External battery charger circuit

[audioguru]

I bought 4 Energizer AA 2500mAh Ni-MH cells that came with a simple charger. The charger is made for charging discharged 2500mAh cells and just has a timer (8 hours) to stop charging. If charged cells are charged then the timer over-charges them for 8 hours and they get very hot. When I charge 600mAh Ni-Cad cells then in 3 hours they are very hot.

 

[blueroomelectronics]

What about simple trickle charging? It works with NiCADs C/20 is safe. What about NiMH?

 

[Hero999]

I thought it was C/10 for NiCADs and C/20 for NiMH?

 

[audioguru]

Energizer recommends a trickle charge current of C/40 for Ni-MH.
Panasonic recommends C/30.

 

[Hero999]

Even so, I find it quite hard to believe that a 2500mA battery can be damaged by charging it at C/10 as it'll only be dissipating 40mW at most.

 

[audioguru]

Even so, I find it quite hard to believe that a 2500mA battery can be damaged by charging it at C/10 as it'll only be dissipating 40mW at most.

How is your math?
C/10 is 0.25A.
The fully charged Ni-MH cell is 1.5V.
The dissipation is 0.25A x 1.5V= 375mW. There are plastic parts inside and the battery is not well ventilated.
The battery manufactures say that a trickle charge of C/10 will shorten the life of the cells.

 

[Hero999]

Dough, I got the decimal point in the wrong place.

 

[Blueteeth]

Hey there,

'Trickle charging' is out I'm afraid, it'll just take to long, I'm really after charge times <=8 hours. And with 2500mA AA's, C/30..thats a long time! I think anything shorter than 4 hours charge time would probably make problems for myself, so a nice safe charge time would be 6-8 hours. Plus that means I don't have to have a super big heatsink on any components for C/2 current (1200mA?)...a nice 500mA would be nice.

Right, I've continued the research, but I don't want to keep this thread going forever, so its time for phase two. As far as I can see, I have three options:

1) Use an 'off the shelf' dedicated charging IC, from maxim, TI or linear. Use the recommended application circuit. Medium part count, hassle with parts. But no real design issues.

2) Build a 'smart' controller similar in operation to the above, except from scratch, using a small 8-pin PIC with its A/D for voltage and temp measurement. Program shouldn't be too complex, its essentially just timers and A/D comparators. Only forseeable problem is the ADC may not have enough resolution, increasing the part count for analogue. (total difference in voltage over the entire charge is 0.25V....for an 8 hour charge its rise time is so slow that a 10-bit ADC will hardly show any difference when the 'hump' hits).

3) Non-intelligent charger. Similar to the one using comparators. Its a simple linear charger, that shuts off when either the dV/dt = 0, or, the temperature rises above a preset. Again, small, simple, low part count, but not full controll over whats happening.

I have idea's for all three of the above, including some clever 'sample/hold' tricks for detecting the roll-off in voltage. Using a PIC 'should' be trivial in this case...internal low freq osc, A/D for voltage/temp and a backup timer. But as always, using a dedicated chip, designed by someone as huge as 'maxim' is preferable.

I think we can all agree that cell temperature should be used as a safety cut-off, whether the voltage has rolled off or not, so either way, the charger WILL stop charging before anything gets too hairy. Albeit, not the primary charger termination point.

I still don't know exactly why many schematics on the net are so damn complicated? Tons of analogue, some logic, even huge microcontrollers...I guess if they are charging 16+ cells, and monitoring each one individually, then its justified, but surely everyone else can see how a lil PIC could handle this and so much more?

Blueteeth

 

[Nigel Goodwin]

Hey there,

'Trickle charging' is out I'm afraid, it'll just take to long, I'm really after charge times <=8 hours. And with 2500mA AA's, C/30..thats a long time! I think anything shorter than 4 hours charge time would probably make problems for myself, so a nice safe charge time would be 6-8 hours. Plus that means I don't have to have a super big heatsink on any components for C/2 current (1200mA?)...a nice 500mA would be nice.

Right, I've continued the research, but I don't want to keep this thread going forever, so its time for phase two. As far as I can see, I have three options:

1) Use an 'off the shelf' dedicated charging IC, from maxim, TI or linear. Use the recommended application circuit. Medium part count, hassle with parts. But no real design issues.

2) Build a 'smart' controller similar in operation to the above, except from scratch, using a small 8-pin PIC with its A/D for voltage and temp measurement. Program shouldn't be too complex, its essentially just timers and A/D comparators. Only forseeable problem is the ADC may not have enough resolution, increasing the part count for analogue. (total difference in voltage over the entire charge is 0.25V....for an 8 hour charge its rise time is so slow that a 10-bit ADC will hardly show any difference when the 'hump' hits).

Did you check the example I posted?, that only used an old PIC with an 8 bit A2D. As your application doesn't require dropping to trickle charging, you could use a simple resistor charger and switch it off when fully charged, an 8 pin PIC (like a 12F875) would be all that's required.

3) Non-intelligent charger. Similar to the one using comparators. Its a simple linear charger, that shuts off when either the dV/dt = 0, or, the temperature rises above a preset. Again, small, simple, low part count, but not full controll over whats happening.

I have idea's for all three of the above, including some clever 'sample/hold' tricks for detecting the roll-off in voltage. Using a PIC 'should' be trivial in this case...internal low freq osc, A/D for voltage/temp and a backup timer. But as always, using a dedicated chip, designed by someone as huge as 'maxim' is preferable.

I think we can all agree that cell temperature should be used as a safety cut-off, whether the voltage has rolled off or not, so either way, the charger WILL stop charging before anything gets too hairy. Albeit, not the primary charger termination point.

I still don't know exactly why many schematics on the net are so damn complicated? Tons of analogue, some logic, even huge microcontrollers...I guess if they are charging 16+ cells, and monitoring each one individually, then its justified, but surely everyone else can see how a lil PIC could handle this and so much more?

Generally commercial solutions tend to be crude and nasty, and shorten battery life - simply because of cost reasons. Designs on the net may well be over-complicated, because the designer is trying to do the best job possible, and they often include dis-charging circuits as well.

 

[Blueteeth]

Nigel, thanks once again for your input

The more I look at that link you posted, the more I am convinced I will just use that idea. The 12F675's are great little work horses, and will be perfect for that.

You are of course, correct about the schematics on the web...I don't why I didn't realise that DIY designers are essentially trying to do what I am, without 'skimping' due to cost. Well, they are either quite complex (rightly so) or...overly basic (old circuits for NiCads).

I will start tinkering today. After the huge amount of research I have done, looking at charge curves etc.. I am not convinced even a 10-bit ADC will cut it for an 'accurate' cut-off for voltage. That said, combined with a precision NTC thermister, a timer, and a semi-controlled current source, I reckon we're on to a winner. Maybe I won't get +98% charge, but it doesn't matter for this app, above 90% is dandy for me, the important thing is cell life.

So, one more quesiton to get the ball rolling... A current source. Charging 3 NiMH cells min 3.0v up to 4.5v with *roughly* 400-500mA. I am not keen on having a large heatsink on a transistor, or a LM317. So either I reduce the current (increase charge time) or find a more efficient method, without stumbling into the world of switch mode power supplies. An LM317 would be more accurate, but surely using a transistor as a current source, it would disapate less power? Your thoughts? bearing in mind, I can choose what PSU to use, wall plug, 5-15V +500mA.

Blueteeth.

Ps, Sorry for dragging this on, but I guess it might be useful to others wanting to design one, exploring the merits of various approaches.

 

[Nigel Goodwin]

Nigel, thanks once again for your input

The more I look at that link you posted, the more I am convinced I will just use that idea. The 12F675's are great little work horses, and will be perfect for that.

You are of course, correct about the schematics on the web...I don't why I didn't realise that DIY designers are essentially trying to do what I am, without 'skimping' due to cost. Well, they are either quite complex (rightly so) or...overly basic (old circuits for NiCads).

I will start tinkering today. After the huge amount of research I have done, looking at charge curves etc.. I am not convinced even a 10-bit ADC will cut it for an 'accurate' cut-off for voltage.

Don't forget you aren't looking for a specific voltage, just a detectable drop in the voltage. You might look at the other thread running about the Oshonsoft PIC charger?.

That said, combined with a precision NTC thermister, a timer, and a semi-controlled current source, I reckon we're on to a winner. Maybe I won't get +98% charge, but it doesn't matter for this app, above 90% is dandy for me, the important thing is cell life.

You should get full charge with the dv/dt charging, the voltage only starts to drop once it's fully charged.

So, one more quesiton to get the ball rolling... A current source. Charging 3 NiMH cells min 3.0v up to 4.5v with *roughly* 400-500mA. I am not keen on having a large heatsink on a transistor, or a LM317. So either I reduce the current (increase charge time) or find a more efficient method, without stumbling into the world of switch mode power supplies. An LM317 would be more accurate, but surely using a transistor as a current source, it would disapate less power?

Makes no difference LM317 or transistor, they still have to dissipate the same amount of power 'somewhere', using PWM could make it less, but at the cost of more complexity.

 

[Blueteeth]

Ahh thanks again nigel,

About the 'specific drop in voltage'. This is only going to be for NiMH, so the drop in voltage I'm loking for, apparently can be as little as 2mV/Cell. That said, I will just look for a flat voltage dv/dt = 0, as it levels off, just like NiCads...but that will be just before said 'voltage drop'.

Now that I feel confident about the 'control/shutoff' part (god bless software) I'm up for getting the current source right, and you've guessed it, my analogue knowledge is crap to say the least, uni skimped on that part. Of course I do know transistor theory etc..

One thing that confuses me, and is probably obvious, but it relates to the power supplys voltage. Say I use an unregulated supply, capable of 600mA. Using an LM317, the input voltage would have to be the maximum voltage required by the cells (4.5V) + the drop out of the reg, 1.5, + an extra volt for good measure...thats 7v. For 400mA, power disipation would be 2.5*0.4 = 1 Watt, . Obviously I can't reduce the current of the power supply, but what would the minimum voltage be? Of course, I need it to be 4.5V or above, perhaps with a transistor + resistor, I would get a smaller voltage drop, of say, 0.5...which would make the disipation 0.5*0.4 = 200mW.

I'm only worrying about this since size is critical, and any 'heatsink' components would have to be the simple clip-on ones.... perhaps i'm over tihnking this, but I've dealt with voltage regulated PSU's for so long, thats its thrown me a bit. Again, this isn't for 'efficiency' purely reducing heat generation and therefore footprint. I wonder if LDO voltage regulators similar in nature to the LM317 could be used? Perhaps with a power supply of roughly 5v? bah I'll get the breadboard and multimeter out...

Blueteeth

May have found a convenient solution:
**broken link removed**

And I made a booboo on the maximum power disipation....input = 5v, minimum output voltage = 3. so it'll be 2*0.4 = 0.8watts? Assuming the cells won't go about 1.5v each (they probably will) and i can get a dropout voltage from the above current reg of <0.5v. A lot of if's and buts, but I knew a MOSFET would be in there somewhere

 

[audioguru]

You decided on having an unregulated input voltage to the LM317 so that it still has enough voltage across it to charge the fully charged 4.5V cells.
But you forgot that the cells are only 3V when charging begins so the LM317 will have 4V across it and at 400mA then would dissipate 1.6W.

The dropout voltage rating is when it has already dropped its output 0.1V.
At 400mA, for the LM317 the typical dropout is 1.8v.
For the LM1086 it is 1.0V.
For the LP8340 it is only 0.2V but it is in a tiny surface-mount package and the datasheet doesn't say its thermal resistance.

 

[Blueteeth]

You are of course completely right there audioguru...

I forgot about the 3v minimum batt pack voltage (1.0v per cell). The LM317 would be good for testing, but I just edited my post (seconds before you posted) with what seems to be a relatively simple - still more parts than a LM317 though. As long as I can avoid a monster heatsink I'll be happy. Although, it is not all bad, as it will be in an aluminuim box (which already contains the original circuit that needs retro-fitting with the charger), so many an extra hole to drill.

Cheers for picking up on that though, my maths isn't what it should be.

BT

 

[Hero999]

Switching supplies don't have to be complex, a Black regulator can provide a constant current output, just remove the zener or keep it there if you want to limit the open circuit voltage.
**broken link removed**
https://www.romanblack.com/smps/a04.htm

No nasty bulky heatsinks or expensive ICs required!

 

[Blueteeth]

Thanks audioguru, and hero999.

I'm 'iffy' about using a switcher, as its wall plug powered, and with all the problems with switching circuits..(this is inside a box woth lots of RF stuff) I tihnk I'll stick to linear. But cheers anyway, that looks like a wicked circuit!

Ok, back to the drawing board. Bizarely, I tihnk Have a good plan for almost everything in the circuit...temp measurement, the PIC code (its in pseudo code at the mo) and voltage measurement using the ADC, as well, as the comparator creating a make-shift slope A/D as a more accurate measurement of batt voltage (initial tests show I may be able to detect 1mv drop, so a 10mV should be easy).

The 'problem' area, is always my weak point. The current source. Again, as I have mentioned, I can choose what 'wall plug' powers upply to use (unregulated, any voltage, I<500mA) and with 3 NiMH AA's, thats min 3.0v, max 4.5V. I sitll want to avoid the wonderful convenience of the LM317, purely because of the dropout voltage...at 500mA, I will certainly need a heatsink on it.

I cannot afford the space for a heatsink, so as hero99 suggested, either a switcher....or, what I'm really after is a MOSFET/transistor constant current source. Now, my analogue knowledge is crap, and I have been doing the research/maths to work out all the specifics.

What makes it 'slightly' harder is, I need 3 different current values from the source: 500mA (full charge, 30-40mA (trickle) and really low (essentailly 'Off' but <5mA is ok). So, after looking around, I was thinking something like this:

https://radiolocation.tripod.com/LEDdimmer/Schematic.jpg
https://radiolocation.tripod.com/LEDdimmer/LEDlampDimmer.html

The voltage divider (the pot) could be replaced with a couple of uC I/O's and resistors, for a simple but effective way to change the current. I have that part down. But I stil don't fully understand how to make this 'high-side' so that the batteries -V terminal is GND meaning the charger, battery and app circuit ALL share the same ground.

Why are PNP's used for high side, and NPN's used for low side? Is it purely because of the voltage required to turn on/off PNP's and NPN's? (or in this case, MOSFETS). I can't say they'll me much of a voltage drop across the C and E, so power disappation in the main transistor/mosfet should be tiny (ergo...NO heatsink). So, could I use an NPN transistor/N-channel MOSFET to source current, or must I abode by the norm and go PNP?

Of anyone can shed some light on this, I'd be grateful. I'm learning, but slowly. And right now, all I have is NPN's, and N-channel (logic level) mosfets.

All I really need is a simple constant current source, with little power disappation (no heatsink for 500mA) and the ability to change the current by the use of resistors tied to logic I/O's (or switching transistors).

Thankyou.

Blueteeth

ps. Just before I make an order, will I need PNP's? or P-channel MOSFET's?

 

[Blueteeth]

Hm, ok,after a bit more 'research' (wiki) I ithnk I know now why PNP, or P-channel MOSFET must be used to source current, high side....is it the voltage required at the base/gate? So Vbe must be +0.6V, or in the case of PNP, -0.6V. So If I used an NPN on the highside, with the battery charging connected to its emitter...the emitter voltage would be roughly the collector voltage...which is connected to VCC, so I would need to boost this above VCC?

Where as a PNP would need a base votlage of Ve - 0.6V so no boosting required?

I'm still confused about how to work out the power disappation in the transistor....obviously say we have 0.5A flowing through it...voltage drop? From VCC (charging power) to GND it'll be PNP, collector-emmiter, battery, power resistor (1-3ohm 5W). At the start, the battery voltage would be roughly 3v. So, for a 6V powersupply...and a 2 ohm resistor, thats 1v across the resistor, 3V across the battery leaving 6-4 = 2v across the transistor? 2*0.5 = 1watt power disappation max. Or should I choose a series resistor with a value to drop most of the voltage?

Again, this is still in the 'LM317 or MOSFET/bipolar' decision stage :/ Any help (Nigel, I'm looking your way) would be appreciated.

Rant over.

Blueteeth

 

[Nigel Goodwin]

I'm still confused about how to work out the power disappation in the transistor....obviously say we have 0.5A flowing through it...voltage drop? From VCC (charging power) to GND it'll be PNP, collector-emmiter, battery, power resistor (1-3ohm 5W). At the start, the battery voltage would be roughly 3v. So, for a 6V powersupply...and a 2 ohm resistor, thats 1v across the resistor, 3V across the battery leaving 6-4 = 2v across the transistor? 2*0.5 = 1watt power disappation max. Or should I choose a series resistor with a value to drop most of the voltage?

Again, this is still in the 'LM317 or MOSFET/bipolar' decision stage :/ Any help (Nigel, I'm looking your way) would be appreciated.

A transistor used in a linear fashion will obviously dissipate quite a bit of heat (as would anything else), which is why people have been suggesting a switching supply option.

 

[Blueteeth]

A transistor used in a linear fashion will obviously dissipate quite a bit of heat (as would anything else), which is why people have been suggesting a switching supply option.

Hmm, I guess I'm a bit stuck then. Either reduce the charging current to a level which would negate the need for a heatsink....or go with SMPS :/ It's funny as I thought the hardest bit would be charge termination...but PIC's are wonderful little things for that. I suppose I *could* compeltely change the design...and have the charger, with current source 'external'.

As the batteries are pemenantly 'in system', in a box, with all its electronics...I would need at the very least three connections form the charger to the box/batteries. Power, Gnd, and the thermistor. but then charge status idication would have to be in this external unit as well....bah.

Hmm perhaps I could go back to an ealier idea, as a comprimise between an adjustable linear reg (configured as a CCS) and a SMPS....An LDO reg? To reduce the voltage across the regualtor as much as possible. Or am I missing the point completely here? Now, I have a small 8-pin PIC in there, which, in fairness would be doing nothing most of the time, so PWM could be used. Also....as I'm measuring battery voltage, I would have to stop the charge current, while this takes place...meaning the transistor/reg would have a bit of time to cool down a bit.

So many options....but I dont want to be limited by the parts I have, so I guess I'll order some PNP's, MOSFET's to do some experimentation. Hell, even 5v supply would give 1.5-0.5 between supply and battery voltage, and 0.5 of that could be 'absorbed' by a resistor. 1V * 0.5V = 0.5W! (won't need a heatsink for that on a TO-220).

I'm starting to think I'm out of my depth here. Nigel...PNP's? yes or no? lol Farnell awaits your reply

BT

 

[Nigel Goodwin]

I'm starting to think I'm out of my depth here. Nigel...PNP's? yes or no? lol Farnell awaits your reply

I would use PNP's, but probably mostly because I rarely use FET's