How do I create a split rail?

[PRR]

> your schematic actually attenuates by a factor of 5 before boosting by a factor of 6.

Note that R9 1Meg is in shunt with R0 20K.

My beads say the Rx R0 R9 loss is 0.16387 or 1/6.1.

The gain amp is 120K/20K or 6.0

The overall gain is 6.0/6.1 or 0.983 (assuming perfect parts). This is -0.147dB, which is really teeny on the ear.

There is an additional loss (in direct-coupled) from R13 100r into R4 1Meg, which is 0.99990(!) or -0.00087dB. There may be this much again in jack, contact, and cord loss.

I say it is ultra close to unity-gain. (But my beads may be wrong.)

I also say that musicians with direct A/B comparison path may be able to spot sub-dB differences without being able to say what the difference is. And usually the louder signal is "better". And that if I was designing an add-on, I would set the gain a hair high of unity just to be sure my-side sounded "better". Changing R3 100K to 105K (100K+4.7K) might do it, or might even be a bit too "obvious". Shunting R1 20K with a spare 1Meg might be a way to go.

But don't go crazy chasing sub-dB differences. You won't entertain the audience with a series of A/B switches. The stuff you put in the loop will have dB errors, or may be over-unity gain to "sound better". Musicians can put an extra dB in the sound instantly with a bit more finger-effort.

[PRR]

You do not need/want "R10" diode. That is for changing batteries in the dark while drunk every night. Or if you have various polarity wall-warts. You "will" make contact backwards, risk the chip, the diode prevents disaster. If you only connect-up once, sober, with clear notion of polarity, just do it.

However "R10" is the place to put a hundreds Ohms resistor, which with C8 (perhaps upped 10X) would be your ripple filter.

[Keppy]

My beads say the Rx R0 R9 loss is 0.16387 or 1/6.1.
(But my beads may be wrong.)

No, you're right. I did the math wrong for the attenuator (20k/100k instead of 20k/120k). Apologies, Sage, it is 6/1 in both stages.

[Sage]

You do not need/want "R10" diode. That is for changing batteries in the dark while drunk every night. Or if you have various polarity wall-warts. You "will" make contact backwards, risk the chip, the diode prevents disaster. If you only connect-up once, sober, with clear notion of polarity, just do it.

However "R10" is the place to put a hundreds Ohms resistor, which with C8 (perhaps upped 10X) would be your ripple filter.

Funny story -- originally R10 was in fact a resistor.  However, I noticed that when the power to the amp was switched off, the amp would emit a high-pitched squeak when the effects loop was on (and the squeak would repeat through the delay pedal).  My guess is that the opamp was oscillating or something as the power supply dropped out.  Putting a diode at R10 was an attempt to prevent that, suggested by someone on these forums, I forget who...

The reason it's still called R10 is because I built the circuit using the multipurpose opamp PCB from Jack Orman and I just labeled everything the same to make it easy to assemble.

New idea: ST Micro has an example circuit in the data sheet for their MC34063 boost converter that takes a 12V DC input and outputs 28V, which seems like it would be perfect for my needs here; I could get 12V from the top of C38 and turn it into 28V or even 30V to power the TL072.  As an added bonus, the MC34063 would act as a regulator, as there is none on the existing 12V line.  Can you think of any reason this wouldn't work?

No, you're right. I did the math wrong for the attenuator (20k/100k instead of 20k/120k). Apologies, Sage, it is 6/1 in both stages.

No sweat, man, I just really appreciate you taking an interest in this.

[slacker]

Yes the boost convertor will do the job, with any sort of convertor like that there is a risk of added noise but with careful filtering they can be a good solution.
A simpler alternative would be the LT1054 http://www.linear.com/product/LT1054 this give you a number of options, it can double and generate negative voltages. It can also be used it the setup shown here http://geofex.com/circuits/+9_to_33.htm which would give you 30ish volts from 12.

[Sage]

I'll try one of those solutions. is the LT1054 interchangeable with the MAX1044? Which of these, including the MC34036, is likely to introduce the least noise into the audio signal?

[slacker]

The LT1054 is arguably better than the MA1044, because you can run the oscillator a higher frequencies keeping it further away from audio, I don't know about the MC34036 sorry. The LT1054 isn't completely interchangeable with the Max1044, pins 1 and 6 have different functions and pin 7 works a bit differently. If the design doesn't use those pins they're interchangeable or with small changes to the circuit either can be used most of the time. The LT1054 can handle inputs up to 15 Volts the Max only goes up to 10.

The LT and the Max are used a lot in stompboxes so help and support round here will be better than for the MC.

[PRR]

> act as a regulator, as there is none on the existing 12V line.  Can you think of any reason this wouldn't work?

I'm not fond of adding oscillators to audio amps; your feelings may differ.

I don't see a need for "regulation". You need "enough" but not "too much". Granted that 12V may be marginal, but 18V should be ample, and 36V could blow the chip. 18V to 35V should be all the same to the circuit. (Check cap ratings.) Any wall-outlet steady enough for the tubes to be happy can be steady enough for the opamp. "Regulators" also cut ripple, but at this low current an R-C network is simpler.

[Sage]

I'm aiming for somewhere between 24V and 30V.  The TL072 might be able to withstand 36V but 30V is the highest nominal.

My issue with taking the 24V that I'm reading off the input to the 18V regulator is that although I'm reading 12.3V on the AC line coming from the transformer, the rating printed on the transformer housing for that particular line is 10V.  I'm worried that under different conditions, that input to the voltage regular could be as low as 20V, in which case I'm not getting much more than I was already getting from the other side of that regulator.

If there's a more reliable way to get to +24V or +30V I'd love the suggestion.

There's another 6.7V line coming out of the transformer and going directly to CN6 and CN7 ... I'm not sure I want to touch that, though, since it's clearly tied into the power supply to the tubes.

Is there another way to get 24V out of a 12V or 18V line?

[Sage]

Is there any info out there on how to get a specific voltage out of the LT1054?  Schematics for using it as a voltage doubler are readily available, but I don't see anything about converting +12V to +30V...

[slacker]

You can't really get specific voltages only roughly multiples of the the input voltage. If you use the circuit I mentioned earlier http://geofex.com/circuits/+9_to_33.htm but with an LT1054 then with 12 Volts the  "+25" stage will give you about 32 Volts. You could stick a couple more diodes in series with the output to drop it to closer to 30 if you wanted.

[Sage]

Cool, thanks.  32V isn't a problem, in fact that probably would be ideal -- I could just stick an lm317 on it and regulate 30V out from there.