nano amp battery circuit for powering up micro controller

Oh, and it needs to be ultra cheap

Sort of like this circuit:

Cheers

Klaus

for some milliseconds is tolerable, it?s the steady state current t hat poses a problem

ouple of years instead of 10

, that will be a very slow running oscillator with extreme low duty cycle w hich will power up the micro and possibly take much lower current

or fed into a CMOS gate to power on the micro, but it is not fully thought through

efore Never done it. But I was thinking some relay that turns on once a day. Some photodiode/ solar cell powered thing?

George H.

There's a one-shot circuit in the LM13700 datasheet that draws no standby current, with its supply pin connected to the coin cell an output pulse from that could deliver enough energy to the main supply bus for the uP to come out of complete cold-and-dark and switch over supply directly to the main coin cell thru a PFET or something.

While it's up it charges a timing capacitor out of a pin and the last task the uP tdoes is axe its own power supply connection to the coin cell by shutting down the PFET. then a low power consumption circuit like a JFET schmitt-trigger arrangement still connected to the coin cell supply then watches the decay of the timing capacitor and triggers the zero-standby-current one-shot again after some interval repeating the cycle

the question is IDK what the power to ground leakage of the LM13700 is when it's off, if that's itself in the 100s of nA it's going to be a problem. zero-standby is probably not truly zero-standby.

I played with it in LTspice. It runs fine, but I'm not so sure about runnin g it on the bench. Q1 would have to have a high beta at nanoamp currents to initiate switching, and leakage currents may hamper the oscillation. There are also some unexpected interactions when you change component values. It is fairly easy to extend it to 1,000 seconds but I doubt it would go much further. You would need a counter to get to one day, which would add to the current drain.

You can play with it yourself. Here's the ASC file. Watch the wrap on the l ast line.

Version 4 SHEET 1 1296 680 WIRE 496 -96 416 -96 WIRE 752 -96 496 -96 WIRE 416 -80 416 -96 WIRE 752 -80 752 -96 WIRE 496 -16 496 -96 WIRE 416 16 416 0 WIRE 752 48 752 0 WIRE 800 48 752 48 WIRE 864 48 800 48 WIRE 944 48 864 48 WIRE 752 96 752 48 WIRE 496 144 496 64 WIRE 672 144 496 144 WIRE 688 144 672 144 WIRE 864 144 864 48 WIRE 944 144 944 48 WIRE 496 192 496 144 WIRE 752 224 752 192 WIRE 752 256 752 224 WIRE 496 304 496 256 WIRE 832 304 816 304 WIRE 864 304 864 208 WIRE 864 304 832 304 WIRE 944 304 944 224 WIRE 944 304 864 304 WIRE 944 320 944 304 WIRE 752 384 752 352 WIRE 944 416 944 400 FLAG 752 384 0 FLAG 496 304 0 FLAG 944 416 0 FLAG 416 16 0 FLAG 672 144 Q1B FLAG 800 48 Q1C FLAG 832 304 Q2B FLAG 752 224 Q1E SYMBOL npn 688 96 R0 SYMATTR InstName Q1 SYMATTR Value 2N3904 SYMBOL pnp 816 352 R180 SYMATTR InstName Q2 SYMATTR Value 2N3906 SYMBOL res 480 -32 R0 SYMATTR InstName R1 SYMATTR Value 1e9 SYMBOL cap 480 192 R0 SYMATTR InstName C1 SYMATTR Value 100pf SYMBOL res 736 -96 R0 SYMATTR InstName R2 SYMATTR Value 4.7e6 SYMBOL res 928 128 R0 SYMATTR InstName R3 SYMATTR Value 1e9 SYMBOL cap 848 144 R0 SYMATTR InstName C2 SYMATTR Value 0.1uf SYMBOL res 928 304 R0 SYMATTR InstName R4 SYMATTR Value 1e9 SYMBOL voltage 416 -96 R0 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 SYMATTR InstName V1 SYMATTR Value 3v TEXT 512 -192 Left 2 !.tran 0 50 0 1m TEXT 512 -224 Left 2 ;'2 seconds Oscillator original TEXT 280 464 Left 2 ;See

Klaus Kragelund wrote

Using very high resistor values makes it sensitive to moisture condensation and dirt on the board. Maybe if it was potted in a good high R stuff?

Would it not be simpler to add a solar cell + supercap or whatever to your microprocessor circuit to supply current on top of the lithium or whatever you use battery?

PS there are many other ways to get that little bit of current too, for example: tuned ciruit to local radio station -> RF -> Ge diode small speaker vibration LF -> transformer -> Ge diode Peltier element Just a piece of wire, mains and RF pickuop

You can use the JFET oscillator like done here to light a LED with a thermocouple to up the voltage

So converting up very low voltage sources is not really a problem. There is energy everywhere.

Although the input pin of a CMOS gate draws very little current, when the input pin is close to the threshold voltage, CMOS gates do draw significant current from their power supply pin. Just think about it - between the supply rails, you have two transistors in series, both turned partly on. You might be able to reduce this - e.g. make the supply voltage be less than the sum of the n-channel and p-channel threshold voltages so that they can't both conduct much (though they do conduct somewhat sub-threshold, which can be useful). With a CD4007 you can insert some high value resistors in series with the transistors, and add positive feedback to make sure that the second stage does not spend much time with its input near its threshold voltage. Then with all of those high impedance nodes you might want to conformal coat it. I think there might be some stuff about this sort of thing in AoE3.

Some of the low-frequency RC or even crystal oscillators built into the microcontrollers have quite low power consumption. Check all of the manufacturers until you find a good one. That might be the best option unless you have a lot of time to spend on your own design.

Use two coin cells. :^)

Is that 200nA typical or max?

I don't think you're going to find a transistor that claims any better, at least until you're paying much more than the micro to guarantee it. A typical 2N7000 will do better, yes, but if you need the datasheet to say so, that's where it hurts.

I'm guessing since coin cells are acceptable, this isn't going above 65C or thereabouts.

Can you run the micro from a lower voltage, possibly at a lower clock frequency?

You aren't going to get a lower current clock oscillator I think, without doing the same trick, subthreshold e.g. 74HC.

At those power levels you may consider an energy harvesting method, too.

Tim

-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Design Website:

I have a micro controller I need to power up once a day

The energy needed for powering up the microcontroller and other circuits for

problem

The micro has a standby current of 200nA so the lifetime will be only a couple of years instead of 10

So I am looking for a circuit to put in between the battery and the micro, that will be a very slow running oscillator with extreme low duty cycle which will power up the micro and possibly take much lower current

Initial thought is a very low discharge capacitor and a high ohmic resistor fed into a CMOS gate to power on the micro, but it is not fully thought through

Just wanted input from maybe somebody at SED that has done such a thing before

Cheers

Klaus

Not sure whether such a solution is in the cards from a budget POV:

If too expensive then the UJT or pseudo-UJT (two transistor) solution might the the only option but that will require really good leakage control.

No way to harvest a few hundred nanowatts?

I haven't used them, but TI nanopower timers quote a headline

If nothing else their app notes and dev boards might give you a few related ideas.

s for some milliseconds is tolerable, it?s the steady state current that poses a problem

couple of years instead of 10

ro, that will be a very slow running oscillator with extreme low duty cycle which will power up the micro and possibly take much lower current

stor fed into a CMOS gate to power on the micro, but it is not fully though t through

before

The product may be placed in a basement, with no light, so solar power will be difficult

Cheers

Klaus

ing it on the bench. Q1 would have to have a high beta at nanoamp currents to initiate switching, and leakage currents may hamper the oscillation. The re are also some unexpected interactions when you change component values. It is fairly easy to extend it to 1,000 seconds but I doubt it would go muc h further. You would need a counter to get to one day, which would add to t he current drain.

last line.

Thanks, I played around a little, and as you state it is not terrible robus t. Changing the high ohmic resistances makes it operate unreliable or even stop

But anyway, when it runs, the simulation shows very low average current. So it's a starting point

Cheers

Klaus

Polution of the board could be a problem. My fellow engineers look funny at me even talking about reistances above 10Mohm

I cannot use harvesting. There is no light, no viration, no temperature gradients

Cheers

Klaus

It could be in the middle of africa, no magnetic fields, and sold over the entire worlds, so fields if used would be different frequencies

Cheers

Klaus

s for some milliseconds is tolerable, it?s the steady state current that poses a problem

couple of years instead of 10

ro, that will be a very slow running oscillator with extreme low duty cycle which will power up the micro and possibly take much lower current

stor fed into a CMOS gate to power on the micro, but it is not fully though t through

I know about the current hogging of logic. I wrote CMOS, to indicate that i t would probably be a FET solution, to keep current down. But the suggested schematics, mentioned elsewhere, is a BJT solution

Cheers

Klaus

We have an ASIC design in the works (Jim Thomson designed a portion of it), so I might ask the designer if he could do nA design in the selected process. So that circuit could just be done in the ASIC, which has more control of impedances and leakages

Yes, coin cells are acceptable. That is what I am using now, for the first draft

I am already, to keep currents low

I did look at standard logic, but numbers are typically 900nA quicent for those IC's

Not really an option, sorry

Cheers

Klaus

ate

A good solution, but the price is much higher than my target

ol.

Yes

Sadly, no

Cheers

Klaus

s for some milliseconds is tolerable, it?s the steady state current that poses a problem

couple of years instead of 10

ro, that will be a very slow running oscillator with extreme low duty cycle which will power up the micro and possibly take much lower current

stor fed into a CMOS gate to power on the micro, but it is not fully though t through

before

Nice solution, but expensive

I will save the link for reference

Cheers

Klaus