim

[Carabus]

Okay, I had some 12v 5W zeners lying around and it seems to work. I will try to implement 1.5KE68A because it's easily available for me in radio shop. Do I need resistor in series with diode, because it gets hot quick? Will resistor affect the suppressing of voltage spikes?

[magic]

For maximum output voltage from a flyback you want maximum voltage on the primary.
Also, any power burnt on the primary is not dissipated on the secondary.

To a certain extent the drain voltage doesn't need to be clamped separately, as the transistor can handle some by itself; it won't last long this way, however (check the datasheet for avalanche ratings; they often give single-pulse figures, rarely repetitive -- and when they do it's only a tiny fraction!).

What about avalanche rated MOSFETs?

Some of them really seem to suggest that they can take anything as long as Tjmax is not exceed (and you know it from transient thermal impedance plots). Not sure if it means they can be used as high power zeners at DC too. But they surely promise to withstand operation with unclamped inductive loads.

[T3sl4co1l]

Okay, I had some 12v 5W zeners lying around and it seems to work. I will try to implement 1.5KE68A because it's easily available for me in radio shop. Do I need resistor in series with diode, because it gets hot quick? Will resistor affect the suppressing of voltage spikes?

Well, you'll need quite a stack of those 12V zeners, but if you don't mind putting them together, no problem.

Resistor, nah.  Mainly just pick a transistor (and thus zener/stack) rating high enough to avoid most of the energy that's supposed to go to the secondary anyway.  Like magic said, you want fairly high voltages here.

What about avalanche rated MOSFETs?

Some of them really seem to suggest that they can take anything as long as Tjmax is not exceed (and you know it from transient thermal impedance plots). Not sure if it means they can be used as high power zeners at DC too. But they surely promise to withstand operation with unclamped inductive loads.

The non-rated types just have inferior handling (in terms of peak current or energy), I think a hotspot issue?, which happens to trade off directly with body diode t_rr.  Rated types usually handle full energy, but not at "full" current: usually the avalanche rating is close to the DC current rating, but it's also nowhere near Id(max,pk).  Due to the lower current, the inductance is also usually quite large and so the pulse relatively slow (but that's fine; find the spot on the transient thermal resistance plot and you should see the energy checks out, and so it stands to reason, the energy would be smaller if the current could be higher*).

*Or, they compromise and don't tell you the ultimate (limiting) current, but inflate the parameter somewhat as a result.  No idea.

I don't recall if it's strictly due to deposited charge (so, the current is irrelevant, even to currents higher than given in the datasheet, as long as pulse width is short enough), or if there's some curve to it?  Or of course, if hotspotting can't be avoided completely, so there's still a limiting current.

Anyways, the mechanism is hot carriers depositing in / disrupting / damaging the gate oxide.  So, I guess, a similar mechanism to EPROM/Flash wear?  (Not sure if it causes Vgs(th) shift in the same way.  But, there's no floating electrode or ion migration for that to happen with either, I guess.)  It's cumulative, so, at low enough currents and duty cycles, maybe it lasts long enough to make a viable product; most likely for anything more frequent than electromechanical switching, it's not going to survive.

Hence why the repetitive figure is so small; I'm not sure what count they test/estimate with, a million pulses maybe?  Whereas the single figure is maybe good for a couple total or something.  Not a thermal figure where it can do it forever at low enough duty (long enough time between pulses) (or, give or take thermal stress anyway).

A maybe giveaway is they never use avalanche in protected MOSFETs; instead they use a zener strapped from G to D.  Fair, that might have to do with the integrated environment -- avalanche in the power section probably sends free charge carriers throughout the die?  (Or at least neighboring sections of the control circuit.)  Plus the channel is more than strong enough to do the job (i.e., Id(max,pk) at least), so it's not like it's a poor solution.

Tim

[CaptDon]

Any TVS or snubber will reduce the energy available to your arc. I would seriously look at a horizontal output specifically rated transistor!!! They are designed for exactly this use and many can withstand 1500 volts collector to emitter. Why waste time perfecting a mosfet circuit when purpose rated transistors exist and are readily available. We were doing this stuff back in the 1970's robbing parts out of solid state televisions that had been blown up by lightning or had dead CRT's. Your drive circuit from the 555 only needs a slight modification to drive the H.O. transistor!! You want the energy in your arc and not wasted heating diodes, resistors, snubbers, TVS, Etc., Etc.......
To get a good fast release on the output transistor include a single diode (1n4007 works) 'Baker Clamp' circuit collector to base on the transistor to prevent the transistor from reaching saturation. It will place a tiny bit of heat into the transistor from not being in hard saturation but in the final analysis may be well worth the time and trouble to get quick turn off of the tranny and more fizz in your arc!!

[Carabus]

Turns out I have a horizontal output transistor (probably from the same TV). S2000N
It is rated for 1500V, but problem is that hfe is only 9, my 555 can output 200mA at best, how can I increase base current so I have 7-8A at the output?

[CaptDon]

May have to use a pre-driver transistor in a darlington configuration. The baker clamp circuit works even better when tied to the pre-driver base of a darlington pair! The single diode drop of the 1n4007 can easily overcome the two-diode drop of the pair of base/emitter junctions in a darlington. There are a lot of TO-3 style darlingtons available up to around 400-500vce but I am unsure if that would be high enough in your circuit. If the arc fails to strike and drain the available flux energy of your flyback then the overshoot at the collector of your output transistor will be tremendous. MOSFET's have a body diode that will eliminate the negative undershoot damage. You may need a 'damper' diode (there was probably one in the t.v. you got the flyback from) since the normal BJT may not include a damper. Some horizontal output specific BJT trannys had an internal damper diode!! Easily determined with an ohm meter measuring from C to E and looking for the conduction of the diode with negative on collector and positive on emitter.

[T3sl4co1l]

Something like this:



The 22R 1W sets output base current.  If a lower supply voltage can be used, it'll waste less power in this resistor (and lower values can be used instead of the 1k's driving the 4401/3).  Alternately, 2N7002 (NPN --> Nch) and BSS84 (PNP --> Pch) can be done, removing the base divider resistors instead just driving the gates directly.  Which works better with a CMOS 555.

Instead of UF4007, place the zener/TVS stack in this place.  The 100R + 0.001uF snubber is kinda take-it-or-leave-it.

Also a MOSFET output is fine, something pulled from a 100W+ flyback supply will probably do (typically 8A 900V or something like that).  Which can be driven from the same circuit actually, heh, but even easier with the 555 directly as your original, or with a gate driver added inbetween.

Tim

[Carabus]

Sorry for a lot of questions, why do we need pnp and npn transistors instead of one?
Also, what is the purpose of 2 diodes?

[magic]

Anyways, the mechanism is hot carriers depositing in / disrupting / damaging the gate oxide.  So, I guess, a similar mechanism to EPROM/Flash wear?  (Not sure if it causes Vgs(th) shift in the same way.  But, there's no floating electrode or ion migration for that to happen with either, I guess.)  It's cumulative, so, at low enough currents and duty cycles, maybe it lasts long enough to make a viable product; most likely for anything more frequent than electromechanical switching, it's not going to survive.

Well, I looked up some application notes and it seems that people who want me to buy their avalanche MOSFETs don't entirely agree with you, whatever it means

APT APT9402
IR AN-1005
ST AN2344

They all agree on two failure mechanisms:
- excessive current triggers an internal parasitic BJT structure, avoided by not exceeding the max avalanche current rating
- excessive power dissipation overheats the die, avoided by keeping temperature under control

APT explicitly states that contrary to initial worries, no long term wear out mechanism has been identified. That was in 1994, and the other appnotes are from the 2000s and say nothing at all.

IR has much more plots in their datasheets and AN-1005 goes into detail on how they are produced, I will have to dig into it. It does involve using the transient thermal impedance plots.

ST cautions against using the transient thermal impedance plots like "certain manufacturers", because avalanche behavior may be different than linear mode. APT shows comparison of avalanche vs linear for one of their devices - avalanche is 2x worse.

IR shows an application example where inductive load is driven with no other clamping or snubbing. Nothing about long term damage.

They all agree that devices need to be specially designed and tested to stand avalanche. APT further elaborates that it involves a tradeoff against gate charge.

[CaptDon]

T3s, that is an excellent driver circuit you provided. I think he wants to modulate the 555 to get a singing or musical arc. Your thoughts on the driver circuit are similar to mine. You have included driving the base both on and off, that will help with the turn-off time to get a good clean 'flyback' without burning power during the linear phase of turn-off. With a hybrid darlington arrangement where the collector of the predriver runs at a fixed low voltage and the collector of the output device is connected to the load I have had excellent luck with the baker clamp. The single drop of a 1N4007 can hold down the maximum base drive into the pre-driver and keep the output transistor out of hard saturation due to the pair of B/E drops in the real darlington or even the hybrid darlington. I have used this in push/pull power supplies at over 1kw with very minimal heatsinks and a fan and little to no damping.

[T3sl4co1l]

T3s, that is an excellent driver circuit you provided. I think he wants to modulate the 555 to get a singing or musical arc.

Yes; so the stuff attached to the 555 can be ignored, and replaced with the existing modulator circuit.  So, just the right-side driver stuff being important.

Amazingly enough, I don't have a modulated oscillator circuit handy, so you'll have to piece things together I'd afraid.

Your thoughts on the driver circuit are similar to mine. You have included driving the base both on and off, that will help with the turn-off time to get a good clean 'flyback' without burning power during the linear phase of turn-off. With a hybrid darlington arrangement where the collector of the predriver runs at a fixed low voltage and the collector of the output device is connected to the load I have had excellent luck with the baker clamp. The single drop of a 1N4007 can hold down the maximum base drive into the pre-driver and keep the output transistor out of hard saturation due to the pair of B/E drops in the real darlington or even the hybrid darlington. I have used this in push/pull power supplies at over 1kw with very minimal heatsinks and a fan and little to no damping.

I don't really care about the Baker clamp here, turn-off will be sharp enough anyway (delayed by some ~1us for these typical BJTs, but t_f ~ 200ns is common enough).

If you want to, mind that base voltage needs to be above diode Vf + transistor Vce(sat) for any clamping to occur.  Classically (e.g. 74LS logic family) this requires a schottky diode (less Vf than a PN diode) wired B to C.  A high voltage diode won't do it across a single transistor: too much Vf and Vce(sat), not enough Vbe.  So we can increase Vbe by adding a follower transistor -- as a darlington -- or tapping the resistor divider to the base.

Tapped resistor (top, don't mind the bottom half):



(Bottom half: by using a matching transistor, in inverted diode-strapped mode, a tighter Vce(sat,clamped) can be had. This... wouldn't work well here, given the high capacitance of another whole-ass power transistor. )

Darlington (left, don't mind the right half):



Mind, this circuit has a tendency to oscillate (the right was an attempt to simulate and explain that observation, but none of the values really look reasonable given the breadboarding?) so be careful.

Fo use a high speed diode -- it will be forward-biased by whatever drive current is, so also deliver at least as much recovery current, slowing turn-off.  1N4007 typically is some microseconds, so it's a very worthwhile change to a UF4007 of some 50ns.

Tim

[T3sl4co1l]

Well, I looked up some application notes and it seems that people who want me to buy their avalanche MOSFETs don't entirely agree with you, whatever it means

APT APT9402

1994, we've had several whole generations since then!  Heh, they show a lateral DMOS; I wonder if the illustrated current path is far enough away from the gate (i.e., channel length and diffusion depth) that this mechanism doesn't occur?  Or that the feature size was large enough (~um?) it didn't matter.  Similarly for the HEXFETs (VDMOS), they show breakdown under the source contact region, away from the gate.

I've certainly had HEXFETs fail under such service, without overheating, so I don't think they're immune as a class.

Here's one that discusses it, in regards to [modern] trench MOS:
https://www.infineon.com/dgdl/Infineon-ApplicationNote_Some_key_facts_about_avalanche-AN-v01_00-EN.pdf?fileId=5546d462584d1d4a0158ba0210977cde

Note that modern processes use fine feature sizes (~250nm I think?), so the gate/source structures are very small/shallow, tightly packed, and this has the effect of making a more nearly uniform source contact with respect to the body/bulk.  So, smoothing out variations in electric field (less crowding under avalanche), and also over such small length scales, diffusion of minority carriers is pretty much total (it's typically over some ~um in Si) so it would make sense that some high-energy carriers can make their way over to the gate oxide and do hijinx there.

Another consequence is, I think gate leakage should be sensibly increased during avalanche?  Not like nearly enough to cause problems with gate drivers, just that it should be measurable (~uA?).  Hmm, I might test that, that sounds neat.

They all agree on two failure mechanisms:
- excessive current triggers an internal parasitic BJT structure, avoided by not exceeding the max avalanche current rating
- excessive power dissipation overheats the die, avoided by keeping temperature under control

Ah yeah, how could I forget the parasitic BJT, there's your ultimate current limit.  Similarly, avalanche is degraded when applied with high dV/dt (since the R and C effects add), though in practice it's almost impossible to apply dangerous dV/dt so at least that's not a big deal.  (Still something to check; not all FETs are created equally.  Make sure not to cheap out on dV/dt.  There's also a dV/dt limit for diodes, though I don't remember what it's actually about, maybe something with surface charges??)

Tim

[magic]

Going too crazy with primary voltage is also crazy because you could damage the secondary. I know nothing about TV flybacks and how robust they are, but here are the basic principles of flyback, if you don't know:

Initially, the FET turns on and applies your input voltage (5V? 12V? 24V? whatever it is) to the primary. The secondary's output goes negative and is blocked by a diode (inside the TV flyback brick, afaik).
Current gradually ramps up through the primary, building a magnetic field in the core.
The FET turns off, flyback effect inverts polarity of voltage on both sides. The output diode start to conduct output current, the FET blocks the sum of your input voltage and the primary winding voltage.

Ideally, you should have some clue of the maximum permissible secondary voltage (HV) and the transformer's turns ratio (n). Then, the MOSFET needs to be able to withstand HV/n primary voltage (on top of your input voltage) because a transformer is a transformer and that's how turns ratio works. At the same time, it may make sense to clamp primary voltage to no more than HV/n in order to protect the secondary from overvoltage.

Another matter is primary leakage inductance which doesn't transfer its magnetic energy to the secondary and always dissipates it as a brief overvoltage on the MOSFET right after turn-off. This needs to be absorbed by the FET or snubbed, and it will surely need to be snubbed if you go BJT because they can't deal with avalanche at all.

1994, we've had several whole generations since then!

Well, yes, and some of them are rated for repetitive avalanche.

I've certainly had HEXFETs fail under such service, without overheating, so I don't think they're immune as a class.

IR never claimed otherwise. But if you look at IRFP250N datasheet, it says:
maximum repetitive avalanche energy: 21mJ, pulse width limited by peak junction temperature, see fig. 11 "transient thermal impedance"

So to fig. 11 we go. 30kHz switching frequency is 30µs period. Avalanche time will be 15µs for 50% duty cycle, 3µs for 10% avalanche duty cycle. (= 90% conduction duty cycle) and so on. We are clearly in the region where the datasheet implies that junction temperature ripple is negligible and permissible pulsed power dissipation is simply the max continuous power dissipation divided by duty cycle. Also, 21mJ·30kHz = 630W, so this limit is way above anything we could practically achieve.

Taken at face value, it seems to imply that OP could completely open circuit the secondary and just burn tens of watts of power in avalanche, subject to sufficient heatsinking.

And again, IR's own application note shows a circuit where some fuel injectors are driven using nothing but the FETs as clamps.

Here's one that discusses it, in regards to [modern] trench MOS:
https://www.infineon.com/dgdl/Infineon-ApplicationNote_Some_key_facts_about_avalanche-AN-v01_00-EN.pdf?fileId=5546d462584d1d4a0158ba0210977cde

Yep, but that's a different manufacturer (yes, I know they bought IR) and different product family and the note ends with stating that Infineon doesn't design their parts for repetitive avalanche due to the tradeoffs involved. Apples to peanuts

[Carabus]

I have a lot of these transformers so burning a few is no big deal I'd like to see what their limits are.

I'm a bit confused. Do I need the primary flyback voltage to be as big as possible, but not damage the transistor, or does the primary coil flyback voltage not affect output so I need it to be as low as possible?
Also earlier I found that if I put, for example, a diode reversed to the coil the output becomes trash. Too bad I don't have any oscilloscope I would like to probe my circuit with modifications at low voltage to see what's going on there.

[magic]

It's a transformer. Secondary voltage is primary voltage multiplied by turns ratio and inverted. Do the math

I played with an automotive ignition coil myself, the nice thing about traditional ignition coils is that they have known 1:100 turns ratio. So you get 25~30kV (about their normal operating voltage) from a 250V MOSFET (actual breakdown voltage is a bit higher than the rating). I used an avalanche rated part (also from IR) but my circuit was a "press the button, get one spark" kind of thing, so I have only avalanched that transistor maybe a few hundred times in total. A 30kHz switcher will obviously exceed this number in a fraction of a second.

TV flyback bricks are more complicated, it is my understanding that they usually contain built-in diodes or diode-capacitor voltage multipliers. When you put too much voltage on the output, the MOSFET will turn on as usual, make the secondary slightly negative (-1.2kV for 12V input and 1:100 transformer) and those diodes and/or capacitors will break down and either fail catastrophically immediately or slowly overheat when this is repeated.

You prevent this failure by clamping either side of the transformer with a zener or something similar. Primary is easier done than the secondary, because it's low voltage and components of your choice with known ratings.

[Carabus]

UPDATE!

So I got myself IRFP460 mosfet and TVS diode 342V and it seems to work very well at 50VDC 20+Amps Hot Big 5-7cm ARC. Also I got myself a scope and did some tests with the primary coil and found that it resonates around 18-19kHz, don't worry I did the tests with 8v 1A by powering the coil and then disconnecting it maximum flyback voltage around 25v. Also found that my 555 duty cycle is 61%
So I am planning on driving flyback at 18.5kHz also reducing the duty cycle to 50%, also adding amplifier before audio input to play music.