Caparitors - Charging in parallel, discharging in series for higher voltage

[neazoi]

Hello,
I am working on a project that requires a very lightweight circuit for harvesting power from a button-cell battery (cr2032 or similar). My circuit requires about 12v but the cell is only 3v.
Since only pulse operation is needed (12v is required at a fraction of time, every a vew seconds), I am thinking ofslowly charge some capacitors in parallel and change their connection to series, to obtain the 12v.

have you got any other ideas of how can I do this, since the weight is very limited (<10 grams) and the battery to be used has low current capability.

 

[albbg]

Did you evaluate to use an energy harvesting IC ? For instance:

http://cds.linear.com/docs/en/datasheet/31291fa.pdf

 

[neazoi]

Did you evaluate to use an energy harvesting IC ? For instance:

http://cds.linear.com/docs/en/datasheet/31291fa.pdf

I was concidering an LTC3225 better, because the button cell has low current capability, but it can provide this current in longer times. If this current can be used to slowly charge two small supercaps, then they can be discharged in series to provide higher voltage and a high current pulse instantly.

This pulse can be used to drive a small class-E beacon.

I am looking for a minimum external components solution (especially not bulky such as coils) because the total weight is limited to less than 10 grams.
What do you think?

 

[D.A.(Tony)Stewart]

Can you specify the Pulse current and thus compute the energy 1/2CV^2 to ensure the battery has enough juice.

The ESR of a CR2032 is in the 1~10k range.

 

[andre_luis]

Had you already considered some circuit topology based on the switched capacitor voltage multiplier ?

 

[neazoi]

I have no idea
The battery to use is the cr1632, for it's low weight and relatively high capacity.
I see that the LTC3129-1 can do directly 12v from the 3v (theoretically) at 30mA.
If I could use some series supercaps (provided that they do not weight much) then it could do the trick.

The pulse will be sent from a single 2n7000 operating as a class-E amplifier at 14MHz (just to get an idea)
If the amplifier can start instantly, so the pulse is quite narrow, then 100uF capacitors could be used instead of supercaps?

 

[D.A.(Tony)Stewart]

CR1632 has 140 mAh @0.1mA @3V =< 420mWh @0.1mA ( neglecting battery sag)
CR2032 has 225 mAh @0.2mA @3V =<675mW @ 0.2mA "

Charge Pump IC adjustable to 12V
ST662ABD

 

[neazoi]

CR1632 has 140 mAh @0.1mA @3V =< 420mWh @0.1mA ( neglecting battery sag)
CR2032 has 225 mAh @0.2mA @3V =<675mW @ 0.2mA "

Charge Pump IC adjustable to 12V
ST662ABD

So instant current of these bateries is too low. The supercaps will take ages to charge...?
Maybe better to use a 470uF 16V tantalum instead of a supercap?
I do not know if so low current will be able to charge this cap (through a dc-dc converter) in sufficient time (30sec-1min or so)

 

[D.A.(Tony)Stewart]

Are you familiar with the Capacitor charge current vs time constant formula?

Ic= C dv /dt or
dt = C dv /Ic = 470uF 12V / 0.1mA = 56.4s

Assume for now the charge pump is 100% efficient the current required to supply 12V@0.1mA = 1.2mW = 0.4mA @3V

This means one discharge per minute would draw 0.4mA minimum. and an ideal charge pump converter would allow the 140mA battery to last about 2 weeks.

and 470uF is hardly much to drive a class E beacon

When choosing a suitable battery for such a project, consider the load power and effective impedance and that your power source should be much lower impedance.

Factor this. Peak power * duty cycle * lifespan or Peak power * time duration to get Watt Seconds or watthours and ensure your battery has more than this to do the job.

I would use 3 LiPo cells @11.1V or 4

 

[dick_freebird]

The trick with your fly-cap charge pump is in getting the
high voltage switching phased so perfectly that nothing is
lost, and getting those high voltage switches to not leak
(and there must be at least one high voltage forward switch
that can stand off the product high voltage full time). You
may find a difficult box between high temp leakage current
(which has a N-fold pump efficiency driven penalty) and
the incoming current.

As I think about this I'm inclined to like a flyback scheme
operating in (very) discontinuous conduction - like let the
cap charge to where it's got useful energy, and pop the
primary one time taking the high voltage through a good
diode. Then let the cap charge... I think a transformer
scheme can be way more efficient than a cap ladder or
bucket brigade pump especially when there's more load
current involved.

 

[neazoi]

The trick with your fly-cap charge pump is in getting the
high voltage switching phased so perfectly that nothing is
lost, and getting those high voltage switches to not leak
(and there must be at least one high voltage forward switch
that can stand off the product high voltage full time). You
may find a difficult box between high temp leakage current
(which has a N-fold pump efficiency driven penalty) and
the incoming current.

As I think about this I'm inclined to like a flyback scheme
operating in (very) discontinuous conduction - like let the
cap charge to where it's got useful energy, and pop the
primary one time taking the high voltage through a good
diode. Then let the cap charge... I think a transformer
scheme can be way more efficient than a cap ladder or
bucket brigade pump especially when there's more load
current involved.

Do you think it would be a good idea to use these transformer-based inverters from the flash circuit of cameras then?
Some circuits can operate at as much as 1.5V. These transformers are tiny, so the output current will be small. But They can effectively charge these 300v 330uF electrolytics for flash operation. Maybe if rewuiring the transformer, or make a custom one, I can achieve the required voltage (or using a zener).
A similar approach would be this one I have tried from my website **broken link removed** there are a series of self oscillating inverters there. The unloaded voltages in some of these are very high.
What if I use them to charge a capacitor (even a supercap, with some form of regulation)?
I have no estimation about their required input current and output current capability, that is the problem.

 

[betwixt]

You will get significantly higher efficiency if you use an inductor based pump but there *may* be a simple solution to your problem. I stress I have not tried this but it just might work...

A 3V cell can power a common MAX3232 interface IC which has both negative and positive charge pumps on-board. If you 'float' the IC and battery then take the output across the + and - reservoir capacitors you should in theory get about 20V. The advantage in this method, (if it works at all!) is that there is no reasonble limit on the size of reservoir capacitor you could use. The data paths inside the IC would be unused.

Just a thought...

Brian.

 

[neazoi]

You will get significantly higher efficiency if you use an inductor based pump but there *may* be a simple solution to your problem. I stress I have not tried this but it just might work...

A 3V cell can power a common MAX3232 interface IC which has both negative and positive charge pumps on-board. If you 'float' the IC and battery then take the output across the + and - reservoir capacitors you should in theory get about 20V. The advantage in this method, (if it works at all!) is that there is no reasonble limit on the size of reservoir capacitor you could use. The data paths inside the IC would be unused.

Just a thought...

Brian.

So
The battery negative should not be connected to the GND of the rest of the circuit, but just the gnd of the IC. The IC GND should not be connected to the GND of the rest of the circuit as well.
C1+ will be the VCC out
C2- will be the GND
C1- and C2+ should be connected together (both in charging time and discharging time of the caps?)

Am I right?
Could I also use the smaller MAX3222/MAX3237/MAX3241, since I do not need the data lines?

 

[betwixt]

Thats correct. Connect the IC (or it's alternatives) as in the data sheet but only make outside connection to the C1+ and C2- pins, using them as the high voltage output.

As I pointed out, I have ever tried it but it seems to be a simple solution. I have seen the C2- output used as a negative voltage source in schematic before while the remainder of the IC was carrying data as normal.

Brian.