Micropower charge pump for 3-6v @ 20uA

[Martin Brown]

I suspect this is a tall order.

But can anyone see a simple way to build a micropower charge pump for
a hobby electronics PIC based device drawing approx 20uA at a voltage
between 2.5 to 6v. The aim is to run it off a single cheap 1.5v
battery rather than a pair of them or an expensive 3v cell. Efficiency
of 20-50% would be OK - 40-100uA drain on the cell. Regulation need
not be good - but its output mustn't go above 6v. Its for a PIC based
sidereal clock for astronomy.

The linear LTC1751 looks like it might come close but the datasheet
says it is only good for 2v and up input voltage. And if the solution
is bigger than the battery then it may as well use a pair of cells.
OTOH It draws so little current.

Thanks for any enlightenment.

Regards,
Martin Brown

 

[Frank Buss]

But can anyone see a simple way to build a micropower charge pump for
a hobby electronics PIC based device drawing approx 20uA at a voltage
between 2.5 to 6v. The aim is to run it off a single cheap 1.5v
battery rather than a pair of them or an expensive 3v cell. Efficiency
of 20-50% would be OK - 40-100uA drain on the cell. Regulation need
not be good - but its output mustn't go above 6v. Its for a PIC based
sidereal clock for astronomy.

You can use the MAX1724 (see e.g. http://www.digikey.com how to buy it).
I've used it for a product design some years ago, which is still produced
and works without problems, so unlike some other Maxim chip looks like this
is one of the more standard products and will be available in future, too.

According to the datasheet, efficiency is about 60% at 1V battery input,
100uA load and 3.3V output. It works down to 0.8V (but needs at least 0.91V
for startup).

 

[John Larkin]

I suspect this is a tall order.

But can anyone see a simple way to build a micropower charge pump for
a hobby electronics PIC based device drawing approx 20uA at a voltage
between 2.5 to 6v. The aim is to run it off a single cheap 1.5v
battery rather than a pair of them or an expensive 3v cell. Efficiency
of 20-50% would be OK - 40-100uA drain on the cell. Regulation need
not be good - but its output mustn't go above 6v. Its for a PIC based
sidereal clock for astronomy.

The linear LTC1751 looks like it might come close but the datasheet
says it is only good for 2v and up input voltage. And if the solution
is bigger than the battery then it may as well use a pair of cells.
OTOH It draws so little current.

Thanks for any enlightenment.

Regards,
Martin Brown

Use a PIC port pin to drive an NPN that in turn drives the inductor of
a boost converter. Let your software pulse it as needed. If you have a
spare adc input, feed back on that, and you're now in the DSP power
supply business!

You could also do a schottky charge pump/C-W multiplier directly off a
port or clock pin, given the efficiency you're willing to accept.
That's not bad... a 4-diode multiplier stack, two sot-23 dual
schottkies and some caps, piggybacked on top the +1.5 rail.

John

 

[Joerg]

I suspect this is a tall order.

But can anyone see a simple way to build a micropower charge pump for
a hobby electronics PIC based device drawing approx 20uA at a voltage
between 2.5 to 6v. The aim is to run it off a single cheap 1.5v
battery rather than a pair of them or an expensive 3v cell. Efficiency
of 20-50% would be OK - 40-100uA drain on the cell. Regulation need
not be good - but its output mustn't go above 6v. Its for a PIC based
sidereal clock for astronomy.

The linear LTC1751 looks like it might come close but the datasheet
says it is only good for 2v and up input voltage. And if the solution
is bigger than the battery then it may as well use a pair of cells.
OTOH It draws so little current.

Thanks for any enlightenment.

Did you check integrated versions such as the TPS61070? Ok, a bit
overkill here and expensive. If it had to be really cheap I'd experiment
with a JFET and tapped inductor (which could be made with a ferrite bead).

Basically you need something that will oscillate well at 1V or less and
doesn't require more than 20-30uA or so of quiescent. No idea if some
tiny Schmitt inverter would do but might be worth a look. 74AUP1G14
maybe. Those are also really cheap. Make an oscillator with it and then
step that up with a ferrite bead transformer, LAN transformer or
whatever you've got flying around, rectify, then regulate if needed.

Of course, if I was in a real rush I'd grab one of the Ge-transistors
I've stashed away but I guess you want something that others can also build.

 

[Joerg]

Use a PIC port pin to drive an NPN that in turn drives the inductor of
a boost converter. Let your software pulse it as needed. If you have a
spare adc input, feed back on that, and you're now in the DSP power
supply business!

You could also do a schottky charge pump/C-W multiplier directly off a
port or clock pin, given the efficiency you're willing to accept.
That's not bad... a 4-diode multiplier stack, two sot-23 dual
schottkies and some caps, piggybacked on top the +1.5 rail.

But, as Jim would put it, how does it start?

 

[Jim Thompson]

But, as Jim would put it, how does it start?

The world runs on smoke and mirrors ;-)

...Jim Thompson

 

[Joerg]

The world runs on smoke and mirrors ;-)

I always thought it was pixie dust.

 

[Jim Thompson]

I always thought it was pixie dust.

Naaah! That variation only applies in Californica and Massatwoshits
;-)

...Jim Thompson

 

[John Larkin]

But, as Jim would put it, how does it start?

Oh. I was assuming the pic ran on 1.5, and the hv was for something
else. My bad.

So, make a blocking-oscillator flyback converter, I guess. Or use a
3-volt lithium.

John

 

[Joerg]

Oh. I was assuming the pic ran on 1.5, and the hv was for something
else. My bad.

So, make a blocking-oscillator flyback converter, I guess. Or use a
3-volt lithium.

Or just do the usual Schmitt inverter thing but instead of your LM5112
use a 71AUP1G14. Haven't tried that one but should work.

 

[John Larkin]

But, as Jim would put it, how does it start?

OK, be pickey.

This version self-starts:

http://img402.imageshack.us/my.php?image=picdcik3.jpg

The boost voltage isn't well defined, so if the pic can't control it
maybe a shunt regulator, possibly a couple of led's, would help.

John

 

[John Larkin]

Did you check integrated versions such as the TPS61070? Ok, a bit
overkill here and expensive. If it had to be really cheap I'd experiment
with a JFET and tapped inductor (which could be made with a ferrite bead).

Basically you need something that will oscillate well at 1V or less and
doesn't require more than 20-30uA or so of quiescent. No idea if some
tiny Schmitt inverter would do but might be worth a look. 74AUP1G14
maybe. Those are also really cheap. Make an oscillator with it and then
step that up with a ferrite bead transformer, LAN transformer or
whatever you've got flying around, rectify, then regulate if needed.

Of course, if I was in a real rush I'd grab one of the Ge-transistors
I've stashed away but I guess you want something that others can also build.

How about this one?

http://img127.imageshack.us/my.php?image=seesawxs8.jpg

It's basically a blocking oscillator. Output is clipped to be 2x the
input voltage. There are lots of small 1:1 inductor/transformer things
in Mouser and Digikey.

John

 

[Spehro Pefhany]

Oh. I was assuming the pic ran on 1.5, and the hv was for something
else. My bad.

1.5V is too low for any PIC I've seen. There are some very low power
micros which have an on-chip bootstrap charge pump that works pretty
much as you describe, but the PIC isn't spec'd for it. It might work
if you used an internal RC oscillator, OSCOUT and the oscillator
actually started, but it wouldn't be guaranteed.

The SN74AUC1G14 is spec'd to operate with an 800mV power supply rail.

So, make a blocking-oscillator flyback converter, I guess. Or use a
3-volt lithium.

John

There are plenty of SOCs that work from single cells (eg. MP3 players)
and IIRC some SMPS chips too.


Best regards,
Spehro Pefhany

 

[Joerg]

OK, be pickey.

This version self-starts:

http://img402.imageshack.us/my.php?image=picdcik3.jpg

The boost voltage isn't well defined, so if the pic can't control it
maybe a shunt regulator, possibly a couple of led's, would help.

I don't know PICs but I would imagine they need something above 2V for a
substantial amount of time (maybe msecs, don't know) to be able to start
anything meaningful in there. Might be a wee problem if the battery is a
well used 1.5V alkaline.

 

[John Larkin]

I don't know PICs but I would imagine they need something above 2V for a
substantial amount of time (maybe msecs, don't know) to be able to start
anything meaningful in there. Might be a wee problem if the battery is a
well used 1.5V alkaline.

This should boost to well over 2 volts at startup. It would peak at 3
volts except for the diode loss.

John

 

[Joerg]

How about this one?

http://img127.imageshack.us/my.php?image=seesawxs8.jpg

It's basically a blocking oscillator. Output is clipped to be 2x the
input voltage. There are lots of small 1:1 inductor/transformer things
in Mouser and Digikey.

Yes, either something like that or a JFET, tapped inductor at the drain,
battery to tap, feedback from the top. As long as it all remains within
Martin's power budget which probably doesn't allow more than 25uA or so
of quiescent. I'd probably start with a 74AUP1G14 because they are
surprisingly cheap.

I found inductors at Digikey, Mouser and other US sources a bit pricey.
There ought to be a source where one can buy things like modem or LAN
transformers for a quarter or so. Once I did a design and despite high
production volume we could not find a source. So we had it custom made
in Taiwan because that was cheaper.

 

[Joerg]

This should boost to well over 2 volts at startup. It would peak at 3
volts except for the diode loss.

But it would first require the PIC to properly start. Worst case would
be a PIC hanging with the port pin on high. Phssst ... pop ... crackle
.... BADABANG.

 

[John Larkin]

But it would first require the PIC to properly start.

No, it doesn't. The LC just behaves that way at powerup. The pic just
has to pump it now and then to keep the cap from discharging.

John

 

[Rich Grise]

Oh. I was assuming the pic ran on 1.5, and the hv was for something
else. My bad.

So, make a blocking-oscillator flyback converter, I guess. Or use a
3-volt lithium.

Why not use one of Watson's LED flashlight circuits?

Cheers!
Rich

 

[Joerg]

No, it doesn't. The LC just behaves that way at powerup. The pic just
has to pump it now and then to keep the cap from discharging.

Well, yes, but for that pumping to happen the PIC must be operational
and well past its POR/BOR boot process.