Re: Sam's HIGH POWER Ignition Coil Driver. Record pulses, driver dead:-(

>  How can I get SCRs to work at higher voltages? I canít find anything
> larger than 600V 50A here. The 50A are more than enough, but I would
> like at least 1000V for my next design.

The fancy way (used in high power high voltage inverters) is to use
optically triggered SCRS (or optoisolators to run a trigger circuit) and
run them in series.

The less fancy way is to just put the second SCR in series with the
first, and connect the gate to the cathode with a medium sized resistor.
What you've basically done is make a 4 layer switch that will breakover
at some voltage (probably around 600-700V). Now, with a suitable high
voltage across both SCR's in series you fire the bottom SCR. It
conducts, so now the full supply voltage is across the top SCR, breaking
it down so it conducts.

Lest you think that this will damage the device, it won't. All the
trigger pulse on the gate really does normally is reduce the breakdown
voltage of the SCR from it's zero gate voltage voltage to something
lower. You can actually make an adjustable relaxation oscillator with a
DC bias on the gate...

You'd probably want some resistors across the SCR's to distribute the
voltages when they are off, and some capacitors to provide enough
current to keep the first SCR on, while the second one is turning on.

Yet another scheme is to stack a bunch of small pulse transformers to
propagate your trigger pulse up the chain. Typical transformers have
isolation voltages of 5000V (for CE/VDE/UL approval purposes)

Take a look at Motorola Application Note AN1045: "Series Triacs in AC
High Voltage Switching Circuits" for some more ideas. (see the moto web
site) Lots of good ideas there, along with calculations of di/dt, dv/dt
and all the important stuff. Granted it is for triacs, but SCR's are
much the same for this purpose.

>  If I put a 50A SCR and a 50A diode on the driver, will the higher
> resistance of them cause loss of input? Should I try to keep their
> values as low as possible (within working limits) to maximise
> efficiency or can I make them bigger?

Devices with bigger current ratings typically have lower resistances (so
that the power from I^2*R losses stays low)... At high currents, watch
out for the power dissipated by the forward voltage drop. At 0.7V for a
typical (non-Schottky) silicon diode, 50A Ifwd would dissipate 35 watts,
not to mention any IR losses...

The problem is, of course, that bigger devices have bigger dies, which
means more gate-cathode capacitance which means you need more energy to
trigger the device. Undertriggering an SCR is sure to lead to short
life... Hit it hard to get it to all turn on, then drop the gate current
to a lower value to keep it turned on, but to minimize the gate/cathode
dissipation. A good way is to charge a cap through a resistor suitable
for the back porch current, then discharge the cap through a smaller SCR
(like an optoisolator SCR?) into the big SCR. The cap discharges in a
hurry, turning on the big device, then the back porch current comes from
the original cap charging resistor. If you have yet another cap behind
that, it will limit the length of the back porch.  
What you wind up with is a sort of ladder
		------ Rbig ------ Rsmall-----Switch ---- Gate
			     |             |  (SCR)
Charging supply            Cbig         Csmall		      Big SCR 
			     |             |

Details left as an exercise for the reader...

>  What voltage are they outputting if they make 15cm long DC sparks?

Assume they are "needle gaps" with breakdown about 1/4 that for uniform
field, so 30 kV/cm * 15 cm / 4 = 100 kV...

If you are discharging a cap into the primary, you probably aren't
getting any boost from "induction". More like a classic car capacitive
discharge ignition system.... They use 1 uF charged to 400V (typical)
for that application. So, given the 100:1 turns ratio on your coils
(really 200:1 because you've got primaries in parallel and secs in
series, right).. I'd say you're dumping in 500V?

For a real thrill... start making a multi staged system.... once you've
got optical triggering, there is no reason to stop with just one tier of
doubled coils.... 10 tiers gets you 1 megavolt...

Jim Lux                               Jet Propulsion Laboratory
ofc: 818/354-2075     114-B16         Mail Stop 161-213
lab: 818/354-2954     161-110         4800 Oak Grove Drive
fax: 818/393-6875                     Pasadena CA 91109