Voltage inverter - why and where to put one?

[Guurf2]

Hi guys what's up?
so i decided to do a Rangemaster-style treble booster with a Germanium PNP transistor in it.
As far as i know that means Positive ground. so i'm familiar with the ways of working that out in terms of powering the pedal and i'm planning to put a negative voltage inverter in the pedal (maxx1044 or similar based).
but i saw something interesting that got me thinking - looking into a Jext Telez Rangelord  (a Great Ge PNP Rangemaster clone) i noticed there's no inverter in there, just a plain old treble booster circuit connected directy to the power jack.
so my question is how does that work? the pedal is clearly daisy-chainable (my friend has one on his board daisy-chained to many negative ground positive voltage pedals and it works without any problem). i don't quite understand how it is possible since i have always been told that you can't power a positive ground and a negative ground pedal off the same daisy chain but with this pedal it somehow works.
can anyone here shed some light on this?
Thank you very much.
Guy.

[R.G.]

It's one of those things where theory and practice seem to be different. They're not different, but people usually don't understand the theory deeply enough to see it.

In any case, electronics theory says that if you have a theoretically perfect power supply, you can define either the + or the - side of the power supply as "ground" with no bad effects on how it all works. This is the idea that the Rangelord relies on to work.

No real-world power supply is perfect according to theory. They all have some internal imperfections.

With those thoughts in mind, we can think about PNP circuits. For a bunch of nit-picky reasons, PNP transistors are a more natural match to a positive ground power supply, just like vacuum tubes are a more natural match to a negative-ground power supply. In the first pedals, PNP germanium was all that was available, and pedals were powered only by batteries. With one battery per pedal (and generally only one pedal per guitar player!!) it didn't matter which side of the battery was "ground".  But when you start powering more than one pedal from a common power supply, which side of the power supply is labeled "ground" does matter. If one pedal attaches the positive side to signal ground, and the next pedal uses the same signal ground, but attaches the negative side of the power supply to signal ground, the power supply has both sides connected to signal ground. This amounts to a dead short across the power supply.

And that is the reason that "positive ground" and "negative ground" pedals must have separate power supplies.

People with a little bit of theoretical education in electronics will sooner or later fall back on the idea that they can simply interchange which side of the power supply they call "ground" and eliminate the problem. This is what the Rangelord designer did. Reversing which power supply lead is called "ground" lets you always connect the negative side to signal ground, and now it's daisy-chainable.

There is much to be admired about this approach. It's certainly cheap, and simple. And it works in many cases. The problem is, it does not work in all cases. This is because there are no "perfect" power supplies. Sometimes the internal resistance or inductance of the power supply makes the PNP circuit oscillate or be very noisy. The chances of it oscillating or making noise get bigger as the circuit gain gets bigger.

It is possible in some cases to put a big capacitor between the + and - sides of the power supply and cure some problems with "reversed ground". And sometimes even this doesn't help.

Bottom line is: it's possible. It works sometimes. But sometimes it doesn't. Another approach is to put a charge-pump style voltage inverter inside the PNP pedal and make -9V from the +9V. This gives a negative power supply to run the pedal, and the incoming power supply negative side can still be daisy chained.  Many pedals do this. This works in all cases where you get the negative voltage generator right.

Putting a charge pump

[merlinb]

Bottom line is: it's possible. It works sometimes. But sometimes it doesn't.

RG, you're the only person I've enountered who had such a problem. Can you remember, did the circuit oscillate in isolation, or only when connected to another pedal with a shared power source?

[R.G.]

I'm not your typical pedal or audio hacker, probably. I've encountered this situation on more than one occasion, but then I've been through a lot of pedal circuits. Actually, I've run into it on quite a few occasions, but what I think of as ordinary techniques of rewiring and power supply improvement cured it in most instances. But there have been a few hard cases. Those last two sentences account for my advice.

As best I can remember, the hard cases involved PNP distortions with high gain, usually the buffered input type, but a couple of fuzz face variants, and one NPN-PNP complementary setup. All the tricks I could muster with good RF technique, rewiring, better grounding techniques, power filtering and decoupling, even tossing sea salt over my left shoulder while standing on my right foot chanting in Welsh. In all cases, putting the power supply "ground" situation back to the more normal positive ground setup promptly fixed the issue.

When I have posted my regular polemic about reversed ground, it's because someone posts either the question about whether reversed ground works and is OK, or posts that they did the reversed ground and now their pedal oscillates or makes no sound at all. I believe that in all instances, putting it back to positive ground made the pedal in question start working.

In my mind, my advice is sound based on my experiences, both on my personal bench and in reported instances on the web. Given that isolated power supplies are becoming so ubiquitous, either as a high quality additional power supply for under US$25 or one channel of the multiple-output power supplies that have become so popular, and that charge pump converters have become so well known, there doesn't seem to be any reason to actively choose to take the chance on reversed ground wiring of a PNP germanium pedal.

As I note every time I post this advice, in most cases, you can make inverted ground wiring work. It works so often that people who haven't run into the hard cases think the technique will always work. As an engineer, my approach is to try to figure out what happens under the surface. If you can cover all situations, it makes sense to do it.

[italianguy63]

I'm not your typical pedal or audio hacker, probably. I've encountered this situation on more than one occasion, but then I've been through a lot of pedal circuits. Actually, I've run into it on quite a few occasions, but what I think of as ordinary techniques of rewiring and power supply improvement cured it in most instances. But there have been a few hard cases. Those last two sentences account for my advice.

As best I can remember, the hard cases involved PNP distortions with high gain, usually the buffered input type, but a couple of fuzz face variants, and one NPN-PNP complementary setup. All the tricks I could muster with good RF technique, rewiring, better grounding techniques, power filtering and decoupling, even tossing sea salt over my left shoulder while standing on my right foot chanting in Welsh. In all cases, putting the power supply "ground" situation back to the more normal positive ground setup promptly fixed the issue.

When I have posted my regular polemic about reversed ground, it's because someone posts either the question about whether reversed ground works and is OK, or posts that they did the reversed ground and now their pedal oscillates or makes no sound at all. I believe that in all instances, putting it back to positive ground made the pedal in question start working.

In my mind, my advice is sound based on my experiences, both on my personal bench and in reported instances on the web. Given that isolated power supplies are becoming so ubiquitous, either as a high quality additional power supply for under US$25 or one channel of the multiple-output power supplies that have become so popular, and that charge pump converters have become so well known, there doesn't seem to be any reason to actively choose to take the chance on reversed ground wiring of a PNP germanium pedal.

As I note every time I post this advice, in most cases, you can make inverted ground wiring work. It works so often that people who haven't run into the hard cases think the technique will always work. As an engineer, my approach is to try to figure out what happens under the surface. If you can cover all situations, it makes sense to do it.

I think this is rock-solid advice.. when I got started making (fuzz) pedals years ago, R.G. gave me the same advice, and I followed it.. Never, have I had a problem.

Never looked back.

MC

[Guurf2]

Ok so thanks everyone youv'e all been very helpful.
no i'm experimenting with adding such an inverter to a FF circuit but it seems i'm not able to eliminate the 60-Cycle hum when using one.
i've been able to eliminate the noise successfully in the past using the simple r/c filter but now when the charge pump is in the circuit it seems the r/c filter isn't helping anymore and the circuit squeals.
anyone got any insights on that?
thanks!
Guy.

[R.G.]

Ok so thanks everyone youv'e all been very helpful.
no i'm experimenting with adding such an inverter to a FF circuit but it seems i'm not able to eliminate the 60-Cycle hum when using one.
i've been able to eliminate the noise successfully in the past using the simple r/c filter but now when the charge pump is in the circuit it seems the r/c filter isn't helping anymore and the circuit squeals.
anyone got any insights on that?
thanks!
Guy.

If I understand correctly, adding an inverter (another transistor? an opamp?) to a FF circuit gives you 60 cycle hum. Is this with or without an external DC power supply? Does it do this when powered by a battery? Does it do this when you short the input signal?

Then, again IIUC, you have used a simple RC filter to eliminate the 60Hz noise. But now, with a charge pump, the circuit squeals.  I don't quite understand the note that the RC filter isn't helping any more. Does this mean the 60Hz hum is back? or that the R/C doesn't help with charge pump noise?

Adding a charge pump is not without its own issues. It's a switching device, and it demands pulses of current from power and ground. To do a charge pump well, you have to keep a "bucket" of charge for it to suck pulses from - a filter cap - as close to the charge pump as you can get it, plus some good high frequency filtering in the form of a ceramic 0.1 or 0.01, again, as close to the input of the charge pump IC as you can get it, and ensure that the CP has its own power and ground wires back to the source of DC power. If it shares a ground wire with the transitor gain circuit, you're going to hear the switching pulses, as the pulses generate noise in the resistance of the wires. This isn't a huge thing - some caps and some thought about wiring.

There is another issue with charge pumps; that is, you want them running at a frequency above audio., If that's true, any noise leakage most likely won't be heard, and any filtering is more effective.

On the other hand, it can be done. I used a MAX1044 pump running in the audio band inside a guitar level volume control in production once. It had no charge pump noise.

[Guurf2]

Ok so thanks everyone youv'e all been very helpful.
no i'm experimenting with adding such an inverter to a FF circuit but it seems i'm not able to eliminate the 60-Cycle hum when using one.
i've been able to eliminate the noise successfully in the past using the simple r/c filter but now when the charge pump is in the circuit it seems the r/c filter isn't helping anymore and the circuit squeals.
anyone got any insights on that?
thanks!
Guy.

If I understand correctly, adding an inverter (another transistor? an opamp?) to a FF circuit gives you 60 cycle hum. Is this with or without an external DC power supply? Does it do this when powered by a battery? Does it do this when you short the input signal?

Then, again IIUC, you have used a simple RC filter to eliminate the 60Hz noise. But now, with a charge pump, the circuit squeals.  I don't quite understand the note that the RC filter isn't helping any more. Does this mean the 60Hz hum is back? or that the R/C doesn't help with charge pump noise?

Adding a charge pump is not without its own issues. It's a switching device, and it demands pulses of current from power and ground. To do a charge pump well, you have to keep a "bucket" of charge for it to suck pulses from - a filter cap - as close to the charge pump as you can get it, plus some good high frequency filtering in the form of a ceramic 0.1 or 0.01, again, as close to the input of the charge pump IC as you can get it, and ensure that the CP has its own power and ground wires back to the source of DC power. If it shares a ground wire with the transitor gain circuit, you're going to hear the switching pulses, as the pulses generate noise in the resistance of the wires. This isn't a huge thing - some caps and some thought about wiring.

There is another issue with charge pumps; that is, you want them running at a frequency above audio., If that's true, any noise leakage most likely won't be heard, and any filtering is more effective.

On the other hand, it can be done. I used a MAX1044 pump running in the audio band inside a guitar level volume control in production once. It had no charge pump noise.

Ok, so as for your questions:
the circuit (now still on a breadboard, powered by an unregulated 9vdc adapter) used to be hum-free before i added the charge pump.
it was hum free thanks to a basic R/C filter i added made of a 100ohm resistor, a 47uF electrolytic cap and another 0.1uF polyester cap.
now added a charge pump (Icl7660s CPAZ with caps - 10uF and 47uF) and i started hearing a squeal/hum that resembles the mains hum many of my circuits used to have before i started fitting them all with R/C filters (like the one mentioned above).
now i do not know if that is an actual 60 Cycle hum or the charge pump noise but all filtering cap methods i tried using did not eliminate the noise.
i tried powering the circuit with a different, regulated 9v adapter and the noise was the same.

Now while waiting for answers i kept on trying different solutions until i found a solution that i think is acceptable.
i changed the Icl7660s to a different ic, th LT1054 and the noise was reduced.
now i removed the R/C filter and the weirdest thing happened - the noise disappeared.
there was still some noise only when i turned the guitar volume to 0 but if i kept listening for another 5 second the noise stopped completely.
for me that's a fine solution, and my thoughts are that these noises are radio interference caused by all the jumper wires on the breadboard and the fact that it is wide open and has no grounded metal enclosure.
do you think i got that right?

Thanks again,
Guy.

[Danich_ivanov]

I've noticed a lot of these questions lately regarding power and pnp transistors. I have quite a few circuits in the works that mix pnp and npn transistors, it's all daizy chainable, no voltage inversion needed, all i do is use ground to power the transistor's collector and "+" to power it's emitter, and bias accordingly. Not sure if all the classic circuits will work this way, but if making a circuit from the ground up, works totally fine.

[italianguy63]

Guurf..  Do you have pin1 and pin8 of the 7660S connected together?  That enables the "frequency boost".

MC

[Guurf2]

Guurf..  Do you have pin1 and pin8 of the 7660S connected together?  That enables the "frequency boost".

MC

Yes sir they are connected

[BluffChill]

People with a little bit of theoretical education in electronics will sooner or later fall back on the idea that they can simply interchange which side of the power supply they call "ground" and eliminate the problem. This is what the Rangelord designer did. Reversing which power supply lead is called "ground" lets you always connect the negative side to signal ground, and now it's daisy-chainable.

Don't you then run into problems if your supply ground and signal ground (ie your I/O jacks) are all connected together?

[R.G.]

No, but I can see where what I wrote could be interpreted that way.

The theoretical idea is that a power supply should be zero impedance for both AC and DC. No real world power supply is, of course, but we can make that mostly true by using capacitors to make both sides of the power supply be tied together for AC signals. With enough capacitance, both positive and negative sides of a power supply can be at the same AC voltage, merely separated by a DC offset. When that's true, you can use either side as "ground". You keep the DC connections for the circuit right, use BIG decoupling caps, and tie either side, positive or negative to the AC signal ground at the input and output jacks, and it should work.

At least that's the simple theory. It works, mostly. In fact is works so often that many people think it always works. But there are some asterisks, footnotes, and special cases.

I wrote and erased several pages of explanation that I won't drag you through. Suffice it to say that using the negative side of a power supply for signal ground is more natural for NPNs, and using the positive side as ground is more natural for PNPs, at least in simple circuits. There are ways around this, but this is the simplest case.