Micropower charge pump for 3-6v @ 20uA

This solution has an admirable minimalism and might be a winner too. OTOH it might suffer from occassionally killing rather than curing. Worth playing around with though and smaller than the more easily explained dpdt switch and series/parallel capacitor.

OTOH at 10-20uA off 3v it is a load equivalent to 200k so with 1uF gives a tau ~200mS. I'd expect a PIC to be fully operational driving outputs well before then - even on slow minimal energy 32kHz clock.

Regards, Martin Brown

What's wrong with the MAX1724? It is not cheap, but has all other attributes you listed and starts from as low as 0.91V

There is an MOQ on the MAX1724, like one reel (5000), but the Iq of 1.5uA is unbeatable.

I used the MAX1675 instead. Read the data sheet carefully. The Iq for the MAX1675 is ~20uA, but is referenced to the output voltage (the chip is bootstrapped), and at low input voltages much more input current is needed to supply that. It surprised me until I read the doc closely. At 1.4Vin, the efficiency drops to ~70% (dep on the inductor) which draws

Clifford Heath.

Unlike for some other chips from Maxim, there is no MOQ at Digikey:

For 20uA applications using a 3V lithium will be cheaper, simplier, more reliable and more compact, I think you are creating a problem which doesn't exist. Charge pumps are more useful for 1.5v batteries (AA,C,D size) when you need more current that a small lithium can't supply, but you don't have that problem. 3V lithiums are only $1 in small quantities (we pay less then 10 cents for them in large quantities)

Yeah, it might be, I do not know PICs well. There are some uC type chips (such as PSoC) that seem to have to rattle off lines of embedded code for their POR/BOR function and draw considerably more than 20uA for a while. IOW they have a firmware-implemented POR/BOR.

Also, you'd be relying on the integrity of the switch, that it comes on in one swift click. If grampa turns it on with trembling hands that might not be the case. Same if the battery rattled after which the unit may not start up again unless you power cycle (could be ok though). It's an interesting method but I wouldn't do it.

For the life of me I would not spec in a single-sourced part from that source. Or as Clint Eastwood would put it: Only over my dead body.

I was surprised how cheap the 74AUP1G14 has become but if $0.10 is still too much you can try with a JFET. However, then you'll need a tapped inductor or some cheap transformer from which to wire up one. Problem is, JFETs are rather pricey as well these days. Well, maybe not for you guys in the UK right now ...

If you don't require this to work on a totally worn cell you might even get away with a BJT plus tapped inductor.

Check out

I've been keeping an eye on this company for a few months - they are beginning to progress beyond me-too products to pushing the state of the art.

As a hobbyist, I hope you're OK with soldering surface mount IC's.