PNP Negative Ground Principles

[Mcentee2]

Wait...come back!

Really, this is something I have yet to understand even after wading through countless posts for hours as to why this is a bad idea and might or might not work sometime.

Particularly to the AMZ page:. http://www.muzique.com/lab/fuzzface.htm

What is the actual, honest to goodness,  basic point of doing this?

The PNP transistors and current flows are exactly the same in both, to me this is still "emitter points to more positive voltage than the collector" - so what is the use of this?

Is it to be able to run this and other negative ground pedals off the same supply, daisy-chained?

If so, then I always thought this wasn't possible anyway, ie the difference is between (e.g.) 9v to 0v (-ve ground), and 0v to -9v (+be ground) therefore you can't wire then with the same supply, have to use isolated or battery for one.

If the intention is to daisy chain, then my understanding is missing something re that last paragraph!

[amz-fx]

so what is the use of this?

It allows you to use the circuit on the same negative ground power supply as other pedals that might be connected to that wall adapter or power supply, instead of requiring a separate supply with the power rail negative with respect to ground. Every Zvex Fuzz Factory uses this power scheme as do lots of other pedals. However, it is possible that a circuit layout or a vintage transistor with quirky characteristics or even just a poorly routed  ground connection can cause oscillation. A good alternative is to use a charge pump chip to generate a negative voltage to power the circuit.

regards, Jack

[Mcentee2]

Thanks - so it is my understanding of the shared power supply that is wrong for -ve and +ve ground pedals.

I'll untangle this mental mess eventually!

[Mcentee2]

Still not getting it.

There are many descriptions and arguments similar to this one that explain why you can't mix the two,

http://stinkfoot.se/archives/532

I can't seem to square this with the AMZ approach ?

[amz-fx]

With a PNP transistor in this type amplifier circuit, the emitter must be more positive than the collector. You can ground the emitter and use a negative voltage on the collector, or you can use a positive voltage on the emitter and ground the collector. Either way, the emitter is at a more positive potential than the collector.

If you are using one power supply that has the negative grounded, then you have to use the AMZ method, or generate a negative voltage rail referenced to the ground from the power supply.  Some circuits even use both a plus and minus voltage as well as a ground (many op designs).

regards, Jack

[Mcentee2]

With a PNP transistor in this type amplifier circuit, the emitter must be more positive than the collector. You can ground the emitter and use a negative voltage on the collector, or you can use a positive voltage on the emitter and ground the collector. Either way, the emitter is at a more positive potential than the collector.

If you are using one power supply that has the negative grounded, then you have to use the AMZ method, or generate a negative voltage rail referenced to the ground from the power supply.  Some circuits even use both a plus and minus voltage as well as a ground (many op designs).

regards, Jack

Ah, ok, so the fuzz face and similar circuits are constructed in a way that "allows" that rearrangement to get relative + and - voltages;

Whereas other types of circuit build can't operate like that regardless and cannot be run on the same supply with just swapped terminals?

ie they do require a discrete +ve or -ve and a ground and can't share a supply with the opposite type.

[R.G.]

Let me present an alternative view.

Well, actually, it's the same view I always express when this comes up. Don't do this unless you have training/experience and steadiness of mind to realize that although many, even most, attempts at making this work do in fact work, there is a noticeable percentage of times when it will not.

If the collective we knew all the reasons for why it sometimes fails, some of us ( I for one ) would write them up in a concise form so pedal hackers would never be troubled again. The inverse is true - I do not know all the ways this can fail, and that is the reason there is not a simple, easy guide to making it work first time every time.

I can hear you thinking "But wait! I've read that... " Yes, you have. Inverted grounding is sneaky. Many times it just works. But inevitably there will come a time when it won't. I was extruded through a formal EE education, and spent decades designing circuits for a living, as well as making and designing pedal circuits on my own for fun. The formal, theoretical side says that because you can happily reverse power and ground for AC purposes because they are the same voltage for AC signal purposes. Problem is, an ideal, theoretical DC power supply has an AC impedance of zero at all frequencies, and no real power supplies do this.

One would think that simply doing a better job of power supply decoupling will make it always work. Sadly, no, it won't. Mostly it works, depending on how well you decouple. Sometimes, it doesn't. This tells me that sometimes it's not the power supply impedance that's at fault.

My conviction on this point comes from three sources: (1) it comes up over and over and over on pedal forums, (2) there is a smaller group of posts from people who have tried inverted grounds and it did not work, and (3) I once expended lots of hours and used every circuit design trick I knew and that I could glean from the circuits pros I worked with and failed to fix my persona hard case.

In all the cases where we've received posts from people who tried inverted grounds and had them fail, putting the power and ground back the "right" way fixed the problem.

This topic gets repeated play because it works mostly. So people who don't do this a lot may not ever run into a hard case that some decoupling or signal-path tinkering can fix up. In their experience, it works every time - and it has, for them. But keep doing it and you'll eventually run into a hard case.

This is a fine thing to d if you're ready and informed enough to just roll with it and go to plans B, C, D, etc. when it hits you. If you can't, take the advice: put in a charge pump chip to generate the inverted voltage and let ground stay ground, or use a separate isolated power supply in one of the several ways that can be done. 

[Mcentee2]

Appreciate the response, R.G., and I have read your similar  replies over the years re what to do and not to do, I accept every word.

However, I think I am asking something slightly different to those countless "help, my PNP -ve ground build doesn't work".

I am actually trying to understand the theory as how it is supposed to work in the first place and why AMZ presents the build as he does.

This is with all the other advice I read about clanging around my brain about not mixing +ve and -ve ground builds.

I just can't quite yet get my head around the theory of why it should work 😀

There is something simple I am missing cometely at a base level to square those two bits of "lore".

[Danich_ivanov]

I'm powering pnp's with ground on the collector and +9v on the emitter. It doesn't work great in all circumstances npn would work in, i would think because base votlage forced to be on the high side (and correct me if i'm wrong on this), but it can replace npn's in some cases, which can come in handy if you want to use less npn's for example, and it will work just fine in a standard "negative center" type circuit.

[amz-fx]

I am actually trying to understand the theory as how it is supposed to work in the first place

Here is a greatly simplified representation of what we are talking about. G is the chassis ground that is common to all other pedals in your signal chain. The circuit in the blue box between the input and output capacitors is essentially floating until you assign one of the power terminals to be ground... in this case G1 on the emitter of Q1 is connected to the chassis ground G when the builder designed the pcb. This requires that the collector of Q1 be connected to a voltage that is more negative than the G1 potential. With a conventional power supply, the voltage output is more POSITIVE than the ground side of the power supply, so the circuit does not work properly. The ground in the power supply has already been assigned by the design to be more negative than the positive output.

The circuit above will work with a battery since the terminals of a battery are floating and either terminal can be connected to ground... so with a battery you connect the positive terminal to ground (G1) and the negative terminal to power the circuit and the fuzz works properly.

The circuit above will work with a wall adapter that is only powering this pedal since the power output of the adapter is floating until connected to the pedal. The plus side is routed to ground and the negative powers the circuit, so the fuzz works properly. No other pedal is on this adapter so it doesn't matter that the plus side has been assigned to ground.

The problem arises if you want to power this circuit from a power supply or wall adapter that is also powering other pedals that are wired for negative ground. You only have two power connections and the more negative side is connected to ground for most of the pedals and the more positive connection is available to power the circuit of the pedal, but this fuzz requires a more NEGATIVE connection on the circuit power so it does not work.

The idea is to flip the connections of the fuzz circuit that are isolated between the input/output capacitors, so that the fuzz can be connected to a power source that is also supplying other pedals:



Note that this is only a problem when you have other pedals connected to the same power supply. A battery, wall adapter, negative voltage output or isolated power supply jack will power the original circuit properly.

(Edited images to improve them.)

[Mcentee2]

I am actually trying to understand the theory as how it is supposed to work in the first place

Here is a greatly simplified representation of what we are talking about. G is the chassis ground that is common to all other pedals in your signal chain. The circuit in the blue box between the input and output capacitors is essentially floating until you assign one of the power terminals to be ground... in this case G1 on the emitter of Q1 is connected to the chassis ground G when the builder designed the pcb. This requires that the collector of Q1 be connected to a voltage that is more negative than the G1 potential. With a conventional power supply, the voltage output is more POSITIVE than the ground side of the power supply, so the circuit does not work properly. The ground in the power supply has already been assigned by the design to be more negative than the positive output.

The circuit above will work with a battery since the terminals of a battery are floating and either terminal can be connected to ground... so with a battery you connect the positive terminal to ground (G1) and the negative terminal to power the circuit and the fuzz works properly.

The circuit above will work with a wall adapter that is only powering this pedal since the power output of the adapter is floating until connected to the pedal. The plus side is routed to ground and the negative powers the circuit, so the fuzz works properly. No other pedal is on this adapter so it doesn't matter that the plus side has been assigned to ground.

The problem arises if you want to power this circuit from a power supply or wall adapter that is also powering other pedals that are wired for negative ground. You only have two power connections and the more negative side is connected to ground for most of the pedals and the more positive connection is available to power the circuit of the pedal, but this fuzz requires a more NEGATIVE connection on the circuit power so it does not work.

The idea is to flip the connections of the fuzz circuit that are isolated between the input/output capacitors, so that the fuzz can be connected to a power source that is also supplying other pedals:



Note that this is only a problem when you have other pedals connected to the same power supply. A battery, wall adapter, negative voltage output or isolated power supply jack will power the original circuit properly.

(Edited images to improve them.)

Ah, lightbulb!

The chassis and input/output grounds are the same as other shared pedals, but only the "internal section" ground is flipped.

In my head I was mis-looking at the flipped diagram and seeing all those 3 grounds the same.

Many thanks, I think I understand how this is supposed to work.

More than that, it is clear in my head that this is not now a +ve ground device, therefore the advice to never mix those and -ve ground on the same PSU doesn't now apply, woohoo!

I also accept and understand all the arguments why it isn't necessarily something that works all the time, but at least I have that basic understanding my belt.

Many thanks

[R.G.]

However, I think I am asking something slightly different to those countless "help, my PNP -ve ground build doesn't work".

I am actually trying to understand the theory as how it is supposed to work in the first place and why AMZ presents the build as he does.

This is with all the other advice I read about clanging around my brain about not mixing +ve and -ve ground builds.

I just can't quite yet get my head around the theory of why it should work 😀

The theory on why it ought to work at all is simple: Theoretically, a DC power supply is a zero impedance source of DC voltage, with no possible change between the two supply voltage terminals. That is, an ideal power supply of 9V is 9V no matter what the circuit does, including pulling mixed AC and DC currents to make AC signals go out an output. The power supply cannot wobble in any way, because it is theoretically perfect.

Any single ended (i.e., not differential) AC output signal can be derived from the circuit by blocking any DC from the circuit's output with a capacitor. This produces an output AC signal, but the output signal must be connected to some "ground" reference to drive any other circuit that follows it. The "outside terminal of the DC blocking cap can be referred to either the positive or negative side of the circuit's power supply by using a resistor to pull it to the + or - side of the circuit's power supply. Either side works, because no AC voltage can exist between the theoretically ideal DC power supply's terminals. All that changes is the perceived DC voltage on the outside end of the DC blocking cap.

An output reference voltage is needed because electricity flows in circuits - there has to be some electrical path from the signal source (in this case, the circuit that is generating the output) through some follow-on load, and back to the source of the electricity that allowed the circuit to amplify, that being our ideal DC voltage source. If you had the circuit under consideration running and amplifying some signal to produce an AC voltage on its output, and you wanted to measure that output, you could attach an appropriate meter to it to measure the output voltage. Problem is, you cannot measure any output voltage at all without connecting TWO meter leads, one to the signal being measured, and the other to some terminal that's electrically connected back to the ideal DC power supply. The meter needs some reference voltage to measure any signal at all.

It's common to use either the positive or negative side of the theoretically ideal power supply to be the meter's reference, so you would attach the second meter lead to the positive or negative power supply terminal. When you do that, the meter gives you a reading of the signal voltage difference from the output signal to the reference voltage - the "ground" as far as the meter is concerned.

If we replace the meter with a second circuit, another piece of equipment driven by the circuit in question, we have to have two leads for it to see an output voltage too. The followup circuitry has its own internal set of requirements, including what reference source of voltage it uses what its "ground", or reference of zero volts is. Unless it has a differential input, it will use some internal DC voltage as a reference voltage of zero volts. This could be either one of its power supplies, or some voltage in the middle. Whichever it uses, the second circuit will expect a signal wiggling around that local "ground".

So now we have two circuits which each have to see a reference point of zero voltage to work properly. If they have independent power supplies, no connection between the power supply for the first circuit and the second, you can pick either the positive or negative side of the first circuit as "ground", and the second circuit won't care. Neither will the first circuit. The first circuit makes an AC output signal which is referenced to BOTH its positive and its negative power supply leads at the same time. You can pick either one as "ground". Add a second circuit with an independent power supply, and it still works fine, because both circuits have their own reference "ground", and don't depend on the other.

In an ideal situation, where each circuit has its own ideal DC power supply and there's no system-wide ground, you don't run into issues with positive ground vs negative ground or inverted ground. Each unit with an independent power supply is its own "power domain" and whatever it says is ground for it is also ground for its output signal.

The problem comes when (1) the power supplies are not perfect, ideal DC sources and (2) you start hooking those circuits into common system-wide grounds and single system-wide power supplies.

The DC issue is easy to understand - If circuit 1 says that the + side of the system-wide power supply is ground and circuit 2 says that the - side is ground, you can easily have the signal cable grounds connected to both the + side of the power supply and the - side of the power supply - a dead short between + and ground. That is exactly what happens when a positive ground effect and a negative ground effect use the same external DC power supply and you plug a signal cable between them. The signal cable shorts the + side of the power supply on the positive ground effect to the  - side of the power supply on the negative ground effect, and tries to force the + and - sides of the single power supply to drive a DC  volltage across a wire.

The AC issues get more complicated. Even within a single circuit, power supply impedance matters. A non-ideal power has some impedance - a combination of resistance, inductance, and capacitance - between its terminals. This means that it cannot supply any current from zero to infinity without its external voltage changing. The theoretical EE world views this as an ideal voltage source in series with some combination of RLC, and this is a good model for most real-world power supplies.

Now the power supply choice of "ground" matters even inside one overall circuit. Our circuits are usually not a single transistor amplifier, but strings of two or more interconnected transistor amplifier/switch circuits that we conceptualize as one bigger "circuit". As with two independently powered circuits above, now it matters which side of the DC power supply each transistor thinks is "ground", because the power supply itself includes some resistance/inductance/capacitance in series with the actual DC power source. So a common emitter amplifier sees the power supply as adding a bit of resistance and inductance in series with the collector resistor, and an emitter follower sees the power supply as adding a bit of resistance in series with the emitter resistor. In isolation, neither is a problem, because the gain will be a little off, but no issues arise. However, start combining these stages, and now there is a shared impedance inside the power supply causing error voltages on both kinds of stages (happens in common base too, but we don't use that one much) and tossing error voltages from emitters to collectors and through biasing networks back into bases, which happily amplify the error voltages.

What happens is that mostly the impedance-caused error voltages don't hurt things too much, and you mostly get what you expected from the circuit. The more transistors, the higher the gain, the more closely connected the gain stages, the higher the probability that something will satisfy Nyquist's Criterion and the circuit will oscillate, perhaps at a frequency which common 20MHz scopes can't even see, given how very good modern transistors are.

So mostly, it works. Even theoretically you're free to swap + and - for what you say is "ground", but there is a string of asterisks and footnotes associated with that. The more circuits, the more gain, the likelier that an asterisk or footnote will get activated and the circuit doesn't work.

Did that help?

[Mcentee2]

Yes, R.G., that was a very informative read, thankyou.

As usual, I now have some useful notes to bounce off and learn more.

[pinkjimiphoton]

basically it works cuz the caps isolate the q's from the power supply. the q's will work as long as theres a voltage difference, so they don't really care.
can work great, with some caveats.
i find i can do great mating buffallo sounds with my volume knob down on my guitar often with this kinda circuit.
requires a big-ass monster motha funkin' cap across the power rails to cure usually. sometimes that won't work, either.
weirdly, it seems to be fine with the guitar full blast. its only when trying to turn them down i've run into issues.

remember, you can have positive voltage or negative voltage or both, and the midpoint between them is ground... there's more than one way to skin a cat.

me? i like charge pumps better. little more work to lay out, but i don't have to try and cram a 22000uf cap in a 1590 a