ACT 28 / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

ACT 28

Move Thread

LAN_403

Proud Mary

Thu Feb 12 2009, 01:32AM

Registered Member #543 Joined: Tue Feb 20 2007, 04:26PM
Location: UK
Posts: 4992

ACT28 is not the point in a dull play where you finally doze off, but a Cold War disc-seal forced-air VHF jammer triode, Va 11kV. Ia 50A, Pout 300kW/50uS

Clearly, in its original iteration, a meaty thyratron would have dumped its load into a pulse forming line ending at the triode anode, but what more modern way of generating the 50us 11kV/50A anode pulses might there be, in keeping with the general low cost of the ACT28 valve itself (unopened boxes Â£24/ US$35).

Steve Conner

Thu Feb 12 2009, 11:10AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

A spark gap and pulse-forming line, I guess. It's not more modern, but it's cheaper than a thyratron. The pulse forming network (PFN) could be as simple as one inductor and one capacitor, giving a half-sinusoidal pulse instead of a square one.

This would be a great valve for abusing without mercy in a VTTC with a staccato controller. (Or with the plate fed by a PFN, and the cathode triggered by a large SCR: no sparkgap needed in this case, and the PFN's stored energy avoids the absurdity of trying to draw 300kW peak power from the mains.) With 50 amps of cathode emission available, I think the results would be terrifying. I want one.

GeordieBoy

Thu Feb 12 2009, 12:04PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

As Steve said the modern solution is still going to involve a pulse forming network and a fast HV switch. The difference might be how the PFN is charged and exactly what is used for the switch. (Possibly also with an HV diode reverse connected across the switch to protect it.)

A modern switch would consist of a stack of series connected MOSFETs or IGBTs. 10 years ago I worked on a MOSFET stack capable of switching 4kv at 30A with a sub 10ns rise-time for a fast burst/transient generator. There is surely better silicon available now.

Depending on the required rate of rise of current (di/dt) a beefy thyristor may or may not be suitable. There is something called safe commutating di/dt. If the current rises too quickly after triggering an SCR it can cause hotspotting and destroy areas of the die. If the current rises more gradually it gives the conduction region time to spread across the die from the regions that trigger first.

This commutating di/dt is the thing that stops SCRs being used effectively in Tesla Coils operating at typical frequencies. The primary current just rises way too quickly.

Harry, is there a particular FM broadcast you have taken a dislike to?!?!?

-Richie,

Steve Conner

Thu Feb 12 2009, 01:07PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

With the SCR in the cathode circuit, it's effectively cascoded with the valve. So it doesn't have to stand the full charge voltage of the PFN: just whatever grid voltage corresponds to cutoff on the valve.

Since we're also using the valve as our RF oscillator, then the di/dt requirement on the SCR is just the di/dt of the burst envelope, which can never be greater than whatever the PFN inductance and charge voltage allows. It's not the di/dt corresponding to the RF current and frequency: the SCR just switches the DC component.

A single SCR should do fine, or one of Finn's crazy OL-BRISG things.

Dr. Dark Current

Thu Feb 12 2009, 01:47PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

GeordieBoy wrote ...

There is something called safe commutating di/dt. If the current rises too quickly after triggering an SCR it can cause hotspotting and destroy areas of the die. If the current rises more gradually it gives the conduction region time to spread across the die from the regions that trigger first.

Sorry for OT, but does this apply also to MOSFETs/IGBTs when turning off inductive load?

GeordieBoy

Thu Feb 12 2009, 02:48PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

Steve wrote:

With the SCR in the cathode circuit, it's effectively cascoded with the valve. So it doesn't have to stand the full charge voltage of the PFN: just whatever grid voltage corresponds to cutoff on the valve.

Ok, thanks for the explanation. I never understood how people got away with such low voltage SCRs in burst mode VTTCs. I remember John Freau once tried to explain it to me, and it seemed akin to pulling yourself up by your bootstraps!

I wasn't sure whether thyristors would be able to withstand the rate of rise of the current out of the PFN, whatever that might be in this application! Or indeed what the latest inverter grade thyristors can tolerate these days.

> Sorry for OT, but does this apply also to MOSFETs/IGBTs when turning off inductive load?

For IGBTs there is usually a maximum rate of turning off the current, otherwise you can get a displacement current through the parasitic thyristor structure and cause it to latch up. Of course latest generation devices are designed to be latchup free by clever doping, irradiation, NPT etc, but only when operated within their specified switching SOA. So if you try to turn off a gross over-current in a few hundred nanoseconds, then latchup is the likely result.

Latch-up is apparently possible with power MOSFETs also, but for slightly different reasons. The fabrication of a MOSFET involves the formation of a parasitic BJT which can get turned on if too much voltage is dropped across the body region. Once turned on, the MOSFET gate no longer controls conduction. Having said that I have designed things like boost converters with incredibly tight layout and the absolute fastest turn-off speed achieveable at peak current and have never experienced modern power MOSFET latching. The body and source are shorted at the time of fabrication in an attempt to prevent this effect - Whilst most documentation says that it is theoretically possible to cause latchup, in practice it requires considerable effort!

FWIW, you can normally consider MOSFETs to be capable of switching off instantly. It is usually the forward-recovery speed of whatever free-wheel diode that picks up the load current that really dictates how quickly you can safely turn off the MOSFET. If the diode is slow or there is any stray inductance, turning off the MOSFET too quickly can cause a damaging voltage spike before the diode clamps the voltage.

There are lots of references about IGBT latchup, but could only find this referring to MOSFET latchup:

See page 2 section entitled "Parasitic Bipolar Transistor"

Anyway, enough. My appologies for being led off topic

-Richie,

Proud Mary

Thu Feb 12 2009, 04:27PM

Registered Member #543 Joined: Tue Feb 20 2007, 04:26PM
Location: UK
Posts: 4992

Thank you all for your interesting and informative reactions to this splendid anti-aircraft relic!

Firstly, and with apologies, tp(max) is 5us, and not 50us, as I reported wrongly above.

The data sheet is singularly spare on data, probably because no general market for the valve was ever envisaged.

What we do see is Vg1 curving up to an incredible 500V+ for Ia(max), when we must assume great amounts of grid current will be drawn.

We have no frequency de-rating data for this valve, but it's reasonable to assume that a triode that will output 300kW at 200Mc/s, will be able deliver substantially more on the long waves favoured by most Tesla men. 500kW peak power perhaps.

But I wonder if I have not so far missed the obvious. This valve barely begins its conduction cycle until Vg1 = 0, so I would guess that total cut-off would lie somewhere between -250V and -1kV. (no figures in data sheet, so to be determined by experiment at reduced power.)

So I am suggesting that the valve could be switched from the grid with 5us pulses where the pulse amplitude is measured from cut-off to the degree of Vg+ chosen to set Ia.

As no figures are given, I would guess a mark-to-space ratio of at the very least 10:1 should be used while the envelope temperature is measured during set-up trials. This would give tp 5us, PRF 20 kc/s. In the simplest forms of AM RT jamming, a wobbulating PRF cycling across 300c/s - 3kc/s could well have been chosen, but this is conjecture on my part.

I am up for a go at this, but think I might try single shot mode as my point of departure, as I have some idea what 300kW RF pulses are like.

I envisage putting the tuned circuit in the grid, and think I might start off with fo = 5Mc/s, so as to keep the inductances of high Q and manageable physical proportions. In single shot mode, Va can simply be stored in some suitably large and fast condenser, until the positive-going pulse is applied to g1 and the valve turns on.

Does this sound plausible gentlemen?

In its original setting as an AA jammer,* tuned by gleaming lengths of silver-plated copper pipe, the frequency must have wandered up and down with the temperature, and the sharp pulses would have generated harmonics from here till Kingdom Come, creating a broad spectrum interference field where frequency instability was probably seen as an advantage!

* ACT28's design frequency suggests to me the device was intended to work against air-to-air inter-pilot RT tactical communication of the period. ACT28A differs to ACT28 in allowing tp(max) = 10us, so operating them in aperiodic pairs, the two of them would have created a dense VHF harmonic-rich interference field. I would guess that the units were deployed in clusters around air fields, ships, and missile silos where enemy pilots of the period would have had to co-ordinate their 'bombing runs' on R/T. But this is all my dreamy conjecture. All I know for certain is that ACT28 was developed as an anti-aircraft jammer.

If a few others were interested, I might (possibly) be able to induce the vendor to come down on the price if we had a multiple order. I have an idea that they might accept Â£100 for six pieces, of which I would buy two for Â£33.34, leaving a shortfall of Â£66.66 if anyone else were interested in the other four pieces.

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob