How do protection diodes get damaged?

[Mark Hammer]

I'm repairing a nice little P90 that I sold to a guy at work that had stopped working; apparently after very little service.  Alittle bit of tinkering and I see that there is some kind of short between V+ and ground.  I go over the copper side of the board with a fine-tooth comb and heavy magnification, looking for shorts or solder bridges, and can't find anything.  One by one, I desolder the chips, and the power line is still shorting to ground.  Finally, I decide to test the 1N4148 protection diode on the power line, that would normally shunt any reverse voltage coming from a wrong-orientation power supply or battery momentarily touching the wrong contacts on the battery snap, and I see the diode is fried - 000mv voltage drop in both directions.  Yank it from the board, and the short disappears, so clearly this is the source of the problem. (Which I'm very happy about, because I promised him I'd bring it in repaired tomorrow).

So how does a diode of that type in that sort of role get fried by the usual sorts of user errors?

[merlinb]

So how does a diode of that type in that sort of role get fried by the usual sorts of user errors?

Reverse polarity power supply (or an AC supply). The diode passes all the current the PSU can deliver and burns out. Really it ought to be a power diode in that position.

[Mark Hammer]

So a 1N4001 is a better/safer candidate?

[digi2t]

So a 1N4001 is a better/safer candidate?

That, or a 1N4004, is what I've always used for protection.

[Keppy]

I use a series diode in the positive supply. If polarity is reversed, it just won't conduct. I use a 5818 Schottky to keep the voltage drop to a minimum.

Several posters on here have voiced the opinion that the reverse biased diode to ground is either A) a relic from when all pedals ran on batteries and reversed connections were brief or B) a plot by pedal manufacturers to allow them to void your warranty for connecting the wrong power supply.

[Govmnt_Lacky]

What is the reverse bias threshold on the 1N400X series diodes?

Just wondering if using these as reverse polarity protection could cause a lower amperage wall wart to continue to provide reverse polarity and burn up! 

[darron]

These days I find in the case of most modern regulated supplies any diode reversed or shorted should be an overwhelming current and turn the supply off regardless of the diode type. But when people use simple transformer power packs or an AC supply it just keeps on cooking. I prefer the series schottky unless I'm trying to do an authentic fuzz or something. Some people swear schottky are more trouble than they are worth too! A lot of them only have a reverse of about 40v

[PRR]

> 1N4148 protection diode

That's not protection.

The diode must be BIGGER than the WORST-CASE wrong supply.

If you have a well-known 9V 1.8A supply, it will try to dump 1.8 Amps the wrong way.

Or as Merlin says, the 12V 5 Amp supply someone found at GoodWill.

The 1N4148 is rated 0.1 Amps. It is 18 or 50 times too small.

(Some smart power supplies will shut-down faced with a near-short.... but you won't be that lucky.)

The 1N4148 *may* clamp a 2nd-best 9V battery. A full-fresh DuraCat Plus really should be able to cook a 1N4148.

And if you are talking worst-case, in this too-electronic world, there is NO limit what some fool will ignorantly plug into.

So you can't just clamp. You also have to throttle the input.

Add a series resistor before the diode. 100 Ohms is a nice starter. For one transistor or opamp the drop will be less than a Volt. P90, perhaps more.

100 Ohms and a 0.1A diode will survive 0.1A*100r= 10 Volts the bad way.

With the equally-cheap 1N4001, 1 Amp, you can go 100V without blowing the diode. The resistor will smoke, which does alert the user that something bad has happened. (Unless you use a 100 Watt resistor in a box bigger than a stomp.)

And since you have a rail bypass cap(?), a 100r resistor in the feed helps the cap fight-down the ripple-crap from mystery supplies.

For non-minimal pedals where 100 Ohms hurts, I think it is time to give-up the reverse diode "protection" idea. The world has become too brutal for such an approach. Use series diode.

[R.G.]

So how does a diode of that type in that sort of role get fried by the usual sorts of user errors?

Overdissipation by over current, or sheer overcurrent burnout - as Merlin correctly notes.

Finally, I decide to test the 1N4148 protection diode on the power line, that would normally shunt any reverse voltage coming from a wrong-orientation power supply or battery momentarily touching the wrong contacts on the battery snap,

The max current rating on a 1N4148 is 100ma continuous, something bigger for single pulse. As Merlin again correctly points out, that should be a power diode. As several people touch on, the 1N400x series is often used, and cheap. They're 1A continuous, and huge currents for single pulses.

However, long term high current pulses do heat them. An AC power supply connected to the DC in or a really high current power supply the wrong way will overheat a 1A diode and eventually either burn it to a short or disconnect it by melting the solder.

1N4001 is 50V reverse voltage. If you hook more than a 50V reverse voltage to your pedal, it won't help much. Ditto with 40V Schottkys in series.

Forward voltage conduction starts at about 0.5V, like for any silicon junction.

The power supply I support professionally is nominally rated for 1.7A continuous, but does a foldback current limit with eventual shutdown and restart every so often. I have tried and failed to kill 1A diodes hooked revers with it. They don't get noticeably warm because of the foldback/retry behavior. This is not the case with simple current limit at 1A. That will usually fry the regulator from heat as opposed to overcurrent if it's not well heat-sunk.

I agree with Keppy - series protection is better. That's the beauty of the series MOSFET switch: http://geofex.com/Article_Folders/mosswitch/mosswitch.htm Series switching so it protects without trying to out-conduct the power supply, with lower drop that most Schottkys can manage at pedal currents.

[davent]

R.G., I've been cheaping out and using your Cheap and Good scheme, where might it fit into the polarity protection hierarchy?

Thanks!

http://www.geofex.com/article_folders/cheapgoodprot.htm

[GibsonGM]

Awesome analysis, guys.  And here we were, all of us thinking we were SAFE with 1N914's!! 
A little closer look shows that to be not quite the case.  Good discussion.

[Mark Hammer]

+1

I'm glad I asked the question.  I was living in a dream world.  Glad I recently bought a bunch of 1N400x diodes.

Incidentally, the fellow I repaired the pedal for told me that "the battery got really hot".  So, the extra tidbit of info here is that, if one applies the wrong power to the circuit, fries the (presumed) "protection" diode, and then connects a battery, that battery will drain PDQ through that created short.

[ashcat_lt]

Which then leaves the question how the heck did he connect a 9V backwards long enough for it to get hot?  I mean, they're unidirectional on purpose, right?  Unless something in the wiring is messed, he must have tried pretty hard!

Frankly, though, you got lucky that the diode failed to short.  The couple that I've seen just exploded and left the reverse power free to plow through the rest of the circuit popping electros and frying chips.

[Mark Hammer]

Just met with the guy, and we suspect he had simply taken a fresh alkaline 9V and accidentally tried to install it with the wrong battery terminal going to the wrong battery snap (and I'll admit that sometimes, if the lighting is just right/wrong, you can look at the snaps - especially the ones where the female snap is curled "in", rather than crowning "out" - and they can be hard to tell apart at a glance).  When he realized his error, he attached the battery "correctly" and found that it got really hot.  Small wonder; the + and - were essentially shorted out and directly connected.

As it turned out, the wisdom conveyed to us in this thread about what sort of diode serves best in this role, was not received until I had replaced the original with another 1N4148.  And quite frankly, after far too much soldering and trouble-shooting the past few days, I was in no mood to take it apart, change out the diode, and reassemble.  So I simply explained to him just now what went wrong, why it went wrong, and what he should strenuously avoid doing in future.  Should the worst happen again, I'll know what diode to install.

[R.G.]

R.G., I've been cheaping out and using your Cheap and Good scheme, where might it fit into the polarity protection hierarchy?

It's close to the MOSFET in operation. It won't stand as much reverse voltage, but as otherwise pretty good.

Just met with the guy, and we suspect he had simply taken a fresh alkaline 9V and accidentally tried to install it with the wrong battery terminal going to the wrong battery snap (and I'll admit that sometimes, if the lighting is just right/wrong, you can look at the snaps - especially the ones where the female snap is curled "in", rather than crowning "out" - and they can be hard to tell apart at a glance).  When he realized his error, he attached the battery "correctly" and found that it got really hot.  Small wonder; the + and - were essentially shorted out and directly connected.

A fresh alkaline battery can sometimes do a couple of amps for a short time. That's enough to fry the insides of the 1N4148/1N448/1N914 signal diode stuff. A 1N400x diode will not be shaken up too badly. But the battery will get hot.

As it turned out, the wisdom conveyed to us in this thread about what sort of diode serves best in this role, was not received until I had replaced the original with another 1N4148.  And quite frankly, after far too much soldering and trouble-shooting the past few days, I was in no mood to take it apart, change out the diode, and reassemble.  So I simply explained to him just now what went wrong, why it went wrong, and what he should strenuously avoid doing in future.  Should the worst happen again, I'll know what diode to install.

Maybe he had a teachable moment. 

[knutolai]

For non-minimal pedals where 100 Ohms hurts, I think it is time to give-up the reverse diode "protection" idea. The world has become too brutal for such an approach. Use series diode.

Would germanium diodes be the best suited for this as the voltage drop is lower than for silicone? Or is there any other shortcomings of the germaniums that prevent them from being suitable?

[darron]

For non-minimal pedals where 100 Ohms hurts, I think it is time to give-up the reverse diode "protection" idea. The world has become too brutal for such an approach. Use series diode.

Would germanium diodes be the best suited for this as the voltage drop is lower than for silicone? Or is there any other shortcomings of the germaniums that prevent them from being suitable?

I've got some fatty russian germanium diodes at home that hold up to a reversed battery without any problem. Most of them are small signal though.

But along your line of thinking, wouldn't a silicon schottky diode be even better with a lower voltage drop still? Why germanium?

[knutolai]

Im pretty clueless on Shottky diodes. Ill check that out. Is there anything else than the forward voltage drop I should look at? Im doing all my powering with a wall wart

[R.G.]

Using a series diode means that the forward drop of the diode is subtracted from the available voltage.
A silicon diode has about 0.6V forward drop at the currents a pedal might use.
A germanium diode has about 0.3V forward drop at the currents a pedal might use.
A Schottky diode is effectively "half a diode", at least in terms of forward drop. Silicon Schottky diodes begin conduction at about 0.3 to 0.4V. GERMANIUM Schottkys would be lower still, but would suffer a lot from the high leakage and temperature problems that plague ordinary germanium diodes. I'm sure these exist, but not sure if they are in active production anywhere.

At one time, losing 0.6V was a disaster if the battery sagged to 7V. Today, I suspect few pedals run from batteries - it's too expensive. And once you're running from a 9.0 to 9.6V power supply, losing the 0.6V of a silicon diode doesn't hurt so much.

Losing 0.3 to 0.4 hurts less.

Losing 50mV to 100mV with a MOSFET or transistor switch hurts even less.

[knutolai]

thx that makes sense. I think Ill go with a standard 1N4001 in series