The LiPo Project

[Dougigs]

This thread is devoted to the development of workable, safe use of Lithium-Polymer batteries in headphone amplifiers. Its end goal will be a PCB that provides smart charging of LiPo stacks of 2 cells (typically 7.2 volts) or 3 cells (typically 11.1 volts) and that monitors the cell voltage and cuts it out of the circuit when it reaches its low-voltage point. Our goal is to put both these functions on the smallest possible PCB, i.e. one of MINT proportions, so that it can be added to absolutely any existing headphone amp.

These cells are available in extremely small size (smaller than 2 9v batteries) in up to 900mAh (compared to 250mAh for the two series 9V), or in a larger size (somewhat larger than 2 9v batteries) in up to 2400mAh.

Another thread, "Li-Polymer batteries for headphone use?" is devoted to the question of whether these cells are good or bad for our applications. If you have questions about the merits of these cells, please visit that thread; this one is for people who believe they have merit and who want to develop applications that can be used by the hobbyist.

On the next post I'll show the technology that we've collected or identified so far.

But for now I'm going to issue a warning that should be repeated frequently in this discussion:

NEVER USE LI-POLYMER CELLS UNLESS THEY ARE BEING RUN BY 1) A DEDICATED CHARGER MADE ONLY FOR THESE CELLS, WHICH CONTAINS A CIRCUIT THAT STOPS CHARGING WHEN FULL VOLTAGE IS REACHED -- OTHER CHARGERS, INCLUDING NIMH CHARGERS, COULD CAUSE AN EXPLOSION -- AND 2) A CIRCUIT IN YOUR AMP THAT CUTS THE BATTERIES COMPLETELY OUT OF THE CIRCUIT WHEN THEY FALL BELOW THEIR SAFE VOLTAGE (TYP. 3V PER CELL -- 9V FOR AN 11.1-V CELL). IF THEY ARE ALLOWED TO DISCHARGE BELOW THIS LEVEL, AT BEST YOU COULD LOSE A $50 BATTERY AND AT WORST YOU COULD CAUSE AN EXPLOSION WHEN YOU TRY TO CHARGE AGAIN. IF YOU ARE AT ALL UNSURE OF WHAT YOU ARE DOING, DO NOT EXPERIMENT WITH THESE CELLS.

 

[Dougigs]

Here are some pics of commercially available Lithium Polymer cells and a small charger, so you get a sense of what we're working with. They're shown next to a MINT amp and an IPod for size reference.

1) A pair of 790 mAh, 3.6v LiPo cells -- these from a last-generation Sony Ericsson cell phone, but of a type that can be bought from most major parts companies. These are the sort of cells that would be used if you wanted to harness this technology for a pocket-sized amp.

2) The commonly available 11.1-v, 2400 mAh 3-cell LiPo stack available from radio-controlled-aircraft hobby dealers. It is sized for a PIMETA-type enclosure. This is what I'm going to start working with for development. I bought it at this location for 23 pounds, but these are available from many places:

http://stores.ebay.com/RevolutionShop

3) A LiPo charger, which costs about $15, on a tiny board. It charges fast (800ma) or slow (400ma), and it stops charging when the 11.1v point is reached. It is powered with anything from 15 to 18v (i.e. most laptop power warts). I was going to try reverse-engineering this unit, but it uses very obscure chips, and a lot more external components than you'd want to use in a headphone amp -- the more modern chips need only a few resistors.

However, many hobbyists may find this unit a pretty good drop-in charging option; I'll be using it while I work on a cutoff circuit. NOTE THAT YOU STILL ABSOLUTELY MUST HAVE A CIRCUIT TO SHUT DOWN THE BATTERIES COMPLETELY IF THEY DROP BELOW THEIR MINIMUM VOLTAGE -- IT IS VERY DANGEROUS TO OMIT THIS.

Here's where I bought this:

http://www.heli-fever.com/product_i...&products_id=76

4) Just so you're aware of what's coming, this shows the latest generation LiPo cell. This is a 950mAh, 3.7v cell -- it's one inch by one inch by 5mm. It comes from a Sony Ericson 750i phone. These cells aren't very available, but developing a charge/discharge board at the smallest possible size because these cells will come on the market, making possible something like a LiPo MINT for those who want it.

 

[Dougigs]

Finally, here are the chips that have been identified by me, Born2bwire Nisbeth, Frenchman and others as being useful for our purposes.

I think that the first priority is to develop a low-power cutoff circuit. Pre-built chargers exist (see above) but there isn't an off-the-shelf cutoff circuit that suits our needs.


1) LOW-BATTERY CUTOFF CIRCUIT

Our goals are 1) a safe, reliable circuit that can fit inside the smallest headamp case and will cut off 2-cell or 3-cell LiPos completely from the circuit when they drop below 3v per cell; 2) A circuit that draws little current (10 microamp range) when the circuit is off; 3) requires as few components as possible, so it can go on a PCB wiht the charger.

So as a starting point for a low-battery cutoff, I think the LTC1841 is the most exciting -- it's a dual micro-power comparator in an 8-pin SMD package with a built-in voltage reference; it consumes 5.7 microamps. A low-battery cutoff circuit is shown on page 12 of this datasheet:

http://www.linear.com/pc/downloadDoc...13,P1694,D3376

On the other hand, we could use a more conventional micro-power op amp, such as the circuit shown below (you wouldn't have to use that particular opamp, as there are plenty of micropower chips more commonly available):

http://www.elecdesign.com/Files/29/6166/Figure_01.gif

Or another nifty option is to use this dedicated chip (which, however, isn't easy to source):

http://pdfserv.maxim-ic.com/en/ds/MAX1614.pdf


2) CHARGER CIRCUIT

This will come next (Unless someone wants to start experimenting now).


Take a look at the MAX846A charger chip:

http://pdfserv.maxim-ic.com/en/ds/MAX846A.pdf

It's quite nice -- simple 16-pin SMD chip, hardly any external components (it's worth seeing if the two 10uf electrolytics can be reduced); wide voltage range. It's a bit hard to obtain.

A more commonly available chip, also very nice, is the bq24105RHLR. It has the advantage of using a linear rather than switchmode charger (so if you're wart-powering your amp while charging you won't listen to switcher hash in your signal). It is also easier to find. Its 20-pin flatpack might be tricky for PCB design and soldering -- anyone with experience using this package should tell all. Look at the schematic on page 2 of this datasheet:

http://focus.ti.com/lit/ds/slus606b/slus606b.pdf

 

[Jazper]

you can order those chips directly from www.maxim-ic.com you know...

 

[Dougigs]

Oh, thanks for pointing that out, Jazper. That makes the MAX1614 much more appealing -- You just need it, two resistors and a MOSFET (and there are lots of applicable MOSFETs in SOIC-8 packages)... and it also offers a momentary-button on/off switch, driven by an internal flip flop.

One problem: It only comes in the 8-pin uMax package, which is 60 per cent smaller than SOIC. Now, I note that RS (and probably other places) sell adapters for this package to DIP, for experimenting.

But has anyone had any DIY experience with these smallest-sized SMD chips? Can they be hand-soldered at all?

 

[Dougigs]

To be specific, that package has lead spacing of 0.65 mm, exactly half that of a SOIC. Any experience with those?

 

[Born2bwire]

The only chips that I haven't been able to do by hand are BGA. Pretty much, if it has leads, it can be done by hand if you have the right tools. I have good results using solder paste, an iron with a very fine tip, and tweezers. Having a magnifier of some sorts is really needed to check the integrity of the job and look for any bridges though.

 

[Jazper]

I've had great success with Maxim chips in my experimenting projects, but I don't think there will be many diyers interested in soldering tiny chips to boards - SOIC-16 is about the worst most people can handle. Remember that not every DIYer has a good soldering station or precision tweezers.

edit:
Also remember that voltage references (and shunt diodes) create a bit of noise so you will want some noise cleaning or isolation on the power supply if you were to implement the LT184x chips. The easiest way I can think of doing this is a set of cascoded fets or a linear regulator (even if it's not in regulation or only dropping the voltage by a volt or so) and a cap to isolate the batteries/power supply

Also please be aware of the maximum voltage the LT184x chips can handle (11-12v) you're going to be running skin of the teeth with it. I suggest looking for another alternative

 

[Dougigs]

Excellent points. I'm also worried about the 50khz oscillator in the charge pump of that Maxim chip, directly driving the gate of the power supply MOSFET. I really want to avoid anything that's going to create noise.

This brings us back to using microcurrent op amps in a simple comparator circuit (see below).

This circuit has the advantages of a) using any micropower op amp in an SO8 package, thus not limiting people to one supplier, and b) being a low-noise, low-hassle circuit.

It could also use any precision voltage reference. In fact, many of these come in SOT-23 packages, which is basically a 3-pin SOIC... that, plus the lack of capacitors, means we can have total flatness. So if we just make pads for a) an SO8 op amp, b) an SO8 MOSFET and c) an SOT-23 precision voltage reference, then there's a lot of flexibility as far as which brands etc.

EDIT: Is a rail-to-rail op amp mandatory in this design?

 

[Dougigs]

If we're using a discrete precision voltage reference with an op amp as in the above circuit, couldn't we eliminate any noise problems created by the voltage reference by bypassing it with a small cap? This approach works just fine with zener diodes in shunt regulator circuits, for example...

 

[tyre]

Quote:

Originally Posted by Dougigs One problem: It only comes in the 8-pin uMax package, which is 60 per cent smaller than SOIC. Now, I note that RS (and probably other places) sell adapters for this package to DIP, for experimenting. But has anyone had any DIY experience with these smallest-sized SMD chips? Can they be hand-soldered at all?

I've soldered an MSOP package opamp before, which is the equivalent size and lead pitch of the Maxim uMax package. It's not particularly difficult. Solder wick works wonders for solder bridges.

This is an interesting project btw. I have several old cellphone batteries laying around that I've wanted to do something with.

 

[cetoole]

This project looks very interesting, I will definitly want to follow this. Personally, I dont think using uMax chips should be a problem, the spacing doesnt look any worse that the SSOP28 ICs, which anyone who has done one of the Guzzler/00940 DAC kits has done with either the PCM2702 or PCM2902, and as it is a much lower pin count, it shouldnt be that hard. Worst case, just flood and suck it, with a little care, it shouldnt be a big deal. Cant wait to see what you come up with.

 

[Dougigs]

Well, I'm not personlly worried about using a teeny chip. But I actually think the comparator/voltage reference/MOSFET circuit is a better one.

( http://www.elecdesign.com/Articles/P...ArticleID=6166 )

For these reasons:

- it's quieter. As neat as that Maxim chip may seem, it contains a noisy charge-pump engine and a voltage refreence that is said to be noisy. The comprarator approach is a simpler, linear circuit.

- It is very solid -- it has a nice hysteresis built in that cuts off the battery at 3.0v per cell (i.e. 9v) and cuts it back in at just under 3.1 v, so you don't get it flipping in and out around the threshold

- it's easier to make even quieter -- I'm assuming we can simply place a small capacitor around the voltage reference

- It uses only a few microamps when it's monitoring the battery.

I'm going to breadboard up that circuit using a regular DIP op amp, to determine the resistor values for R1-R4 for a 9V cutoff. If you look further down in the article, if provides the formulas for determining some of the values, but I can't see how to solve for R1 and R2... if someone smarter can work it out, that would be great.

(I'd also like to figure out the 5.6V cutoff resistor values, for circuits using 2 cellphone batteries)

I'll work them out by hooking up a variable PSU and twiddling the trimpots, unless anyone can do the math.

 

[Jazper]

Quote:

Originally Posted by Dougigs If we're using a discrete precision voltage reference with an op amp as in the above circuit, couldn't we eliminate any noise problems created by the voltage reference by bypassing it with a small cap? This approach works just fine with zener diodes in shunt regulator circuits, for example...

That might work, but it would be better to isolate it properly, an LM317 doesn't cost THAT much...

 

[guzzler]

Quote:

Originally Posted by Jazper I've had great success with Maxim chips in my experimenting projects, but I don't think there will be many diyers interested in soldering tiny chips to boards - SOIC-16 is about the worst most people can handle. Remember that not every DIYer has a good soldering station or precision tweezers.

Eh? Your own USB DAC has a smaller chip than that, SSOP. It's 0.65mm pitch as well, and I agree that soldering it is fine; just take the time to line it up properly.

As always, my standard SMD soldering link:

http://www.infidigm.net/articles/solder/

Enjoy, I'm following this one