Half bridge protection measures

[Sk1ppy]

Hi. Recently I constructed a half bridge flyback driver based mostly off of Mads' Flyback driver, and so far it works great. However, I do have a few concerns regarding the half bridge itself.

Originally, I constructed the circuit identically, with the addition of dual 18v zener diodes between the gate and source to clamp and excessive voltage on the gates. I also used irfp260 mosfets. This worked great on 30v, but unsuprisingly when i fed it 60v from a variac while (foolishly) operating near resonance I unsurprisingly blew the MOSFETS. The gate diodes and the zeners. I replaced the FETS with irfp460s, hoping that the higher voltage rating would provide incresed resiliance to potential voltage spikes.  Unfortunately, my gate resistors were blown and I had to remove them. My bridge currently works, but it draws about 30 watts when idle near resonance, which it did not do before. Could this be due to the lack of gate resistors?  Or maybe I blew my protection diodes as well and my multimeter didn't detect it?

Any ideas would be appreciated.

Secondly, my ultimate goal is to operate with rectified mains (170VDC) at resonance. What additional measures can I take to protect my MOSFETS in rather extreme operating condition?  From my research it seems like I would want to Add a spark gap to my flyback, increase primary turns, and maybe add snubber capacitors

Thanks

[John123]

What about something like this which trips overcurrent shutdown when a threshold has been reached? It's not my design and I never got around to trying it, but it could save the switching devices.



Then there's the feedback design to give zero current switching, this could be what's causing the fets to blow during resonant operation if they're switching high reactive currents during a switching transition. Not sure how to include current limit to this one unfortunately.


Are you using the parasites of the flyback to give resonant operation or an external capacitor and/or coil? Also I heard some people say to remove the core spacers, but in my case I find it gives me the needed current limiting to pull arcs without massive currents being drawn.

There was one design on youtube which used an sg3525 or tl494 bridge and the guy had come up with cycle by cycle current limiting, but he was too uppity to share the schematics.

Edit: Lack of gate resistors could be a problem, maybe the gate drive transformer became part of the resonant tank somehow! With my little 50v half bridge I always used 10 ohm minimum on the gates of IRFP260's along with the anti series zeners.

Can you scope the waveform of the gate drive transformer outputs? Maybe it's somehow partially damaged the gate driver chips, also increase the size of the gate drive blocking cap to at least 2uF as 0.22uF made my gate drive waveforms a bit lackluster at regular flyback frequencies (40-50khz was resonant for mine IIRC).

Definitely no bad connections or punctured transistor insulators? What about corona discharge around the gate drive transformer across the isolation boundary.

[Sk1ppy]

I have a feeling that the gate resistors could have something to do with the issue, so I'm swinging by the electronics store as soon as possible to get replacements. As for that pll circuit, it looks very interesting. Up until now I've never seen a schematic for a cd4046 flyback, and haven't felt that comfortable designing my own due to the lack of info on the chip.  As far as "resonance" goes, I probably was not hitting the resonant frequency of the flyback, and was probably on some harmonic of it. I will have to do continued experimentation to improve my results.

During my research, I found that some H bridge circuits used in sstcs have schottky diodes placed after the fets which disables the internal body diode, as the recovery time isn't fast enough and can lead to shootthrough. As a replacement, they use ultrafast diodes like the ones I already have. I don't know whether this would help, as i'm operating at lower frequencies where the recovery time may not be a problem.

Thanks for the suggestions

[John123]

I have a feeling that the gate resistors could have something to do with the issue, so I'm swinging by the electronics store as soon as possible to get replacements. As for that pll circuit, it looks very interesting. Up until now I've never seen a schematic for a cd4046 flyback, and haven't felt that comfortable designing my own due to the lack of info on the chip.  As far as "resonance" goes, I probably was not hitting the resonant frequency of the flyback, and was probably on some harmonic of it. I will have to do continued experimentation to improve my results.

During my research, I found that some H bridge circuits used in sstcs have schottky diodes placed after the fets which disables the internal body diode, as the recovery time isn't fast enough and can lead to shootthrough. As a replacement, they use ultrafast diodes like the ones I already have. I don't know whether this would help, as i'm operating at lower frequencies where the recovery time may not be a problem.

Thanks for the suggestions

For the gate resistors and zeners you could just use what you've got laying around, anything between 11-18v would be fine for zeners and they don't even need to be matching. Gate resistors just cobble some together to make 10-30 ohms. Those decoupling capacitors near the gate drive IC, do you have them right next to the IC pins? Without them and some extra electrolytic caps from my junk bin my gate drive waveforms looked very ringy when the bridge was on.

The resonance on mine was found by putting the arc electrodes about 2 inches apart and sweeping the frequency knob until it jumped the gap giving the highest output (this was with around 24v). Without a safety gap it has a habit of destroying the transformer as it looks for other ways to complete the hv circuit.

I've not tried disabling the internal diodes on a flyback driver either, but I heard it can cut down on mosfet heating.

Back around 2014 there was a group of experimenters on youtube who cracked what you're trying to do, resonant offline (115v-240v) bridge with cycle by cycle current limiting but they decided to keep the schematics a secret. 

The notes on this schematic are pretty useful such as 3-6v for every turn on the primary coil (right click open image in new tab for better text readability).

[Sk1ppy]

Right now I have 18v zeners between the gate and source, and I had 10ohm gate resistors.  I suspect that the inductive kick of the flyback when extinguishing an arc shot through and blew my fets - although they died closed circuit, and the bang was rather impressive (at least on one fet.)  I will have to do more experimentation, and ramp up voltage with the driver tuned further away from resonance. The PLL driver has piqued my interest quite a bit. I like the automstic tracking of resonance - and the ability to audio modulate seems like a big plus.

[Sk1ppy]

With regards to the driver IC capacitors -- I have 10uf electrolytics connected from vcc to gnd on the IC, and my waveforms are all very clean

[John123]

Right now I have 18v zeners between the gate and source, and I had 10ohm gate resistors.  I suspect that the inductive kick of the flyback when extinguishing an arc shot through and blew my fets - although they died closed circuit, and the bang was rather impressive (at least on one fet.)  I will have to do more experimentation, and ramp up voltage with the driver tuned further away from resonance. The PLL driver has piqued my interest quite a bit. I like the automstic tracking of resonance - and the ability to audio modulate seems like a big plus.

Actually I think Mads had a PLL design on his website which pretty much used the circuit you're using now but with a few extra parts, let me see if I can find it.

I think this might be it http://kaizerpowerelectronics.dk/tesla-coils/kaizer-sstc-ii/


The top half is a self oscillating oscillator, the bottom parts are for making sstcs sound like spark gap tesla coils by running it in bursts. Not sure how it would sound with a flyback however, possibly like an 8bit chiptune/midi. For the antenna loop a bit of wire around the core once.

I'm sure Mads will be along at some point and he can advise if its suitable for use with flybacks.

[Sk1ppy]

I see no reason as to why it would not work. Fundamentally however, it does essentially the same thing as the PLL, but with less control as the drive signal is created only with feedback.  Ultimately my issue isn't really driver topology as the TL494 + driver IC pair does a fantastic job. I'm more concerned about keeping my FETS safe as I pump huge amounts of power in. I suppose at some point I may have to invest in IGBT bricks...

[John123]

I see no reason as to why it would not work. Fundamentally however, it does essentially the same thing as the PLL, but with less control as the drive signal is created only with feedback.  Ultimately my issue isn't really driver topology as the TL494 + driver IC pair does a fantastic job. I'm more concerned about keeping my FETS safe as I pump huge amounts of power in. I suppose at some point I may have to invest in IGBT bricks...

It looks like most over current protection trips use some kind of current transformer on the primary to trip the drive IC or the gate drive enable lines (ignore the 555 timer).



The thing with PLL is you can tune it and "lock onto" where the MOSFETs only switch when the primary current has fallen to zero, this can prevent certain mosfet failures because they don't need to hard switch when high primary resonant reactive currents are passing through them. Have you still got the stock core spacers in the core to make a core gap? When I took them out I got all sorts of random high current runaway when I tried to draw arcs.

I swear there was a circuit out there which combined what you've got now with PLL, but I can't seem to find it. All I can think of is google image searching for "pll sstc tl494 4046".

I didn't think resonant action in a bridge design subjected the fets to any over voltage condition, shouldn't they always see the supply rail max? Feeling brave enough to scope around the primary when its driving a flyback?

[Sk1ppy]

From what I understand, resonance would actually be easier on the switches due to switching at zero voltage -- but with resonance also comes greater voltages on the secondary, which i would assume would lead to more intense kickback on the primary when the arc extinguishes. As for the probing the primary while operating, I think it would be allright, but to be safe ill keep the voltage on the bridge low.

I still have the spacers in my transformer, and from what I understand is that they can be removed, which decreases mosfet heatingx but makes core saturation easier to achieve.

[John123]

From what I understand, resonance would actually be easier on the switches due to switching at zero voltage -- but with resonance also comes greater voltages on the secondary, which i would assume would lead to more intense kickback on the primary when the arc extinguishes. As for the probing the primary while operating, I think it would be allright, but to be safe ill keep the voltage on the bridge low.

I still have the spacers in my transformer, and from what I understand is that they can be removed, which decreases mosfet heatingx but makes core saturation easier to achieve.

Times like this would make a x100 probe very handy! I'm always scared of arcs jumping between the secondary to primary when I connect my scope probe up to a HT circuit with insulation being pushed to an inch of its life.

I had bad luck taking the spacers out with this kind of driver, everyone kept saying to remove them but when I did it acted like the core was saturated or the primary inductance suddenly dropped making current shoot up without any output.

Wonder if Mads has scoped the peak drain voltages when a bridge is being used to drive a flyback at resonance (this is actually his forum).

BTW I'm not an expert by any means, I just dabble with things until they sort of work lol. Hopefully a more experienced forum member can give you better help.

[Sk1ppy]

No worries about being inexperienced - I myself am rather new to the world of high frequency power electronics. Previously, I only ever really dabbled in iron core stuff (Especially MOTS), and so my knowledge was rather limited. To be honest, I'm surprised my perfboard soldering isn't complete garbage.

[John123]

No worries about being inexperienced - I myself am rather new to the world of high frequency power electronics. Previously, I only ever really dabbled in iron core stuff (Especially MOTS), and so my knowledge was rather limited. To be honest, I'm surprised my perfboard soldering isn't complete garbage.

Got any pics you could share of both sides of your circuit board? Maybe I could spot something that you might of missed. I built that driver by Jan Martis and haven't managed to destroy any component over than flybacks from overvoltage, although I only took it up to 48v.

Were you getting big arcs before it blew up? Secondly how is the input power being measured and is anything getting hot? Could be some kind of reactive current draw you're seeing.

Edit: Yet another thought, how much dead time is there between each gate driver firing? I vaguely remember someone telling me on 4hv that too much dead time with a gate drive transformer is not good as it lets the gates wonder/float momentarily. Then when bridge voltage is applied the reverse transfer capacitance charges the gates up and either causes shoot through or overvolts them, I think just enough deadtime to avoid cross conduction is needed but no more.

To check for deadtime shoot through apply voltage to the bridge with no primary coil connected and start the driver up, now connect a low resistance resistor (1 ohm is more than enough) in series with the bridge and scope across it, momentary shoot through will appear as current spikes even though it shouldn't be drawing any current.

Edit2: Yep, just had a look for it but it's yet another great 4hv thread lost to the sands of time.

[Sk1ppy]

Sadly I do not have any pictures at this time, and it might be a day or two before I can get some sent.

With regards to deadtime, the TL494 has a built in minimum 5% deadtime, and I have connected pin 4 (Deadtime control) to ground, which minimizes the deadtime.

Right now I am measuring power draw via the display on my variable power supply. I am operating around 32vdc, which is as high as my supply goes.

In terms of arc length, I got very impressive arcs (I estimate upwards of 6 inches) before I blew the half bridge

[John123]

Sadly I do not have any pictures at this time, and it might be a day or two before I can get some sent.

With regards to deadtime, the TL494 has a built in minimum 5% deadtime, and I have connected pin 4 (Deadtime control) to ground, which minimizes the deadtime.

Right now I am measuring power draw via the display on my variable power supply. I am operating around 32vdc, which is as high as my supply goes.

In terms of arc length, I got very impressive arcs (I estimate upwards of 6 inches) before I blew the half bridge

Wow, was that with just 32v and are you using extra resonant capacitance or just the parasitics of the flyback? Don't think I ever got 6 inch arcs even at 48v relying on parasitic capacitance alone.

[Sk1ppy]

I was simply using the parasitic capacitance of the primary coil. I believe I was operating around 70v when i got those arcs (and blew the bridge)

[John123]

I was simply using the parasitic capacitance of the primary coil. I believe I was operating around 70v when i got those arcs (and blew the bridge)

Still very impressive, out of interest place an ohm meter across the flyback secondary coil and measure resistance in both directions. These drivers often overvolt and destroy the flybacks internal diodes at these voltages, you might now have an AC flyback.

Biggest drawn arc I ever got was around 1ft from a ZVS driver, biggest jump the gap arc was around 5 inches from a single ended true flyback driver (transformer arced over soon after).

[Sk1ppy]

One foot?  That is very impressive!  I'm sure that flyback was not having a very good time haha. Ill see if the diodes are blown in the secondary.  Soon I think i will wind a transformer of my own more suited to the abuse.

[John123]

One foot?  That is very impressive!  I'm sure that flyback was not having a very good time haha. Ill see if the diodes are blown in the secondary.  Soon I think i will wind a transformer of my own more suited to the abuse.

Yeah (48v battery bank input voltage), I was crapping myself thinking the fets were going to explode so I put the driver in a closed box. I literally couldn't wind a thick enough primary coil to prevent it getting hot, the resonant capacitor and fets didn't heat too much but the axial leads for the capacitor did. My chicken stick grew to 4ft during that, wasn't taking any chances with a 1ft long 1 inch thick channel of white plasma!

Anyway sorry to derail your thread, just thought of something else. Maybe the gate drive IC's could be damaged but it only shows up when they're having to fight the reverse transfer capacitance combined with the flyback being connected?

[Sk1ppy]

I doubt it. I think the problem boils down to how my half bridge itself is constructed. So far my driver works flawlessly. I think I'm going to increase the number of turns on the primary to bump up the impedence which will hopefully limit current and thus heating.