Turning an SCR _off_

You can get Gate turn off (GTO) SCRs but they're not very common (probably expensive!).

A common method of commutating an SCR is to use a second SCR and string a capacitor between the two. As the SCR to be switched off is conducting, its anode is at low voltage, the commutating SCR is off so it's end of the capacitor is pulled up to supply voltage via it's load resistor. Firing the commutating SCR pulls it's end of the capacitor to ground - the other end of which is charged more negative so it drives the original SCR below the ground level turning it off.

See Fig 1b.

Or parallel it with a MOSFET, sort of per your link from your other post.

If its for high voltage it may be difficult to find a HV MOSFET with low enough RDSon.

Hmm...

Looks like, for some reason, Agent doesn't allow parentheses in a link.

Anyway, if you copy the whole thing and paste it into your browser's "address" window it should work.

Well, it does for me anyway. :-) JF

do?

Use the GTO SCR or make an analog of SCR from two BGTs.

Vladimir Vassilevsky DSP and Mixed Signal Consultant

Not so. Furthermore, an SCR will have a higher drop than a single BJT. Check out my suggestion to use Zetex BJTs in the other thread.

Pulling current out of an SCR gate can turn it off, with a few caveats. First, the reverse gate voltage must be limited to 5V maximum. Second, it seems that one must sink as much current thru the gate as is flowing in the anode (draw a discrete PNP / NPN diagram for an SCR for verification). Third, the turnoff dV/dT on the anode must be limited so that does not turn the SCR back on. As far as reliability goes, any semiconductor device can be zapped into a 2 or more terminal short at least as good as metal.

Is there an issue with current crowding when you do this? IIRC that's one reason for the dI/dt limitations of SCRs.

Cheers,

Phil Hobbs

Don't think so. AFAIK it comes from the limited firing front propagation speed which means that if current rises faster than the conducting area grows then current density may rises to a point where your SCR is magically transmutated into a blob of fused sand, hence the dI/dt limit.

-- Thanks, Fred.

In case the OP is still interested in GTO thyristors, there's a tutorial on them in the same folder - and loads of other interesting stuff by backspacing the URL even further.

I believe it will generally work, but you might have to reverse bias the gate (not just short it to the cathode) and draw a significant fraction of the load current out of it (which might not be possible without damage).

IOW, they're not specified for this, so you won't be able to easily prove it will work reliably.

If you really want this function, why not just connect two transistors as a synthetic SCR and then you can shunt one of the bases to the emitter and definitely shut it off? By including a series resistor you can limit the current required to something halfway reasonable.

Really, I don't see much in the way of disadvantages of the proposed synthetic depletion mode IGBT, nor do I see how an SCR would be better in any way.

Best regards, Spehro Pefhany

Only board space, and none of the methods for treating an SCR would be remotely better.

Yup! :-)

Does it have to be a SCR, there are SCS's silicon Controlled Switches.

Cheers

In that case an IGBT may be the device of choice.

Yet another case of why i like a URL to be all by itself on a single line.

There's always struck me as being something iffy about the IGBT - its still a MOSFET but driving a PNP power transistor - which AFAIK can't be saturated.