Need feedback on this circuit (flyback driver)

[Domo]

Hello all,
I would like to request feedback on this design, maybe I am missing something that i am not aware about, and I do have few questions about it..

Basically, taking rectified AC voltage (170 VDC) feeding to R1/Q1, which acts as voltage regulator on source end. D2 prevents ringing voltage to return to Q1. D3 is a snubber diode. Q2 controls freq and PWM, which is listed nest to it. Q1,Q2 can handle 450V@20A. D2 and D3 can handle 1200V@8A

The question that i have, with D3 being in place, do I need to worry about ringing voltage? If so, what voltages can i expect? 1kvpp? 2kvpp?

[PhotonLab]

Hello,
first remove Q1 (is not working the way you think, i.e. a current regulator) and D2, they are not necessary for the purpose of making spark
D3 must be a fast diode if you really want a snubber network, diode + capacitor, there's some math involved, add a 10 kOhm resistor betwen gate and GND.
Cheers
M.
 

[Anders Mikkelsen]

D3 will kill the flyback pulses on the primary and therefore also strongly limit the output voltage. RCD snubbers can work here, but the most common approach for simple drivers is to just let the flyback pulse go as high as it does, and dimension the MOSFET to handle it. This limits how high you can go with the input voltage without exceeding the breakdown voltage of your IRFPs.

Making a linear regulator with MOSFETs as shown in your diagram will lead to a lot of dissipation and likely blown transistors as MOSFETs often have very limited tolerance for linear mode operation at higher voltages (above some tens of volts). I don't recommend running anything directly off mains, especially if you have limited experience with switching circuits. Both due to the high voltage, lack of isolation from mains, high fault energy available and limited tolerance for faults. A power supply of some 20 - 40 V will be much better and safer here.

[Domo]

The purpose of Q1 is a voltage regulator, it worked on smaller scale (lower current, but still over 100V DC). Purpose of it is to reduce voltage to coil, as some flybacks operate on 90V or 110V. Q1 and Q2 are on heatsink with fan.

D3 must be a fast diode if you really want a snubber network, diode + capacitor, there's some math involved, add a 10 kOhm resistor betwen gate and GND.
Cheers
M.

D3 (and D2) will be RHRP8120 on heatsink.

D3 will kill the flyback pulses on the primary and therefore also strongly limit the output voltage. RCD snubbers can work here, but the most common approach for simple drivers is to just let the flyback pulse go as high as it does, and dimension the MOSFET to handle it. This limits how high you can go with the input voltage without exceeding the breakdown voltage of your IRFPs.

That is my concern, wont the ringing voltage be 2x, 5x of input voltage? IRFP460 can only handle 500V..

Making a linear regulator with MOSFETs as shown in your diagram will lead to a lot of dissipation and likely blown transistors as MOSFETs often have very limited tolerance for linear mode operation at higher voltages (above some tens of volts). I don't recommend running anything directly off mains, especially if you have limited experience with switching circuits. Both due to the high voltage, lack of isolation from mains, high fault energy available and limited tolerance for faults. A power supply of some 20 - 40 V will be much better and safer here.

Would you be able to recommend some sort of solid state regulation that does not involve transformer?

[klugesmith]

>>That is my concern, wont the ringing voltage be 2x, 5x of input voltage? IRFP460 can only handle 500V.

That's the whole point of original TV-set flyback circuits (and ignition circuits).   The primary voltage needs to "ring" to magnitude several times greater than power input voltage.   High output voltage depends on that;   the secondary/primary voltage ratio is the same in all phases of switching cycle.  As Anders says, D3 will clamp the primary winding voltage when switch is off.  Magnetic flux will decline slowly, as the primary current goes through D3 instead of the switch.

In the simplest model (of normal circuit), primary voltage with switch open is determined by secondary voltage at which the output rectifier diode conducts.  Operating with no load, the primary and secondary voltages "ring up" to the voltage at which something breaks down, and that voltage (divided by turns count) determines magnetic dF/dt.

It looks like you're on the way to running transient analysis in a simulation program.  That will be informative.  Start with a resistive load, or rectifier diode + HV capacitor, on the secondary side.

[Anders Mikkelsen]

The voltage will ring up to a value determined by the sum of the supply voltage, the reflected secondary voltage (load voltage divided by transformer ratio) and the turn-off time multiplied by the transformer leakage inductance. This can easily be over ten times the supply voltage when unloaded. For reference, TVs often used a BU508a for flyback drive, rated for 900 V Vcb and 500 ns turn-off time, along with clamping and damping of the overshoot. Your MOSFET is ten times as fast as this, so without any clamping or snubbing I wouldn't go above some 30 - 40 V with an IRFP460.

When using a MOSFET as a linear regulator, the problem is not so much power dissipation but device destruction due to local thermal runaway in parts of the die. Making a linear regulator that drops more than 50 V is very difficult, and I would avoid it at all, and if I had to I would use a linear rated MOSFET like an IXYS L2 series. Note that these are some ten times as expensive as normal MOSFETs. You can potentially use a buck converter to power it, but then it would be simpler to just run the flyback from a halfbridge if you insist it has to be powered directly from rectified mains.

From the questions you are asking, I would strongly recommend you to avoid circuits connected directly to mains. The fault energy is dangerous, any error will lead to immediate destruction of multiple components, and the circuit can not be practically or safely probed with oscilloscopes. Is there any reason you can't use a low voltage supply to power it?

What are you planning to accomplish with the circuit? This sounds like a classical XY problem.

[Domo]

From the questions you are asking, I would strongly recommend you to avoid circuits connected directly to mains. The fault energy is dangerous, any error will lead to immediate destruction of multiple components, and the circuit can not be practically or safely probed with oscilloscopes. Is there any reason you can't use a low voltage supply to power it?

The citcuit posted above is not complete, There is few more components included, but for simplicity, I just posted something simple. There is a 2A fuse, EMI filter, and I was planning an putting something for current limit before reaching rectifier. I would not be putting direct unprotected AC line into my project. I may be crazy, but not stupid .

When using a MOSFET as a linear regulator, the problem is not so much power dissipation but device destruction due to local thermal runaway in parts of the die. Making a linear regulator that drops more than 50 V is very difficult, and I would avoid it at all, and if I had to I would use a linear rated MOSFET like an IXYS L2 series. Note that these are some ten times as expensive as normal MOSFETs. You can potentially use a buck converter to power it, but then it would be simpler to just run the flyback from a halfbridge if you insist it has to be powered directly from rectified mains.

Purpose of that regulator using mosfet, is that some TV flybacks are designed to have B+ of 90V. For example, i have HR 8293 https://www.hrdiemen.com/reparation/flyback/scheme/8293 and if I put in 170VDC (Rectified US AC line), that is over 170W in transformer that was designed for 90W. If there is any other way to use solid state components to regulate 170VDC to at most 130 VDC. I would be glad to do it . I try to stay away from transformers, as they can be bulky and heavy, that's why I try to find a way using solid state.

What are you planning to accomplish with the circuit? This sounds like a classical XY problem.

For now, I want to make HV arc, even a Jacob's ladder for starters. Im aware it can be accomplished with lower voltages and current, but in future I will be needing higher current output.

[klugesmith]

Anders wrote:
>>The voltage will ring up to a value determined by the sum of the supply voltage, the reflected secondary voltage (load voltage divided by transformer ratio) and the turn-off time multiplied by the transformer leakage inductance. This can easily be over ten times the supply voltage when unloaded.

Thank you for presenting a model fancier than mine.   We know that real flybacks (and engine ignitions) must not be immediately damaged by voltage stress from running unloaded.

How about if Domo experimentally runs his driver circuit with 15 or 30 volts DC, and uses an oscilloscope to measure the maximum voltage on Q2, to get that "ring up" factor with actual transformer and gate drive.  He could see how D3 kills the ring up.