Geofex voltage double with transistor switching... how does it work?

[Sage]

The instructions for the Geofex voltage doubler contain this note at the bottom:

Better, wire in the transistor switching circuit shown at left. The stereo jack switches the base drive of a transistor, and that transistor switches the battery into and out of the circuit. The high-current pulses can now go on wires that do NOT share the signal ground, and the whining is avoided.

I don't understand how this works.  Where do the high-current pulses go?  To the tip of the stereo jack?  I want to adapt this to a dedicated power supply; I don't need the battery switching because there is no battery, but I'm interested in diverting the noise from the oscillator.

[Keppy]

First, note that the tip and ring connections appear to be reversed as shown. Tip carries signal, ring connects to transistor base through 6.8k, and sleeve is ground. I believe the idea then is to run the charge pump ground (including decoupling cap C2) to a separate wire that connects to the jack sleeve. Power and signal ground are thus joined at the jack sleeve, but the charge pump switching current is not on the ground wire throughout the rest of the circuit.

[highwater]

The normal "stereo jack trick" uses the sleeve of the plug to connect the battery's ground to actual ground - the ring terminal and the sleeve terminal both touch the same part of a mono plug, and the plug connects the battery to ground. With no plug inserted, the battery ground isn't connected to anything at all. Note that if someone uses a stereo cable, the circuit also won't get power.

In this circuit, the power to the MAX1044 is switched instead. When a mono plug is inserted, the transistor's base is grounded through the 6k8 resistor and the sleeve (with biasing applied through the 1k). This causes/allows the transistor to conduct from emitter to collector, providing power to the MAX1044.

The high-current pulses still go to ground, just as they would with the normal version. The difference is that they don't need to go through the input's ground connection to do so - the input is often the most critical part of the circuit, noise-wise.

In any case, it sounds like you aren't interested in switching the power at all. Don't bother with the transistor (or the stereo jack), just use the first scheme, but with the power jack where the battery is shown - the transistor is to solve problems that only battery-users have. You do still have to keep your grounds clean, but that's a matter of layout, not schematic.

[PRR]

As highwater says, there is a rude trick to avoid a battery switch, using the extra finger in a stereo jack. (And as Keppy says, R.G.'s draftsman drew it a bit wrong.)

For a "simple" audio amplifier (or distorter), this usually works OK.

When you have a chomp-chomp-chomp power cruncher in the box, routing the glitchy power line through the same jack as your precious signal just begs for trouble. See below.

[italianguy63]

Paul.  A chomp-chomp-chomp?  Really?   



MC

[Sage]

As highwater says, there is a rude trick to avoid a battery switch, using the extra finger in a stereo jack. (And as Keppy says, R.G.'s draftsman drew it a bit wrong.)

For a "simple" audio amplifier (or distorter), this usually works OK.

When you have a chomp-chomp-chomp power cruncher in the box, routing the glitchy power line through the same jack as your precious signal just begs for trouble. See below.

Best schematic ever.  Thanks, Paul. 

[midwayfair]

This is wonderful. 

[Sage]

Forgive the threadcromancy, but as a followup to this:

If I use this charge pump just to double or triple my voltage:


...and use that to power this opamp dual-buffer circuit:



How do I keep the high-current pulses out of my signal path?  Is it even possible?  Do I need separate grounds for the charge pump and the opamp?

[armdnrdy]

Just put a 15pf cap across pins 2 and 7 of your LT1054 to raise the frequency above (normal) human hearing range. It's in the data sheet.

[Sage]

Just put a 15pf cap across pins 2 and 7 of your LT1054 to raise the frequency above (normal) human hearing range. It's in the data sheet.

Yeah, I saw that, but I've got a dog and I don't want to bother him.

[armdnrdy]

Well then stop wagging him. 

[MrStab]

i always make sure the power supply filtering + charge pump ground go down their own wire, meeting the signal ground at a single point. it's in keeping with the reasoning behind the transistor switch.

the frequency boost method that Larry mentioned probably makes this pointless in most situations, but every so often you might come across a "compatibility issue" with pedals which have an LFO in them. "heterodyning". i don't know the ins & outs of all that myself, but i do know there are numerous reports of charge pump pedals (with frequencies above the audio range, according to the schematics) interacting with delays etc. to introduce unwanted noises. i can't tell if this ground isolation would prevent that, but it couldn't hurt.

be paranoid. it gives you an excuse to make more stuff. lol

[Sage]

i always make sure the power supply filtering + charge pump ground go down their own wire, meeting the signal ground at a single point. it's in keeping with the reasoning behind the transistor switch.

the frequency boost method that Larry mentioned probably makes this pointless in most situations, but every so often you might come across a "compatibility issue" with pedals which have an LFO in them. "heterodyning". i don't know the ins & outs of all that myself, but i do know there are numerous reports of charge pump pedals (with frequencies above the audio range, according to the schematics) interacting with delays etc. to introduce unwanted noises. i can't tell if this ground isolation would prevent that, but it couldn't hurt.

be paranoid. it gives you an excuse to make more stuff. lol

I'm laying all this out on a PCB.  If I do this, but both ground wires are connected at even a single point, does it really make a difference?  If so, why?

[Keppy]

the frequency boost method that Larry mentioned probably makes this pointless in most situations, but every so often you might come across a "compatibility issue" with pedals which have an LFO in them. "heterodyning". i don't know the ins & outs of all that myself, but i do know there are numerous reports of charge pump pedals (with frequencies above the audio range, according to the schematics) interacting with delays etc. to introduce unwanted noises. i can't tell if this ground isolation would prevent that, but it couldn't hurt.

Heterodyning is when two signals interact in such a way that you hear the sum and difference frequencies between them. For example, say you have a 25k charge pump oscillator and a 26k delay clock. You could end up with some 51k (inaudible) and some 1k (BAD!). Since heterodyning depends not on the frequency of an individual signal but on the interaction between two signals, raising the frequency of the charge pump could make the problem better or worse or neither, depending on the specific frequencies involved.

Keeping the oscillator/clock signals far away from audio signals AND from each other is the answer. Also, low pass filtering the signal to remove supersonic frequencies might help in some designs.

This is not an issue with LFOs, since they're below audio frequencies. It's only an issue at audio frequencies and up, so designs using charge pumps, delay clocks, and any digital circuitry are where you tend to encounter it.

[Keppy]

I'm laying all this out on a PCB.  If I do this, but both ground wires are connected at even a single point, does it really make a difference?  If so, why?

Wires and PCB traces can be thought of as extremely low value resistors. Since they're resistors, any current flowing through them HAS to produce a voltage, according to Ohm's law. That means that different points on your continuous ground actually have different voltage potential, so you want to avoid having your signal ground carry power supply current, especially from a noisy charge pump. R.G. has said all of this much more eloquently than me, many times. Merlin's Valve Wizard site also has lots of great information on grounding, presented very well.

[Sage]

So what you're telling me is that instead of doing this:


I should instead be doing this:


Is that the gist of it?

[Keppy]

I'm not sure what the 100R/47uF power filter is for, but otherwise yes. The second diagram is better practice than the first. Running a separate ground wire for the power section is slightly better still. If you don't have an oscillator or clock in the circuit you're using, the danger of charge pump whine is low to begin with, so I would personally use the scheme in your second diagram.

What's up with that filter? It seems unnecessary to me, but if it's there for an extra stage of power filtering I'd expect the opamp to be powered from that junction rather than the 4007/100uF junction.

[Sage]

What's up with that filter? It seems unnecessary to me, but if it's there for an extra stage of power filtering I'd expect the opamp to be powered from that junction rather than the 4007/100uF junction.

It may well be unnecessary, I don't know.  The LT1054 shown above is slightly simplified; I'm actually using twice the diodes and caps in that power stage in order to multiply a 12V input up to roughly 36V (34V when you account for voltage loss).  Then I take that 4007/100uF junction into an LM317L and use a resistive divider to output 30V.

From there, I'm powering this circuit, which I've adapted from Jack Orman's multipurpose dual-opamp PCB:


I'm using this circuit as a buffered effects loop.  My amp didn't come with one, and this is a mod.  The dual-ganged R3 potentiometer allows me to attenuate the input to the buffer while simultaneously boosting the return signal back to unity gain.  Otherwise, it's a straightforward pair of buffers.  One goes from Input to Send, and the other goes from Return to Output.

The power filtering in this part of the circuit is, as I said, copied from another source, and it's a bit of a black box to me, as I'm an amateur at this.  If you have any suggestions as to what I should eliminate, I'd love to hear it.

[tubegeek]

FTFY, Tex.

[PRR]

> not sure what the 100R/47uF power filter is for

It is a standard detail. Usually some parts of the amp are more sensitive than others to power trash. We give them extra filtering. If the load is small, R-C filtering is ample.

Here someone assumed the TL072 has very good power trash rejection, but trash at the input Vref bias will of course go right through. True, we already have R11 C7 filtering the Vref. But another R and C is cheap insurance. Since the total load is 20K, maybe 100r could be larger for more filtering, but what the heck.

However the TL072 PSRR is good at 50Hz-120Hz, but declines, and may be stunningly poor at 20KHz. And the proposed 100r at R10 could carry the total load of the TL072 power with about 0.5V drop. So I'd be tempted to move TL072 + power pin over to + of C8, so this R-C also cleans the TL072 power.