High voltage switching

[dgta]

Looking for possible solutions to switching B+ for delay purposes. Need to switch 470-500vdc supplies in several amps, all CT secondaries.

Has anyone come up with a simple HV switching scheme that has proven robust and reliable? MOSFET, IGBT, one of the new MOS optorelays? I'm not adverse to mechanical relays, but I don't know any compact ones with contacts rated for 500vdc.

[nigelwright7557]

Your going to get some serious arcing if you use a relay at that power.

You can get MOSFETs with S-D voltages up to 1500 volts.

[tomchr]

How about using one transformer for the HV and one for everything else? You could then switch with a relay on the primary.

As pointed out above, you can get MOSFETs that can handle the voltage (across source-drain), but bear in mind that the gate oxide can typically only handle about 20 V. I.e. Vds may be 1 kV, but Vgs is only 20 V. So if you use a high-side switch you need a fancier (floating) gate driver. I'm not a big fan of low-side switches.
You might be able to use a resistive divider and another MOSFET as the gate drive for the main switch.

~Tom

[dgta]

Ooops, sorry about the duplicate thread. Not sure why it double posted, I musta goofed.

Anyway, thanks guys, I understand all the considerations involved. My question was a little more practical. I don't want to design and test a switch right now, so I wanted to see if anyone has either 1) a commercially available switch/circuit I can buy or 2) a diy circuit that has been tested and proven reliable in use. I don't want to reinvent the wheel if I don't have to.

And no, an additional transformer for filaments won't fly in any of the amps I have (lack of space and aesthetics) and neither will an Amperite thermal delay.

[astouffer]

You could look into vacuum relays. A quick ebay search turned up some reasonable prices.

[LinuksGuru]

High-voltage thyristor will be optimal solution, will cost just few bucks. Google for T-Reg power supply, or Rubli board, they all provide B+ delay circuits.

[M Gregg]

Originally Posted by LinuksGuru

High-voltage thyristor will be optimal solution, will cost just few bucks. Google for T-Reg power supply, or Rubli board, they all provide B+ delay circuits.

I would have thought that you could switch on with a thyristor, however if the latch current is exceeded then the gate will not effect power off unless you use GTO or a contact. Remember that the gap of the contact in a switch or relay is the important factor so you can series contacts for voltage "break" and parallel for current. You could also bypass the HV contacts with a polypropylene capacitor, however this should not be done if its failure could be a safety problem in which case a HV mov can be used across the load side of the contacts.

Regards
M. Gregg

[dgta]

I like that Rubli board. You need one per channel, but still seems like a pretty robust idea. If it fails either short or open, no damage to anything else. That's good.

Anyone actually using that Rubli circuit? Or one like it?

The vacuum relays also a good idea. I shoulda thought of that.

[janneman]

Originally Posted by M Gregg

[snip]. Remember that the gap of the contact in a switch or relay is the important factor so you can series contacts for voltage "break" and parallel for current. [snip]Regards
M. Gregg

There is a difficulty with putting contacts in series to increase voltage widthstanding. Generally the contacts will not open at exactly the same moment, so you end up with the first contact that opens having to cope with the full voltage.

jan didden

[oshifis]

I used solid-state relays for switching HV in my tube amp. In general, it is better to switch AC than DC because of arcing. Advantage of the SSR is that it switches at the zero crossing of the cycle. There is an internal optocoupler, the control is an internal LED. I connected it at the secondary of the mains transformer just before the rectifier diodes. It worked fine.
The end of the story: I reverted to a mechanical relay, with 4 contacts parallel. No arcing whatsoever. One silicon device less