Battery-powered Pi?

Over on http://www.raspberrymod.com/viewtopic.php?f=60&t=640&sid=b006909c691d78e07788ff03dbf8526c I wrote a little piece about Lithium-Polymer batteries and the rising trend to integrate Li-Po chargers on microcontroller boards.

This doesn't help at all with Malcolm's question of course, but I'd like to ask a supplementary question in case someone here knows the answer.  Is there a 2-cell Li-Po charger module (2 x 3.7V = 7.4V) available from anywhere, suitable for integration into 5V-powered equipment?

@ Morgaine,

Li-Po batteries are fussy things and you might suffer from tolerance stack up or balance problems if you charge in series (although lot's of battery packs must do this).

Why not use a boost converter to generate the 5V you need from the battery's 4V or so?

Boost converters are simple, small, cheap and quite efficient.

If you can give me a  spec I'll suggest something specific.

Michael Kellett

Many thanks for the suggestions, I guess I'll start with an off-the-shelf charger, I may well get a longer run-time than with my 6V battery.

Now how to change the status here to answered ??

Malcolm,

   At the top of the original post, just under the title, there should be

a link you can click on that says "mark as assumed answered".

Also, at the bottom of each response, there are two boxes you can

click, one for "correct answer" and one for "helpful answer".

Any correct or helpful answer you click on will get automatically mentioned

in the original post.

I'm using this one with Raspberry Pi very successfully: http://www.amazon.co.uk/gp/product/B005JSG7GE/ref=oh_details_o03_s00_i01

Sadly out of stock at Amazon (and all the other suppliers I looked at seem to link back to Amazon!).

@Michael:  Indeed, Li-Po batteries are very fussy in their charging requirements, which is why I shied away from a simple series arrangement with two floating single-cell chargers.  I strongly suspect that there be dragons in that area.  That's why I asked whether anyone knew of a ready commercial module that is specifically designed to keep two Li-Po cells charged up as a unit

I've noticed in 10.8V Li-Ion LED torches (not quite the same as Li-Po, but share some features with them) that the built-in charger has connections to the two inner links between the 3 Li-Ion cells, so there must be something clever going on.

Your suggestion of using a boost converter from a single Li-Po cell up to 5V is an excellent one though.  If the conversion is efficient enough and the component cost less than that of a two-cell charger, it seems the more sensible way to go.

Morgaine.

What commercial RC LIPO chargers seem to do is to discharge the cells that come too close to the 4.2V maximum. This is able to top all of them off to 4.2V without overcharging the cells that reach 4.2V first. This is not efficient. The one I have can only discharge about 100mA, which charging could be done at up to 2A, so when you want to top off the best cell but the worst cell reaches 4.2V, you'll have to keep discharging it at 100mA while only charging the good cell at 100mA (the bad cell cannot handle any more charge).

Floating chargers works slightly better.

Ideal would be a circuit that could selectively charge every separate cell. This is slightly tricky, I don't know of any chips that aid in building this.

The best I've seen is a chip that allows you to charge an inductor using ANY cell, and then using the charged inductor to continue to charge the whole pack. This is energy efficient, and does not waste the energy that needs to be discarded from the cell that is topped off first. So things don't get hot.

Anyway, a boost from a single cell is the easiest configuration. Charge by "buck" converter from 5V, and discharge using a boost converter from the 3.3-4.2V battery voltage.

A boost converter is likely to have a diode in the power-flow part. This causes a 0.4V minimum voltage drop, so in effect the converter has to deliver 5.4V of which only 5V can be used at the output. This is a significant part of the losses of a boost converter.

If my memory serves (I can hunt it down - the article not my memory), Circuit Cellar had a design article in which someone had designed a balanced charging system, the batteries were lithium but high capacity ones such as those used for vehicles etc. The design though could be scaled down, the balancing circuit detected those batteries that were at capacity and disconnected that bank from the charging circuit, which of course is pretty logical, though would require a string/bank of more than just two batteries for that particular design.

I suppose you could have a combination of a constant current source and an active load and switch through a solid state (4066 or similar) the charge so it could be in series up to a certain charge point and then switched over to a single battery at that moment. There are plenty of battery charger/conditioner/fuel guage IC's around that could provide the charging algorithm and a a small microcontroller could be the 'brains' and perform the switching and possibly some other house keeping duties. a FET or IGBT could provide the switch and active load function.

Colin

Colin, Switching batteries between series and parallel is completely nontrivial. If you're going to make a parallel battery pack you can manually ensure they are at the same charge state (e.g. at 4.2V) before connecting them together, but doing as you suggest, when one cell is full, and others aren't then connecting them in parallel is likely to result in bad things happening.

Roger, I wasn't intending to suggest switching from serial to parallel I was suggesting switching from batteries in series to charged batteries being isolated from the 'string' via the switching mechanism. So battery_bank_initial =  b1 --- b2 --- b3        one of those 3 batteries becomes charged the others not so - battery_bank_part_charged = bX --- bX --- (bZ)(isolated battery bZ). .

That suggestion is almost as complex as switching from parallel to series.

It does solve the "large currents because you connect non-equal-batteries" problem.

Hmmm.... Interesting indeed. For each cell, the "incoming" connection is the "bottom of the next-higher" cell. That needs to be connected either to the TOP or the BOTTOM of the current cell. You can use logic level N-FET for the connection to the bottom of the current cell. You have 4.2V available for when this is neccessary. You can use a P-FET for the connection to the top of the current cell. You have (at least) -3.0V available for that when this is neccesary. If you use a weak pulldown on this, you might be able to recover from "drained-too-far" cells.

You could use an attiny45 to control each individual cell. Each attiny measures its internal 1.1V reference voltage with respect to its own VCC (the cell voltage). If that drops below 268 (out of 1023, the full range of the ADC), the cell voltage is at 4.2V. Very interesting project. :-)

(A dual optocoupler, or maybe some capacitors can provide level shifting communications channels with the "main" processor).

Hmm. Attiny processors (like the '44 and '45) have increased in price 25% over the last two months. Weird!

That suggestion is almost as complex as switching from parallel to series.

It does solve the "large currents because you connect non-equal-batteries" problem.

Hmmm.... Interesting indeed. For each cell, the "incoming" connection is the "bottom of the next-higher" cell. That needs to be connected either to the TOP or the BOTTOM of the current cell. You can use logic level N-FET for the connection to the bottom of the current cell. You have 4.2V available for when this is neccessary. You can use a P-FET for the connection to the top of the current cell. You have (at least) -3.0V available for that when this is neccesary. If you use a weak pulldown on this, you might be able to recover from "drained-too-far" cells.

You could use an attiny45 to control each individual cell. Each attiny measures its internal 1.1V reference voltage with respect to its own VCC (the cell voltage). If that drops below 268 (out of 1023, the full range of the ADC), the cell voltage is at 4.2V. Very interesting project. :-)

(A dual optocoupler, or maybe some capacitors can provide level shifting communications channels with the "main" processor).

Hmm. Attiny processors (like the '44 and '45) have increased in price 25% over the last two months. Weird!

For a high efficiency small boost convertor you could consider the TI63020 (Farnell 1815784) - synchronous so no series diode, very fast so tiny inductor. Downside is that it's in a package that some will find difficult to prototype and it is not very cheap.

Michael Kellett