Sometimes I don't know where my head is. The obvious thing to do on polarity protection was staring me in the face and I didn't even think of it.

I'll post a schemo on GEO as soon as I can draw it out, but it goes like this: a PNP transistor is a good positive voltage pass device. Put the emitter to the voltage source, the collector to the effect and pull down on the base with enough current to saturate it and you have a good on/off switch that withstands reverse voltage.

The problem is how to turn it on when the polarity is right. No problem - use an NPN with its emitter to ground, its collector through a resistor to the base of the PNP in the power lead. The NPN's base goes to the power source through a resistor. Now when the power supply is correct, it turns on the NPN, which turns on the PNP, and power is passed through. If the power source is reversed, the NPN is turned off, the PNP and the NPN both don't conduct, and you have a switched-off protective switch up to the Vceo of the PNP, about 40V with easily available 2N3906's.

The NPN transistor could be damaged by high reverse voltages, so you can clamp its reverse voltage by splitting the base resistor into two resistors, and putting a 1N4148 diode from the NPN emitter to the junction of the split NPN base resistors. Now reverse voltage on the NPN base emitter is clamped by the diode to 0.7V, and the diode is reverse biased when the power supply is the correct direction. At some voltage, the diode will break over - which is OK, it limits the current into the NPN base by regulating the drive voltage.

So - two ordinary transistors, three resistors and one diode get you polarity protection that loses less than about 30mV forward and conducts not at all reverse until you get to about 40V. It saves buying a BS170 MOSFET, which seems to be a huge problem to at least someone whenever polarity protection comes up.

Actually, this is a synchronous rectifier. If you feed AC in and have a filter on the output, it will make the right polarity out of AC. Not full wave nor regulating, but it won't kill the circuit.

I'll do schemos asap.

OK, it's up on the web page now.

Thanks VERY much R.G.  

Kleber AG

I (briefly) tried something simpler which seemed to work. The PNP was flipped: Input to collector, 10k base resistor to ground, and output from emitter.

Possible explanation?
The collector/base form a PN junction, which is forward-biased by the 10k base resistor under correct polarity. The depletion layer thins and electrons can now jump from emitter->collector. If polarity is reversed, PN junction becomes reverse-biased, depletion layer thickens to block collector->emitter flow.

-Joe

Hi.  I've only been thinking about polarity protection for a while.  Could someone summarise the classes of sensitivity in different devices?  I mostly build distortions....

Would it be true that transistors (including Darlington pairs, FETS, MOSFETS) are always ok?

Op-amps (eg TL series) seem to be easily destroyed. (?)

How do things like CD4049s stand up?

thanks for any info..

By the way, I simply use the simple in-line diode (1N4004) whenever I use an op-amp or CD series device.  Will that save me from 9V reverse polarity?

QuoteWould it be true that transistors (including Darlington pairs, FETS, MOSFETS) are always ok?

No.

If you read "When Good Opamps Go Bad" at GEO, you can read some about the degradation of noise performance in bipolar transistors if the base emitter junction is reverse broken *even once*. Noise may not keep a transistor from working in a fuzz, but it sure makes it sound bad. Beyond that, if the circuit resistors happen to offer a low-resistance path through a circuit, it is possible to burn out bonding wires on low-signal devices.

It is true that discrete circuits are not as prone to destruction this way as IC's of several families, but it's never good practice. That's one reason I put that reverse diode in the polarity protection circuit - it's there to keep the reverse voltage from harming the NPN, even when the circuit is intended to be reversed. It's more reasonable to say that discrete devices are more tolerant, and you get away with it more often.

It's really never OK to reverse bias a circuit unless it's designed for that to happen without harm. You may get lucky, even be lucky almost all the time, but that's not how you prevent problems.

QuoteOp-amps (eg TL series) seem to be easily destroyed. (?)

They are. ICs as a rule have tiny, tiny devices, much smaller in silicon area than discretes, and they are correspondingly easier to damage. Reversing the power on most ICs forward biases the substrate isolation diode and simultaneously turns on every pathway through the chip. This usually burns out a bonding wire or fuses a metal layer on the chip.

QuoteHow do things like CD4049s stand up?

As badly as analog ICs, for the same reasons.

There aren't any "always OK" devices for reverse polarity in the usual kinds of circuits. There is always the potential for excess current through unexpected pathways to burn out something delicate. That is why we design protection circuits.

QuoteI (briefly) tried something simpler which seemed to work. The PNP was flipped: Input to collector, 10k base resistor to ground, and output from emitter.

Possible explanation?
The collector/base form a PN junction, which is forward-biased by the 10k base resistor under correct polarity. The depletion layer thins and electrons can now jump from emitter->collector. If polarity is reversed, PN junction becomes reverse-biased, depletion layer thickens to block collector->emitter flow.  

That's the PNP dual of another version I tried, with an NPN in the ground lead.

You are running the PNP in inverted mode, which isn't necessary. You can leave the PNP with the emitter to the supply, collector to the circuit, and a 10K to ground. When power is normal, base current flows in the 10K, turns on the transistor. When power is reversed, the base-emitter is reverse biased, and nothing runs - except that in normal operation, the base emitter breaks over at about 7V.

It is likely that it works better in reverse mode even though the reverse mode current gain is quite low because the collector base junction which is acting like a base-emitter has a much higher breakdown voltage than the real base-emitter, so it doesn't break over like a non-inverted one would. You could get much the same result by putting a 1N4148 in series with the base lead to protect it from reverse breakover.

You also need to be sure that you get the pass device saturated, or you'll have it clipping off some voltage. That's what I did with the NPN - it has enough gain to really bang the PNP on regardless of the battery voltage.

For series diode protection you're better off with a Schottky diode like a 1N5818 (or 5817 or 5819) because the forward voltage is about 0.4 less than a 1N4004.

Your better off with a Schottky for a reverse diode from power to ground for the same reason. If the Schottky starts conducting before anything else, it's going to be the only thing that absorbs the jolt.

Take care,
-Peter

The only difference with the emitter to the supply was -0.4v on the collector with the power leads reversed.  When I flipped it around, it went down to zero.

I would trust the brute-force method more, with both transistors, especially for larger supplies. Something like this could be used in a lantern-battery powered portable amp, etc. Pretty cool! Maybe a battery-low/battery reverse indicator(s) could be snuck in there.

-Joe

QuoteMaybe a battery-low/battery reverse indicator(s) could be snuck in there.

Hey - there's a good idea. I'll ponder it a bit.