Univibe power supply using charge pump & regulator - best practice

[aion]

I'm working on building a Univibe that runs on 9V, with a charge pump that boosts it to 18V regulated down to 15V, and had some questions about theoretical best practices.

There are a few Univibe projects out there that use the charge pump setup, as opposed to requiring 18VDC input or using rectified AC. Some of these (e.g. PedalPCB and Madbean) run the audio off of the 18V rail and the LFO off the 15V. However, the current draw of the LFO/lamp varies signficantly throughout the sweep, so the unregulated 18V rail fluctuates in voltage.

I'm wondering whether it's better in theory to run the audio portion at 18V and LFO at 15V, or to run them both at 15V. Both methods have been done successfully, so I'm thinking less about viability and more just about optimizing.

In the 18V/15V corner, the advantage is that the audio runs at higher headroom and has better isolation from the LFO. However, I'm not sure 3V is enough of a difference for headroom to be a factor, and the original unit shared the same 15-16V rail across the whole circuit (though the source supply itself was significantly different in design).

In the both-15V corner, the advantage is that the power supply is regulated across the whole circuit and the voltage shouldn't vary appreciably across the LFO sweep. However, I'm not sure if the static voltage supplied by the ancient LM7815 is a qualitative improvement over a clean(er) but fluctuating voltage.

Maybe there's a third option that's even better - e.g. a more modern regulator with lower noise, or using small resistors off the +15V supply to create two sub-supplies with better isolation. But right now I'm at the stage where I've got time and opportunity to wonder whether it could be improved from what's been done in the past, so I thought I'd fire up a discussion.

[antonis]

I'm not sure if the static voltage supplied by the ancient LM7815 is a qualitative improvement over a clean(er) but fluctuating voltage.

A capacitance multiplier should "clean" enough 7815's voltage..
(in case of really needed..)

[PRR]

"improved".... is there an actual problem?

18V supply: this is guitar-level work. 3V may be ample. The 12V-15V-18V level may be more about the incandescent lamp than audio.

The designers of 1979 were not fools. The LM7815 may be older than some people here, but does not suck.

The plans I am seeing run the Lamp and LFO on the raw DC, and the audio through R-C-R-C filtering.

Opamp audio would ignore most any crap on the supply. The oldest plans show discrete transistor buffer, gain, and splitter. IMHO it would be easier to transpose to chips than to figure the PSRR of these classic but unfashionable circuits.

[POTL]

"improved".... is there an actual problem?

18V supply: this is guitar-level work. 3V may be ample. The 12V-15V-18V level may be more about the incandescent lamp than audio.

The designers of 1979 were not fools. The LM7815 may be older than some people here, but does not suck.

The plans I am seeing run the Lamp and LFO on the raw DC, and the audio through R-C-R-C filtering.

Opamp audio would ignore most any crap on the supply. The oldest plans show discrete transistor buffer, gain, and splitter. IMHO it would be easier to transpose to chips than to figure the PSRR of these classic but unfashionable circuits.

can op amps replace the BJT univibe discrete circuitry. I remember simulating the depths/easyvibe style schema and the notch filters didn't match the univibe schema.

[Rob Strand]

I'd prefer not to have the audio supply rails fluctuating.

The original LFO had unregulated supplies so to some degree it is known to work.

The Univibe lamp drive is largely (not 100%) current mode so it doesn't care much about the supply voltage.

To me the best set-up alternatives would be:
- run everything off the regulated 15V
- run everything off the regulated 15V except the *lamp*, which you wire back to the 18V in.

[aion]

The designers of 1979 were not fools. The LM7815 may be older than some people here, but does not suck.

It's good to get some pushback on that. I have heard the 78XX series trashed many times in the recent past. One of my weaknesses is that I tend to collect that sort of info in my head without asking what the context was and whether it applies to a given use case.

To me the best set-up alternatives would be:
- run everything off the regulated 15V
- run everything off the regulated 15V except the *lamp*, which you wire back to the 18V in.

Would it be any improvement to add a second regulator for the audio supply so that both are at a steady 15V but isolated from each other? 78L15 (TO-92) for audio, and 7815 (TO-220) for the LFO to handle the much larger current requirement?

[antonis]

IMHO, only in case of 7815 (TO-220) heating effects dominate its load regulation..
(shouldn't happen here, but..)

[ElectricDruid]

The designers of 1979 were not fools. The LM7815 may be older than some people here, but does not suck.

It's good to get some pushback on that. I have heard the 78XX series trashed many times in the recent past. One of my weaknesses is that I tend to collect that sort of info in my head without asking what the context was and whether it applies to a given use case.

I've run DACs from supplies regulated by 78xx parts and that's probably not to be recommended. There are modern alternatives that are less noisy for sensitive applications like that. For day-to-day stuff, yeah, fine. I still use them all the time. As ever, it depends on the specifics of the job at hand.

Would it be any improvement to add a second regulator for the audio supply so that both are at a steady 15V but isolated from each other? 78L15 (TO-92) for audio, and 7815 (TO-220) for the LFO to handle the much larger current requirement?

In theory, you'd probably guess that it's better. If it *actually* makes any significant difference is an open question that can only be answered by experiment.

[PRR]

I still don't know what actual problem we have. Hiss? Hum? Thump? IMD?

[aion]

I still don't know what actual problem we have. Hiss? Hum? Thump? IMD?

Right now it's purely a question of theory since I haven't built one yet!

The 9V charge pump method is a relatively new way of powering the Univibe (compared to the original late-60s circuit, I mean) and has less of a history in the DIY scene. There's 20+ years of discussion here to learn from, but most DIY iterations of the Univibe have used an 18VDC adapter, an AC power supply with rectification, or 120V wall power with a transformer as in the original, each of which would avoid the issues I'm wondering about.

The charge pump method has been done successfully a few times, but I'd seen mention of certain side effects (such as the unregulated charge pump output sagging with the LFO) and they were often dismissed as not a big deal. So I wanted to understand whether those really were inconsequential rather than just inaudible, or whether they were worth revisiting and possibly being improved on.

[Elektrojänis]

In the 18V/15V corner, the advantage is that the audio runs at higher headroom and has better isolation from the LFO. However, I'm not sure 3V is enough of a difference for headroom to be a factor, and the original unit shared the same 15-16V rail across the whole circuit (though the source supply itself was significantly different in design).

18V supply would have about 1.6dB more headroom than 15V if the circuit can swing from rail to rail. So in practice the difference will be a bit more. I would worry a lot more about the LFO contaminating the audio path than for that little headroom improvement. And if you need that maybe about 2 dB more headroom, you could just attenuate the input.

[Rob Strand]

Good layout can help, like separating the ground and power for the audio and LFO/lamp.   You could also use an RC filter on the LFO, perhaps not on the lamp.   Large-ish cap on the regulator output.

The idea of powering the lamp from the unregulated supply is a way to off-load power dissipation from the 15V regulator.  It also removes one source of thump.   It depends if you want to use TO-220 or TO-92 package regulators.   In general the TO-220 regulators are superior.