Li-Ion Pack Build From Scratch For CrazyBike2

So I'm tired of hauling around lead bricks for power, and had an unexpected opportunity to get enough Li-Ion cells to give me the range I would really like to have, assuming I can make them work without blowing myself up.

I'm currently collecting the rest of the information and parts I need to build the packs from. I've been reading about doing this for a couple of years now, and back at the beginning even started collecting "dead" laptop batteries, when I could get them for free, so I could harvest the still-good cells to build packs from. Unfortunately there usually were few if any cells worth saving, by the time I was given them, due to the time they sat around uncharged. So that idea went nowhere fast, as an ebike power pack. (I did build a better pack for my own laptop, though).

I'm figuring on 36V 10Ah packs for now, which will be paralleled to run the bike from. Each pack will take 50 (or 60, depending on sag under load) cells, so that should give me 5 or 6 packs. This will make it easier to find places on the bike to put them. It's possible I may go with 12V packs instead, to make it easier to series them for 48V or higher later on.

I run 45-ish pounds of 36V 18Ah SLA currently, using used server-UPS batteries. That gets me, with judicious use of pedals, 15 miles at 12-16MPH with frequent stops/starts for traffic controls. That's using my old motor, a 300W-ish, maybe 350, 2-pole brushed PMDC powerchair motor. I've now got a much heftier motor at 650W, also a powerchair motor, which I'm still working out the gearing for. It's a 4-pole brushed PMDC, with way more torque, and a totally different efficiency curve I'm still figuring out.

Typically with the old motor I pulled up to at least 20A battery current during acceleration depending on the gear I was in and the slope. Average current would be more like 4A to 10A. Terrain here is pretty flat, but there are some hills, and most of the big ones are very long, up to several miles at small grades, but often with a section of higher-grade here and there; two notable ones are enough that I would have to walk this bike up to the top rather than ride it, as even with the motor (the old one) I wouldn't be able to reach the top without either running out of battery or overheating the motor or both.

The new motor pulls probably more instantaneous current, but I haven't got a meter that will read higher than that right now. However, that current is pulled for a much shorter time, as it takes much less time to get up to speed with it. Average, it pulls around 4A-8A doing the same thing the other motor did. Not sure why it would be less (I expected it to be more, due to 4 poles).


The controller is a customized 2QD, so in theory I could do regen braking if I work out a clutch for the motor (so I can still coast) and take out the freewheel in the post-motor drivetrain. Right now the pedals and motor both feed into the main regular 21-speed bicycle drivetrain. The new motor has so much torque it can actually pop the chain over the teeth of the rear sprockets in higher gears. Not a problem I ever expected to have, so I'm still rethinking the drivetrain to possibly fix that. (I have a newer chain and cassette to try first).


On to the batteries (which I've not yet got in hand, but want to work out stuff before they arrive if I can):
They're 18650 sized, most of them Sanyo UR18650F, spec sheet here:
http://www.datasheetarchive.com/UR18650F-datasheet.html
or here in case that isn't working at some point:
http://www.houseofbatteries.com/pdf/UR18650F
The rest are Sony but I don't remember their p/n.

So they're nominally 3.7V 2.1Ah cells; I'll just round that down to 3.6V 2Ah and be happy for any extra.

Charging current is listed as 1C @ 4.2V, for 2.5 hours.

Assuming 3.6V start voltage and 3.0V LVC, I could draw up to 2C out of them to still get around 2Ah. I'd rather draw much less from any one cell, since I need range more than I need massive power for any length of time, excluding those rare hills.


I'd prefer to charge in the same configuration I discharge in, so I'd like to build a BMS for it. I'm still reading thru the fairly long threads that contain lots of BMS information, and I will probably end up building one of them from scratch on perfboard, using whatever parts I have laying around here. (I can't afford to buy or make PCBs, or else I'd get the ones from TPPacks and populate them with closest equivalents I have in my junkboxes. I've already spent my month's grocery money on the batteries themselves, plus some will go for the Turnigy watt meter if the order goes thru).

At first, to test the cells, I'll be charging them one at a time with a Sorenson linear adjustable voltage/current 60VDC 1.5A current-limited lab power supply, which I use right now to charge the third SLA in my pack (a 24V 3A wheelchair charger is used for the other two). Later, I will use the same supply to charge them as packs, at it's higher voltage. Yeah, it'll take at least all night to charge if I fully discharge the packs, but it's the best one I've got right now.

I have a bunch of other linear power supplies that I can probably convert to chargers, but they're not current-limited yet, and most are much lower voltage (one is only 5v, but up to 200A).

I also have a bunch of 4V-6V celphone chargers that I am considering turning into a portable charger system, if I have enough of them that are isolated so I can series their outputs, to use for opportunity trickle charging. Got a bunch of higher-voltage (12V-15V) laptop power supplies that can supply lots more current, if I can series those, to use instead. Have to work out a BMS before I can use either method, though.


Any constructive criticism, questions, or suggestions are welcome.

I have roughly 300 total batteries, split among Sanyo URF18650F & Sony Energytech STG US18650GR G2F types. I have not yet had enough time to sort thru them and test which of them have what capacity, but there should be enough of them to give me at least 36V 30Ah, and probably 36V 60Ah. There are also around 50 known bad ones, some of which are dead as nails but some of which have some capacity but not much; they don't charge to full voltage.

Regarding this thread:
http://www.endless-sphere.com/forums/viewtopic.php?f=8&t=13892&start=0
I now have a Ping V1(?) Signalab BMS courtesy of JulesA, which will need repair before it can be then modified to possibly use with these cells. I would have to figure a way (if there is one) to fool it's LVC and HVC to match the cells I have. Might not be possible, but my brain is pondering it while I'm not looking. There is at least one more damaged BMS on it's way at some point to potentially rebuild into one for this pack.

It is same kind as the one pictured here:
http://www.endless-sphere.com/forums/download/file.php?id=20570

I found these threads that have given me some information, probably enough to determine if it is possible to modify this BMS:
http://www.endless-sphere.com/forums/viewtopic.php?f=14&t=14773&start=0
http://www.endless-sphere.com/forums/viewtopic.php?f=14&t=8841&start=0

The Sanyo info I've found does not state what exact chemistry they are, but the Sony data states it's material is Li-Co with Graphite. According to a post in the schematic thread, the chips on that BMS are designed for Li-Co but the HVC/LVC voltages it uses per it's data sheet are not the same as the ones on the Sanyo or Sony battery data sheets.

So I have to come up with a a way to fool the BMS chips into thinking they are getting a higher voltage than they really are by a specific offset amount during charge, so HVC triggers a few tenths of a volt early, and that they are getting a lower voltage than they really are during discharge, so LVC triggers about a volt early. I think that if I could push ground on the chips up (during discharge) or down (during charge) relative to the actual battery voltages, it would do the trick. I just need time to sit down and figure out if that is possible.


If I end up with enough working BMS units I'll split this into multiple packs to swap out or add in parallel for range, but it is currently planned as a single pack of the max size I can manage. At this moment, that is at 36V, but might go to 48V after I do some experiments using my SLA at that voltage on CrazyBike2.

I'm still finding and reading pack build threads on the forum; any input or links are welcome.


(oh, and a funny link I found while looking for a PDF data sheet for the Sony cells: http://www.centrix-intl.com/details.asp?productid=6052 where they show an originally $4.99 price for one, say it's 80% off, and then say the current price is $8.99. )

Not much progress here yet. Still trying to find enough time to test the cells, sort them, and then find a good way to assemble the packs.

A new problem with doing it is that the Sorenson I had been using to charge the batteries on CrazyBike2 before, and was going to use to charge up the NiMH packs (from deardancer) and test and charge the Li cells with, just sparked and smoked when I turned it on last night, even though it wasn't connected to any load yet. It's a linear supply, and does not require an external load to operate, so something must've died in it during power-on-surge. A transistor burned and a resistor smoked, and a couple of traces on the board's cardedge connector vaporised, too.

I've got three others (only 40V models) that are not working for various reasons, so I will need to check their boards for the values of parts, and then see if I can trace out enough of the circuit to ensure replacing the blown parts is all that will be needed, so I don't smoke more stuff when I turn it back on again.

Maybe I'll get the others fixed while I'm at it.

I did have another idea for charging the NiMH, though: I already had two little NiMH chargers for AA batteries, and I found another for a couple bucks at a thrift store, so now I could charge up to twelve cells at a time. All three chargers will charge either two or four cells at once. They are all microcontrollers or dedicated chip chargers, so they should be safe to use with any NiMH cells.

I am considering putting charging taps on every cell in the NiMH packs, in connectors of four cells, and one of two cells. Then put a mating connector wired to the contacts on each little charger so I could not only plug in to these cells, I could also still use them for the AA and AAA cells I already use them for (for my flashlights/headlights/taillights, etc. on the non-ebikes).

Then I would be able to charge three sets of cells in the packs at a time. On the 36V pack, that's 30 cells, so two charge cycles for 6 of the 7.5 sets in it, and one more charge cycle for the other set plus 2 cells. On the 24V pack that's 20 cells, so only two charge cycles total (one for 3 of the sets, and one for the 2 full and 1 half set).

The whole pack will not get that warm this way, and though I will have to manually plug and unplug the charging connectors after each cycle, it will let me safely charge the packs.

Well, assuming I don't burn out the chargers doing it; they only put out less than 500mA charge current. Well, one of them can put out almost an amp charging current if it's only charging two cells. I'll try it out with the cheap thrift store one first. It's the model down from one I already had, and is an Olympus BC-100.



I'm also considering taking a bunch of old celphones I have that are broken in various ways, and wiring them up to charge cells. They're all isolated from each other, so I could theoretically charge the pack "in place", with charging taps coming out of it for each cell, and a charging connector. It would never be a fast charge, but it would be possible, though a lot of work.

I would need to take the little BMS board off each celphone battery, and lay them all out on something flat. Then solder wires from the cell connection tabs to the interconnect that would go to my pack cells. More wires from each board's input connection tabs to the celphones themselves.

If I have enough celphones that actually give me a display (many have broken screens or even broken-off flip tops) then I can even monitor the charging process via the little battery meter on the displays. Not much of a meter, but better than nothing.

Well, the above theory was nice, but if the batteries are really low (like say, 0.75V, since I forgot which cells were which that I had sitting out), the little celphone battery BMS boards vaporize silicon. The parts are so small I can't get the camera to focus on the board well enough to take a pic, but there is just a tiny crack in the main chip on the board.

So, for anyone ever trying this, don't try to charge a DEAD cell this way.

My eye still hurts from the dog scratch, so I had to stop before I got the *right* cell soldered to a different board to test this idea. Typing is easy with one eye but manipulating a soldering iron is dangerous.

After some more research, and digging into some old laptop battery packs I have around here that happen to have the exact same Sony 18650GR's in them, I found that they charge the batteries up to 4.15V to 4.20V, and let them discharge down to around 3.0V to 3.2V, depending on model or brand. That's the same as the Ping BMS does (or should), since it's based around a chip designed for those voltages.

So...I decided I would start with 16 of the not-exactly-toast but not-real-good cells, charge them up to abour 3.5V individually with the Sorenson (doesn't take long even at 250mA current limit), and then solder them into a 16-cell series string.

Then, since I didn't have a connector to mate to the one on that Ping BMS, I decided to remove the connector itslef, and directly solder some small-gauge teflon wire I had from a piece of test equipment directly to each spot on there. First I stripped only one end of each wire, then soldered that end to one hole of the cell-connection points.

Then I began at the most positive end, which appears to be farthest from pin1, and started soldering them to the cell wires on the battery string. I ran into a problem at the 4th and 5th ones in that required unsoldering the balancing shunting resistor on those two channels, as they began shunting as soon as the cells were touched with the wire, and I could feel the heat immediately (I had my wrist and forearm on the BMS to keep it from moving around, and to feel for heat like this). Not a lot of heat but noticeable and there should not have been any.

I don't know why it shunts on those two yet, but it might be fixable once I figure out why.

I continued until the whole pack is hooked up, without further issues.

No voltage drops across any shunting resistors other than the two I disconnected, so nothing else is yet misbehaving.

Set the Sorensen series'd power supplies to total of 65V (64V for 4.0V per cell plus 1V for losses in the BMS and stuff I expected). Current limit set at 500mA, to be safe-ish. Hooked up power positive to the pack positive, power negative to charger negative. No current draw registered on the Sorensons' meters.

Found that the charging FET was being kept turned off by the optos (probably by the two bad channels), so I turned it on via the power from the VCC pin of the logic chip next to it, and charging began.

At around 3.8V on a cell, the resistors began getting voltage drops on them and warming ever so slightly, and at 3.95V they had about 1.8V drops on the resistors and were quite warm. Current flow had dropped to about 200mA at this point, but the BMS was still shunting current from the cells even after I disconnected the power supply.

Either I did something wrong when hooking up the BMS to the cells (possible, despite reading all I could find on ES about how to hook up the board to a pack/charger/load), or something is definitely screwy about this particular BMS (no surprise, as it was reported known dead but failure unknown), or...HVC is really under 4.0V for these Ping BMS. (V2.x?)

I left it for about 5 minutes, until every cell was less than 3.95V, but it continued to shunt at around 1.8V on every resistor. Didn't calculate out the current that's pulling out of each cell, but it shouldn't be doing it at all AFAIK, so rather than risking the cells getting completely drained if they didn't stop (and me forgetting I was running the experiment), I decided I'd better just take the BMS off.

Note that the batteries never changed temperature at all.

So I disconnected the wires from the cell ends, just snipping each one leaving a little to connect to later when I want to retry things, and then remembered I hadn't taken any pics of it to share. Oh, well.

It did do it's apparent job, up to the point where it forgot to stop doing it. But since I also wanted the BMS for it's LVC as much as it's HVC/balancing functions, I'll have to figure out what's wrong with it before I can use it.


Just to see how charged they all were, I put a little 1/3HP PMDC brushed weedeater motor on there, which is designed to be run from AC and just has a diode in series with the power cord to the motor. I bypassed the diode and just ran the motor with no load on it from this pack, which started at 62V before hooking it up to the motor.

The motor instantly spun up but not very torque-ly (not a word, but hey), and the voltage dropped to 59.5V and held there, with just a tiny pre-hold sag for an instant to 56V. I let it run for about 30 seconds without any change in voltage showing up, then stopped. I did not measure the current, as I was too lazy (and tired, still feel a bit sick) to go get the wattmeter off the bike and use it for this testing, although I should have. Probably will later. As in, maybe tomorrow.

Also should have used the wattmeter to measure power into the batteries as they charged, but I didn't think of it till I was already started, and that's when I just sighed and decided to skip it.

If a LiCo battery has been sitting below 2.8v resting, toss it, it's definitely damaged and both a liability to you and the pack. If a battery won't hold a charge at 4.2v, toss it.

It only takes 1 failure in a big parallel string group to take the whole parallel group out. (depending on the failure mode)

The trick to making a pack like this work is in finding and culling the bad cells before they cause the pack to die from just sitting for a couple days.


To initially balance the cells, perform the tinning on each cell with an iron as hot as you can get, and solder with as low of melting point as you can find. (lead free is a joke). Solder some tiny little scrap wire to every positive end, and negative end. Twist together all the positive ends, and twist together all the negative ends. Let this sit for a few hours, and the SOC will normalize between all the cells, and you will start out with every cell in balance before you assemble the pack.

Good luck!

Hi,

I'd dump the BMS's and charge each parallel string with individual DC-DC convertors and use CellLogs for LV Monitoring. You could even use parallel strings with different chemistries:
http://endless-sphere.com/forums/viewtopic.php?f=14&t=4215

Charge li-ion efficiently for cheap and balancing? go here

Click to expand...

http://endless-sphere.com/forums/viewtopic.php?f=14&t=9728&p=150780

GGoodrum said:

What I'm working on now is a modular setup that uses a bunch of 20A/3.7V DC-DC convertors. Here's what this looks like:

Click to expand...

GGoodrum said:

The input to these can be any DC supply, from 33-72V. I'm using a couple of very compact Vicor 48V supplies, each driving 8 of the convertors. Anyway, once I get this complete, I'll probably just use this most of the time, as it will charge at a higher rate than anything else I have.

Click to expand...

http://endless-sphere.com/forums/viewtopic.php?f=14&t=2824

Why DC-DC (like the Artesyn) are perfect for charging li-ion like the A123

The idea with the Artesyn, (Vicor, Lambda, or any other brands ) DC-DC converter is that from the same big power source, you can get many lower and divided voltages that are independent and isolated. They are very compact and small and easy to use. It's like having many single cells charger, but with alot more power than others availlable on the known market like only 2A. They are adjustable and can charge at 4.2V or 3.7V for any chemistry li-ion depending on how you adjust their voltage output. generally they are fully protected and will recover automaticly in a second if shorted. they are very stable and efficient. the artesyn i plan to use the BXB150 serie is 84% efficient! so for cahrging 30A per parallel group i will only dissipate 17.7W. Let say
212W of heat for a 1330W charging!!

After reading alot of comments about BMS and article about the charge of many li-ion string cells, i made my own conclusion since i plan to regulary use my ebike and that i dont want some complications when charging:

-Charge them individually with multiple isolated sources.. forget the cell balance with only one big charger and a balancer.. that's to complicate and the balancer usually balance at max 250 or 500mA so if we want to charge a packs with many serie/parallel cells at current alot higher than 500mA, like 10 or 30A, the balancer will not do the job correctly and will not be able to follow the charge and balance efficiently at the same time!...some circuit like in the dewalt pack, reduce the charge current if one cell goes at higher voltage .. but the charge time increase alot!..

By charging each cells with a charger that will top each of those A123 at 3.7V, i'm sure that no cell will overcharge and that they will be perfectly balanced. The only protection circuit that i will need is a LVC to cutt the controller when the lower cell will reach the 2.7V level...

Seriously, spending 120$ + few buck for the main transfo and activation of the dc-dc circuit.. let say 200$ total for a 12s perfect charger up to 30A !!! i conclude that this is the BEST way to get the maximum performance of the A123 for the use we need! :wink:

Click to expand...

liveforphysics said:

If a LiCo battery has been sitting below 2.8v resting, toss it, it's definitely damaged and both a liability to you and the pack. If a battery won't hold a charge at 4.2v, toss it.

Click to expand...

I won't put them in the bike's power pack, but I will probably still use some of them individually for other things. The dead ones (less than 2V) I may carefully trickle-recharge just to see what they will do. I did this to a handful, and some do hold a charge afterwards, most don't keep much of the charge. They make good soldering-practice units. Outside, in the backyard.

It only takes 1 failure in a big parallel string group to take the whole parallel group out. (depending on the failure mode)

Click to expand...

And what a failure, too, if it's a cell short. Though maybe not until I try to recharge the pack again.

The trick to making a pack like this work is in finding and culling the bad cells before they cause the pack to die from just sitting for a couple days.

Click to expand...

That's part of what I've been doing as I have time these last couple months or more since I got them--individually charging them with the Sorensen, then seeing what they do just sitting, then under a load. I've only checked out about a tenth of the cells so far, though.

To initially balance the cells, perform the tinning on each cell with an iron as hot as you can get, and solder with as low of melting point as you can find. (lead free is a joke).

Click to expand...

The hottest I have is a "50W" adjustable temperature iron. But it tins a spot on the positive ends in less than a second, and the negative in about two seconds. The solder I'm using is at least two and probably three decades old, and is most definitely not lead-free. It works very well.

Solder some tiny little scrap wire to every positive end, and negative end. Twist together all the positive ends, and twist together all the negative ends. Let this sit for a few hours, and the SOC will normalize between all the cells, and you will start out with every cell in balance before you assemble the pack.

Click to expand...

I hadn't wanted to do that for risk of high currents between cells, but I realized during more reading not too long back that any high current would be a very short duration, and probably would not cause any heating, if the cells are within some reasonable percentage of each other, SOC-wise.

Good luck!

Click to expand...

Thanks--I am pretty sure I need it.

MitchJi said:

I'd dump the BMS's and charge each parallel string with individual DC-DC convertors and use CellLogs for LV Monitoring.

Click to expand...

If someone donates some used CellLogs and DC-DC converters, I'd be happy to. (remember--I don't *just* build stuff out of junk because I *want* to--it's usually my only source for affordable parts!)

At the moment I don't even have a BMS (that works) to charge with--just the damaged one I discussed above, which I might be able to fix (but even if I do, it still won't charge these cells fully, as it starts shunting just before 4V, and from what I can find I need to charge to 4.1-4.2V). Kind of odd, since I had thought the spec sheet for the controlling chips on the BMS gave LiCo type HVC/LVC specs.

I am working on an idea to use some broken celphones to charge each parallel string, however I first have to determine the operation of the celphone battery's BMS board during charge, and then use it's signals to wire out to a much bigger MOSFET to control current flow from a much more powerful charging supply, so it can charge more than one cell at a time without blowing up. Right now I don't have enough time to put into that (or most of the other projects I want to explore).

Alternately I have a bunch of laptop packs whose BMS might be able to charge cells up correctly, but they'll only do maybe 3 parallel cells at best, and 4 series cells. Many of them can't be used without the laptop they were for, as their little BMS needs to talk to the laptop to be told to start charging, etc. Some will charge if power is hooked up, and cut off charge when it thinks they're full. Those I can use.

I may be able to simply rig up a comparator circuit to control a MOSFET, and use the 5V output some computer power supplies to then build some little current-limited buck converters to get it to 4.1V that will be disconnected from the cells at full charge by the comparator / MOSFET.

So right now I'm still just sort of playing around with the cells, until I figure out a charging solution.

Hi,

If someone donates some used CellLogs and DC-DC converters, I'd be happy to. (remember--I don't *just* build stuff out of junk because I *want* to--it's usually my only source for affordable parts!)

Click to expand...

You scrounged the Sorenson's. Maybe you can find some DC-DC converters.

Would two of these be affordable?:

Best damn 6S lithium monitor on the planet for $15

Click to expand...

http://www.endless-sphere.com/forums/viewtopic.php?f=14&t=9229

MitchJi said:

You scrounged the Sorenson's. Maybe you can find some DC-DC converters.

Click to expand...

Probably eventually. Could be a long time, though. The Sorensens came from a friend who was forced to move and couldn't keep everything, so a lot of the stuff that didnt' actually work, like these, I ended up with for helping him move the rest of it.

If I can build a current limiter for it, and figure out a way to lower it's output voltage significantly, I even have a 200A 5V supply (but it has to sit on the ground because it's too heavy for me to lift to my table!). Then I just need to parallel *all* the cells to charge them quick.

Would two of these be affordable?:

Best damn 6S lithium monitor on the planet for $15

Click to expand...

http://www.endless-sphere.com/forums/viewtopic.php?f=14&t=9229

Click to expand...

Eventually, but not till I can scrape up more $$. I haven't worked off what I owe a couple friends for the Fusin on DayGlo Avenger yet. Might manage to do that in the next few days, though, then at least I will be even instead of in the hole.

But since those are only 6s, I'll probably need three to keep track of 48V worth, if I end up going that route (since 48V would require less current from them to accomplish the same power output; easier on the cells). (assuming 3.7V nominal each)

Be very careful with those cells, especially the cells which won't hold voltage. Treat them each like little bottle rockets, because believe me they are...

Hi,

I hope you don't mind another wild idea.

The Gary/Richard BMS (v4 almost ready) sells for $60.00 plus about $60.00 worth of parts (16 channels) from Mouser. I think Andy charges about $200 for an assembled and tested 16 channel BMS.

Assuming you have the knowledge to do this (it sounds like you do) I bet if you post an offer in the For Sale Forum to assemble and test one or two in exchange for the parts for another BMS you can get one for no cash outlay.

MitchJi said:

I hope you don't mind another wild idea.

Click to expand...

I am all up for wild ideas.

The Gary/Richard BMS (v4 almost ready) sells for $60.00 plus about $60.00 worth of parts (16 channels) from Mouser. I think Andy charges about $200 for an assembled and tested 16 channel BMS.

Click to expand...

Yeah, this is the BMS I have looked at most often, as if I had time and found all the parts (I have a LOT of stuff in old SMPSs I could use for parts for it) I could build this on perfboard. I just never have enough time to dedicate to it when I'm feeling well enough or am not worn out.

Assuming you have the knowledge to do this (it sounds like you do) I bet if you post an offer in the For Sale Forum to assemble and test one or two in exchange for the parts for another BMS you can get one for no cash outlay.

Click to expand...

That is definitely a thought. The catch being whoever wanted to have me build them would have to be patient; I am not quick at getting things done, even when there is a good reason to hurry on it.

It's one reason I don't have any side-businesses; I have enough trouble working at my day job, which nowadays is only from 1 to 4 days per week, with up to 6 hours a day if I'm lucky. Last few months I haven't even been doing much in the way of job-hunting, which used to take up a significant amount of my time--it's depressing.

Anyway, I definitely appreciate the idea--I will likely do this, and post some of the other things I had intended to do before, as well, including DC-DC converters from old celphone/etc wall power supplies, tested to different voltages and output powers (if I can get them all tested).

grindz145 said:

Be very careful with those cells, especially the cells which won't hold voltage. Treat them each like little bottle rockets, because believe me they are...

Click to expand...

Yeah--I've seen enough videos about that! Seems that they tend to die when recharged, so I'd be most careful during those times.

One of the battery cases I considered was a small aluminum beer keg, and another an old fire extinguisher. I have a handful of those around, and at least one has a broken valve-end on it (came from the alley, I think, years ago).

Where did you get the batteries, if you don't mind me asking? Have you thought about scraps of PVC pipe from a construction site for container? Although the fire extinguishers might put folks at ease as you drive by. :wink: Have you thought about holding a sign at the end of an exit ramp? You could say something like, "need cash for volts" Some of those guys make bank.

I actually have some PVC pipe here in scraps already, but the inside diameter is pretty large compared to the cells. They'd rattle around a lot.

The cells came from a for-sale post here on ES toward the end of last year, as "new pulls" from unused battery packs for (I think) medical equipment. I think I recall them being intended for an EV project but not getting used, and thus offered for sale, with at least 50 of them known bad and about 300 theoretically good.

http://www.endless-sphere.com/forums/viewtopic.php?p=242212#p242212

Amber,

I have 1 spare CellLog and a handful of BM6 units which you are welcome to. Im sure someone else has another celllog8 to spare and with a little mod (to ensure consistent voltage from all cells in the pack) 2 of the cell log 8s will cover your 16s pack.

The BM6 units are already modified so when the beeper goes off they trigger an optocoupler output which can be linked to form an automatic LVC/HVC.

The celllog8s would be a better fit and give you more useable info in a nicer and easier to use package.

Let me know.

-Mike

PM'd

I am getting closer, at least BMS-wise. Thanks to AndyH I now have most of a pair of 16s v2.6 F/G BMS that I can build up from bare boards and parts.

I have a lot of the parts myself, for resistors, caps, the little diodes and such, and good-enough substitutes for the power transistors for shunting.

I've put a list of the parts I still need here:
http://www.endless-sphere.com/forums/viewtopic.php?f=8&t=18333&start=0
in hopes I might save a bit on the shipping if one person happens to have all of the parts I need, and can spare them at their cost plus envelope-shipping to me.

If no one does by the time I've built the boards up (probably take a while) and saved up some extra $$, I'll order from whichever place has the cheapest total including shipping for all the parts.

I've still gotta build the actual cell packs, but even that has to wait till I'm done testing. I'm so slow.

I think using the celllog stuff on a ride would be a better solution than carrying the BMS around, and then only use the BMS for charging. Might save some wear and tear on it.

I've got most of the parts for the v2.6 BMS now; I forget which ones I am missing but will post a new needed list in this thread:
http://www.endless-sphere.com/forums/viewtopic.php?f=8&t=18333
once I get a chance to go back thru the parts list vs. what I have. Maybe Saturday, I hope. Some of them I think I can substitute stuff I have; others I am pretty sure I'll be buying from someone (preferably someone here on ES rather than Mouser/Digikey/etc.).


I might've already posted this before in a different thread (can't remember and can't find it now), but I have a charger for the pack(s) now, too:


Sorenson DCS 55-55; 0-55V at 0-55A current-limiting. On indefinite loan from a friend, until I come up with something more suitable or he needs it.

Had to change the 3-phase twist-lock NEMA plug for a dryer-style 3-prong 2-phase NEMA, as that's all I have in my house it'll work with (well, the stove, but I didn't have a spare cord for the 50A stove type, and I did for the 30A dryer plug).


I've also been working very hard in what "spare" time I had to find friends and friends of friends to work for enough to earn "extra" money to cover enough other things I was way behind on and to get some luxury items, like that Lyen controller + 9C motor kit, and also to finally pay off (over a few months) the TS pack Mechanix was selling, which should work well on either CrazyBike2 or on the cargo trike.

It's 10 TS 60Ah LiFePO4 cells, so about 33V nominal; sufficient for CB2's already over-voltaged-at-36V powerchair motor drive. Should be less sag than the SLA, with much more usable Wh out of it, and not very much more weight (about 50lbs TS vs 45lbs SLA. Also, in theory they'll fit better on the frame, as their shape and size in theory allow me to also bolt them compressed between plates on the front frame as well as the center inter-frame section the SLA currently reside in. On a cargo trike like ARTOO, they'd just get a dedicated battery compartment or two, near the bottom center of the front 1/3 of the trike (tadpole) or rear 1/3 (delta).

Should be here soon.


Now to get that BMS built....

The TS 60Ah cells arrived today; a bit lighter than I posted, at 42lbs for 10, rather than 50lbs. All at 3.31-3.32V.


Gotta dig out some bolts, make connecting cables, and then perhaps test them out on DayGlo Avenger with the 9C after I finish fixing the caps in the controller.
http://www.endless-sphere.com/forums/viewtopic.php?p=284045#p284045

Won't have a BMS yet but they should be ok for a quick "what can they do" test.

Before I charge them up again or use them deeply I'll want to finish the v2.6 BMS LiFePO4 versions, at least for the LVC/HVC part if not the shunting. Gotta finish making the list of what I still need for them, and post that back in this thread:
http://www.endless-sphere.com/forums/viewtopic.php?f=8&t=18333&start=0

Then I *really* need to finish up testing of the LiCo cells and their BMS, and make a pack or packs of them, too.

amberwolf i have ypedals old cells. try buying andersons and a single cell charger. works well and its cheap :wink:

Perhaps, but I think I've spent all the money I can afford for the moment.

I actually have a good pack charger for it now, a Sorenson 55V 55A (adjustable voltage and current limits) lab supply; I still need to finish the F/G BMS though. I did test charge the cells to see if they would charge with it, watching with a meter on each one, for a short time, nowhere near enough to top them off I suspect.

Didn't yet discharge test it as a pack on a bike yet, because I didn't have the right bolts and washers and stuff, although I do now (since December, I think it was). Still have to finish making all the interconnects and then do a test run with it. At the rate I'm going with various projects (currently working 6 days a week), I'll be able to do that by next Christmas. Pretty stupid considering I spent way more than I should have (or really could afford) on them, and haven't even used them at all yet.

Plus I think that to really use them effectively I need to get at least 5 more of them, maybe 6, to get a 48V nominal pack voltage. (and I can't afford to do that)

Kinda depends on if I use them with a hubmotor bike or with a chain-drive I can just gear differently for the lower voltage (which is what I intended them for when I got them).

So, after lots and lots of stuff going on for more than a year, I finally got back to this thing (actually, bakc to both battery packs).

TS stuff I have noted here:
http://endless-sphere.com/forums/viewtopic.php?p=466685#p466685
in my new bike's build thread.


I have managed to lose all the notes I'd made about what cells were what, and so I am retesting these LiCo laptop-style cells. Tonight I sorted them by voltage, and most of them are still good (more than 2/3). I did more htan just a go-nogo sort becuase it also shows me which ones are more capable of holding a charge over time, since all the non-0V ones had been charged up to 4.2V, IIRC, during my previous testing. It's been probably a least a year or more since then, so those still holding most of that charge are pretty good, I guess.

The green ones (Sony) are marked as column G, and the red ones (Sanyo) are marked as column R. I sorted them by tenths of a volt difference, basically, with exceptions at the low end.

Voltage G R
0.0-1.0V 81 8
1.0-3.0V 25 10
3.0-3.2V 6
3.2-3.3V 8
3.3-3.4V 6
3.4-3.5V 7 2
3.5-3.6V 2 6
3.6-3.7V 22 6
3.7-3.8V 66 26
3.8-3.9V 8 44

The red Sanyo cells appear to be better cells--a higher proportion of them are still good at a higher SOC. Unfortunately I have far fewer of them, only about 1/3 of the total of these cells.

Dead cells, and probably unusable cells:
View attachment 1


The good stuff:


Anyway, I think that I will probably end up simply taking everyting that is 3.0V and above, solder them up in 1SxP groups (based on grouping above) with small wires, and charging them all up to 4.2V. Then I'll parallel them all to get them all to the same SOC.

Once that's done, I can then discharge each one with the same load (not sure what, yet) via a wattmeter, and note the capacity on each. Then I can see if I have enough of them with enough capacity to build a testable traction pack from, for experimental purposes (since I don't really need it for transportation anymore, but it would be nice to have as a backup pack, and I'd really like to see for myself if these laptop-cell packs are a viable solution or not, and how well they perform).

when you do test them i think this should be added to the build :wink: