Reverse Polarity Protection 1N5817 or 1N4001

[R.G.]

@amptramp

I don't know any other industry that has stayed with a legacy power requirement for more than five decades.

... there are good reasons to move away from 9V for some pedals. For others, it doesn't really make a difference, so we keep what we have. Inertia is a powerful force.

You'd be shoveling sand against the tide. This is one of those things you can't fix. It has to wear away, like eroding stone. Have you ever contemplated how different things would be if the early amplifier designers had used a single dual triode as a differential input, instead of a single ended amplifier input? Noise issues we struggle with would never have existed. Or that the "Japanese transistor radio" revolution in the early/mid 1960s had not happened, with the resultant rise in availability of the PP3/9V battery? The earliest pedal makers used AA or AAA cells. Then we got silicon, and the 9V battery was handy, and now the most coveted vintage effects demand 9V, so every older commercial pedal used 9V, and then the pedal hacker/solderer/cottage industry would not have been tied to 9V, and then...

The first entity selling something sets the standard. That's how Boss got us the center-negative polarity on adapter input plugs. But it is a truism that you simply can't rewrite history, nor change every one of the every user's minds. So the whole I'm-a-new-pedal-seller industry simply must use 9V because of the history, and there is no one with the clout to do things differently, excepting pedal makers who had to do it for other reasons - the digital people who needed vastly more power.

Sigh. Like I said, the unspoken "standard" is eroding over time, as the BUMS (Blind Urge to Mod Syndrome) crowd expands to using 12V and 18V to "improve" their 9V pedals. In that way, the standard is eroding. But you can't un-make thevintage fuzz face, nor can you un-convince people of what fuzz sounded good 50 years ago. You probably have to wait 20 years for the majority of those marginal players to age out of playing.

BTW: Is there something about power supply filtering and polarity protection methods in the Wiki of FAQs? I'd be willing to write something about the basics if someone would be willing to read the thing and correct mistakes and such. Don't want to confuse novices with poorly written half-truths.

I tried to do a lot of that with all of the "how tos" at Geofex, but the internet runs on new bodies, so the newbies go only to the newest pages, ignore the FAQs, and don't read the stickies. Teaching novices is hard. Getting their attention is even harder.

[POTL]

Hi
Thanks to everyone who responded, today I checked 1n5817 - this method on two mock-ups of three, my multimeter showed a drop of about 0.7V (what?), The layouts were powered by batteries.
The tone has really changed, but not globally, the low and high frequencies have slightly decreased.
In general, the difference is small, but my fuzzy model obviously does not like it, it starts to give oscillations when it drops below 8.6.
I decided to give up this protection.
The only way I came up with this is the detailed information next to the power socket.
In general, my observations - circuits using operational amplifiers do not pay attention to falling, when working with field effect transistors and germanium, the difference is more palpable.


Thanks for the advice.
In any case, I'm sure that the information voluminously submitted at work with one section of the scheme will be practical for other users.
This is much more convenient than ripping it to pieces from large sources.
This led me to think that it would be great if experienced users created a FAQ on different parts of the scheme, with brief but useful information.
For example, much can be pulled from my previous topic about power filters, there are links to calculators and a detailed explanation of the principle of operation of all components =)

[Phoenix]

today I checked 1n5817 - this method on two mock-ups of three, my multimeter showed a drop of about 0.7V (what?)

Did you purchase your 1N5817's from a reliable source like Digikey, Mouser, element14/Farnell/Newark/CPC, RS Components, Smallbear, etc. Or did you get them from eBay or Alibaba or something? 1N5817's don't drop that much voltage until about 1.5A, and that's only an instantaneous rating, not continuous. Sounds to me like you've got yourself some 1N400* that have been mis-labelled or you've been sold fakes.

it starts to give oscillations when it drops below 8.6.

I don't mean to insult, but if the design is intolerant of a 400mV drop, it is a bad design and needs revision. There's no reason such a small voltage drop should cause oscillation. You may need further decoupling, or to work on your layout.

[Fancy Lime]

@ R.G.
True, baby birds want to be fed directly to the mouth and would starve in front of a buffet. If that were not so, the we wouldn't need teachers in school, just books. Although that can be a bit frustrating, all is not in vain, I learned pretty much all I know about electronics from your and Mark's and Aron's and Craig Anderton's and Tim Escobedo's and Joe Davisson's and Jack Orman's articles, posts and schematics and I greatly appreciate the great work you guys have assembled there. But it would of course have been easier to just go to a forum and ask for directions from the senseis.

@ POTL
Greg is right, if your 1N5817 really drops 0.7V, then its not a 1N5817. And if your fuzz circuit starts to oscillate when supply voltage drops, there is something wrong with the fuzz circuit, not the supply protection. That should never cause a fuzz to oscillate. And it would also mean that an aging battery or simply using a different wall wart than you did when prototyping might cause oscillation. You need to get rid of this problem first, else the whole thing is practically unusable, imho. Have you accidentally built an oscillator into your fuzz, that is controlled via the input voltage and if voltage is high the oscillation is supersonic? Happens quite easily in a high gain circuit when there is positive feedback where it should not be (either by faulty design or short circuit on the board). If you could post the schematic, that may make the problem easier to diagnose.

Cheers,
Andy

[POTL]

It seems that I'm taking my words back =)
Today I decided to once again measured the voltage change when connecting 1N5817 and 1N4001
1N5817 ate about 0.1-0.12V
1N4001 ate about 0.4-0.45V
Either my multimeter deceived me, or the last time I measured it wrong.
Well, if my fuzz will work well, I definitely choose the side 1N5817
Once again, thank you all!

[POTL]

@ R.G.
True, baby birds want to be fed directly to the mouth and would starve in front of a buffet. If that were not so, the we wouldn't need teachers in school, just books. Although that can be a bit frustrating, all is not in vain, I learned pretty much all I know about electronics from your and Mark's and Aron's and Craig Anderton's and Tim Escobedo's and Joe Davisson's and Jack Orman's articles, posts and schematics and I greatly appreciate the great work you guys have assembled there. But it would of course have been easier to just go to a forum and ask for directions from the senseis.

@ POTL
Greg is right, if your 1N5817 really drops 0.7V, then its not a 1N5817. And if your fuzz circuit starts to oscillate when supply voltage drops, there is something wrong with the fuzz circuit, not the supply protection. That should never cause a fuzz to oscillate. And it would also mean that an aging battery or simply using a different wall wart than you did when prototyping might cause oscillation. You need to get rid of this problem first, else the whole thing is practically unusable, imho. Have you accidentally built an oscillator into your fuzz, that is controlled via the input voltage and if voltage is high the oscillation is supersonic? Happens quite easily in a high gain circuit when there is positive feedback where it should not be (either by faulty design or short circuit on the board). If you could post the schematic, that may make the problem easier to diagnose.

Cheers,
Andy

I fully agree with your request to RG, I would add to this list Brian Wampler.
My fuzz is a germanium Tone Bender MKII with a switch to Fuzz Face / Tone Bender MKI
In Fuzz Face mode, he does not pay attention to the voltage drop, in TB MKII mode
he begins to create such sounds

I used this scheme


The only change I made was the replacement of the resistor at the base of the first transistor, instead of the nominal value of 100K, I installed 22K.
This rating proved to be the best, it lowered the volume of the TM MKII to the FF level, while the level of the gain was the same, so it allowed to reduce noise (although it's worth trying to install a shunt capacitor in front of the base of the right transistor).
What characteristically I collected and other fuzz, the problem only occurs with TB MKII.
I used Russian transistors MP42B and GT402 (letters).
They sound great, easily accessible, cheap, in principle they can be obtained for free, because people who do not make music stompboxes just throw them away or give away for free.

I recently traded 3PDT for 10 GT402
And the other day I have to bring a box of different germanium transistors 

[Plexi]

Using this thread, may I ask:
I've been using 1n5819 and 1N4007

I'm wasting my time?

1N400* series: as I read, the difference is in the amount of voltage inverse they can reach before breakdown (smoke like an old train..)
Normal Vf is ~0.7V

1N4001 = 50V
1N4004 = 400V
1N4007 = 1000V

About the 1N4148: Vf 1V

What about the 1N581* series?

[bluebunny]

20/30/40V for 1N5817/8/9.

[lars-musik]

I'm a bit late to the party but the 1N5817-scheme is not necessarily as perfect as assumed.

I already commented in another thread (here) that I bought a whole bunch of an SMD variant of that diode (datasheet) because I thought that would be perfect – super small, low voltage drop, cheap – but then a customer just tried any PSU he had lying around and hooked an (suspected) AC source to the pedal. The diode didn't burn through but opened up and took a precious LT1054 with it. Now I made myself some pcbs for RGs cheap and reliable reverse protection. But I must say: one diode was easier!

[Fancy Lime]

How would AC harm the reverse polarity protection? How can a diode open to reverse polarity without burning through?

I had a (maybe) similar case, when a customer kept returning a pedal with burned out op-amps (bloated and scorched) despite power protection. He insists he only ever used a correctly polarized 9V PSU. Yet I could never replicate the problem because I never managed to burn the thing with any of my PSUs. I suspected a faulty PSU back then that put out a lot more than 9V. That is the only reason I could think of because the pedal had reverse polarity but no over-voltage protection.

Andy

[Rob Strand]

The short answer is:
- series protection use Schottky for lower drop.
- parallel protection use silicon rectifiers like 1N4004 - as in most cases you are wasting your money using a Schottky.

[Phoenix]

I'm a bit late to the party but the 1N5817-scheme is not necessarily as perfect as assumed.

I already commented in another thread (here) that I bought a whole bunch of an SMD variant of that diode (datasheet) because I thought that would be perfect – super small, low voltage drop, cheap – but then a customer just tried any PSU he had lying around and hooked an (suspected) AC source to the pedal. The diode didn't burn through but opened up and took a precious LT1054 with it. Now I made myself some pcbs for RGs cheap and reliable reverse protection. But I must say: one diode was easier!

How would AC harm the reverse polarity protection? How can a diode open to reverse polarity without burning through?

The power supply was an AC wall-wart, so was small, and therefore had poor regulation. Remember, a 9VAC transformer actually puts out ~12.7V peak, and higher off-load. An LT1054 has a max voltage of 15VDC, so it would only take 18% regulation to exceed its max voltage and burn it out. Once it had gone short circuit, it would destroy the schottky.
Typical AC wall-warts have regulation as poor as 30%, so this is not at all surprising.

If my logic is correct, then RG's cheap but good circuit wouldn't have protected the device in this instance, because it does not have over-voltage protection, however, with the addition of a couple more parts, RG's circuit can be made to also provide very robust over-voltage protection too.

[Fancy Lime]

The short answer is:
- series protection use Schottky for lower drop.
- parallel protection use silicon rectifiers like 1N4004 - as in most cases you are wasting your money using a Schottky.

Where do you buy your diodes? Where I usually buy, a 1N5817 costs a whopping 0.06€ more than a 1N4001. I would have to buy over 100 diodes to save enough money to buy one beer at a bar by going with the cheaper diode. Granted, even in series mode using a 1N400x is not going to have an audible effect in most cases. But when voltage drop is a concern and you want to avoid more expensive and/or complicated protection measures, I think the advantages of series protection by far outweigh the minuscule savings of going parallel with a 1N400x.


@Phoenix
Ah, that explains it. in the end, the dilemma with protection is how much to trust the user. Are we nice and try to make it absolutely idiot proof or are we cynical and say "if you cannot or will not follow clear instructions about which power supply to use, you do not deserve to play this thing anyway". Unfortunately, in my experience at least, many musicians are complete technical (or electrical) dyslexics who struggle with the concept that electricity comes in different forms and strengths. I once spent quite a while explaining to a guitar player that "having different looking connectors" is not the only difference between a wall power outlet and a 9V Boss-connector.   

Andy

[R.G.]

How would AC harm the reverse polarity protection? How can a diode open to reverse polarity without burning through?

When the "don't use AC" issue came to light, it came up with customers from my day job calling in to tell us that our extensively-protected and tested DC power supply for pedals was killing their pedals. I spent quite a lot of time doing lab work to find out what was happening. I'm sure other people might have found this out first, but to the best of my knowledge I brought this issue to light in the pedal community. It works like this.

AC-only supplies are simple transformers, not active power supplies with current limiters. An actively regulated supply will limit its output current to some value simply to protect itself, and its load. Transformers do not.

So a transformer feeding a circuit with a reverse-polarity diode feeds the circuit a half-sine wave in the forward direction, but is "shorted" by the reverse protection diode in the reverse direction. The only limit on how much current flows in the diode is the diode's forward resistance, which is deliberately as small as it can reasonably be made, the resistance of the lead wires, and the resistances of the transformer windings. These are all, in fact, as small as they can reasonably be made. So the reverse current in the diode is not  just the limited current that a reversed regulated DC supply will produce at limiting, but "large".

That's OK for a while, as the pulsed current rating of a silicon diode is quite a bit larger than its DC average current rating. But it does cause heating, and the diode heats up. The diode's thermal time constant is short, as it's small and low mass, so in a short time, the diode heats quite a bit. A couple of things can happen. One is that the diode can de-solder itself. I've seen this happen in a couple of experiments. Another possibility is that the very hot diode can exceed the thermal rating of the silicon chip inside the diode, and the junction will get over the 150-200C that the junction can stand, and you'll get large enough currents in the chip that it will punch through.

When that happens, the diode conducts as a relative short in both directions. This doubles the currents going through it, as it now has large currents in both directions. So the diode gets even hotter.

What can happen then is heavily dependent on tiny differences in the setup - a kind of "Butterfly Effect" for thermal disaster. The diode can simply sit there heating, and char a spot on the PCB under it. If the transformer feeding it has enough current capability into a short, it can even burn through the PCB. Mostly the transformers are small and can't quite do this. It can unsolder itself in the second phase of thermal disaster. Or it can burn through the bonding wires or open the wire-to-chip solder connections on the chip itself. In any case, it can burn open.

While the diode is shorted, it has been protecting the circuit from the reverse polarity half-cycles. When it opens, it is no longer part of the circuit and the AC starts killing chips and other polarized things in the rest of the circuit. How far this goes depends on how long they heat. Opamps die pretty quickly. Other things take longer.

I did experiments with out power supplies and diodes, leaving them reverse collected for long periods of time. The diodes only got a bit warm, as the protection clamped the currents to the supply max, the supply shut down and waited to retry. I could never damage a diode with our supplies. I connected diodes to 9Vac 1A wall warts, and the would smoke in minutes.

The real crux of the problem is that the plug on the 9Vac adapters as used by L!&# 6 is a 2.5mm center pin, 5.5mm outer barrel, and this can be forced into most DC 2.1mm/5.5mm sockets. Grab the wrong plug and plug it into your daisy chain, and you start killing pedals.

[Fancy Lime]

Hi R.G.

thanks for the detailed explanation. Just to be clear: does that apply to both single diode protection schemes, series and parallel, or just parallel? For the parallel case I see how the scenario you describe would play out. But for series protection there should be no current through the diode during the reverse polarity half-cycle of the AC, correct? And during the forward polarity half-cycle the current is limited by the effective load resistance of the circuit, isn't it?

Thanks,
Andy

[R.G.]

thanks for the detailed explanation. Just to be clear: does that apply to both single diode protection schemes, series and parallel, or just parallel?

Yes, just the reverse-parallel situation. For series diodes, the polarity is always in the correct direction, even though with a 9Vac pedal, the peak voltage is larger than the "9v", about 12.7V nominal.

The only objection to a series diode is the voltage drop. Schottky diodes are good for having a lower voltage drop, but they have a reverse breakover level of as little as 20V. Not a big deal, but still there.

The point of more complicated series protection schemes is that they can give the same series protection without the forward drop. The MOSFET series scheme can be as low a voltage drop as you want to pay for the MOSFET, and the "Cheap But Good" bipolar scheme got down in the tens of millivolts, although it also has its own reverse voltage breakdown issues.

I really liked the idea of an active MOSFET bridge, because polarity is no longer a concern, full stop. All variants just work, given that you can provide too much voltage with an AC power supply fed to it and full wave rectified to the peak of the AC. There are still those issues of millivolts of ground offset, but hey, nothing is perfect.

[POTL]

The short answer is:
- series protection use Schottky for lower drop.
- parallel protection use silicon rectifiers like 1N4004 - as in most cases you are wasting your money using a Schottky.

Often I hear that the Schottky is more expensive, but I see the price tag at 1N4001-0,05 $ / 1N5817-0,07 $, when buying several hundred the price will be equal to half.
Funny difference in price.

[Fancy Lime]

Once we're at the "complicated but awesome" circuit protection topic: I was wondering why not use two relays that are controlled by transistors in such a way that with no voltage the circuit is cut off from the power supply (+ and - leads) and then two relays switch the appropriate power input lead to the right rail of the circuit, depending on polarity of the power supply. Expensive, bulky, complicated and (unlike diode or mosfet bridges) too slow to deal with AC, but on the other hand: DC power can be connected either way round and there is (nominally) no resistance in the ground wire. Not that this is really practical or necessary but it should be possible. Yet I have never seen this used and normally it seems that if it's complicated and could be sold as "more advanced" and it's at all possible, someone is using it, no matter how unnecessary.

Cheers,
Andy

[anotherjim]

I think there is a hidden catch with the series diode scheme. You may have gone to the trouble and expensive of getting a really well regulated, smoothed, low noise DC supply. Does the series diode in the pedal +9v affect the supply at the pedal in any way, apart from the diode voltage drop?

The supply will almost certainly be using a reasonable amount of capacitance on it's outputs. When the pedal circuit is sucking DC current, the series protection diode is forward biased and as much current as it needs will pass. But, what about AC signal current? We have put a rectifier in there.

Stop and consider all the ancient & not so ancient pedal circuits that have zero power rail capacitance built in. They worked fine with a battery, because that comes with a lot of capacitance built in for free. They work ok too if a DC supply with sufficient capacitance isn't too far away and may also benefit from other pedals on the same DC that do have supply capacitors. However, the existence of the series protection diode means that signal currents (though they may be tiny) are blocked in one direction by the diode. This can easily result in audible distortion. That distortion can be really horrible where signal currents are not small, such as 9v practice amps.

I think series protected pedals must have relatively large local supply capacitors to handle all the signal current needs. 100uF is good without taking too much space and use a 25v rating so it won't mind an 18V supply accidentally or deliberately employed.
Chip amps need a lot more C. At least 1000uF will keep them happy.

[Fancy Lime]

Hi Jim,

you mean an additional cap between the rails before the diode? Or just the "normal" one after the diode (and series R), which imho should always be there, no matter what the protection scheme.

Andy