UVICS 3 power supply/charge pump questions

[Morocotopo]

This topic:

http://www.diystompboxes.com/smfforum/index.php?topic=97217.msg849204#msg849204

got me thinking about powering my Neovibe with an LT1054 charge pump and, looking around Geofex, I saw that R.G. incorporated the LT1054 charge pump into the UVICS 3 project. I happened to have a 1054 charge pump in my PCB cemetery/drawer. Pulled it out, but the voltage sags to  around 12V when loaded with the Neovibe. Mine has the in jack ring/transistor trick to avoid whining getting into the audio, a 5p6 cap from CAP+ to OSC pins to raise the switching freq, and uses 1N4148 diodes.

Questions:
- I see that R.G. uses 1N5817 diodes. What´s the advantage?
- In the transistor, there´s a 1K R from base to emitter, as per R.G.´s original article. But on one of my projects (a voltage quadrupler to get 30-36V with a 7660) I had to remove it to get it to deliver the target voltage. The idea to remove it came from some page I can´t remember now. Should I do the same here? Or is the transistor trick not doable here (more current involved)?

Well, those are my questions. I´ll do some breadboard testing myself.

[R.G.]

- I see that R.G. uses 1N5817 diodes. What´s the advantage?

1N5817 is a Schottky diode, and has a lower forward voltage drop, about 0.4V instead of 0.7. Not a big deal, but why give away the voltage you're trying to get?

- In the transistor, there´s a 1K R from base to emitter, as per R.G.´s original article. But on one of my projects (a voltage quadrupler to get 30-36V with a 7660) I had to remove it to get it to deliver the target voltage. The idea to remove it came from some page I can´t remember now. Should I do the same here? Or is the transistor trick not doable here (more current involved)?

You can change the resistor from base to emitter to almost any value from 1K up. That resistor is there just to make really sure the transistor goes off if there is no drive on its base. If the transistor leaks a little,  you could get a partially-on situation. The smaller the resistor, the more leakage it immunizes the transistor against. I learned transistors at a time when one had to worry about that, and it held on. Modern transistors are so good that you can make do with 10K, 100K, 1M, even an open circuit in most cases.

If 1K doesn't work for you, increase it to a bigger value, or use an open circuit and see if your transistor is non-leaky enough to work that way. It almost certainly is. I'm just a transistor Luddite.

[Morocotopo]

OK, tried it with the 1K R removed but get around 14,5 - 15V. Not enough (current?) to make the LFO oscillate. Or the lamp to light. More experiments to do.

[R.G.]

So, I managed to confuse myself. Exactly what transistor are we talking about?

[Morocotopo]

What model/type of transistor? A 2N3906.

EDIT: clipped the 5p6 cap, to see if changing the switching frecuency had some effect. No dice. I´ll have to breadboard this to experiment more. Luckily the 1054 is socketed, so I can remove it and move it to the breadboard.
More details: The out cap is a 220 uF, the in one 2u2.

The datasheet for the 1054 states a switching frequency of 25 KHz. Maybe the transistor trick isn´t necessary here, making the whine too high to be audible? We´ll see. I´m off to dinner.