10kv, 10ma, 1ms, switch

You can't do it. RC = 2 mS You'll need to change your values.

Check here for the switching:

The cheapest approach would be to stack say 15x 1000V 1A Teccor SCRs, these are cheap, and use say 10K gate-cathode pulldowns. Then you would have to wind a 15 secondary pulse transformer to fire them- your current is so small it won't make any difference how the stack collapses. The tricky part here is that you don't have enough "holding" current to maintain SCR conduction, and the way around that is to inject a continuous current into the gate which makes the SCR act like a high beta transistor or artificially boost the conduction current. There are several ways to do that- one is make the transformer with large enough VT product to sustain a 1ms pulsewidth, and a second is catch the top stack SCR anode with current injection from a low voltage supply through an HV diode that supplies the required holding current when the stack has collapsed. Remove the pulse or remove the LV supply current and the SCRs recover provided the voltage recovery on the 0.002u is not too fast. But if that's the case, other things can be done to get around that. It is basically a Radio Shack grade project.

I've seen a similar approach using stacked MOSFETS. But I believe it was fairly slow, in that each gate was turned on in a cascaded fashion.

Cheers

I'm sure you can find a thyratron rated for 10kV or more without needing an absurd current level (most thyratrons that size are going to be in the kiloampere range...). If nothing else you can build an ignitron, though that kinda seems overkill.

More common thermionic devices, hum a sweep tube might do that (possibly with an external high voltage seal, most are rated for only 7.5kV peak, dunno if that's when the inside or outside arcs over). Sweep tubes are cheap and plentiful and most can sink several amps peak. Of course, you need to apply control grid voltage for the duration of the pulse, and keep it negatively biased the rest of the time.

Tim

-- Deep Fryer: a very philosophical monk. Website:

I don't think SCRs or thyratrons would be well suited for dealing with such small currents. As for MOSFETs, they can certainly be made to work quickly. For convenience, consider a series stack of optically-coupled SSRs made from MOSFETs. Aromat NAiS makes the AQY214EH, a 400V part with the H-type 5kV insulation. This has 0.5ms switch-on time and 0.08ms off time. And there's a slower AQV254H with 1.8ms on time. DigiKey has both of them in stock. [ NAiS also offers 1kV and 1.5kV relays (the AQV259 and AQV258), but sadly they're limited to modest 1.5kV insulation ratings. Ditto for all their 600V parts. Maybe another manufacturer? ]

One healthy property of FETs, and a serious advantage in this case, is their tolerance of avalanche breakdown, provided only that the die power dissipation is kept below the specified limits. This is important, because the switch-on and off times won't be exactly the same for all of the series-connected MOSFET switches, exposing some to an overvoltage for a fraction of a millisecond.

Back to the SSR's official 5000V input-output insulation rating. One solution would be to modify the 10kV setup to +5kV and -5kV.

--
 Thanks,
    - Win

The photovoltaic stuff is much too slow and too expensive. Something like this would be cheap and provide pull-enty of isolation: View in a fixed-width font such as Courier.

. . . . 15 x 1N4007 . 50mA . HV>---+-------[1M]-----------+------|

First a little math: Time constant T=R*C or T=(10^6 ohms)*(2*10^-9 farads) = 2*10^-3 seconds. That means it would take *two* milliseconds to discharge about

63 percent of the energy. Therefore, it will be impossible to "dump most of that 100mj into a 1Megohm load in 1ms".

There are a number of relays that can act in the millisecond region (but those cannot stand off even 500V), and ways to impliment a mechanical switch (which could withstand the 10KV) for closure that is controllable to give 1mSec closure within a few percent on timing.

An SCR might work, but how do you propose to shut it off? It would stay on until the current drops below its internal threshold in the tens of microamp region. One could do that by forcing the gate negative (but no more than 5V negative so that the junction does not zener), but one must extract all of the anoe current until the SCR itself goes off (all of the stored charges inside are removed).

BTW, a relay *is* solid state (metal) !

Except for the mercury kind.

Best regards, Spehro Pefhany

--
"it's the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

I forgot; Mercury is a *planet*.

Wanna see one of its moons?

Best regards, Spehro Pefhany

--
"it's the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

No, it's not. The part that does the actual switching is the air (or vacuum) gap. ;-)

Cheers! Rich

1. Relays generally eplloy air or vacuum so they can hardly be called solid state.

1. If you use a discharge device you will need to commutate it or turn it off. This is often done by using a second device and a capacitor in power control circuits. Anther way which is often used in pulse radar sets is to use resonate the output capacitor.

Personally I'd use a trigered spark gap, but that's just me.

Another realistic possibility is to take a lower voltage pulse (say 500 volts) and just step it up to 10KV. This makes the parts on the primary at a more tractable voltage level. with a transformer.

There are some high voltage shunt regulators but I'm not sure if they are available.

6HZ5 is rated at 6KV 7235 is rated at 10KV PD500 is rated at 25KV.

These are all from a parametric search on using the TDSL program from

There are probably much more modern devices available.

You could maybe used a marx-generator type topology, using mosfets instead of sparkgaps for the switches.