Crowbar circuit - zener "biasing" resistor.

[Diabolical Artificer]

I'm knocking up a crowbar protection circuit like the one in the pic . It needs to "fire" at around 7v. This goes on the OP of a LM317 regulated 6.3v PSU ( so C1 and SD1 arn't present) in case of the regulator failure. It's a heater supply.

The only suitable - ish zeners I have are 500mW and 5W 6v8, so from what I've read in the AOE is the voltage at which the circuit "fires" is Vz + 1 diode drop, IE 0.4v. I presume this is from trhe SCR. I'm using a C106D SCR.

My main question concerns R1, which I assumed was to limit current through the zener, but which I've read is a pull down R to keep the gate of the SCR low before the zener conducts. On the breadboard I've killed a couple of zeners, so want to work out R1's value.

If we ignor for the moment R1's actual function and want to find the value of a resistor to stop a zener exceeding it's rated power rating then we want a resistor that passes enough current to turn it on, but limit it's on current. Is that correct? I've read 10% or the zener's rated power is about right. In this case ( if we use a 500mW zener) 0.5/6.8 = 0.073, but 10% of that is 7mA, surely not enough to turn the zener "on"; datasheet says 20mA for test current.

This is further complicated, though not really, in that we need to have a bit of left over current for in this case the SCR's gate, which AFAIK is 15 - 200uA.

Am I over complicating things here and talking out of my hat? Any suggestions welcome.

Andy.

[Refugee]

I would question the connections on the SCR.
When the zena conducts and triggers the SCR it should shunt the zena and resistor so that nothing flows in the zena/resistor.
It should never blow the zena.
Test the SCR on an analogue meter such as an AVO on low ohms.
It should trigger when it is the right way round if the gate is touched on the right probe and reset when the probes are touched together.
You can also test it with a battery or bench power supply and a bulb.

[Mark Hennessy]

Have a look here to learn about SCRs: https://en.wikipedia.org/wiki/Silicon_co..._rectifier

As shown, you can make one from two "normal" transistors. I've done that a fair few times over the years - it's occasionally a useful building block. I find it useful to look at the transistor circuit to understand the operation.

Between the gate and the cathode, you have a PN junction. If you put about 0.6V across it then, as you'd expect for any forward-biased PN junction, current flows. This fires the SCR and latches. However, the last thing you want from a crowbar is "nuisance triggering", which is why R1 and C2 are there to hold the gate close to the cathode. Otherwise, noise can cause the SCR can trigger...

R1 can't limit the current seen by the zener. As you try to increase the voltage across a PN junction, it simply passes more current, doing its best to hold the voltage at 0.6-0.7V. It's just a diode, effectively...

So as you exceed the nominal 7.4V given by the combination of the zener and gate-cathode junction, the current will rise and rise... But as Ref suggests, the SCR should trigger and short the whole circuit before any harm can come to the zener. So if you've damaged zeners, there's something amiss somewhere. Perhaps your chosen SCR is a slow one, perhaps you've got a mistake in the construction - who knows? But if you're worried, you could put 100 ohms or so in series with the zener. It will raise the threshold slightly, but that shouldn't be an issue during development...

As part of a PSU course I run, I get folks to build up one of these. But they connect an incandescent light bulb in series with the crowbar (in place of the fuse), then slowly advance the output of a bench PSU until the crowbar fires - which simply lights up the lamp with no drama - and then they can record the voltage from the PSU that caused the crowbar to fire. During the debugging of this, I'd recommend a similar approach.

[Kalee20]

The circuit should be fine as-is! The resistor and capacitor between gate and cathode are to stop 'nuisance tripping' and also to guarantee that the Zener diode is well round its 'knee' before the SCR fires. Some Zeners are a bit soggy and could pass several microamps at well below their Zener voltage - which could make the SCR fire prematurely.

There's nothing really limiting the Zener current if the input goes overvoltage. But hopefully the SCR will fire within a couple of microseconds anyway. Let's face it - if the Zener fails short-circuit, it's only going to give the SCR more of a kick, and protect the load even more...

Be sure to use a chunky SCR that doesn't fail open-circuit before the fuse blows. You need one rated to pass maybe 5x the fuse rating. A more thorough analysis would compare the I-squared-T rating for the fuse and SCR.

[Diabolical Artificer]

Thanks all. Misread the circuit. The zener's died on an old breadboard, I suspect the tracks insulation has about had it.

Thanks for the explanation Mark, good idea to use a lamp instead of fuse.

How does one work out the value of R1? In the Art of Electronics crowbar circuit, which is more or less identical R1 is 68 ohm. Big disparity between 68r and 1k. Or is it a case or suck and see?

Andy.

[Refugee]

Some designers may well only limit the zena current to leave the dissipation to the maximum for the zrna and rely on the thyristor firing and bringing the volts down quickly.

[Mark Hennessy]

Power dissipation in a crowbar zener is a red herring - it's all over so very quickly.

To answer the question about R1, we need to think about the zener diode. Specifically, look at the graph of voltage vs current: https://www.digikey.com/en/maker/blogs/z...c61e2640f3

Notice particularly the "knee" that you get as the zener approaches its operating voltage.

So as the voltage across a zener depends on the current applied, you'd choose the operating current accordingly - certainly, an operating current that is sufficiently around the "knee" is good. Say, at least 5mA, for example. Some might say 10mA. All diodes are different.

Remembering that we need 0.6V across R1 when the SCR is to trigger, and we'd like 5mA to be flowing at that point, that tells us that R1 should be 120 ohms.

But this is very approximate. As well as the basic voltage tolerance, the minimum recommended current for zener diodes varies a lot from diode to diode within the same family (as there are actually 2 distinct mechanisms used in zener diodes, according to their rated voltage), so look at the datasheets to get a feel for this (e.g. page 5 of this: https://www.vishay.com/docs/85607/bzx85.pdf)

Also, that ignores the current taken by the gate of the SCR. You'd need to check the chosen SCR's datasheet to decide if that's low enough to ignore...

In short, experiment and see what happens

[Diabolical Artificer]

Thanks Mark, concise answer as always.

Andy.