assembling a 80V/15A A123 charger from 75#s of telecom scrap

TheWizard

TheWizard

1 mW
Joined
Dec 15, 2010
Messages
14
Location
Phoenix, AZ
Preface:
Last week I salvaged a Chinese AOWA 36V pedal-elec. After stripping off the cheezy plastic panelling, I found a very solid bike that was fun to "drive" (the cranks are just above the ground and are 155mm long with a 1:1 gear ratio, making "riding" a bit of a joke) but the SLA batteries are holding it back - as well the wimpy 18A controller and thin phase wires on the 9C motor, but those are the easy things to fix.

I'm gonna cut off the SLA bracketry, bend and weld a rack to hold a 7s3p A123 EV pack just forward of the cranks.
That will get the heavy weight nice and low. And it will make room for welding in a top-tube that will run from the seat-stay junction to the far edge of the front wheel, turning this moped into a 72V 20Ah cycle-truck!

But first I need to assemble a charger for my A123 7s3p packs. I bought 3 of them, betting that e-biking will be a serial addiction! The plan is to cut the bus bars and rewire them from 24V (7s3p) to 72V (21s).

I trolled the junkyard today and took a second look at a well-aged heap of telecom wreckage. I removed 3 of these 24V 15A power supplies from some 19" racks:
http://www.ebay.com/itm/SOLA-CVDC-POWER-SUPPLY-28-1203-2-24VDC-4-5A-15A-OUT-NICE-USED-TAKEOUT-M-O-/160756813551
The item pictured is in much nicer shape than what I recovered, but it was a fair wager at scrap price. They weigh about 25 lbs a piece! All american :)

My phone is being a goon so I'll have to upload pics of the under-side later. The only hidden pieces down there are 2 pots in-line with the transformer and an inductor/shunt/wire-wrapped-rod thingy. No protection of any kind (maybe inside the big rectifier unit?), I'll be adding some fuses

Am I wrong in assuming that I could wire the outputs in series like people do to make HV chargers out of multiple SMPS's?
I'm reading through this but I haven't really groked the circuit yet
http://www.google.com/patents/US2804588
http://www.google.com/patents/US2806199
http://www.ieeta.pt/~alex/docs/ApplicationNotes/DC%20Power%20Supply%20Handbook.pdf

Looking at those big ole caps, they are rated 30Vdc & 40Vdc. If I tweak the windings on the inductor/wire-thingy, do you all think there's enough headroom to get an additional 6V out of each supply to get a 86V charge voltage?

1.21 Jiggawatts! err, 1.2Kw charger....time to string a new power-run in the garage.

What kind of charge times could I expect? Is that a just a math problem or is there some LiFe/A123 chemistry-voodoo there?
 
usually the lifepo4 is considered 3V nominal. so a 24V lifepo4 pack is 8S or 8 cells in series. 72V is 24S, and the voltage is 88V when fully charged.

the BMS will be 8S, or 12S, or 16S, or 24S for lifepo4, and chargers 30V, 45V, 60V, 90V or close to that.
 
As long as the output side is isolated from the input side (you can check this with an ohmmeter, when the PSU is not connected to anything), then you should be able to series them without a problem.

But keep in mind those things probably aren't regulated very well, and likely have a lot of "extra" voltage over the 24V when they aren't loaded down some, and may sag a bit under the 24V when they are fully loaded.

I wouldn't recommend messing with the transformer windings; I'd just use them as they are. Unwinding and rewinding a transformer is not all that hard in theory, but sometimes the way they are assembled it is difficult to actually put them back together if you take it apart--or else the windings are impossible to unwind without damaging their enamel and having to rewind the whole thing with new wire. :( Often, it's not that big a deal...but it can still be time-consuming.


I have a number of much older teletype PSUs that are somewhat similar linear PSUs, and probably weigh as much as yours (maybe more), and they don't regulate quite as perfectly as I'd like. Mine use various regulation methods, and some don't actually "regulate" at all-they simply convert the AC voltage from wall-level to equipment-level, rectify it, filter it kinda flat, and that's pretty much it. They expect that the load will be fairly constant from the attached equipment, and designed the PSU to output just enough voltage under that exact load to keep it within tolerances.

As chargers, they may not do what you want, since the load is constantly changing, and the voltage will probably keep going up as the load decreases.

If yours are actually using DC-DC converters (whcih they might be), then they may well have some decent regulation, and possibly even an easy way to cut them off once the charge current drops below some preset level, or voltage reaches a certain point. (you'd build an external circuit for the detection, and have it trip the "enable" line on the DC-DC units)


Just curious--which junkyard? Apache Reclamation?
 
well shoot, I misread that label, I thought it said CVCC, not CVDC...now I have a boat anchor. I'm not sure how much the current will sag as it reaches voltage, but that defeats the whole purpose of building a monster like this.
They came from Davis Salvage.
 
They're certainly still good as power supplies, and may even be good as chargers. If they have DC-DC units inside them then you may be able to do a lot more with them than you think, if you can find the pinout for the DC-DC and if they have an "adjust".


Regarding "current sag as voltage rises", well, they will *all* do that--it's the nature of Ohm's law and such. The internal resistance of the battery would have to drop just as fast as the voltage rises in order for the current to remain the same. Otherwise, you'd have to raise the charger's output voltage well above the final charge voltage for the pack to keep the current the same all the way to end-of-charge.

If you do that, I don't know how the cell chemistry will react, so it might cause issues (or might not).

But definitely if you do that you can't use a "final charge voltage" as a simple cutoff, like a standard HVC (at least, not until you measure what that should be for a full pack under these conditions). You also can't use a final charge current as the termination limit, as you won't know what that is, either, until it's measured for a known-full-pack. You'd have to use coulomb-counting, as it's called in many BMS design documents, or basically measuring the Ah put back into the pack vs what you took out, plus known charging inefficiencies of the particular chemistry.

So you have to start with a pack in a known-full or known-empty state (or some other known-capacity-level), then always measure Ah in and out as it's charged and discharged. Once you are sure what your termination voltages or currents are to match up with a known full state, you can then use those instead of coulomb-counting to terminate the charge, for a simpler cutoff method.


So it's much simpler to use a standard charging method that terminates at a specific voltage, if you're building your own charger. And that means that current is going to have to drop during charging, as it nears termination voltage. How much it drops depends entirely on the chemistry of the cells, and how it's internal resistance reacts to becoming full.
 
thanks guys. I did not read about columb counting in any of the charger+BMS designs I had looked through - the ones I saw monitored each cell and would shunt-resist current away from them once they approached HVC, and then kill-signal the charger once all cells were being shunted.

I think building a columb counter would be worth doing as a first try, probably easier than trying to build a multi-lane BMS
 
FWIW, the coulomb-counting is only part of the BMS--it is just a charge monitor for the whole pack, and doesn't balance any cells. You'd still need the "multi-lane" balancer if you ahve cells that don't all match internal resisntaces and capacities and the like, if you use a lot of their capacity. (they might never get out of balance if you don't ever fully charge them to absolute max and never discharge them too close to empty, but it depends on the individual cells).

Either way, you would still want the "multi-lane" part of a BMS taht montitors HVC and LVC for each individual cell, separately.

If you look at Linear and TI and Maxim and other BMS-chip makers, they have appnotes on their assorted chips that sometimes describe coulomb-counting; that's where I first ran into the term and the idea, before I came to ES, actually, and still had some pretty half-baked ideas on how to motorize my bikes. :oops:


Right now, I only use a full BMS on my Vpower/Cammy CC pack. The RC LiPo and other experimental packs of similar chemistry I am using only bulk charging (with the charger only going as high in voltage as the whole pack should ever get to), and occasional balance checking of the cells, while manually monitoring main pack voltage during use for pack-level LVC, and now a Methods' LVC/HVC system to prevent disaster. (unfortuantely his units aren't available in the cell voltage range of A123 or other LiFePO4, only for LiCo and LiMn and NMC types, basically, AFAICR)

But am knowledgeable enough about the packs and trust the cells enough to do this, and can fix whatever might go wrong. I don't reocmmend this to anyone that doesn't feel absolutely sure that they can deal with the consequences of the stuff that can go wrong this way. ;)
 
Ok, I think I've finally figured these old supplies out and have a way to get them to charge my packs.
(apologies for not filing this thread under the battery section in the first place)

The mains go into a step-down center-tapped transformer to get 24vac and the isolation needed for paralleling the supplies. in series with that transformer is a ferroresonant transformer to regulate the voltage at a constant 24v regardless of load/supply fluctuations. from the 1st transformer's center tap and the regulator, the regulated 24vac then goes through two ancient diodes to get rectified 24vdc. That is piped through the big filter caps and a "choke" inductor to squelch out the AC ripple. The supply was hooked up to a rack of radio equipment, so I'd think the output would have to be clean.

what I'm still wondering is if that choke also limits the max current to 15A, of if that was merely the rating on the transformers. from the prismatic cell spec sheet, the batteries have a 300A max charge rate, but I've never hooked up a depleted cell to see how much current it would actually try to sink by account of its own negative resistance. If it just sucks down 300A in the absence of a switcher circuit, I fear the cells could destroy any transformer-based power supply!

Since I know the supplies are fixed at 24V a piece, I can count on getting the 3 of them to always make 72v when in series. apparently I had the wrong stats for my pack's charge voltages - the bulk stage should nominally charge at 75.6v and the absorption stage should be around 71.4v . So my thought is to add a trimmable 5v/15a SMPS I have to the string to make a 77v charger, just above the bulk voltage. Then, I can build a current-detection circuit that will kill the 5v supply once the current starts to drop off after the pack reaches the bulk-stage voltage. dropping the 5v supply will cut the charge voltage down to a 72v level for the absorb stage, which I could eventually shutdown with a timer or amp-hour counter.

Using these things might be a bit of hassle maybe, but I'm finding that single-unit 1kw+ supplies (whether transformer or transistor based) are pretty durn pricey, even from surplus warehouses and auctions. Buying a charger that costs about half as much as my batteries (and is unserviceable by me) just grinds my gears. now I see why those little Meanwells are so popular...
 
A choke only helps limit current *change*, to smooth out current spikes in the same way that capacitors smooth out voltage spikes.
 
You must log in or register to reply here.
Back
Top