Modding a charger for higher current?

[swbluto]

Hello, I'm looking for a 300 watt charger that can supply a voltage of 24 volts or thereabouts. I'm thinking a current limited 24 volt power supply may do it, but I'm looking for compact, light ones BUT above all, compact. I found a relatively compact 100 watt power supply set for 24 volts, so I was wondering, would it be possible to modify it to output something closer to 300 watts? How would one do that? I'm thinking I'd find the current shunt, add solder / do whatever to reduce it's resistance, and incrementally increase the current output and find out what parts are getting hotter than usual and then increase cooling as needed. Maybe better heat-sinks? Fans?

The idea is that I want to charge my 300 watt-hour a123 battery in under an hour. I don't care about "balancing", so a straightforward power supply would work well. Purpose-suited chargers would work, as well, I just haven't had much luck finding 300 watt 24volt chargers.

 

[fechter]

I don't think you can mod a 100w supply into a 300w one. They can usually barely handle their rated power.

You might try one of those Meanwell power supplies. 320W in a reasonably small package.
Sorry, it's a long thread: http://endless-sphere.com/forums/viewtopic.php?f=14&t=4125

 

[mwkeefer]

swbluto,

fechter is right, forget over running the 100w charger... for 300w, get yourself an S-350-24 which you can fairly easily modify into a CC/CV charger set for 320w to 350w (13.3A - 14.58A). This will be a reliable system and should last quite a while... the meanwells are also available in S-145-24 models which are about 2/3 the size of an S-350 model and they can too be modified into a charger - they are only passively cooled though, I guess with active cooling you could maybe bump up to 200w without killing one.

The trick now is getting the "S" model and being sure it's not a PS model... the PS do hiccup mode limiting and though were workign (many of us, myself included) on getting them to do normal constant current mode, even for a front end current limiter via PWM and FETs (same as a normal DC motor PWM controller) so that all unit types can be used and will no longer require shunt modifications.

If you can tell me the series count of A123 cells, the AH capacity and the C rating for discharge and charge I will be happy to point you on the proper steps to take to get one of the meanwell style units working for your < 1hr charge time.

Another thing about A123 cells, I think it was Luke (LiveForPhysics) who figured out they have a "Rapid Charge Protocol" where using a voltage much higher than the maximum cell voltage is used for the first 90-95%. This means you may be able to get the charger to run faster (with resistor mod to ensure fan stays on)...

Regards,
Mike

 

[swbluto]

Thanks for the responses. This seems encouraging! Some quick data: I'm looking at a 8S5P a123 pack, or "24 volts 11 ah".

I noticed that some laptop chargers are rated at 80W and they aren't larger than a 2" thick hersey bar. Are there any suitable ~low 80-200 watt supplies that are *compact* that I could parallel for increased charger wattage? Ideally, I want to put these chargers in my pockets for weight distribution benefits, and I could stuff 2,3 or 4 laptop-sized ones in my cavernous coat pockets. I really want to avoid putting the charger in my backpack.

What should I be looking to see if power supplies/chargers are parallel-able? I could put schottky diodes on the output so that they wouldn't sink current, if that would be a problem. Do they have to be isolated? And, if so, do I test them for isolation by measuring resistance between the input and output and infinity means isolated?

 

[dnmun]

ping has a 100W model he sells with his packs. i tested one and it put out 4.3A at 30V and it has a little fan in the end and a lot of heat sink aluminum jammed onto the top of the inductor. also has a thermistor to shut it off if the inductor gets too hot.

my original picture is on a dead disk, but here is one on my google album of that 24V brick from ping. i think he calls it the model A

http://picasaweb.google.com/rdnmun/ChargerPictures#5444611633679358514

 

[jag]

Thanks for the responses. This seems encouraging! Some quick data: I'm looking at a 8S5P a123 pack, or "24 volts 11 ah".

I noticed that some laptop chargers are rated at 80W and they aren't larger than a 2" thick hersey bar. Are there any suitable ~low 80-200 watt supplies that are *compact* that I could parallel for increased charger wattage? Ideally, I want to put these chargers in my pockets for weight distribution benefits, and I could stuff 2,3 or 4 laptop-sized ones in my cavernous coat pockets. I really want to avoid putting the charger in my backpack.

What should I be looking to see if power supplys/chargers are parallel-able? I could put schottky diodes on the output so that they wouldn't sink current, if that would be a problem. Do they have to be isolated? And, if so, do I test them for isolation by measuring resistance between the input and output and infinity means isolated?

200W power bricks are unusual. The Shuttle Zen model came with one.

Like you say laptop supplies are easier to obtain.
My Thinkpad T series powerbrick weighs a bit less than 200g, and outputs 16V at 5A. For a 1h charger you would need 4 of these and adjust the voltage down 2V (or depend on the BMS to cut). At 800g it is not that much less than a Mean Well 320 (which weight 1.1 or 1.2kg I think)

Toshiba Qosmio also use 15V suppolies that you can series 2 to obtain 30V. I think they are higher power too. Don;'t know weight.

Laptop chargers can often be bought for about $10 on ebay.

A Meanwell is expensive retail, but can be had around $30 on ebay.

 

[jag]

B.t.w. I have also been thinking if there is a way to shed charger weight, yet have portable on-the-road quick charging. The other day I asked the power supply expert in our lab if I can somehow make a charger using the windings in the motor as an inductor/transformer. (Since that is the heavy component of a charger.) She thought it was a stupid idea.

 

[swbluto]

Thanks for the responses. This seems encouraging! Some quick data: I'm looking at a 8S5P a123 pack, or "24 volts 11 ah".

I noticed that some laptop chargers are rated at 80W and they aren't larger than a 2" thick hersey bar. Are there any suitable ~low 80-200 watt supplies that are *compact* that I could parallel for increased charger wattage? Ideally, I want to put these chargers in my pockets for weight distribution benefits, and I could stuff 2,3 or 4 laptop-sized ones in my cavernous coat pockets. I really want to avoid putting the charger in my backpack.

What should I be looking to see if power supplys/chargers are parallel-able? I could put schottky diodes on the output so that they wouldn't sink current, if that would be a problem. Do they have to be isolated? And, if so, do I test them for isolation by measuring resistance between the input and output and infinity means isolated?

200W power bricks are unusual. The Shuttle Zen model came with one.

Like you say laptop supplies are easier to obtain.
My Thinkpad T series powerbrick weighs a bit less than 200g, and outputs 16V at 5A. For a 1h charger you would need 4 of these and adjust the voltage down 2V (or depend on the BMS to cut). At 800g it is not that much less than a Mean Well 320 (which weight 1.1 or 1.2kg I think)

Toshiba Qosmio also use 15V suppolies that you can series 2 to obtain 30V. I think they are higher power too. Don;'t know weight.

Laptop chargers can often be bought for about $10 on ebay.

Well, if a laptop charger could do the job, than 2 hours would be fine if the alternative (meanswell) is something like the size of a piece of paper except half the width. To put them in series, though, would require isolated outputs which I don't know is the case. I'll have to test some power bricks I have laying around - however, a voltmeter handy way to test for isolation? I guess I'll have to google it.

 

[bearing]

handy way to test for isolation?

Put a resistor between ground of the two and measure the voltage between grounds.

 

[kZs0lt]

I am very interested in this thread, because I am also looking for solutions to quick charge my 250-300wh batteries. A portable charger would be great.
I have a 50w 6s RC charger. So I have configured my packs multiple of 6. I have a 26650 konion pack of 18s2p which is configured as 6s6p for charging. this way is like 13-15Ah so it takes more than 6hr to charge. Terminal voltage is 24.6v for the konions. I made recently a backup battery of a123 M1 cells, same 182p config, 6s6p while charging 12-13Ah, terminal voltage 21.6v

One of the options would be to buy a 24v 350w meanwell power supply and set the voltage properly. Can't find them though for $30 and shipment is expensive.
In the meantime I am running my RC charger from a computer XT power supply which balances it too. If I need a quicker charge, I wire a big 200w 24v~ trafo in parallel, rectified yields about 33v, and can give out 12-15amps @ 20-25v getting very hot. I then set up the charger to charge in parallel (2.xA) and to beep when it finishes. I stay around and when it beeps, I disconnect the trafo and let the RC charger finish the work.

But the PC power supply is not very portable and the trafo is very heavy. I was thinking of using instead the trafo a laptop power supply, but my best one says can deliver only 4.5A @ 20v, 90w. Not bad for that little guy, if it can really live up to the specs(cheap chinese IBM charger). It should make 4.5ACC additionally and could reduce my charging time to 2 hours. But terminal voltage is only 20v, I thought I could wire it up to the power supply's 5v output to bump it into the 25v region. Power supply's 5v output should be good for 20A. But I need to check if everything is isolated. 5v output to RC charger output and to laptop charger output. Also need to check if the laptop charger can really put out 4.5amps continuously for at least 2hours. Not an easy task, I don't think it was designed to charge 300wh at one time, full output.

 

[jag]

Well, if a laptop charger could do the job, than 2 hours would be fine if the alternative (meanswell) is something like the size of a piece of paper except half the width. To put them in series, though, would require isolated outputs which I don't know is the case. I'll have to test some power bricks I have laying around - however, a voltmeter handy way to test for isolation? I guess I'll have to google it.

The Meanwell is not that big, but definitely bigger than two laptop charger.

You can series two laptop chargers w/o any modifications so you get 2x15=30V.
Or you can look for a 30V powerbrick. Some printers use higher voltage. Enterprise switches also. But current limit is usually lower, so two laptop bricks is close to as small and light you can get for a given power I think.

(B.t.w. o parallel two for more current, one may influence the current regulation circuitry of the other, but it may also just work. Especially under load.)

The possibly bigger question is how the charger behaves when the load exceeds what it can supply at the set regulator voltage. What we want here is constant current mode. Other possibilities are: Shutting down completely, cycling on and off, or getting very hot. (Dell had 100W supplies that would get hot enough that they apparently caused some fires. Maybe only to people who had them in their bed or somesuch though. However was enough concern to issue an updated, but heaver model. I have both and never had a problem).

I will try a slightly different charging arrangement: I'll use two laptop supplies in series, and then these in series with a battery charger (NiCd in this case, but this is irrelevant. In this case the hope is that the charge current control will work as usual in the NiCd charger and the laptop supplies will just boost voltage (compensating for extra cells I've added in series to the pack). Possible problems is that the NiCd charger (which has built in cutoff points for CC 1.8A mode to trickle charge) does not trigger at the right cutoff point any more.

 

[swbluto]

Yeah, what happens when it exceeds the specified current rating is of interest. Ideally it should limit the current, but I suppose the current limiter on regular laptops is within the laptop's battery's BMS so it isn't "that smart". If that's the case, poo. I'm also wondering how well it would handle the specified rating - I don't know why they would over-rate the supply since the supply itself isn't typically being marketed ("Step right up lady and gents, right here we have a laptop with a not 3, not 4, not 4.5, but a fully whopping 5 amp charger!"), but maybe that's just for "peak loads" the computer may have. That'd be disappointing.

Anyways, I'll have to power up my laptop charger and measure how it reacts to a 3 ohm load, and then possibly measure how it reacts to a short circuit. I don't know if I want to short-circuit it, though - I've fried a battery BMS doing that. :|

Also, to the size of the meanwell, the ebay auction at http://cgi.ebay.com/27V-DC-13A-351W-Meanwell-Switching-Power-Supply_W0QQitemZ220566021444QQcmdZViewItemQQptZLH_DefaultDomain_0?hash=item335ac26544 lists it as 225 x 114 x50 (mm) . According to google, that's 8.8"x4.49" inches and a regular piece of paper is 11"x8.5" so it does appear to be roughly the size of half a piece of paper.

Also, I found out the laptop supplies I have and the ping charger are all output isolated. Apparently that's the standard for things that plug into the wall, which is good news.

 

[amberwolf]

One thing to beware of with laptop chargers is that a lot of them get dangerously hot when the laptop is being used at it's full capacity *plus* charging a totally discharged battery. Some of the Sony ones I've seen have gotten hot enough to distort the purple sparkly paint they sometimes use, and many brands I've seen the stick-on labels with the ratings and stuff have distorted/shrunk/peeled away from heat. Toshiba probably was the best about it, in that they usually didn't get that hot during full-power use plus charging, so they might be rated higher overall. Some of the worst are the little tiny HP / Compaq bricks, which sometimes can't even power the laptop at full power while charging a battery.



I've opened a few for experiments or repair that have had "lightly browned" PCBs over large surface areas, indicating severe overheating.

Most of my experience with laptop adapters used *on laptops* is 3+ years old, so newer ones might be different, but it used to be common to see that sort of stuff. I've used plenty of adapters for non-intended purposes since then though, and some can take some short-term hefty abuse before they meltdown or smoke out, including even the Sonys that get really hot.

Some of them have fuses in them, and so will be reasonably easy to fix if you just overcurrent them and blow that, but most of the ones I opened that overheated blew capacitors and then MOSFETs, sometimes burning windings in the switching transformer.

 

[heathyoung]

Yep - laptop chargers run HOT. I modded a dell one to run a EC6, works well but its slooow ~ good for charging while at work, very small and portable, and balances on every charge. They run a very simple shut-down on overcurrent, its easy to fool but also equally easy to set fire to these.

 

[swbluto]

It sounds like laptop power supplies don't have current regulation? Well, that kind of sucks. What would I need to add external current regulation? I wonder how well linear regulation would work. I suppose I could use switching regulation but, jeez, it'd be ideal if that was inside of the supply itself. Nothing like introducing even more inductors.

 

[swbluto]

Okay, I'm testing an IBM Thinkpad laptop power supply, and this is what I found. It's rated at 16V 4.5A and has a floating voltage of 16.77 volts.


With a 3.97 ohm load, I measured 16.02 volts at the terminal of the load. That would imply a current of 4.03 amps. The hottest spot on the laptop power supply reached 88.5 degrees fahrenheit.

With a 2.44 ohm load, I measured 8.23 volts at the load which implies a current of 3.4 Amps. The hottest spot on the power supply reached 104 degrees fahrenheit.

So, it appears IBM Thinkpad power supplies do current regulation, however, it's not at the rated current. It appears that it reduces the current limit depending on how much resistance there is. The lower the resistance, the lower the current limit. I shall now do ultra-low resistance testing to see how low it really goes.

ADDED:

So I put a .88 ohm load on it, and interesting behavior emerged. First, it drops to .1 output volts and then it gradually rises over the period of 5 seconds to a voltage of 1.12 volts, and then drops down again to .1 volts and it repeats this cycle. Apparently, it's checking if the "short condition" has been removed by ramping up the voltage and it resets if it hasn't, and checks again and again and again. It appears to use laptop chargers, I'll have to adjust the load somehow to get the optimum output current.

WAIT... let me retract that. It regulates current upto a certain amount. It appears that it's able to decrease the voltage by at least half before "shutting off" and since the lifepo4 batteries will be at 20 volts or so when discharged and the nominal voltage of 2 chargers in series is 32 volts, it appears it should work fine though not at a full 4.5 amps when "mostly" or "fully" discharged. Yippee! Now, I just kind of wonder how two adapters in series will react when it needs to reduce the voltage. Will each adapter decrease evenly, or will one decrease significantly more than the other, or will there be a "dance"/oscillation from two independent feedback loops in series?

Well, I could have each charge their own string of cells if series would be problematic. I'd have to, however, introduce a cutout switch for each string of cells.

 

[amberwolf]

Well, I could have each charge their own string of cells if series would be problematic. I'd have to, however, introduce a cutout switch for each string of cells.

no, just connect each charger across it's part of the pack.

 

[swbluto]

Well, yeah, that's what I was thinking about doing.

The thing is, though, is that if the power supply is just allowed to "fully charge" the battery, a cell will inevitably become overcharged and go too high a voltage. Ideally, this is what you'd a balancer for to prevent. But, I'm simple minded and I already got a single cell charger array at home for the occasional balancing I may need. So, I need to have a circuit that cuts out the charger when a cell reaches too high a voltage (The "cutoff switch" I refer to is usually a mosfet on a BMS, and I think it's usually a p-mosfet.). With a single bulk charger, or both supplies in series, I would only need one cutoff switch or one BMS. For two independent chargers, I'd need a cutoff switch for each, so that'd be two cutoff switch.

In regards to the cutoff switch, I have no idea how I'm going to implement that. I'll have to figure out how to use p-mosfets, assuming that's the "magic".

Edit: So I found out how one would use a p-mosfet but... I think the only mosfets that I've seen on the ping's bms were connected to ground, so I think the charger mosfet was just a regular n-mosfet. It seems p-mosfets are suited for "high side" mosfets, but I don't believe they have anywhere near the cost/performance advantages of n-mosfets in high power applications.

 

[amberwolf]

If you cut off the ground side you can use N-channels. That's one of the main reasons the ground side is what gets cut off on most BMS--it's just easier to do since you don't have to have a charge pump to raise the gate above the V+, nor do you have to use P-channel FETs (which are harder to find in as high a ratings as Ns, and usually more expensive).

However, you can skip FETs entirely, and their losses (and short-circuit failure mode), if you instead use a relay on each one. If there is a particular voltage you want it to always cut off at, just put a comparator circuit in there that cuts power to the relay coil, opening the charge circuit, as soon as it's done.

You can even use a zener biasing a darlington transistor to do the comparing and relay control, instead of an op-amp, to keep component costs and counts lower. The zener would be setup so that once the pack reaches a certain voltage, it begins conducting and changes the bias voltage on the transistor's base enough to turn it off and drop the relay out.

You could also have two regular transistors, one of which is a PNP that is turned *on* by the bias voltage provided by the zener conducting (so that it develops a voltage across the resistor from base to V+), which then shorts out another resistor that had been providing bias voltage to the second PNP, when the first one's collector/emitter are turned on.

 

[swbluto]

If you cut off the ground side you can use N-channels. That's one of the main reasons the ground side is what gets cut off on most BMS--it's just easier to do since you don't have to have a charge pump to raise the gate above the V+, nor do you have to use P-channel FETs (which are harder to find in as high a ratings as Ns, and usually more expensive).

However, you can skip FETs entirely, and their losses (and short-circuit failure mode), if you instead use a relay on each one. If there is a particular voltage you want it to always cut off at, just put a comparator circuit in there that cuts power to the relay coil, opening the charge circuit, as soon as it's done.

You can even use a zener biasing a darlington transistor to do the comparing and relay control, instead of an op-amp, to keep component costs and counts lower. The zener would be setup so that once the pack reaches a certain voltage, it begins conducting and changes the bias voltage on the transistor's base enough to turn it off and drop the relay out.

You could also have two regular transistors, one of which is a PNP that is turned *on* by the bias voltage provided by the zener conducting (so that it develops a voltage across the resistor from base to V+), which then shorts out another resistor that had been providing bias voltage to the second PNP, when the first one's collector/emitter are turned on.

Thanks, dude. I have the control and detection figured out, I just wanted a "fool proof" way to stop the charging. Now that you mention it, I don't think I'd want to use mosfets for the charging fet because of the possible short-circuit failure mode. Overcharging cells in the case of a mosfet failure would be a fairly expensive failure.

So it sounds like relays are the way. They seem kind of big, though. I wonder if solid-state relays would work?

Gee. I'm looking at relays and it seems the managably sized ones are generally rated at 24 VDC or 12 VDC. I'm thinking that can't be good.

 

[amberwolf]

What kind of physical size are you looking for?

I'm sure SS relays could work, but I don't have direct experience using them yet. I suspect that since they are still semiconductors inside (AFAIK) they would still be vulnerable to the short-circuit failure mode, unless they have some other internal protection against that (internal fuse or polyfuse or something).


FWIW, I have a couple of boards out of an old Sony VTR that are full of SS relays, but I have no idea what their specs are. There are only four pins; the input is on pins 3 and 4, load switched on pins 1 and 2, as far as I can trace the circuit. No one has ever replied to my queries at parts distributors or at IR (the manufacturers). (actually, Mouser replied but with a blank quote form, which made no sense, so when I asked what they meant by that they meant they didn't have any of them, which wasn't what i'd asked....). Sony might have the info, but as I don't know the part number or model of the original machine, they can't help.

If you happen to run across any specs for older SS relays, this is what they are marked:
International Rectifier Solid State Relay D2N 202BD
International Rectifier Solid State Relay D2W 5002
Both versions are black, with white markings. Plastic casing, potted module. About 1" x 1" x 1.25" IIRC.
The markings include "SS Relay" in an old font commonly used on sci-fi shows in the early 70's. Below that is the little "I diode-symbol R" logo of International Rectifier.
There are a number of different (presumably) lot or date codes on them, including "8C10 D", "1E26 C", "1H06 D", "1G29 C".

I'm sure they work, but when I tried 5V to switch one, nothing happened at the load end, so either I traced it wrong or it takes more voltage, and I didn't want to smoke them if I could first find specs.

If you find out the specs for them, and they turn out to be usefully-rated for your project, I could send you some for a few bucks (or less) plus postage costs (probably fit in a padded envelope with a few stamps).

 

[swbluto]

http://www.epanorama.net/circuits/semiconductor_relays.html

Here they document properties of solid-state relays and to quip:

Typical failure mode is output short circuit

Gee. Well, do bipolar transistors usually fail "open"? I could use those... I think.

Well, wait.. it appears "power transistors" cost on the order of 8 dollars and they typically have an hfe of 12, meaning I would probably need another transistor in a darlington configuration just to drive it, plus I'd lose at least 10% of the current just on driving the power transistor. Hmmmmmm.... I guess it's back to relays.


Let me think here. If I position the relay at the battery's ground during charging, then the voltage across the relay is nearly 0 while conducting so that's within a 24volt rating. What about right before contact? The voltage from the charger should dissipate through the battery, shouldn't it, and then the voltage the relay would see would be 0 or certainly less than 12, right?

 

[jag]

Gee. Well, do bipolar transistors usually fail "open"? I could use those... I think.

My biploars usually short circuit... I thought short circuit was the usual consequence of the secondary breakdown failure in bipolars.

 

[amberwolf]

Transistors fairly often fail shorted too, at least until they blow their faces off (like MOSFETs) if running at high currents. At low currents they'd stay intact and fail to do what you need them to.


Voltage across any open portion of a circuit is the full voltage of the circuit. So until the relay contacts close, they have the full potential across them. Once they close, they have effectively zero across them.

You can save some wear on them by connecting the charger only after you have turned on the relay(s), if they are only there to provide the *disconnect* for safety when done charging or at powerfailure, etc.

 

[swbluto]

Here's the type of circuit structure I'm thinking of for relays. All three circuits shown have a negative discharge and charge wire; the top left one is discharging and the measured voltage across the relay is 0v. This simulates powering a bike or something else.

The top right one with is when the relay is turned off and the charger is attached to the battery. The measured voltage across the relay is the charger's floating voltage minus the battery's voltage, which in this example is 1 volt (25 floating, 24 batt voltage).

The bottom left one is when the relay is turned on. I didn't have an internal resistance in this example, so it shows 1 volt but with an thevinin resistance("Total circuit resistance") of .5 ohms, it's actually closer to .166 volts.

View attachment relayChargerVoltageEvaluations.JPG

So it seems like it should remain within ratings with this set up.

And, I have good luck. I found a relay in my stash that's lightweight and seemingly small for a relay, and it has a constant current rating of 7 amps at 30 VDC. So, it seems like it'd be good for a 4.5 amp charger which I'm guessing is what my laptop power supplies should be. For an hour charger, I should use 9amps or so, but I'll make due with what I have and upgrade as I go along (I happen to have 2 spare laptop chargers that I don't use; the laptop it goes to is defunct and I have other chargers around the house I could use to use the laptop is I need to.).

Also interesting to note, the relay takes 20 milli-amps to turn on. That's really good! Quite a bit less than what a transistor would've required. However, it takes >17 volts to turn it on (or off, depending on how it's wired) and I'll need to interface that with a 5 volt signal. I'm guessing an op-amp is in my future? Hopefully *that* doesn't have a short-circuit failure mode(or, rather, it's unlikely to fail - or I should wire the relay so that if the op-amp does fail, it "fails safe".)