Limits of a 7660 bipolar-from-a-9v

[Mark Hammer]

I wanted to throw the Craig Anderton EPFM II ring modulator into a pedal that I could run off a 9v adapter.  The EPFM book states that current draw is +/-11ma, using a 4739 dual op-amp and LM565.  I have omitted the 4739 and will use a lower-power dual op-amp.  I wanted to use a 7660 (or 7662) to convert +9v into +/-9v.  There are three aspects to this:

1) Whether the 7660/62 itself can supply enough current to run the circuit.

2) Whether the 7660/62 can supply enough current using a single 9v battery.

3) Whether the use of a lower-power op-amp, or some other circuit change, will permit the circuit to obey the constraints imposed by the 7660.

I forget where, but I was under the impression that one should nto plan on using a 7660 bipolar adapter circuit if the current draw will be greater than 10ma.  Since the circuit if perfed, I have the flexibility to go with a pair of batteries OR a 7660-based voltage invertor.  I just wanted a little more convenience without inviting risk of unstable performance.

One related question:  Is there a difference between brands with repect to current capability?  I ask because I have some JRC7660s and some Intersil chips too.

Thanks in advance.  And don't be too shy to send me to a document at Geofex that says everything I need to know and should have noticed the first time around!

[frequencycentral]

This one? http://www.geofex.com/circuits/+9_to_-9.htm

Though it doesn't mention ma. 1044/7660 are 20ma out IIR. I'm powering tube plates from them at ~70 volts. My uneducated guess is that you'll be OK.


EDIT: The LT1054 puts out more ma though! http://oh3tr.ele.tut.fi/~ftp/pa/cly10amp1/lt1054.pdf

[R.G.]

Freq's advice is correct, but as usual, I'll amplify on it.

1) Whether the 7660/62 itself can supply enough current to run the circuit.

Probably. It's hard to say a specific current number about the charge pump converters. They're best though of as a voltage source in series with a resistor. The stock Max1044 which is all I can remember off the top of my head looks like an 80 ohm resistor in series with the putative voltage. Pull 10ma, it sags 0.8V. Pull 100ma and it sags 8V. The theory is discussed in switched capacitor filter sendups, where a cap to ground is switched first to an input voltage, then flipped to an output voltage by an SPDT switch. The cap then looks like a resistor, as it lets only 1/frequency "buckets" of charge move from input to output.

However, even if there is a problem with a single charge pump converter, there is an out. You can simply parallel charge pump converters, all pumping into an output cap. This has the effect of lowering the impedance by the factor of the number of converters, or increasing the current by the same amount.

Lastly, the 7660 is going to give you pernicious whining. It runs at audio frequencies. You will need to either force it faster by putting an ultrasonic clock on the oscillator input or super-filter the output. Two or more charge pumps in parallel really should be run synchronously or antisynchronously to help eliminate beat frequencies from their clocks from putting audio range junk on the power supply and ground.

Finally, you simply must keep the high current pulses going to and from the charge pump off the wires leading to your circuit. Take a separate wire from your battery + and - to the charge pump and only connect it to circuit ground at the output. Otherwise you're likely to hear the I*R drop in the wires as whine.

2) Whether the 7660/62 can supply enough current using a single 9v battery.

Yes. It's about 80%+ efficient, done properly. So for an 11ma load on +9 and on -9, the + side uses 11ma and the - side uses 11ma/0.80= 13.75ma. You're paying 2.75ma for the privilege of getting 11ma inverted to a negative voltage. Total battery current for this simplistic example is then about 24.75ma, and a 160ma-hr 9V battery should give about 6.5Hr of service. Many 9Vs are able to produce more than 160ma-hr, so they'll last proportionately longer.

3) Whether the use of a lower-power op-amp, or some other circuit change, will permit the circuit to obey the constraints imposed by the 7660.

Battery life will get better by the factor of lower current used. Whipping in a 20ma LED or two will hurt things, obviously.

I forget where, but I was under the impression that one should nto plan on using a 7660 bipolar adapter circuit if the current draw will be greater than 10ma.  Since the circuit if perfed, I have the flexibility to go with a pair of batteries OR a 7660-based voltage invertor.  I just wanted a little more convenience without inviting risk of unstable performance.

It's more like if the current is under 10ma it's OK to proceed. Otherwise, you need to calculate.

One related question:  Is there a difference between brands with repect to current capability?  I ask because I have some JRC7660s and some Intersil chips too.

I doubt there'll be much difference.

Best advice? Leave space for a second one and pre-figure-out a way to drive them with an ultrasonic signal on the clocks, perhaps a CMOS 555 or a simple CMOS oscillator if you have other CMOS on the board. That way you're sure you can add more current and keep the frequency above audio.

Thanks in advance.  And don't be too shy to send me to a document at Geofex that says everything I need to know and should have noticed the first time around!

Would I do that?

[Mark Hammer]

Thanks, gents.  Much obliged.  Given that the ring modulator already has an issue with minimizing audible whine from the on-board VCO, I think that seriously filtering what I get from the 7660 is imperative.  On the other hand, I recently purchased some momentary X-wing DPDTs that I thought were latching, and was pondering making this pedal a momentary "punch-in" effect where you pressed for as long as you wanted the effect, then lifted your foot when done.  As such, concerns about ongoing whining and other audible artifacts is reduced.

I take it that one of the reasons why folks using a 7660 to power up their Mu-trons/Neutrons haven't voiced complaints about whining is because it's a filter?  Or is the potential whining something that shows up on the output, quite apart from the filter action?