Safe auto sensing charger for serial LIPO packs?

[vintage1]

I knocked this up in an idle moment.

The idea is that you equip a pack with a multipole charging connector

This is completely separate and lower current than the main power connector.

A pair of pins are connected by a resistor that sets the charge current.

The number of cells is defined by the number of pins occupied on the connector.

Its a brute force system. The current is set by the constant current MOSFET at the bottom of the schematic. This current flows down the totem pole of clamps irrespective of whether a pack is connected or not.

Each cell is clamped to a maximum of 4.2v, and the LEDS indicate the clamped condition.

If a pack is plugged in. LEDS wll go out until cells reach the correct voltage.

Unconnected clamps will display the clamped condition.

'All LEDS ON" means a charged pack, or no pack at all, is connected.

The various diodes are to protect against malconnections.

I'd appreciate feedback on this - I won't get round to building it, but its offered as an experimental starting point for anyone who wants to develop such a system, and as a potential way of standardising on pack charging circuitry with complete auto-sensing.

It could be done a lot better with fast switching digital circuitry, but thats beynd my skills these days.

Semiconductor part numbers are not given, mainly because I don't know the useful ones these days.

The Op-amp that controls the current sensing needs to be able to operate with its inputs at or near ground potential: I seem to remember P-FET input types can do this reliably.

[Mr.RC-CAM]

As the series cells climb up the ladder, how does the Vref behave? It would need to be offset by 4.2V at each higher cell, or virtual ground would need to be tricked. The voltage divisors on the Vref and the comparitors' return path must be your solution to this, but I got lost trying to understand the overall mechanism. How is this done?

With the 1N4148 isolated MOSFETS, maybe a high value pulldown resistor should be added to the gates.

I could not recognize any circuitry that provides the initial low trickle charge current for fully discharged packs (<2.7V per cell). Do you have this covered?

What did you have planned for the Vcc voltage value? Something a few volts higher than the max pack voltage?

Are there advantages to your method versus using a traditional LiPO charger circuit with isolated balancers like Suzanne's? I like the idea of the latter, since it would ensure balanced cells under all conditions. Or so I think.

I would enjoy hearing how the project turns out. Please keep us posted.

RC-CAM

[vintage1]

As the series cells climb up the ladder, how does the Vref behave? It would need to be offset by 4.2V at each higher cell, or virtual ground would need to be tricked. The voltage divisors on the Vref and the comparitors' return path must be your solution to this, but I got lost trying to understand the overall mechanism. How is this done?

It all comes out right in the end. Do the maths!

With the 1N4148 isolated MOSFETS, maybe a high value pulldown resistor should be added to the gates.

Good point. The diodes will probably leak enough, but a gate to source resistor no bad thing.


I could not recognize any circuitry that provides the initial low trickle charge current for fully discharged packs (<2.7V per cell). Do you have this covered?

Nope. Don't let em get that low!

Would be possible to ramp down current with some voltage detection if needs be.


What did you have planned for the Vcc voltage value? Something a few volts higher than the max pack voltage?


basicaly yes. For 5s maybe 25-30v or so.


Are there advantages to your method versus using a traditional LiPO charger circuit with isolated balancers like Suzanne's? I like the idea of the latter, since it would ensure balanced cells under all conditions. Or so I think.

Its basically the same thing, except the current sense on the multiple connector. Taking having to dial up current out of the equation is a big step towards idiot proofing.

You could essentially replace the totem poles with suzannes circuitry.



I would enjoy hearing how the project turns out. Please keep us posted.

RC-CAM

I won't be building it. Its just a discussion document. To get ideas and hope that someone like Astro or Suzanne takes heed.

[Comatose]

Vintage, I'm a digital type guy so the implementaton doesn't strike my fancy, but having a pair of resistors to tell the charger "hey, I'm a 4s pack, charge me at an amp" is brilliant, and cheap. If you could get the system adopted by the speed control manufacturers, and the pack manufacturers, then the "i burned down my house charging lipos" would largely become a thing of the past. The cost would be extremely minimal, as all of the afforementioned have to have A/D circuits anyway.

Would you propose doing this with a new style of connector (2 power pins, and then four or two resistor pins (common ground) or with two connectors? The former would be perfect if you could bring some of the connector manufacturers around, the latter would be easy to implement.

[vintage1]

Go and do it digitally!

The real thing as you say is to get just enough info IN THE PACK at super light weight to allow an intelligent charger to 'do the right thing'. Resistors are light and cheap. If no resitor detected well - charger defaults to 'manual mode'.

My gut feeling is to make the charge plugs standard, and let people put whatever gold plated MegaBux Corp connectors they want on the actual POWER leads, so even if these are - as shown - duplicated by the charge leads, its no big deal.

The presence of voltages on the pins indicates that cells are present: No need to tell the charger how many, since it has access to every one individually anyway. open circuit cells (ones where here isn't a cell) get treated as 'fully charged' anyway, and thats it.

You could get more sophistocated with digital I am sure.

Astro could modify one of their chargers to do all this very simply I would imagine.

[Hovertime]

The birth of Safe&Smart charging standard!!! :)
Vintage1-please consider sending PM's to all the high players to have a look here.

[vintage1]

I left a marker in the batteries forum, but so far no one has followed it.

I find that as questions come up elsewhere, if people refer back to here, the idea sort of will naturally take off.

If its a good idea whose time has come, it will eventually happen.

Of course it will be trumpeted as invented by XYZ pack vendor in conjunction with PQR charger manufacturer, but who cares? If it saves someone from a skin graft, its worth it.

[chimchim]

Good idea vintage1. Had a similar thought last spring. If the battery/charger mfgrs. jump on board, I would suggest a chip solution which could store all sorts of pack related information. Cell count, 1C charge current, history of last 50 charge/discharge cycles ( sampling every x seconds during periods of higher than leakage current ) including voltage, current, temperature. Then interface with a charger to analyze the packs performance. Keeping the data with the battery would allow you to use multiple chargers and still track each pack.

Steve

[Dan Baldwin]

I like a lot of the way you've done things. It's an intriguing idea, and my first thought was similar to Comatose's, but then it wouldn't be a balancer. I don't really like the idea of dumping all that heat if you have a 4 or 5 cell charger and are only charging 2 cells, especially if it was running off a 12 volt source, as virtually all new chargers must do. On the other hand, a person could build a charger that works with the packs they fly. For me 2 cell and 3 cell would be all I would ever want.

It looks like you're using a 5.6 volt zener to derive your 4.2 volts reference. It doesn't seem to me that that would be accurate enough. Of course if you've seen any of my threads, you know that I use a TL431 for all voltage references, which would eliminate one of the op amps.

Dan

[vintage1]

I like a lot of the way you've done things. It's an intriguing idea, and my first thought was similar to Comatose's, but then it wouldn't be a balancer. I don't really like the idea of dumping all that heat if you have a 4 or 5 cell charger and are only charging 2 cells, especially if it was running off a 12 volt source, as virtually all new chargers must do. On the other hand, a person could build a charger that works with the packs they fly. For me 2 cell and 3 cell would be all I would ever want.

It looks like you're using a 5.6 volt zener to derive your 4.2 volts reference. It doesn't seem to me that that would be accurate enough. Of course if you've seen any of my threads, you know that I use a TL431 for all voltage references, which would eliminate one of the op amps.

Dan

All good points Dan.

Its over 20 years since I did any electronic design in anger, and I think you can see that in the design. I was strictly analogue, and switched to software when I coldn't make a decent living doing it :D

I agree on ALL your points, and that is ecxatly the sort of comments I wnated to hear. Pull it to pieces, till the bare bones of what is worthwhile are left, and then rebuild the idea using modern technology.

In no particular order, I used a 5.6v zener cos they are stable temperatire wise, and the pot sets the ref voltage to 4.2v. Even then you have resistor tlerances on each clamp...plus op amp input offsets, so I would not expect teh final out put to be better than maybe +-50mv accutaret on clamping. Digital gubbins would improve this. No reason not to have half a dozen sampled A to D inputs coming of the chain...


Yes, lots of heat generated by the analogue design. Switching clamps a la Suzanne would be better. I don't know how to design with them. Anyway, if the clamps stop dissipating heat, it gets passed down to the single linear current regulator. Unless you go totally switching everywhere, you are stick with teh heat.

I make no apologies - this was indeed a brute force and ignirance design.

I wanted to show how it COULD be done, in no way is this put forward as the best or only way to do it.

Chimchim: I think that when (and I am sure we will) see LIPOS used in things like cars, then you will indeed see huge batteries assembled from 'smart packs' with inbuilt history exactly as you describe - and more probably - things like overtemperature sensors and lord knows what else, but our market is price sensitive and small. And weigh conscious.

I feel that a simple passive component like a resistor and a multipole socket is really all we can afford to add. Maybe as someone else suggested a second resistor to ID pack cell count. But I don't think its necessary.

The design basically addresses three issues

- Pack balancing on series packs can be made perfect with taps. Other systems already do this, and although I am by no means convinced its that necessary, it's included anyway.

- User setting of pack charge current is elminated, at the onboard cost and weight penalty of two pins in a connector and a resistor.

- cell count detection is essentially automatic, based on the presence of cells on the charge plug pins.

The only issues unaddressed are over discharge, and pack over temperature.

Short (sic!) of puttung a semiconductor switch in series with the pack to switch it off, I can't see a simple cheap and efficient way to do that. Its probably better handled in the ESC. And if a pack can go up in flames just veing left on te seat of a car in the sun, there isn't much one can do to protect against THAT electronically. micro fan on te pack maybe ? :)




.

[Dan Baldwin]

All good points Dan.

Its over 20 years since I did any electronic design in anger, and I think you can see that in the design. I was strictly analogue, and switched to software when I coldn't make a decent living doing it :D


You should never design in anger.


I agree on ALL your points, and that is ecxatly the sort of comments I wnated to hear. Pull it to pieces, till the bare bones of what is worthwhile are left, and then rebuild the idea using modern technology.


Okay, I have some more rending and tearing comments. One problem is that I don't see any end of charge indication, and I don't see an easy way to implement it. If all LEDs are on, it only tells you that the charge has reached the constant voltage stage.



Yes, lots of heat generated by the analogue design. Switching clamps a la Suzanne would be better. I don't know how to design with them. Anyway, if the clamps stop dissipating heat, it gets passed down to the single linear current regulator. Unless you go totally switching everywhere, you are stick with teh heat.


You're right. The heat in this circuit will be going somewhere; either into the balancers, or into the current regulator. Probably better to distribute it over several transistors rather than concentrate it in one. Switching clamps like the Suzanne balancers don't eliminate or even lessen the heating. They just move it. Instead of the transistor getting hot, some resistors do. When you need to dump power, there isn't an efficient way to do it.


I make no apologies - this was indeed a brute force and ignirance design.
I wanted to show how it COULD be done, in no way is this put forward as the best or only way to do it.



Good lord I should hope that you make no apologies. Thank you for posting this information, and even if the exact schematic isn't used, I think the underlying ideas may well be.


- Pack balancing on series packs can be made perfect with taps. Other systems already do this, and although I am by no means convinced its that necessary, it's included anyway.

- User setting of pack charge current is elminated, at the onboard cost and weight penalty of two pins in a connector and a resistor.

- cell count detection is essentially automatic, based on the presence of cells on the charge plug pins.



As far as pack balancing, I agree with you. I don't balance my packs, but I posted a high current balancer for those who do want to balance their packs. At present I don't have any lipoly packs with taps for checking or balancing, but in the future I plan to buy packs that have taps.


Dan

[vintage1]

I think Suzannes balancers are in fact switchers and don't get that hot.

The theory on tis charger is that you wait ill all the LED's are on, and the pack is charged.

It may be only 95% charged, but what the heck...

Personally I don';t reel we need to hammer the last eelectron into a LIPO for most cases. A simple charger that simply gets to 4.2v and holds that across the cells is probably good enough for the sort of market a charger like this is aimed at.

I.e. a big black box, you plug your packs into, and when all the lights come on, fly the pack.

Thats the 'vision' of what a 'sports' charger should be.

[Dan Baldwin]

I think Suzannes balancers are in fact switchers and don't get that hot.

The theory on tis charger is that you wait ill all the LED's are on, and the pack is charged.

It may be only 95% charged, but what the heck...

Personally I don';t reel we need to hammer the last eelectron into a LIPO for most cases. A simple charger that simply gets to 4.2v and holds that across the cells is probably good enough for the sort of market a charger like this is aimed at.

I.e. a big black box, you plug your packs into, and when all the lights come on, fly the pack.

Thats the 'vision' of what a 'sports' charger should be.

Yes, the Suzanne balancers are switchers, but the power that is being dumped in order to clamp the voltage, is dumped into an array of resistors. The Suzanne balancers have no ability to do anything with that power other than to turn it into heat. If you are clamping at 200 ma you have to get rid of .84 watts, and that .84 watts is turned into heat, even with the Suzanne balancers. To quote a very wise man-do the math.

The charts that I've seen show that a pack has about 80-85% capacity when it enters the constant voltage part of the charge, and some high capacity cells (whos name will not be mentioned) even less. Of course that's still not bad, and you could also just wait for a given length of time after all the LED's came on and you could get whatever level of charge you wanted. I'm just used to having a nice bright LED tell me that I'm not done charging yet. Call me spoilt if you must.


Dan

[vintage1]

Mmm. Should be possible to monitor the difference between clamp current and tail current on the lowest cell, and when that is below some fraction of the charge current, light a green light.

Or perhaps monitor each clamp and light the lamp only when its passing 99% of the tail current..

[Dan Baldwin]

Mmm. Should be possible to monitor the difference between clamp current and tail current on the lowest cell, and when that is below some fraction of the charge current, light a green light.

Or perhaps monitor each clamp and light the lamp only when its passing 99% of the tail current..


Yup. That should work. You could add a current sense resistor to each clamp, and compare the voltage on that resistor to the voltage on the current source. Seems like it would take another op amp per cell minimum though.

Dan

[hul]

this circuit could be used as a balancer by leaving out the part that sets the current and by setting Vref a bit higher, right?

Hans

[vintage1]

Yes. It was overtaken by 'charge guards' and the like.

The 'suzanne' pack balancer is better though.

[hul]

The 'suzanne' pack balancer is better though.
better in what way? I didn't have much luck with it.

I don't see a combined balancer / charge guard out there (except Thunder Power's if used with their charger).
The existing charge guards don't stop my charger and burn out when they try to.

I think this circuit has the potential to be a universal balancer / charge guard without getting too complicated. Would need an additional opamp (they come in pairs anyway) per cell, a power FET and a few other bits.

Hans

[vintage1]

Well, feel free to have a crack at it...I would have preferred it not to be a bunch of linear power hungry stuff, but whilst it may be brute force, it should work...

I don't see why it needs an extra opamp though. What the opamps are is essentially balancing clamps anyway in conjunction with the FETS.

[hul]

I would use this as a balancer (set at ~4.23V) not as a charger. No issue with power hunger that way, but clearly not what was intended originally.
The second opamp interrupts charging when cell voltage goes too high (~4.30V).

Hans

[vintage1]

But it never does. The clamp prevents this?

[hul]

I was going to limit balancing current to about 500mA or so to keep the FETs cool and stop the charger instead.

Hans

[vintage1]

Good idea.
Sorry for being obtuse.

[hul]

built one of the voltage clamps for testing and found that they're not suitable as balancers because gate-source voltage needs to be higher than cell voltage (4.2V) even for moderate currents (for the FET I used anyway; MTP3055). More than cell voltage is not available for the balancer of the top cell in a pack.

Hans

[vintage1]

...

[vintage1]

You are right.

The easiest thing to do is to have a slightly higher power supply for the drive circuitry.

You could simply put a FET, with drain and gate connected together, in series above the top FET.

Or use lower on voltage FET.

[hul]

could probably do that, but I wanted to use it as a balancer with no external supply.

Hans

[vintage1]

Oh. I apologize.

Replace the FET's with a NPN power transistors then - 3055 should work. Cheap and plentiful.

You should put a base emitter resistor in as well to limit 'off' leakage current, and be careful that the current gain is adequate - darlington pair might be in order.

[hul]

it's getting quite similar to Dan Baldwin's balancer then. Perhaps I should just keep using that and concentrate on building a charge guard that can cope with my charger.

[hul]

Schulze makes this 'safe auto sensing charger for Lipo packs':
http://www.schulze-elektronik-gmbh.com/lipoc-e.htm

This is what it says on their website about their balancing cable (which is permanently attached to the pack):
This cable features three more pins/leads than usual. The electronics of the Schulze LiPoCard use these additional terminals to detect the type of battery connected to it, and the permissible charge current for that battery type.

Sounds just like what Vintage1 had in mind.

Hans

[hul]

This cable features three more pins/leads than usual. The electronics of the Schulze LiPoCard use these additional terminals to detect the type of battery connected to it, and the permissible charge current for that battery type.
they use a resistor to set charge current, 1 ohm per mA; the third pin sets battery type, LiIon or LiPo

Hans