Reverse Polarity Protection and Overvoltage Protection

[gjcamann]

So I was kindly pointed to http://www.geofex.com/article_folders/cheapgoodprot.htm after someone reviewed a schematic with the dumb series diode polarity protection and some pathetic attempt at a zenier overvoltage protection. After studying this I decided RG was on to something, but I wanted to add overvoltage protection too. I've come up with the following, it does give up 1/10th of a volt drop which is acceptable in my book. And I know it's protecting on the negative side (which I guess is kind of like a protestant/catholic thing for some) - but I feel it's right. I'd like to get another set of eyes on it to see if anyone has helpful comments or can spot any flaws. This is going on my first ever PCB, a footswitch, power filtering, LED board for use with future hobby pedal builds.




Here is what it does, if the voltage exceeds the desired value, the transistor opens up, also if the voltage goes negative the transistor opens up. You can see voltage across the load drop to zero once the voltage exceeds 18V. To adjust the cutoff voltage, simply select a new zenier (that is 1/2 the cutoff voltage). See plot below.


Thanks for everyone's help. And thanks RG for the initial circuit.

[R O Tiree]

The advantage of RG's circuit is that it supplies BDV with reference to GND. In this new circuit, your RLoad (main circuit) is across V+ and some other, not-quite-GND potential. I had a discussion with RG about this a few weeks ago, when I had a similarly hare-brained idea involving MOSFET protection. If you can keep circuit ground and true, signal GND separate it can work; otherwise it can be pop/hum city. Here's what he wrote (I'm sure he won't mind me quoting from our PMs):

If you have one pedal, opening the ground lead is a usable way to do things, but there are almost no single pedal setups any more. That means that any situation that introduces even a few millivolts into the difference between power input ground and signal ground will be impressed on the other pedals in the chain because at least one of them will have its signal and power grounds tied together. So opening the ground side is just not something that's workable unless your pedal is designed to allow its power ground and signal ground to be both (a) different and (b) inverted in the sense that signal ground is below power ground for negative ground pedals.

And, next day...

Thanks for the prompt reply, RG. I realised shortly after writing that, if a P-channel drops the supply voltage by a few mV, then an N-channel will probably raise the circuit GND by a similar amount for the same reason. That's going to make things messy, as you said.

Yep. The #1 complaint with Neovibes is from people who neglect the advice to use an insulating power input jack and so have huge hum and noise issues from the ground at the power jack trying to short the negative side diode in the bridge. Even a few millivolts of DC will show up as intractible pops which will not yield to normal debugging schemes.

The thing is, just how much should one pander to people's stupidity? The circuit we ended up with in your other thread protects your circuit up to about 20VDC (with a 1W 100ohm resistor ahead of the Zener), AC and reverse polarity. If someone is that determined to screw things up that they exceed those limits then, frankly, I think that they deserve to bear the cost of repair. After all, much above those limits and we start to get close to the limits of what the semiconductors both in the main circuit and the protection circuit will cope with. Where should we stop? When someone connects it direct to the mains? That's starting to get into the realms of Darwinism in Action.

[GibsonGM]

The thing is, just how much should one pander to people's stupidity? The circuit we ended up with in your other thread protects your circuit up to about 20VDC (with a 1W 100ohm resistor ahead of the Zener), AC and reverse polarity. If someone is that determined to screw things up that they exceed those limits then, frankly, I think that they deserve to bear the cost of repair. After all, much above those limits and we start to get close to the limits of what the semiconductors both in the main circuit and the protection circuit will cope with. Where should we stop? When someone connects it direct to the mains? That's starting to get into the realms of Darwinism in Action.

Clap clap clap )   Well said.

The circuit is GREAT!  But I wouldn't *bother*, if you know what I mean. For reasons R O just stated.   We're all musicians, electronics tinkers (and some actual gurus).  We all know to connect a battery the right way, and to look at the wall wart to make sure it has the right pinout.   

For me, the simple protection diode is more than enough, if I even bother to include it.  If you had a mult-effect unit, with special, advanced digital processing components in it, maybe something like this would be cool.  But at our level, I can't find a real need.  I've NEVER had a power issue with anything I've built, since I consistently use the "Boss" scheme. 

Thanks for tossing that together, though, because one day, we may NEED that level of protection!!  You can never know enough, and that was a nice exercise in forcing that transistor to do what you want.

[gjcamann]

Damnit! Well back to the drawing board. Stand by the PNP version next week. Thanks for the reasoning between high-side switching.

While I understand that the effort in this has gone well beyond the likely usefulness of it. But it's become a fun project in it's own right so I will merrily continue down this path for the sake of accomplishment and occupation. (If I wasn't doing this i'd just be sitting around watching duck dynasty)

If I was going to bother with RG's brilliant polarity protection, I might as well add OVP, since it's only a couple more components - once I figure it out.

Thanks again.

[gjcamann]

OK, here's the high side switching, I think it's actually a simpler circuit. I appreciate everyones helpful advise up this point and would like just a bit more if anyone cares to take the time.

[R O Tiree]

I've been messing around and entering numbers in a spreadsheet to work out V_EB.  I think that you will blow transistors in a heartbeat.  The 3906 is limited to -5V for that parameter and you exceed that in both transistors as soon as you flip polarity at anything exceeding -5V (ish). OK, that -5V is a "minimum" figure and you'll probably get away with it at -9V supply, but -30V will make it go "pop!" <Cue the magic smoke>

[PRR]

> The 3906 is limited to -5V for that parameter and you exceed that

Many SPICE models don't model emitter reverse breakdown.

The most likely breakdown is 7V. The 5V on the spec is the low-end of likely breakdowns.

I'm not *sure* it will blow-up. Current may be limited by resistors and load (but load may conduct better backward). If power is not exceeded, reverse breakdown is not grossly damaging.

I do think it is a tough problem.

[R O Tiree]

> Many SPICE models don't model emitter reverse breakdown.

I hear you, Paul, and I'm ever mindful of your sage advice to take SPICE simulations with a pinch of salt unless you have a pretty clear idea of what to expect... Hence why I pulled up the datasheet to look for bear-traps.

> If power is not exceeded, reverse breakdown is not grossly damaging.

I remember reading something from RG a while ago about transistors getting potentially noisy after only a short while "reversed", and I built a white-noise generator some years ago which relied on that principle. I wonder what noise on the power supply might do? Smoothing/filtering caps on the supply would probably get rid of it, but that then leads to another problem with even RG's clever circuit... as it stands, RG's circuit will cope with an AC supply. Put caps after it to GND and the reverse voltage across the 3906 just about doubles on the Vsupply down-swing. I'm not entirely sure why, though... certainly, the collector stays high and positive (stored charge in the caps) and the emitter goes negative (as Vsupply goes negative). The base of that transistor more or less tracks 0.7V below the collector voltage (C-B PN junction now acting as a forward biased diode and current passing via the 4M7 and thence to GND through the other resistors?), according to my sim so, with an AC supply of up to 30V amplitude, which George is trying to protect against, V_EB gets to very nearly -60V at its worst.

[gjcamann]

Oh man. Thanks so much guys! I think that explains a few of the pieces in RG's circuit. I guess it still needs a bit of work. I'll have to check the voltages across all junctions.