LiPo Balancer - Schematics & PCB

[Suzanne]

This is a switching shunt regulator that will show if there is an imbalance in your pack. It will also automatically rebalance the cells at the end of each charging cycle.

Each device must be adjusted so that the red light goes on at 4.15 Volts. One device must be paralleled to each cell group. If you have a 10s4p pack, you need 10 devices. For a 2s1p you need 2.

The max. current available for re-balancing is set by R1A..R1D. These are SMD metal film resistors rated for 0.25W each. 100 Ohms for each give a max. balancing current of 165mA, which is plenty for normal packs, because a pack protected by these devices will not develop a major imbalance in the first place.

Up to 10A are possible with a suitable PCB layout, but then you need fairly big resistors for R1A..R1D, as each one would dissipate up to 10 Watts. Also, for higher currents you will want to connect Pins 1..3 and Pins 4-8 of Q1.

If there is a huge imbalance, the re-balancing current may not be sufficient to rebalance the pack during one charge, and one or more red lights will go on permanently. This is an alarm condition because it means that some cells are too high in voltage and that the charging must be limited to less than 165mA or stopped immediately. The overcharged cells will automatically be discharged to 4.15V and the balancing process will continue during the next charge. After some charges, the pack will be fully balanced.

To completely balance a pack with unknown cell status in one go, slow charge at 100-150 mA. You may use a current limited power supply or even a NiCd charger set to 100mA formatting mode. As some cells become fully charged, their balancers will activate and their red lights will flash. All cells are fully charged and balanced,
when all red lights are flashing.

The circuit draws less than 10 Microamps form each cell while idle.

Please note:

This circuit is intended for keeping a pack balanced when charging it with a suitable LiPo charger. With the listed parts, it can absorb up to 165mA charge current, which is sufficient for the intended purpose. However, if you overcharge your pack by setting the wrong number of cells or by using an unsuitable charger, you will still damage your cells and risk a fire. To be safe, watch your pack while charging. If one or more LEDs stay on permanently, these cells are being overcharged, and the charging process must be stopped immediately, or current must be reduced to less than 165mA to avoid damage and possibly fire.

[Suzanne]

This is a single sided board for SMD components. It is 1.0 by 0.75 inches and weighs 1 gram. It is intended to be taped to each cell.

[devmonkey]

Suzanne,

This is an excellent solution to the balancing problem. The true elegance comes from being able to peak each cell/parrallel pack with a single existing charger, plus having some visual indication of cell balance/charge completion.

My only concern would be variance in the reference divider due to temperature of the cells. This probably wouldn't hurt too much since all the resistors would vary at the same rate.

People using this circuit would find the mAh the charger says it delivers to teh pack is greater than the actual figure since some will have been burnt by the balancing resistors on a cell that peaked early. However it will not alter the value by much in a nearly balanced pack and I'm not really sure how accurate these measures are to start with.

I don't think that the 165mA max balance current is a problem if new cells are fitted with this device when assembled into packs - all the cells I have had from li-poly factories have been exactly 3.80v. With a properly configured 4.2v per cell charger the fully charged balance current will only be 2mA.

In my view the only reason for uprating the current balancing of the unit would be to deal with over voltage conditions from the charger (e.g. faulty charger or wrongly set cell count). 10watt resistors would weigh lots more.

Do you have any ideas on how to protect the cells against short circuit. The momentary high current destroys the cells internally, I wonder if some current ramp device might work.

Cheers

Joe

[Suzanne]

My only concern would be variance in the reference divider due to temperature of the cells. This probably wouldn't hurt too much since all the resistors would vary at the same rate.

I use 1% metal film resistors.


I don't think that the 165mA max balance current is a problem if new cells are fitted with this device when assembled into packs

That's the idea, yes.


10watt resistors would weigh lots more.

In this case I would put the devices outside the pack, i.e., connect only when charging.


Do you have any ideas on how to protect the cells against short circuit.


Basically you need a shunt to measure the current and a few FETs in parallel to pass/break the current. Plus some amplifier / voltage reference / glue logic. Not really difficult. But people don't want it because it reduces voltage under load and introduces an additional point of failure into the system.

[Happy|Harry]

Could you build it with larger components just for use when charging?

phil

[Suzanne]

Phil,

yes, this shouldn't be a problem. The MAX 921 is available in a DIP-8 package, and logic level FETs area vailable in the TO-220 package. If size and weight is not an issue, you may as well replace R1 with a 3.9 Ohm / 5W type to increase the balancing current to more than 1A.

You could easily build this version on a prototyping board.

OTOH, I have just received confirmation from the board manufacturer that the PC-boards should be available in a few days. They will be made from lightweight 0.8mm laminate, tinned, CNC cut, etc. Very professional.

If there is enough interest, I can ship some to the US for $3 (each) plus postage.

[Happy|Harry]

Thanks Suzanne, as i plan only to use it when charging with my home made CC/CV chargers(electron_head's design)
the larger design is of no matter. Also what's a prototyping board?(sorry if this is a stupid question but i'm relatively
new to electronics)

phil

[Suzanne]

This is what it looks like:

[RMFISH]

Suzanne,
What type and value would C1 be?. 47 Microfarad?. What voltage? Electrolitick?.

Thank you for the information,
Robert.

[Michael Moritko]

Suzanne: Why is the balancing voltage set at 4.15 rather than 4.2 volts? Also, could you clarify what you mean by connecting pins 1..3 and 4 - 8 of Q1 for higher currents? I seem to recall the new Astro LiPo charger pulses the LiPo cells as they approach a fully charged state. If I in fact remember that correctly, will your balancing circuit work correctly with the Astro charger? In a 5S4P pack, the charger is looking for 21 volts before it will stop charging (5 times 4.2). If your balancing circuit allows each cell (or parallel group of cells) to reach only 4.15 volts instead of 4.2, will the LiPo charger ever sense a fully charged pack and cut off the charging current, or, in the case of the 5S4P pack, will the charger see a maximum of only 20.75 volts and never completely terminate the charge?

Michael

[Suzanne]

Originally posted by RMFISH
[B]Suzanne,
What type and value would C1 be?

47nF (0.047uF) should be a suitable starting point. It is not critical. You may want to adjust it for the desired blinking frequency.

[Suzanne]

Why is the balancing voltage set at 4.15 rather than 4.2 volts?
Simply a safety measure. Not everyone has a super precise voltmeter. 4.2 is ok if you don't need the additional safety margin.


Also, could you clarify what you mean by connecting pins 1..3 and 4 - 8 of Q1 for higher currents?


Pin 1..3 are all connected to source, Pin 4..8 are drain. For higher currents, you should connect all pins. For 165mA, one pin is sufficient. I have since changed my PCB layout so that the pins are connected.


I seem to recall the new Astro LiPo charger pulses the LiPo cells as they approach a fully charged state. If I in fact remember that correctly, will your balancing circuit work correctly with the Astro charger? In a 5S4P pack, the charger is looking for 21 volts before it will stop charging (5 times 4.2). If your balancing circuit allows each cell (or parallel group of cells) to reach only 4.15 volts instead of 4.2, will the LiPo charger ever sense a fully charged pack and cut off the charging current, or, in the case of the 5S4P pack, will the charger see a maximum of only 20.75 volts and never completely terminate the charge?


This depends on many factors and it doesn't really mattter. The cells are safe either way. The normal behaviour would be that the charger does not terminate and that all lights are flashing, indicating a fully charged and balanced pack.

If you don't want this, just set the threshold to 4.2V or slightly above. In this case the charger will terminate as usual.

[Suzanne]

I just got confirmation from the board manufacturer that the first batch of circuit boards is in the mail.

Should be here Friday.

If anyone's interested, please let me know.

[Happy|Harry]

Originally posted by Suzanne
Simply a safety measure. Not everyone has a super precise voltmeter. 4.2 is ok if you don't need the additional safety margin.


as far as i've read each cell voltage can be as much as 4.25V(4.2V +/- 0.05V) and still be within the safety margin.
so a setting of 4.2V for the balancing voltage should be ok :)

phil

[Suzanne]

Phil,

Yes. 4.2V should still be a safe limit. After I have assembled the first batch, I will experiment with various settings and report here.

Unfortunately, my distributor just informed me that the lead time for the MAX921ESA is 6 weeks...

Well, patience....

[Suzanne]

Just a quick update for those who are interested - I have built a few samples of the LiPo-Balancers and fitted them to one of my packs. They work best when adjusted for a threshold of 4,20 Volts.

Finally I can SEE what happens when I charge my packs...

As one of the tests, I have deliberately overcharged the pack to 4,27V per cell, then removed the charger. Within a few minutes, the balancers brought each cell down to 4,20V and then stopped. Great.

The first batch will go to my local hobby shop. They have preorders for as many as I can make.

[DNA]

Hi Suzanne,

I'm not sure I'm reading your last post correctly.

It looks like you're saying that you overcharged each 4.2v cell
"to 8,27V per cell". Or nearly double the voltage there should be
on each cell.

How is that possible without ballooning or venting the lipos?

[Suzanne]

DNA,

thanks for the correction. This is a typo and should read "4.27 V per cell".

I will edit my post to remove the error.

[vintage1]

Elegant, simple, effective light and probably pretty cheap.

Top marks Suzanne. Its nice to see the art of analogue circuit design is not dead.

[gtstubbs]

Suzanne
Great solution to the problem of balance.

I was wondering whether this might work as a simple solution/ protection to overcharging, leaving aside applying huge overvoltage.

The PCB would mount as the Kockam PCBs do but with a 3.6v zener and red led (with a drop of 0.6v) = 4.2 volts. The idea is simple overvoltage protection and full charge indication via the led.

Do you think this would work?

[Suzanne]

This will definitely not work, for several reasons.

a) A zener diode needs more than 50mV to switch from fully off to fully on

b) A red LED drops approx. 1.6V, not 0.6V.

c) Even if you find a suitable zener diode, Idle current draw would be several orders of magnitude too high. The LiPo cells would be discharged by the protection circuit, leading to major imbalance after a few days.

d) Discharge current would be limited to 20 or 50mA, depending on the LED. This is not sufficient.

e) Precision of such a simple circuit is in the range of 5%. What you need is 0,2%.

I could go on, but you get the idea...

[Happy|Harry]

suzanne have the parts you require arrived yet? as id be interested in trying a few of these.

phil

[Suzanne]

Harry,

the parts arrived on Friday, so I finished everything between Friday night and Saturday morning. 8 hours later, all balancers were already sold.

I haven't even enough left for my own packs.

Some went to my local hobby shop in Stuttgart, Germany. They had pre-ordered as many as I would give them.

If you're lucky, they will still have some on Monday. Their telephone number in Germany is +49 711 292704. Ask for Ingo Tannert, he speaks english. They also make packs to order, complete with balancers.

I will make another batch of 100 pieces, if there is enough interest.

Or would anybody be interested in manufacturing them? I can provide the CAD files and instructions.

[Larry3215]

How much are they going to cost?

I might be interested.

Larry

[Suzanne]

End user price in Germany is 15 Euro each.

If I ship larger quantities to the USA, price will be $12 each.

[luc]

here is a way to avoid imbalance, without any costly PCBs.
It applies mainly for big packs, when weight is not so much a concern, but can also be done on small packs:
Let's suppose you need a 3S2P pack, made of Irate 2200. Instead of building 2 packs 3S, build 3 packs 2P.
Then you only need a multiplug to discharge the packs in serie, and another one to charge the packs in //.
The charger will consider then that you have only one cell of 6 times the capacity. The balancing of the cells , in //, is automatic, as long as the charge level is not too high. It can be safely checked after charge.
Another advantage is that if you need for another apllication 4S,5S...you just have to add another 2P pack in serie, eventually building a new multiplug connector.
This system is very safe, and avoids expensive PCB...

[RD Blakeslee]

luc's way has been my way for over a year now ...

- RD

[luc]

that's fine, RD but there is a big difference:

one year ago, big batteries (only kok 2170 and 3270) were only able to deliver 3C, which means applications were rather poor: only big slow-flyers (quarter scale, hifgh wing or so) could be done.
Today irates/Tanic are able to deliver 10C. A 3S2P pack weighs 260g, yet is able to deliver 44A and peaks above 60A. This gives a wide open new market for these cells (3D planes, scale electric conversions possible, big motorized sailplanes, EDFs,....). So promoting it now can be very useful to plenty users...

[Suzanne]

This system is very safe, and avoids expensive PCB...

Luc,

this is indeed a very good way to solve the problem of cell imbalance for small packs, and definitely much cheaper than adding balancers to each cell.

For large packs, it seems less desireable to me, because

a) Charging time is multiplied by the number of parallel cells. At 4.20 Volts, many commercial chargers are limited to 2 or 3 Amps. This way, a Joker pack would take 30 to 50 hours to charge. You would need 80Amps to charge it at 1C...

b) The connectors become quite complicated. Imagine all the cables and connectors for a 10s4p pack! Having a single pack with one connector is much more convenient for the average user.

c) I have become so used to those blinking lights in my packs...immediate feedback...peace of mind, it's hard to describe. It "feels" so much better being able to see what's happening.

[t.nollett]

silly question but how do you secure your boards to the cells as shrink wrap would damage the boards but they need keeping dry
and free from moisture. do you just use clear tape.

t.nollett

[Suzanne]

They come with double sided tape already attached to the back. You simply tape one to each parallel group of cells, connect the red and black wires, and put clear shrink wrap around everything. They are all SMD, quite flat, no parts sticking out.

Just make sure you use transparent shrink wrap so you can see the red lights.

[RD Blakeslee]

luc, You are correct, larger, higher-C cells make individual cell or parallel cell charging applicable to bigger airplanes these days.

Suzanne, I think large packs will soon be unneccesary - see post No. 100 in the thread:

http://www.rcgroups.com/forums/showthread.php?s=&postid=1476707#post1476707

[gedaso]

Suzanne, any news or progress on the pcb's ?

[t.nollett]

i just got 25 and they are just the right size for my 3sxx of etec 1200's. should help save my packs as a number of my older ones have had cells fail. adds 4g to a 3s pack.

[pgleesonuk]

Hey t.nollett
How did you contact Suzanne to get these balancers?
Just via this thread, or this there somewhere that sells them?

I would like to get some for my new Lipos.

Thanks

Paul

BTW: I live not so far from you in Crick, just off Junc 18, M1
What do you use your lipos in?

[Suzanne]

Originally posted by gedaso
Suzanne, any news or progress on the pcb's ?

Balancers are in stock. If you wish to build your own, I can sell you just the PCBs, they are in stock, too.

I'm checking this board only once or twice a week, so response to PM may take a bit.

[mark_q]

To charge a 10S4P pack, would I set this up as 10 sets of 4P packs? (as opposed to 4 sets of 10S)

Then wire one cell balancer to each of the 10, 4P packs?

Now plug the whole set together in series and charge at the 4P rate (eg: LP1300 x 4 = 5.2 amps)?

Right? :confused: Wrong?

[luc]

Yep, but if you charge (top) your 10 packs in //, you don't even need a PCB.
For not taking too long, begin the charge in serie and end it in .//. That's the way I do on my big packs

[t.nollett]

what motor have you that takes 37 volts.

t.n

also i only build small packs 3s1p then harness them in parallel
got 4 packs of 2s1p in my x-ite[zagi wing].

luc where do you get your cells.

[mark_q]

t.n. - No motor I have will take 37 volts! It is just an example.

Thanks for your replies.

So, luc, tell me more. If I charge my cells in //, I don't need a balancer? Then you say start in series, so, I plug everything together charge to completion then change to // and charge again to completion? This will balance my cells?

Thanks for any help here - Mark

[t.nollett]

if you split a 10s4p into 10 x 4p packs,[1300 cells] each one would need charging at 5.2 amps x 10 packs= long time
so to save time charge in series to 80-90% [if you had a charger that takes 10s] then spilt them into parallel packs and charge them as parallel only to 100%, would be ideal if you had a 50 amp charger:D
the reason i use the lipo balancers is because i'm lazy. just like to plug them in and wait for the beep beep beep done:)

t.n

[luc]

yep, in serie close to the enda nd in // to top them.
here is a pic with a 3S2P pack (3 packs of 2P Irate 2200)) with a miltiplug in serie and another pack 3S2P (IRATE 2600) with a multiplug in //.
perfect matching, always. O course not totally automatic (have to switch plugs), but nothing is perfect...:)

[pgleesonuk]

Using the excellent Lipo Balancers from "Suzanne", here is how I will configure all my packs.

The first picture is an E-tech 1200 3S1P pack with a connecter attached to tap from each cell.

These connectors are from multiplex, part no.s 787028 (Plug) 787033 (Socket) they are small and polarised and one connectors is enough for a 4S pack.

Paul

[pgleesonuk]

Second picture shows 4 Lipo Balancers wired to the matching connecter attached to the pack while charging.

It is a neater solution to have lipo balancers for each pack, I can't afford it! connectors are 2 euros, for up to a 4S pack, lipo balancers are 10 euros each.

If you want to charge more than 4S, you can simply add more connectors, and make more sets of 4 balancers

Paul

[RC_Pile-it]

Originally posted by pgleesonuk
These connectors are from multiplex, part no.s 787028 (Plug) 787033 (Socket) they are small and polarised and one connectors is enough for a 4S pack.

Paul, do you have all of the ground/earth/black/negative leads from the cells connected to a single pin on the connector? Also, would you see any reason why a 4-pin Deans would not work for a 3S1P pack? The Multiplex connectors don't seem to be readily available here in the States.

(I'm a mess without a spell checker)

[t.nollett]

good idea paul, where did you buy them.

t.n

[pgleesonuk]

Deans 4 pin connectors would be fine for a 3S pack

Each balancer is connected across a single cell.
They do not have all the black leads together.

Here is a diagram to hopefully make that clear.

In the UK, the muiltiplex connectors came as part of wing lead kit, but I have asked to order the connectors by themselves from Gordon at www.modelspot.com

Paul

[pgleesonuk]

Here is the diagram from Suzanne

[t.nollett]

that's where i got my evo9, must give him a call, also all cells are not the same as i found when flying my nippy powered mini geebee the bec on my cc 10 amp controller is set at 5volts for my 2s packs, i charge 3 packs on a harness and know they are equally charged, all are new packs but one pack drops below 5v under load at take off causing the bec to cut in and seems low in power the other two are great. must have a bad cell. striped it down and it was only .03v out of ballance.

t.n

[pgleesonuk]

t.n

The balancers work very well. I made the cells out of balance by discharging one slighty. The Lipo balancers worked well.

But only after I connected them to a current limited supply. My Apache 4 cell charger is very careful, and cuts the charge at 12.5V.

Paul

[mark_q]

Can anyone tell where to get 4 pole Deans in the US?

Thanks - Mark

[RC_Pile-it]

Originally posted by mark_q
Can anyone tell where to get 4 pole Deans in the US?


My LHS has them on the rack. They should also be available online. If you can't find them, try here:

Tower Hobbies (http://www2.towerhobbies.com/cgi-bin/wti0001p?&I=LXKX42&P=7)

[HELModels]

Suzanne,
This is neat, but I'm struggling with a few simple things.

If I have 2 cell pack, like a kokam 340mah/7.4v, or some other brand, I will need 2 of these balancers? Or 1 balancer per 2 cell pack?

Also, If I were to make this thing using your diagrams, are these components so small that the 25W soldering Iron I have would be too cumbersome? How do you solder on such a small components without making a mess of everything?

Thanks :)

[RMFISH]

Electrostorch,
I would say that you need some practice with a Weller Temperature controlled soldering iron. From the look of the boards in the pictures, all of the components are SMT and the standard Radio Shack 30W soldering iron will not be suitable.
Robert.

[HELModels]

I asked myself Why? The answer is that you need a way to sink or dissipate heat. How? Use an aligator clip on component to suck up the heat and use a very fine point(exacto blade?) on the soldering iron and use some Well-Aimed component cooling spray. Even if I used the $65.00 iron, heat sinking is still necessary.

I'm still not sure if one of these circuits is for 2 3.7v cells or 1 3.7v cell.

plus, Wouldnt my charger for 2 cells have a balancing circuit built in?

FMA lipo-202

[KreAture]

Great to see someone besides me enjoying EE designing too suzanne. I love PIC's though and might have done something similar using 1 PIC and X FET's for a x-cell pack.

Some of the better protection-boards have balancing buildt in though... May be cool :)

[RC_Pile-it]

Originally posted by pgleesonuk
But only after I connected them to a current limited supply. My Apache 4 cell charger is very careful, and cuts the charge at 12.5V.


Paul,

Are you saying that the balancers will not work correctly with the Apache? If the Apache stops the charge at 4.17 vpc, the balancers will not all start blinking before the charge is complete. If, however, 1 or more cells are more than .03v out of balance, the balancer(s) for that cell(s) should kick in and start blinking, if I'm reading this correctly.

Also, what are you using for a "current limited" supply? Could a ni-cad charger with a low milliamp rating be used? If so, what voltage would be needed?

One last question regarding the use of the multiplex connectors. What gauge wire do you need inside the pack? Since the balancers are only absorbing 165ma, would standard 26 gauge servo wire work? Or maybe 22 gauge would be more appropriate?

[HELModels]

Yeah, I can read too. Forgive me for asking for clarification amongst so many great and humble minds.

How would you "balance" one 3.7v cell anyway? "Balance" implies 2 regardless of your electron knowledge.

How about that!?

[luc]

you don't balance one cell, you balance cells to the same voltage. You can balance 3 cells charging them in //, as 1 cell. That's the example of the pack on the right, with 1 multiplug in //.

[pgleesonuk]

Originally posted by RC_Pile-it
Paul,

Are you saying that the balancers will not work correctly with the Apache? If the Apache stops the charge at 4.17 vpc, the balancers will not all start blinking before the charge is complete. If, however, 1 or more cells are more than .03v out of balance, the balancer(s) for that cell(s) should kick in and start blinking, if I'm reading this correctly.

Also, what are you using for a "current limited" supply? Could a ni-cad charger with a low milliamp rating be used? If so, what voltage would be needed?

One last question regarding the use of the multiplex connectors. What gauge wire do you need inside the pack? Since the balancers are only absorbing 165ma, would standard 26 gauge servo wire work? Or maybe 22 gauge would be more appropriate?

That's right. None of the balancers blink. I even tried putting an imbalance in the pack. (about 0.1v) still not enough.

However, when I connect the pack to a current limited 200ma power supply, after a while the balancers to their job.

I made my own current limited supply using a LM317T regulator and a couple of resistors. Very simple, I'll post a diagram if you want. To fully charger a 3S pack you need a DC supply of at least 13.85V (3x 4.2 +1.25(LM317T loss)) 4S needs 18V

I used servo wires to the multiplex connectors

Regards

Paul

[Suzanne]

Originally posted by ElectroStorch
If I have 2 cell pack, like a kokam 340mah/7.4v, or some other brand, I will need 2 of these balancers? Or 1 balancer per 2 cell pack?


For a 2s pack you need 2 balancers. Each balancer protects one cell or a group of PARALLEL cells.

So for a 2s2p pack, you need also 2 balancers, but for a 4s1p, you need 4 of them.


Also, If I were to make this thing using your diagrams, are these components so small that the 25W soldering Iron I have would be too cumbersome? How do you solder on such a small components without making a mess of everything?


If you have no prior experience with SMT, I suggest that you use standard components to build the balancers. The MAX921 is available in a DIP-8 package, and the 7401 or 7456 FET can be substituted with any 0,5A or better logic level n-channel enhancement FET.

Of course this will make the balancers larger and heavier, but if you built them into a charging harness and not into your pack, this won't matter.

Good luck!


@Paul,

when a charger cuts off early (i.e., at 12.5V), chances are that none of the cells goes above 4,20 Volts, so there is nothing to do for the balancers. Even 0,1V imbalance means that one cell is at 4.1 and two others at 4.2, so still nothing to do.

Try to discharge two cells to 4.1Volts, leaving the third one fully charged. You will then see that the third balancer will start to flash when you recharge the pack with your charger. As far as I know, the balancers work with any brand of charger. I can see no reason why they shouldn't.

Most of the chargers I have tested so far are quite conservative with the cutoff voltage, except for the Spectra and the Orbit, which is very precise due to software calibration. So, unless there is a major imbalance in your pack, I would expect exactly the behaviour that you described.

Over time, the imbalance will grow to the point were at least one of the balancers will start to flash near the end of charge, doing it's thing. It would be interesting for me to know after how many cycles this happens with your charger.

Obviously, the lower the cutoff voltage, the more cycles it takes to reach an imbalance where the balancers must become active.

Might be one more reason why manufacturers design an extra 0,1 Volts (or more) safety margins into their chargers.

The other extreme:

I use modified chargers that work with a 0,02% precision voltage reference. They cut off very very precisely at 4,20 Volts per cell. At least one balancer flashes for a few seconds whenever I charge any of my packs, telling me that a tiny imbalance is there even after a single cycle.

[Suzanne]

Originally posted by ElectroStorch
I asked myself Why? The answer is that you need a way to sink or dissipate heat.


I do not use a heat sink for SMT work. Each solder joint takes less than a second, so the components are safe.


Wouldnt my charger for 2 cells have a balancing circuit built in?


No.

I haven't seen a charger with integrated balancing logic yet. Also, I believe that the balancing logic should be part of the pack, not part of the charger. The problem is the number of wires required. For a 10s pack you would need a total of 11 wires (plus, minus, and nine center taps). Not very practical.

[pgleesonuk]

Here is the circuit to build a limited current supply.

I use 2 LM317's in parallel and 3W resistors as depending on the number of cells and supply voltage you may need to disipate some power.

For 200ma you need a resistor R1 of 6.2 ohms. I parallel a 12ohm and at 15 ohm resistor to get 6.7 ohm for about 190ma.

Note that the leg order on a LM317 regulator are not like a lot of other regulators!

I also included a 200ma fuse, just in case something goes wrong.

Always watch your lipo's as they charge!

Hope this may be of use to someone.

Paul

[KreAture]

Do not charge LiPos with that circuit. You need it to be BOTH voltage limited AND current limited for it to be safe.

If you use the normal LM circuit as a voltage-supply, set to 4.2v unloaded and use a transistor to limit according to a serial resistor on the battery - pole you have a good, cheap charger that will stop charging by itself simply as the current reaches cell leakage-current.

[pgleesonuk]

This circuit is ONLY to be used with a pack equiped with "suzanne's" lipo balancers.

It is not a charger by itself. Read back in the thread and this might become clearer

Paul

[KreAture]

OK, but still... I'd use a voltage-limiter too. That way it will stop charging about when it should...

[HELModels]

Suzanne,
You've been very helpful. Yes, I do intend to rig up a charging harness and will use the jumbo components. This does diminish the beauty of SMT and your design somewhat, but should serve my purposes.

SMT are tiny and I found a .pdf file for amateurs which will help if I eventually try with SMT

In the mean time One More Question:

reading the schematic- R2,R1A-R1D are valued at 100R? Is that 100 Ohms, what is the R?

R3 is 3M3, Is that 3.3 Mega Ohms?

R5 is 1M 1%, Is that 1 Mega 1% precise value?

R4 is 2M7, Is that 2.7 Mega?

[Suzanne]

Originally posted by ElectroStorch
[B]Suzanne,
You've been very helpful. Yes, I do intend to rig up a charging harness and will use the jumbo components. This does diminish the beauty of SMT and your design somewhat, but should serve my purposes.

SMT are tiny and I found a .pdf file for amateurs which will help if I eventually try with SMT

In the mean time One More Question:

reading the schematic- R2,R1A-R1D are valued at 100R? Is that 100 Ohms, what is the R?


R stands for Ohms. If you use standard components, I suggest that you replace all four resistors R1A-R1D with a single resistor of suitable power rating. For example, a 10 Ohms, 2 Watts resistor will give a max. balancing current of 420mA. The FET can switch several Amps, so no problem there.

R2 should be around 220 Ohms and sets the LED current.

R3 is 3M3, Is that 3.3 Mega Ohms?

R5 is 1M 1%, Is that 1 Mega 1% precise value?

R4 is 2M7, Is that 2.7 Mega?

yes, yes, and yes.

:-)

[houfek]

Ok...... I've read through the thread.....

So there is no way to use Suzanne's balancer on old series packs without taking them apart and adding the center taps...correct ?

Jim

[luc]

correct.

[KreAture]

Hmm, And idea suzanne... Add another LED or two that triggers on lov voltage. That way one can use the circuit as a safety monitor under discharge too. If any cell gets below a certain level, a led may light up. The led shouldn't drain the pack if set up correctly so it would need the pack to be abowe safe voltage to light up, while at the same time not light up untill it's closing in on the same limit.

Man, did that make sence ?

Over 4.1x volts = balancer in action
Less than 3.1 volts = low batt warning
Less than 3 volts = NO low batt warning anymore to avoid killing cell.

That made more sence :)

[HELModels]

I currently have little ability to design this myself, but I have always wondered if it is possible to build a circuit which could go between the ESC and Li-Poly pack for the purpose of tricking the ESC into cutting off at 6.0V when ESC is fixed to cutoff at 4.8V.

Somehow design it so when the pack drops to around 6 V, the add on circuit suddenly gets stingy and only delivers 4.8V or someother programmed value, causing the ESC to safely cut off.

One problem with that would be the same problem I am having with the Balancer:

These little chips are hard to get in small quantities!!! Forget finding them at RadioShack

An N-Channel logic level FET in TO-220 style?
MAX921CPA(Dip-8 version)?

[gedaso]

Try Futurlec for microprocessors etc in small quantities. They have max921 for USD $2.90 each. http://www.futurlec.com/

[Suzanne]

Originally posted by KreAture
[B]Hmm, And idea suzanne... Add another LED or two that triggers on lov voltage.

That was my first thought too, but then I decided against it because the balancers go into the pack and you wouldn't be able to see the LED in flight anyway.

Instead I have build special LiPo Low-Voltage monitors which work for 2 to 10 cells. They come either with a superbright LED or a piezo beeper and are user settable to 6,9,12,15,18,21,24,27, or 30V threshold. Supply voltage range is 5-42 Volts.

They are all SMT (except for the LED/buzzer) and weigh 1.5 grams. They are $15 each.

[Suzanne]

Originally posted by ElectroStorch
An N-Channel logic level FET in TO-220 style?
MAX921CPA(Dip-8 version)?

The IRF 3706 or IRF 3708 are suitable logic level FETs in a TO220 package. Way overkill for the balancers, but easier to handle than the tiny SMT packages.

But you will most likely find that all the parts are ridiculously expensive in small quantities.

In lots of 10, material costs in Germany are $10. In lots of 200, this drops to $5 for the current version of the balancers.

I buy larger lots (usually hundreds or thousands), so I pay considerably less. It's a certain risk, however, as I have now the equivalent of a few thousand bucks sitting on my workbench (PCBs, MAX921, IRF7401, LM4040, MAX985, LEDs, and all the other components for the balancers and the voltage monitors). But if all goes well, it will pay for a few more 10s4p LiPo packs for my Joker...

[KreAture]

cool suzanne.
I made my own buzzer too. The WIMP (Where Is My Plane?)
Search the forums and u will find it. Can't be made much smaller hehe. (I cheat though and make it microcontroller based. I love those little buggers.)

[HELModels]

O.K. I'm gradually locating these parts and learning bits and pieces at a time.

Looking at the data sheet for IFR3706 in TO-220, I see 4 leads:

1)Gate
2)Drain
3)Source
4)Drain

Looking at the schematic for Balancer, I see Q1 has 3 connections? LED side, LIPO Neg side, and R3/out.

How do the component leads correspond to the schematic?

TO-220 has 4 leads, would I just leave 4th connection open?

Thanks for your patience, Suzanne.

[HELModels]

O.K. I'm gradually locating these parts and learning bits and pieces at a time.

Looking at the data sheet for IFR3706 in TO-220, I see 4 leads:

1)Gate
2)Drain
3)Source
4)Drain

Looking at the schematic for Balancer, I see Q1 has 3 connections? LED side, LIPO Neg side, and R3/out.

Here is my Guess at how the leads correspond to the schematic:

R3/OUT connection = Gate?

LED side = Drain?

LIPO Neg side = Source?

TO-220 has 4 leads, should I connect the 2 drains together as you suggested(use all the pins for higher current) for using IRF7455?

Thanks for your patience, Suzanne.

[tekochip]

A TO-220 is huge. You'd be better off using a different FET that's still in an SMD package. This circuit just bleeds off current through the resistors, so you won't need a super-low ON resistance (like a motor controller does) or great dissipation.

[HELModels]

My only reason for using large components is soldering ease. I plan to use only when charging, thus weight is not the issue.

If functionality is reduced or might introduce some unwanted behavior in the circuit or risk any pack which might be hooked up, then maybe you could suggest another n-ch logic level FET in a package other than SMT/microscopic/nano size.

I have a pack which lost a cell and just replacing the bad cell without balancing is only inviting more trouble.

[Suzanne]

R3/OUT connection = Gate?

LED side = Drain?

LIPO Neg side = Source?


correct.


TO-220 has 4 leads, should I connect the 2 drains together as you suggested(use all the pins for higher current) for using IRF7455?


TO-220 should have only 3 leads, but if you have 4 leads (2 drains), you may connect either one or both, it doesn't matter for the low current involved.

[HELModels]

Suzanne,
I found a Power MOSFET rated at 60V dss, 0.19R Rds(on), 55A Id. It comes in a D2PAK which I should be able to solder. Comparing the data sheets for this FET and the one in your schematics, I see that the minimum Gate threshold voltage for IFR7455 is 0.2V - 2.0V and the one I have is 2 - 4V.

I dont know much about these FETs but it seems to me that Gate threshold is the minimum voltage which turns this thing on?

If that is so, then is 2V too high a threshold for the Balancer circuit?

[Suzanne]

Originally posted by ElectroStorch
Suzanne,
I found a Power MOSFET rated at 60V dss, 0.19R Rds(on), 55A Id. It comes in a D2PAK which I should be able to solder. Comparing the data sheets for this FET and the one in your schematics, I see that the minimum Gate threshold voltage for IFR7455 is 0.2V - 2.0V and the one I have is 2 - 4V.

I dont know much about these FETs but it seems to me that Gate threshold is the minimum voltage which turns this thing on?

If that is so, then is 2V too high a threshold for the Balancer circuit?

It should work, as the comparator used has rail-to-rail output swing, i.e., 4.2V to the gate of Q1. Since your FET is so powerful and size is not an issue, I suggest to use a 4.7 Ohms, 5 Watts resistor for R1. This will increase the max. balancing current to approx. 1A, increasing the protection level in case of mishandling or charger misconfiguration.

[Pontus Claesson]

Hello!

Maybe this question asked before....but do you think the MAX 965 can work instead of the MAX 921?
The difference is a little less precise voltage ref. (+/- 1,5%)
and the output is open drain.
The IC require typical 8 µA and a pull-up is needed for the
gate and therfore, some more µA will be sourced through the pull up. However, I will use it on a 4p10s and on 3p9s pack so I can afford some consumption, but the MAX 921 cost 2x as much as MAX 965. What do you pay for MAX 965? I have to pay 8EUR/pc!

[ClausT]

I have been able to get some MAX 931 from Farnell (~€4) - trade off is 2% precision instead of 1%, otherwise similar chip.

[Suzanne]

Originally posted by Pontus Claesson
do you think the MAX 965 can work instead of the MAX 921?


Yes, but keep in mind that the internal voltage reference is 1.235 Volts instead of 1.182 Volts, so you will have to modify the input voltage divider accordingly.


What do you pay for MAX 965? I have to pay 8EUR/pc!

I pay 2 Euro for the MAX921, and that is for the extended temperature version. I haven't tried the 965, but it should be considerably cheaper. Of course, I buy fairly large quantities.

You can buy 10 balancers, calibrated and ready to work, for 105 Euro including shipping to anywhere in the world. No point in paying more for the parts alone...

From your description it seems that you would need 19 MAX921ESA to do what you want. I can send you these for 60 Euro including postage and packing. PM me, if you want them, I have a few hundred in stock.

[HELModels]

Suzanne,
What a trip! I finally got all the parts and assembled everything onto a copper clad board. I put 2 balancers side by side and then had no Lipo to test it with so I used a 6.0V nicad to briefly test the overvoltage warning feature. This worked like a champ after I realized I had the Drain and Source reversed. I had to do some cutting of the board and lots of desoldering, but it is wired correctly. I did burn out an LED though by hooking up a high voltage and letting it sit until LED burned out. Could this be because the resistors in parallel are too small?

The problem for me now is how to adjust the potenitometer. When I test the resistance on either side I get a difference, but I dont know how big this difference should be. Is there a ballpark figure for what the resistance should be on either side of the pot?

Another question is I could only get 3.9 Ohm 5W resistors and this is mixed in parallel with 4 of the 10 Ohm 1/4W resistors. From what I have been forced to learn from this exercise, the load is primarily on the 10 ohm resistors since 10/4 equals 2.5 and this less than 3.9 Ohms. Arent the 4 10 Ohm resistors taking the brunt of the load and should therefore have higher Watt rating? If I just reduce the R1A - R1D in half would I double the balancing current? Should I also make sure the resistor in front of the LED matches the value of R1A?

Please excuse my lack of knowledge and thanks for all your help.

I've included a picture of some very sloppy soldering, which is why I was forced to use bigger components.

http://lister3.home.mindspring.com/copper.jpg

[Suzanne]

I did burn out an LED though by hooking up a high voltage and letting it sit until LED burned out. Could this be because the resistors in parallel are too small?

The LED has its own 220 Ohms or 470 Ohms resistor in series, so it won't burn out.


The problem for me now is how to adjust the potenitometer. When I test the resistance on either side I get a difference, but I dont know how big this difference should be. Is there a ballpark figure for what the resistance should be on either side of the pot?


you want to set the pot so that the balancer starts to blink at 4.21 Volts. You will need a precision voltage reference or very clean power supply and a digital voltmeter to make this adjustment. If you lack either of those, just set the pot to the center position. This is not perfect, but acceptable, as long as the resistors used for the voltage divider are 1% types or better.


Another question is I could only get 3.9 Ohm 5W resistors and this is mixed in parallel with 4 of the 10 Ohm 1/4W resistors. From what I have been forced to learn from this exercise, the load is primarily on the 10 ohm resistors since 10/4 equals 2.5 and this less than 3.9 Ohms. Arent the 4 10 Ohm resistors taking the brunt of the load and should therefore have higher Watt rating? If I just reduce the R1A - R1D in half would I double the balancing current? Should I also make sure the resistor in front of the LED matches the value of R1A?


Please remove the 10 Ohms resistors alltogether. You already have a max. balancing current of more than 1A with only the 3.9 Ohms. The 10 Ohms resistors would have to be rated for 2 Watts each, which they aren't.

The resistor in front of the LED limits the current through the LED to approx. 10mA. It has no other purpose. 220 or 470 Ohms are what you want.

Good luck!

[HELModels]

Suzanne,
Youre right about needing some Luck. I think I might have goofed with the 2.7M resistor on the Lipo + side. What I actually got was 2.74M @ 1% 2.74 is out of range by 2% +/- 1%:(
Is it possible to adjust the pot to make up for the .04M extra resitance I introduced?

I hooked up a 4.8V nicad and it discharged to the point where LED starts blinking. I probed the +/- side of the battery hook up and the voltage is up/down up/down and hard to read, but I adjusted the pot so that in that garbled readout was 4.21V. I realize this is not very accurate and if I had a better grasp of how this circuit is designed to work, then I would know what I am dealing with. Plus, I really dont want to risk a Lipo until I do understand this circuit a little better and how it actually operates.

For instance, Why are the pot and the 2 resistors on either side such different values? There is a difference of 1.7M on either side of the potentiometer. Is the voltage going into pin 3(IN+) of MAX921 2.64V, assuming lipo is at 4.2? and the remainder of the voltage goes on upto the FET is at 1.56V?

I'm going to fix those resistors R1A-D and R2. Like you said earlier 165ma of balancing current should be fine for new packs.

Thanks Again for your patience.

[KreAture]

Hmm. I just got a brand new 3s 1500 Kokam pack and after charging it I measured each cell:
4.28
3.89
4.28

The center cell is way off and the two others are overcharged :(

Is this normal for a brand new pack, not being balanced I mean ?

[Dave Hederich]

Originally posted by KreAture
Is this normal for a brand new pack, not being balanced I mean ?

According to AstroBob, it's not unusual:

http://www.rcgroups.com/forums/showthread.php?s=&threadid=184420

[luc]

not normal at all. All cells usually come at 3.8V and are balanced. Try to increase the voltage of the middle cell charging it only and monitor the discharge to see if imbalance reoccurs. If it does, bring it back to where you purchased it...

[KreAture]

Yeh
Have already charged middle cell 140 mAh... Still going at 0.7A chargerate at 4.18v now.

[DNA]

If you can measure each cell after you charged it,

"each cell:
4.28
3.89
4.28"

why didn't you measure each cell BEFORE you charged it?

[KreAture]

I did.
They were level before... (0.01v apart.)

[luc]

strange...maybe you have a bad solder in between...

[DNA]

Was that a factory assembled pack of Kokam 1500s?

If so, I'm wondering how you were able to measure the voltage
of each cell.

I would do what luc said and return the cells or pack to where
you bought it, before this happens again.

http://www.rcgroups.com/forums/showthread.php?s=&threadid=180434&perpage=15&pagenumber=1

[DNA]

Hello Suzanne,

Would it be possible to use your balancers along with a Kokam
Safety Guard on a pack of cells?

It seems this would be the ultimate solution to protecting lipo
cells while being charged. The Safety Guard would protect the
PACK from being overcharged, while your balancers would
balance each cell in the pack. It could be attached to the charger
and only used while charging instead of being permanently
attached to the pack, providing of course you have taps on the
pack to each cell.

[KreAture]

It was a factory assembled pack, yes. I measured it by using 2 needle-probes on my multimeter.

[KreAture]

I can't have measured it right before charging. No way could the cells have been so close. I think the middle cell must have been as good as flat before charging. Glad the Triton starts slowly and ramps up as it goes.

They are keeping within 20 mV of eachother now. 500 mAh out into the discharge cycle. Looking good. If I decide to keep the pack, I'll definately add center-taps for easier check of cell balancing. I will probably make myself a balancer I can hook on while charging. That should take care of the problem.