300 VDC regulator @ 3 amps

[Sparks-a-million]

Post bridge rectifier we have a 300Vdc rail voltage used to drive Mofets or IGBTS that produce the primary oscillations for a high voltage coil driving a corona cell.  The problem occurs when the incoming 120 or 230v changes.  Because we run the corona cell at a resonant freq it draws very little current but when the incoming voltage changes the current draw can increase heating up the corona cell and causing a cascade failure.  Need direction on a 300Vdc regulator that can withstand up to 3 amps of current.  Note the changes in line voltage are almost always momentary and lower line voltage is not a problem.  It’s only a problem when it increases over the 120 or 230 Vac.

[Sparks-a-million]

Here is the schematic with the regulator section at the top after the bridge rectifier.
It's good at regulating the voltage with a varying load but not with a varying input.  Our goal is to clamp down any voltage spikes over 300 volts in a way that doesn't generate too much heat.

[klugesmith]

Looks like a regulator, all right.  Is the regulation OK when line voltage varies slowly, as with a variac?

If the problem is fast-rising high voltage spikes, you might benefit from an off-the-shelf line filter on AC side of the rectifier.   Maybe small high-frequency capacitors in parallel with main capacitors on raw DC bus.   

Some other reader might have experience and advice about passive overvoltage-protection devices (TVS etc), as found in multi-outlet protective power strips.  Does your corona gadget go longer between failures if plugged into a power strip?

p.s. Applause for your schematic that includes objects like thermal pads.  I work with schematics that show small rectangles for mechanical parts, ROHS and serial number stickers, and one representing the bare circuit board. So those part numbers are automatically included in Bill of Materials.

[Sparks-a-million]

Thank you for the thoughtful response.  Momentary spikes are addressed in the power supply section, not shown, with MOVs, Varistors and Thermistors.  Trouble occurs when the mains rise, after other commercial  loads shut off or turn down.  The slower rise causes the cell to heat up changing the resonant frequency.  For other heat events we have a thermistor, on the cell, that shuts off power to the high voltage section, allowing the cell to cool off.  However the problematic rise in line voltage starts an off resonant condition that pulls excess current and can blow fuses, MOSFETs and IGBTs.  The regulator as shown works ok with a varying load but not so well with a varying line voltage.  The components also get hot quickly by trying to reduce the 300 Vdc down too much.

[Twospoons]

Looks to me like the current in R28 is going to vary with line voltage, and this excess will end up shunted through ZD1. This, plus the dynamic resistance of ZD1, is going to push your reference voltage around. Any variation in the reference voltage is going to be multiplied by ~300/9 and show up on the output.

Attached is a 300 volt regulator I designed for a vacuum tube phono pre-amp.  It uses an LM431 for a reference, and uses it to control a PNP pass transistor. I found the PNP topology to work better.  The zeners are only there to setup stable biasing for the LM431 and the level translating transistor.

Looking at it now I can't recall the reason for the stacked zeners - probably some iterative thing I didn't take out. R7 could probably be doubled and go straight to ground.

[Sparks-a-million]

Looks to me like the current in R28 is going to vary with line voltage, and this excess will end up shunted through ZD1. This, plus the dynamic resistance of ZD1, is going to push your reference voltage around. Any variation in the reference voltage is going to be multiplied by ~300/9 and show up on the output.

Attached is a 300 volt regulator I designed for a vacuum tube phono pre-amp.  It uses an LM431 for a reference, and uses it to control a PNP pass transistor. I found the PNP topology to work better.  The zeners are only there to setup stable biasing for the LM431 and the level translating transistor.

Looking at it now I can't recall the reason for the stacked zeners - probably some iterative thing I didn't take out. R7 could probably be doubled and go straight to ground.

Thanks Two Spoons. That's exactly what we are seeing.  I think it's worth a try, we'd have to replace Q1 and 2 with components that could handle the current but I'll see what the breadboard says...  Our Load could potentially draw two amps but allowing for headroom our goal is to be able to provide 3 amps across the regulator. 

[Twospoons]

Yes, you would need much bigger transistors for 3A - my pre-amp only needs 50mA.  Bear in mind that my design was optimized for audio work - so I was aiming for as much line rejection as possible over a really wide band. This dictated my transistor choices.  The simulations I ran were giving me 100dB of line rejection from 10Hz to over 100kHz.

You will also find the solution using the LM431 will have better temperature stability.

There is one other approach you could take : use a boost converter to create 300VDC from 120VAC.  Its more complicated, but you wont need monster heatsinks.

[Weston]

Could you change the driving transformer so a 400V DC input would be ok?

As Twospoons said, a low loss way of dealing with this would be to have a boost converter input to boost the line voltage to a constant 400V. Basically the same as the power factor correction circuits used in most higher power commercial switch mode power supplies. You might not care about power factor, but it could provide a constant DC bus voltage regardless of the AC line voltage. It could even be universal input, working on 120V and 240V ac line with no circuit changes or switch.

3A at 300V is 900W, which is a lot of power, but there is a wide range of PFC controller ICs you could use. A linear regulator is at these power levels is also going to be complicated and need a large heatsink.

The linear regulator you provided a schematic of does not seem to be close to 900W. If a lower power is ok there are highly integrated PFC ICs that only need the power inductor and a small number of additional components. Here is one example from Power Integrations: https://www.digikey.com/en/products/detail/power-integrations/PFS7528H/5583101

[Sparks-a-million]

Thanks Weston:
Is that device a power factor regulator that also does voltage.  I was looking over the datasheet and it looks like it can handle the current load but what happens when the input voltage changes. I wasn't clear on.  Seems that regulating output voltage at 2 to 3 amps isn't as simple as one would think, is it?

[Twospoons]

PFC controllers are designed to work with rectified but unfiltered mains ( no filter caps). So the input voltage is varying from zero to mains peak at twice the line frequency, while the output is held constant at a voltage guaranteed to be above mains peak, typically 400V.  Given this is the case, variation of the average input voltage has no effect on the output. The best thing about using a PFC is the current draw from the mains is sinusoidal, instead of the horrendous peaks that a rectifier + capacitor produce.  Given that you are aiming for ~900W this will make your utilities company much happier.  Its also easier on the rectifiers, and will help a lot if you have to pass any mains harmonic standards for a commercial product.