An "open source hardware" smart charger?

[dulridge]

For the record, here's what I'd want to charge with it.

NiCd
D cells (4Ah)
C cells (Can't remember)
AA cells (about 500mAh but don't really care about those)
Stick of 3 Sub-C (not too bothered about this)

NiMH
D
C
AA
AAA
Pack of 6 Sub-C (Optional) - Tamiya connector.

Li-ion (Optional)
RCR2 - protected
CR123 - unprotected and protected
17670 - unprotected.
18650 - unprotected.

Lead acid (optional)
6V 5Ah SLA
12V 4Ah SLA
Car and motorcycle batteries occasionally (Very optional)

The 12V stuff needs more than a 12V PSU at which point cheapo conversions of PC PSU's are no use. If the -5V and -12V rails produced any useful current one could mess around with those but current probably too low to be useful for large capacity batteries.

A module for each cell/battery type?

Pulling a random AT PSU off the pile, it'll give me 8A at +12V and 0.5A at -12V and -5V and 20A at 5V. It might even actually produce those currents though I'd knock 25% off for safety. Melting PSU's smell bad!

That gives us 6 real A at 12V and given the price of PCs with AT PSU's in them getting a couple for their PSU's would be no problem. A design allowing multiple PSU's to be plugged in for silly currents would be nice though I suppose at least some components have to be specced for the maximum total amperage.

32A at 2V or thereabouts should let you charge 8 NiMH cells independently at 4A. Will anyone want more than that, either current or number of cells.

If we pull that, or most of it, from the 5V rail (As PC PSU's usually produce more than half their rated current at 5V) we still have the 12V rail for the things that need more volts.

I imagine the controller won't draw significant current at any voltage.

More random thoughts. As you may have guessed, i like the idea a lot.

I take the point about a computer being an option rather than a necessity. Makes sense and is a few less fan running unless you want them to.

 

[Peepsalot]

I imagine the controller won't draw significant current at any voltage.

Yeah I'm thinking maybe a few mA at most? If someone was using some 12V-only power supply, it might not be unreasonable to just use a simple two resistor voltage divider to power it.

 

[koala]

If your using PIC then you need a voltage regulator because 12V maybe too high for PICs and many other CMOS type controller.

 

[Skavoovie]

Pulling a random AT PSU off the pile, it'll give me 8A at +12V and 0.5A at -12V and -5V and 20A at 5V. It might even actually produce those currents though I'd knock 25% off for safety. Melting PSU's smell bad!

That gives us 6 real A at 12V and given the price of PCs with AT PSU's in them getting a couple for their PSU's would be no problem. A design allowing multiple PSU's to be plugged in for silly currents would be nice though I suppose at least some components have to be specced for the maximum total amperage.

In my opinion, a PC power supply is a REALLY unsafe choice. Computer power supply manufacturers grossly overstate their power output in a climate of more-or-less no industry regulation.

Many are rated for peak output at 25 degrees C (not a realistic operating temperature). Any non-computer use of PC power supplies I would de-rate to around 25% to (maybe) 50% of their rated output. They make a handy, cheap DIY bench supply, but not for higher power applications.

I recommend sticking to a 12V nominal input (10~14V). The poor man's best method of obtaining a high current, low noise 12V power supply is still an lead acid car battery (in a well ventilated area only). Good quality switching power supplies would be the other option, but aren't cheap. Or you cut back the power requirements and use a wall wart.

PIC 16HVxxx MCUs can run off of 12V, but regulators are cheap and plentiful; don't skimp here. Though, the PIC 16HV785 looks like it might fit the bill. Here's an app note for using it in a lead acid charger: http://ww1.microchip.com/downloads/en/AppNotes/01015a.pdf
And another app note for a Li/Ni chemistry charger: http://ww1.microchip.com/downloads/en/AppNotes/01012b.pdf

As far as a display goes, I'd love to put one of these on there: http://www.sparkfun.com/commerce/product_info.php?products_id=569

Would need an MCU with SPI like the good old PIC16F877A, though. SPI would also make it easy to add the USB/RS232 interface.

 

[VidPro]

you can make a simple PSU of high power, with a center tapped 25v 3-6amp transformer, a 25amp bridge rectifyer and a couple of huge caps.
i make them often to replace various car, and gell cell chargers (when the consumer crap fails over time).

with a center tapped high voltage transformer you can either get major voltage, or divide it to 1/2 and one side is going up while the other is going down (sine wave), this makes for very easy to clean up power.
http://www.partsexpress.com/pe/showdetl.cfm?&Partnumber=120-1076 (12v 4.5A AC wall wart $18)
http://www.fairradio.com/catalog.php?mode=view&categoryid=5515 (36v CT 3A transformer low on page $9)
http://www.skycraftsurplus.com/index.asp?PageAction=VIEWPROD&ProdID=370 (24V Ct 4a $14)
http://www.radioshack.com/product/index.jsp?productId=2102703 (25.2+v CT 2A $11)

i always use the overrated 25amp bridge, because putting in parts that are JUST SPEC, dont last long enough. if you use "3amp" diodes and you pull 3 amps or even 2 they fail over time.
http://www.goldmine-elec-products.com/prodinfo.asp?number=G16228 (25A bridge $1)
http://www.radioshack.com/product/index.jsp?productId=2062584 (25A 50V bridge $3)

by using a fat capacitor instead of "just enough" the cap doesnt "cycle" as much, and caps do not live forever, they have a mtbf like everything. so its cleaner, can handle surge better, and lasts longer.
http://www.goldmine-elec-products.com/prodinfo.asp?number=G9121 (3900uf 20v caps $1 x2)
http://www.radioshack.com/product/index.jsp?productId=2102508 (4700uf 35v cap $5 x2)
http://parts.digikey.com/1/parts/51270-capacitor-10000uf-25v-elec-tsha-ecos1ea103da.html (10,000uf 25v $4)

a PSU of that type can be built and live for 20+ YEARS, and for about $40, volt regulate some 5v out of it at little cost, and you would have a much better supply than a computer supply. the bridge can be sinked to the container, and is overrated anyways, so it lives forever, even if everything else doesnt.

or we can put a bunch of switching junk in there and it can whine and moan, cost more, and needs fans and fails over time , either way is fine with me.


COMP PSUs
the 12V rails on computer supplys were always limited, but nowdays thanks to Crossfire (2 vid cards) and extra 12V to the motherboard the 12V has a lot of juice. If all this stuff can not be split off of a high voltage supply, then the 5V rails on PSUs have quite a bit of juice. if you need 15+ volts you cant get that from a computer PSU because if you use the 12 AND the 5 , it makes 7, because they have common grounds. so some tricks would be needed to get higher voltages. if you used 1/3rd of the max ratings it would still be some heafty amps, if you used 1/2 the speced maxes. continuously the things will croak, a computer Rarely uses the total power of its psu.
One thing that testing showed about computer PSUs is if you load one rail a lot, and not the other rails, the voltage on the rails will change, probably not enough for secondarily regulated things, but that might be important. the computer PSU uses the same base transformer for different rails.
http://www.newegg.com/Product/Product.aspx?Item=N82E16817226001 (Duel 12v PSU $80)

CHEAP BENCH SUPPLY
for 3 amps with a max of 30Volts DC a HY3003 is only 89$ and it has controls and meters and no fan, its a cheap bench supply, but it works
http://www.elexp.com/tst_3003.htm (0-30V DC 0-3A meters $89)

If you can get away with a high voltage original source and split off channels from it in series (i donno how) there is more efficiency (less change from the wall) and less amps.

 

[altis]

May I suggest that the regulator boards are based on the single Eurocard format (100 x 160mm) with DIN 41612 type-H connectors. These use Faston-type spades so the external connection could be via a backplane or wired push-on connectors. Each contact is rated at 15A at 20 celcius.
http://www.farnell.com/datasheets/81500.pdf
http://en.wikipedia.org/wiki/Eurocard

I suggest the following pinout:
Z04: Input -ve (earthed a CEP)
D06: Input +ve (9-36 volts)
Z08: SCL (I2C clock)
D10: SDA (I2C data)
Z12: Logic 0V (earthed at CEP)
D14: /ID0 (channel ID bit 0, link to logic 0V)
Z16: /ID1 (channel ID bit 1)
D18: /ID2 (channel ID bit 2)
Z20: /ID3 (channel ID bit 3)
D22: Output +ve
Z24: Sense +ve
D26: Thermistor +ve
Z28: Thermistor -ve
D30: Sense -ve
Z32: Output -ve (directly connected to input -ve)

CEP = Common Earth Point. The separate sense terminals allow the voltage to be directly measured on the cell. Power for the logic is derived from the 9-36V input. However, it is isolated from it so that no current flows thru the logic 0V connection. Logic on the board communicates with the main controller using the I2C bus. The four ID bits allow for 16 channels in any one system. This save having to put a bitswitch on each board to identify it.

 

[Skavoovie]

On second look, a PIC 18F4553/4558 may be a better choice.

It seems to have a much larger knowledge base for bootloader support, on-board USB for simplicity's sake, a bump from 10-bit to 12-bit A/Ds (still not enough for voltage, current, and temperature per cell given 8, but a mux would work), and 5 hardware PWMs (4 hardware-controlled cells and 4 software-controlled makes sense to me as I can't seem to find an 8 channel driver that would work; the last channel could be used for display back-lighting or something.)

you can make a simple PSU of high power, with a center tapped 25v 3-6amp transformer, a 25amp bridge rectifyer and a couple of huge caps.
i make them often to replace various car, and gell cell chargers (when the consumer crap fails over time).
...
a PSU of that type can be built and live for 20+ YEARS, and for about $40, volt regulate some 5v out of it at little cost, and you would have a much better supply than a computer supply. the bridge can be sinked to the container, and is overrated anyways, so it lives forever, even if everything else doesnt.

or we can put a bunch of switching junk in there and it can whine and moan, cost more, and needs fans and fails over time , either way is fine with me.

Even with a (or a few) 10,000 uF capacitor(s), won't the ripple voltage be far too large for a "precision" LiOn charger? Or are you just counting on regulation elsewhere?

 

[Peepsalot]

May I suggest that the regulator boards are based on the single Eurocard format (100 x 160mm) with DIN 41612 type-H connectors. These use Faston-type spades so the external connection could be via a backplane or wired push-on connectors. Each contact is rated at 15A at 20 celcius.
http://www.farnell.com/datasheets/81500.pdf
http://en.wikipedia.org/wiki/Eurocard

I have never heard of these, are they very widely available?

 

[VidPro]

Even with a (or a few) 10,000 uF capacitor(s), won't the ripple voltage be far too large for a "precision" LiOn charger? Or are you just counting on regulation elsewhere?

i donno, i rarely scope anything, and all of this is beyond my skill level. and as far as ripple and spikes when charging a battery, its a Battery its like a capacitor itself, i dont know any reason to get all precision with it, unless that effects the computer junk.

plus there are pulse and desulfator and all that other junk that uses spiking and pulsing and even high voltage pulsing as an advantage for batteries, supposed to break up bad formations (if you believe it).

for precision on the li-ion, tapering the final charge, and not going over the voltage, not trying to fast charge (because the cell wont accept it) and 50-50 pulse charging, which can go fast.
if nasty dirty power would effect any of that, then that is what i would see as a problem.

on the other hand i know many reasons to overrate the parts HIGHLY, if a 3amp mosfet will do the job, a 6amp one will do the job longer, and mabey costs 20c more. i suppose if the skill level of the people doing it is high enough, making the switching power supply itself WITH deescent highly overrated parts, would be easier than the scematic for the microcontrollers pcb.

if you use BOTH HALVES of a center tap transformer properly connected up, the power doesnt fluxuate as much, . if its not used as a center tapped, then there is more cleanup. (well thats what they tell me, i am not sure)

i do know walwarts of all types fail over time, usually because of cheap parts and continual use. computer PSUs are reasonably reliable then they blow a cap, or short out . i really dont think you can do 5billion cycles on a cap, and if your gonna then it should be 4x what is assumed to be needed, instead of lets see how we can engeneer it into 5mm of space. IF all the PARTS used are Way over the specs, there is no having to Think about one part handling what another part does.

question:
does a switching power supply pulse a cap Many many more times than a 60Htz transformer power supply?
are they using this fast whiney switching method and the induction so they dont have to put massive caps and transformers in?
and Does it work? cause various switching power supplies are not lasting as long as beasty hunks of overweight old school.

what about the dirty switching supplies we already have ? the b-900 psu, why did mine start getting worse ONLY after much time?
why have i had a fire inside my computer PSU? and when i searched the web, one guy had a fire that left his computer.
why are caps blowing up, and why did they (recenetally) put a cross hatch pattern on electrolytics to reduce fragging on failure?
why are flashlight curcuits failing faster than the leds? usually over driving and over powering?
Those are things i would like to know more about, when the old school beasts i have have not done that.

 

[altis]

I have never heard of these, are they very widely available?

Yes:

http://www.newark.com/jsp/Interconn...09+06+115+2921/displayProduct.jsp?sku=30C5376

 

[Peepsalot]

Yes:

http://www.newark.com/jsp/Interconn...09+06+115+2921/displayProduct.jsp?sku=30C5376

$15 seems pretty ridiculous to me for a connector. And that's just the male connector! So it will cost ~$30 just for connectors, per modules. I think that is way too much. Also, who the heck just has a subrack laying around. That is going to be another big chunk of money for people to lay down.

You say all the contacts can handle 15A, but I think we really only need that kind of current on two of those contacts. The other ones will have negligible current.

 

[Peepsalot]

i do know walwarts of all types fail over time, usually because of cheap parts and continual use. computer PSUs are reasonably reliable then they blow a cap, or short out . i really dont think you can do 5billion cycles on a cap, and if your gonna then it should be 4x what is assumed to be needed, instead of lets see how we can engeneer it into 5mm of space. IF all the PARTS used are Way over the specs, there is no having to Think about one part handling what another part does.

I don't know a whole lot about this, but I have seen this "all solid" capacitor that Gigabyte uses in some of their products. They say that they can last about 6 times longer than standard electrolytics.
http://www.gigabyte.com.tw/FileList/NewTech/2006_motherboard_newtech/article_02_all_solid.htm

I have also read that tantalum capacitors have a much longer life than electrolytics. Maybe the "all solid" is just a marketing word for tantalum? Or are they something completely different?

 

[VidPro]

I don't know a whole lot about this, but I have seen this "all solid" capacitor that Gigabyte uses in some of their products. They say that they can last about 6 times longer than standard electrolytics.
http://www.gigabyte.com.tw/FileList/NewTech/2006_motherboard_newtech/article_02_all_solid.htm

I have also read that tantalum capacitors have a much longer life than electrolytics. Maybe the "all solid" is just a marketing word for tantalum? Or are they something completely different?

hmm and so answers the question of why some motherboards die and some dont, and why there are those types of caps on expencive video cards.
so we know we can get them for the switching stuff, but can you get big mondo ones, for cheap power supplies? for reasonable prices?
and what about the "super caps" those are reasonably cheap, not as wonderfull as they sounded but smaller than electrolytic.