HV B+ switch

[db!]

Has anyone here ever tried using a high voltage mosfet or IGBT to control the B+? I'd like to use multi-secondary transformers, but controlling the B+ (To warm up the filaments) has always been an issue since HV switches are uncommon. If anyone has any suggestions, they'd be greatly appreciated!

[metalsculptor]

Supply your heaters from a separate transformer switch the B+ transformer via a primary side relay.

Alternatively if the AC side of the rectifier is only a few hundred volts use a mains rated relay (240V mains not 120V) to switch the AC. Some of these are good for 600V.

finally use a HV relay, not that uncommon old microwave oven had them also some transmitters. I have seen them for sale.

If it was me I would use the first option that is common in valve transmitters. Multi secondary transformers were characteristic of consumer valve applications they were rare in industrial and commercial applications.

Of course an IGBT would work but you would need a high side driver which would work with a DC load or switch B- which is unconventional and possibly hazardous.

[db!]

All of my projects currently do use separate B+ and filament transformers. Most of my B+ transformers already have the filament windings, but I am unable to use them due to the B+ control problem. Not to mention, the extra transformer takes up a lot of room on the chassis and adds to the weight. Filament transformers aren't exactly cheap either, they usually add $40 extra to the cost of the project (Total is about $250ish). The space, weight and cost issue has also restricted me to SS rectification since the rectifier tube needs it's own 5V transformer.

I honestly don't have much experience with IGBTs so I'm not quite sure what is involved. Would applying the voltage to the gate not switch it on completely?

I am also considering using an SCR or TRIAC with it's own trigger supply to switch the B+. Would this be a better alternative?

Any suggestions are welcome and appreciated!

[metalsculptor]

Originally Posted by db!

I honestly don't have much experience with IGBTs so I'm not quite sure what is involved. Would applying the voltage to the gate not switch it on completely?
Any suggestions are welcome and appreciated!

Yes 15V on the gate of an IGBT will reduce the voltage drop to a couple of volts but that 15V is referred to the drain which goes from floating to B+ upon switching the device making the gate become b+ + 13V The traditional way of handling this was to employ one floating power supply per high side channel, that was before IGBTs and charge driven bootstrap high side drivers the only problem with these is that they need constant switching to stay charged. Do a search for high side driver to gain an understanding of what I am talking about.

That is why I suggested using a relay on the AC side of the rectifier, what is the AC voltage? and are you using a bridge rectifier or a centre tapped one?

I would avoid SCR's they have a nasty switching spike every cycle and they need some inductance before a filter capacitor they are fine for noisy industrial PSU's but need quite a bit of work to keep the noise out of sensitive circuits.

[db!]

Gotcha, thanks for the explanation! The AC voltage for my upcoming build will be around 550VAC @ 230mA, future projects will have an even higher B+ (700-1kV AC).

Last but not least, would a solid state relay be appropriate?

[metalsculptor]

Do not use a solid state relay unless it is on the DC side and you do not mind it being a once per power up switch, I doubt it would have enough blocking voltage either it is really a triac packaged with an opto coupler. I believe I have seen DC SSR's but have never used one.

You could use an SCR on the DC supply which would avoid the switching spikes but once it is turned on there is no way to turn it off again without removing power also it has the same high side issues as the IGBT except that a small pulse transformer could be used to turn on the SCR. I would have a quick scrounge for 80's vintage microwave ovens first I have pulled HV relays from them which would easily handle your load. Buying a HV relay is easy but the cost would exceed the price of a heater transformer. Another issue with SCR's is the higher voltage ones are usually higher power devices with higher holding current, if the load drops below the holding current they can switch off.

One more issue all switches have high dV/dt which means a substantial current surge as the filter capacitor charges, some form of soft start such as an NTC device in series with the filter capacitor would help reduce this.

[BobEllis]

Originally Posted by metalsculptor

Yes 15V on the gate of an IGBT will reduce the voltage drop to a couple of volts but that 15V is referred to the drain which goes from floating to B+ upon switching the device making the gate become b+ + 13V The traditional way of handling this was to employ one floating power supply per high side channel, that was before IGBTs and charge driven bootstrap high side drivers the only problem with these is that they need constant switching to stay charged. Do a search for high side driver to gain an understanding of what I am talking about.
...

Why not use the IGBT or mosfet in the 0V side of the B+ supply, so all you need is 13V to turn it on? A voltage doubler on the filament windings should be able to provide this nicely.

Another option would be to use the IGBT or mosfet as a regulator and accept the 13V loss in B+. Use a string of zeners as a reference voltage.

Two ways to turn it on, simple: Size the current limiting resistor and capacitor bypassing the zeners to delay turn on as long as you want. Not accurate timing, but may work well enough.

More complex, but will reset immediately if power is interrupted: Control the voltage reference string current with a HV capable transistor at the ground end of the string. Drive it with a 555 timer circuit again powered off the filament supply. the switching transistor would only need to handle a few mA when on, but would need to be able to stand B+ in its off state.

[db!]

Originally Posted by metalsculptor

Do not use a solid state relay unless it is on the DC side and you do not mind it being a once per power up switch, I doubt it would have enough blocking voltage either it is really a triac packaged with an opto coupler. I believe I have seen DC SSR's but have never used one.

Most of the solid state relays I have found are optically coupled mosfets, not TRIACS or SCRs. I think just found one that might fit the bill:

RELAY 1000V 0.9A ISOPLUS264 - CPC1988J

I am thinking of powering it off the filament supply since it is isolated and can allow the filament supply to float.

[metalsculptor]

Originally Posted by BobEllis

Why not use the IGBT or mosfet in the 0V side of the B+ supply, so all you need is 13V to turn it on? A voltage doubler on the filament windings should be able to provide this nicely.

I suggested that in the post before with usual safety warnings.

Now you have got me thinking, not a good thing

Use a small 555 based oscillator running at 15V and maybe 25Khz referenced to b- put a small high voltage capacitor and 1000V fast recovery diode in series with the output of the oscillator and the gate of the IGBT with a reverse connected diode from the gate to emitter of the IGBT and a 15v zener diode as well.

When the oscillator is powered up it will try charging the gate capacitance through the small capacitor but as the IGBT turns on the emitter voltage will rise stealing gate drive (miller effect) making the IGBT work as a soft starter as well, eventually the IGBT will be fully turned on, If a high value resistor is connected across the gate to emitter then the supply can be turned off by stopping the oscillator. There are a few variants of this circuit some with slow turn on and fast turn off most use pulse transformers for speed but capacitor drive is better in this application.

If the off the shelf SSR will work with DC then by all means use it.