what's the reason using charge pumps as doublers instead of making +/-?

[darron]

I've been playing with some TC7660S charge pumps recently to help make any tricky voltages that ICs might need. They look like a slightly nicer spec'd version of the max1044.



I'm wondering, why do people prefer to make voltage doublers to get +17/18V when it seems better to make a split supply +/- 9v?

I think that this would be better for a few of reasons:



- Less components required for a better biasing scheme.
- Less components required to run charge pump. No diodes. Also, more headroom as no diode dropout.
- Higher current possible. Since the charge pump is only doing the negative supply you are pulling half through it. - This one was the big one for me on my last project as running 18V instead of +/-9V was just outside of the IC's range.
- And since lower current is being used, capacitance sizes can be roughly halved for the same amount of ripple.




Are there any good points for/against each side?

Thanks people.

[R.G.]

Sadly, it's probably because they saw a schematic somewhere on the internet (which of course makes it One True Way) and are copying that.

There are other reasons to do this, but they're subtler than most "designers" do.

[amptramp]

One other reason to avoid charge pumps - they pump noise into the circuitry because they charge capacitors with large currents then dump the voltage into another cap.  There are charge pumps that operate above audio frequencies, but you could still run into beat frequency heterodyning with delay elements which have internal oscillators.  There is no current limiting in a charge pump except for the semiconductor switch and capacitor resistance limitations.  I have a ±12 volt supply that I run things from and I find that some circuit problems go away if you have a ground and both positive and negative supplies.

If you have a simple flyback converter set up as a voltage inverter, you get better control and less noise reflected back onto the power lines because current doesn't change all that quickly in an inductor.  But inductors have to be specified for the frequency and current and the circuit design is more complicated due to the need for feedback control.  A good tutorial on how to design a basic voltage inverting supply might go a long way towards improving noise performance and battery life.  Also, if this kind of article is to be effective, sample designs and part numbers including order numbers for the more popular suppliers just might make a few people adopt a better design. 

[darron]

charge pumps definitely used to scare the crap out of me. it used to feel like adding potential problems:

- some of them are only rated right to the voltages we use, 9v, 10v, - so these would need protection. even from a fresh battery. this TC7660S i've been playing with is rated to 12 from memory but would still need protection
- they are NOT forgiving. usually the first thing in a circuit to go. the slightest bit over the tiny current these ones allow will pop them
- clock noise, as said, make sure you get one that will have the frequency boost. there are common pin-for-pin drop-ins that don't. i think the other 7660 series might not? at higher frequency and low ESR caps i've had little problems. a small value series resistor if you're using a common power supply might make me feel a bit more comfortable. i think it's also less problematic if you have a constant current draw that doesn't fluctuate which should be what our little audio circuits mostly are.

[midwayfair]

You save components in the charge pump circuit, but you might pay for it elsewhere. Well, in discrete designs anyway: Your emitters now need a resistor to -9V, and FETs also get a bit more complicated (Gus did a split rail JFET design). I agree though, in op amp designs, the split rail seems to be almost always superior and I'm not smart enough to say when it couldn't be used. You don't lose as much voltage for one thing.

One thing I've always wondered is why BBD designs incorporating charge pumps don't use the negative voltage inverter to create -15V and run the chips on the voltage they were meant to run on. I don't know if it makes a difference to be honest, but I do wonder if it might help with noise.

[darron]

Cool. Didn't see Gug's design. Would like to see a mu-amp/microamp arrangement!

[slacker]

It's probably because most of the time people are starting from an existing circuit and converting it to 18 Volts so it's easier to go the Voltage doubler route than convert the circuit to +-9 Volts.

[StephenGiles]

And of course, dare I say it - the desire to house effects in tiny boxes!!

[Mark Hammer]

There are two conversations going on here.  One is the decision to use charge pumps instead of a bipolar supply, period.  When Craig Anderton was king, and EPFM the working Bible, it was assumed that effects would use a +/-9v bipolar supply.  That was perfectly acceptable when FX were built into some custom enclosure or rack, or people were willing to use toggles to turn on two batteries at once.

As standards started to turn towards more populated pedalboards, powered by supplies that had single-ended 9V outputs, and used single-ended 2.1mm jacks, the move was on to convert 9v and maintain the convenience of the existing supplies and jacks.  If there were "stereo" barrel connectors that could disable two batteries, and all pedals came with little unobtrusive slide-switches to turn a pair of batteries off, then we would see more designs adopting bipolar supplies.  Instead, what we have seen is the co-opting of 2.1mm jacks to higher single-ended input voltages, and the use of internal charge pumps to make more fruitful use of 9v supplies, when feasible.

The other conversation going on is whether charge pump chips are most appropriately used as voltage doublers (or triplers, in some cases), as opposed to voltage converters/inverters, that yield a bipolar supply from a single-ended one.

A curiosity-based question, though, does the extra capacitance involved in a multiplied voltage from a charge pump provide any additional smoothing of the clock-related noise it generates?

[amptramp]

There is an app note floating around somewhere about using an LM386 as a voltage splitter that will give you half the input voltage at the output, which can carry a lot more current than a charge pump.  Use an 18 VDC wall wart and one of these to establish ground as its output and you have ±9 VDC which gives you enough headroom for almost anything.  The design is shown here:

http://www.edn.com/design/consumer/4429444/Simple-PS-voltage-splitters-based-on-audio-amplifiers--Part-3-

This voltage splitter works too for lower currents:

http://www.instructables.com/id/Split-rail-power-supply-from-a-single-rail-supply/

[PRR]

> better to make a split supply +/- 9v?

This often means messing with Wall Outlet power.

Considering the number of shorted resistors and caps we see here daily, I'm not sure that wall-power is good to play with. Smoking a 1/4W resistor is a lot safer than smoking the wires inside your wall while spraying sparks on your carpet.

Obviously people do manage. But even with decades of practice, recently I got a 24V transformer backward (output tried to be 3,000V, couldn't, and the lights got dim....), and forgot an extra diode ("400V" turned out 800V, on 450V caps).

This was maybe easier back when you could buy 12VAC wall-warts. I built a lot of stuff (correctly) with those things. But all AC-output wall-warts are gone out of style. (A 10W wart may waste 3W of useless heat, all the time, so the energy-nits pick on them.) While there's shiploads of old AC warts at places like Jameco, *today*, if you expect your build to last until 2024 then expendable parts that quit production in 2010 may not be a good foundation.

[darron]

A curiosity-based question, though, does the extra capacitance involved in a multiplied voltage from a charge pump provide any additional smoothing of the clock-related noise it generates?

Tricky. Here's a snippet from the data sheet I was looking at:

If the impedances of C1 and C2 are relatively high at the pump frequency (refer to Figure 2) compared to the value of RL, there will be a substantial difference in voltages V1 and V2. Therefore, it is desirable not only to make C2 as large as possible to eliminate output voltage ripple, but also to employ a correspondingly large value for C1 in order to achieve maximum efficiency of operation.

From this data sheet http://docs-asia.electrocomponents.com/webdocs/11b5/0900766b811b5bb8.pdf on this site http://australia.rs-online.com/web/p/charge-pumps/7747882/?searchTerm=7747882&relevancy-data=636F3D3126696E3D4931384E525353746F636B4E756D6265724D504E266C753D656E266D6D3D6D61746368616C6C26706D3D5E5C647B367D247C5E5C647B377D247C5E5C647B31307D2426706F3D313426736E3D592673743D52535F53544F434B5F4E554D4245522677633D4E4F4E45267573743D3737343738383226






We don't have any problems getting transformer packs here in Australia at all. Yet...

I can see the arguments for switch mode! Most people wouldn't give much thought to how much power a product uses when sitting idle/not in use. Maybe lots of manufacturers wouldn't either? And efficiency when running makes you not feel so bad. They need to consider exceptions for the audio guys though for situations like this and also breaking ground loops. OT, the government was giving out free power saving boards that will detect your TV current draw turning on/off and turn on/off all of your set top boxes etc. They have heft 2400W relays that warm up quite a bit themselves. They need to come in and quickly professionally install them. I told the guy "no thanks, it's all switch mode, I'd rather let it all idle than turn everything back on manually."   ... him: "huh? this saves electricity"




That split rail design looks very simple and neat, but if there was 24v available to play with already it wouldn't be much of a problem, but I'd probably rather something like that than a charge pump/inverter.



The current standard 9V positive-outside plug looks like it's around to stay for a while. For most purposes it's not so bad! Not many distortions etc. sound good running at 18v anyway

12VDC would have given a little more play. 12/24VAC would have been VERY useful. I bet some manufacturer would make tiny 12VAC to 9VDC regulated DIP8 ICs just for pedal guys to save room     Sorry... Rambling. Someone ought to make a split rail project thead... or website...!