from March 7, top of page 51, looks prophetic now!
Trackman417 said:
GCinDC said:
charged to 96v as usual at work w/ modded bmsbatt charger, then rode in extreme wind using 5.6ah.
towards the end, the CA tot voltage dropped to 81V under load.
then at rest and the 20s pack paralleled up to 5s, cell one is looking ugly and has clearly been getting hammered, esp the weakest ones that make it up:
i never mentioned it, but the new 4s pack I've got in series now is a 25c pack, so it may be acting like a bully in there. :lol:
I would say my last fair wells to that cell. That is pretty scary, looking at how low that cell is compared to every-other cell. Couldn't that one cell be the majority of all of the voltage sag problems you have? I would get rid of it at the sake of keeping your house. I don't need the risk of your gloves catching on fire, then where would I be? 
and yes, trackman, the gloves are gone. they were melted under my helmet...
so let's back up and talk about what caused the fire. we may never know, but a few possibilities are:
- the psu or bc168 blew up and started the fire. unlikely. the bc168 shows almost no damage. as soon as the toilet case underneath melted, it fell to the ground and there's little burn marks on it, only melted mush on the bottom. the psu looked exploded, but it was wedged between the bench and the bike, and sat there grilling for the entire cookout.
- a cell could have been punctured. certainly possible but this was an old pack (except for the 4s group that sat on top) that had 400 or so cycles on it, and has been sitting in the same config for a year. and while the frame is steel, there are no sharp edges on the inside. and i've had no major recent wipeouts. still a possibility...
- the balance harness which was charging the the pack @ 8A. it could have shorted, but would that have resulted in a fire? all other shorts i've experienced result in vaporized wires and broken connections.
EDIT: just before posting this, i saw BigMoose's post about the db25! i humbly accept the advice for all future wiring. but BM, is it still likely such a short/failure catches the pack on fire? duh, i guess so. i keep thinking of the balance wiring like sizzleers under that much current, and i guess exposed, they'll catch everything else of fire..
- there could have been a broken wire in the balance harness that i replaced a month ago. the electrical tape looks like hell, but i cut, stripped and soldered fresh wire underneath each. no cold soldering bs.
but if a balance wire failed, how would a cell get overcharged using this charging method? it would be understandable if i were charging via the power leads and a broken balance wire could no longer provide a signal about its voltage. instead, if a balance wire were broken, the cell wouldn't get charged, and then couldn't overcharge.
- a cell overcharge (4.5V+?) does seem to be the likely explanation... esp considering the evidence of how puffed out the bottom pack was, to have widened the steel frame. (see the pack remains vid about
here).
so how could a cell have been overcharged?
let's recall that this pack has been punished:
- i charged it at 10A/1C in freezing temps (with bulk charger).
- i discharged it at 60A in freezing temps.
- i regularly balance charged it to 4.2V using a cheap unconfigurable rc charger, when paralleled to 5s, before i got bc168.
- i undercharged it (to 4v/cell when bulk charging to 96V for 24s) and then rode beyond 80% capacity in freezing temps.
poor treatment to be sure, but nothing to CAUSE FIRE.
so now let's consider the critical phase, the last couple weeks...
as requoted at the top of this thread, on March 6, i charged only to 96V (probably more like 95.6V) then rode home using 5.6Ah and noticed at one point the CA voltage dropping to 81V.
what about the individual cells? well, since i sold my lipo buzzers, i didn't monitor the cell voltages under load! and this is probably the critical failure. (i'd been waiting for replacement voltage monitors to get in stock at HK!). and instead i had only two bm6s left (one of which was unreliable, reading cell #2 at .5V too HIGH and caused general distrust of the devices), shown attached here to monitor:
so inadequate monitoring at a time that was critical.
but when i got home, i checked all the voltages at rest (never posted this):
[youtube]2os0YDOtz4U[/youtube]
if you watched, you may have noticed that only three cells were checked... that's cause one of the wires broke off my discharge monitor harness:
so for the past two weeks i've not been able to check that cell group.... NOTE: this was NOT used for charging...
anyway, then i checked the three cell groups individually with the more accurate bc168, and found some cells lower than others, but nothing lower than 4.3V. and then i paralleled all of them, at which point they read:
since about then (3/11 actually), i ONLY charged using the bc168 at work and at home, and i would parallel all cells immediately after the ride, so i wouldn't notice any discrepancy between cell groups. (i did however out of curiosity always first check the 4s pack voltages, then the 5s8p voltages, before paralleling them both via the bc168 board - the 4s pack voltages were always closer and somewhat higher than the 5s pack).
this monday, i upped the bc168 charge to 8A. it had been at 5A before that. and it was set to charge to 4.18V/cell.
but each time i'd hop on the bike in the morning, the charger left on all night, the CA would read 99.9V. and by the time i'd start up the hill at mass ave, after .4ah or so, the CA voltage would dip to 84V. that's quite a lot of sag for lipo, and if we consider that the new 4s 25c pack sagged little, then the 20s pack was sagging a LOT more, and most likely one or two cells in particular!
and on March 6, for example, if the total voltage dropped to 81V, and we speculate that the new 25c 4s pack only dropped to 3.7v/cell (x 4 = 14.8v) and we subtract that from the total voltage (81-14.8V=66.2), and the divide 66.2 by 20 cells, the average cell voltage is 3.31V, which i guess is tolerable, except that we know from my vid of there was a voltage delta of .4V AT REST (87V), god knows what the delta was under 60A load...
for posterity, the data from the vid shows (italics for unknown/calculated values):
A1 3.3
A2 3.7
A3 3.65
A4 3.72
A5 3.72
18.09
B1
3.31
B2 3.73
B3 3.7
B4 3.68
B5 3.69
18.11, delta: 0.40V
C1
3.36
C2 3.71
C3 3.69
C4 3.705
C5 3.705
18.17, delta: 0.351V
D1
3.3
D2 3.7
D3 3.65
D4 3.72
D5 3.72
18.09, delta: 0.40V
the fact that while the packs have ~0.4V of imbalance AT REST, each cell row is fairly well balanced... #1 at 3.3V, #2,#4, #5 at 3.7V, and #3 at 3.65V...
again, this was two weeks ago, and we don't know what's been going on with the first group since i couldn't monitor it separately..
hmm. i thought i was narrowing in on the cause. but now i'm not so sure.
i guess the important question i was getting at: how low did the voltage of Cell 1 get UNDER LOAD? its voltage popped way up above the others when charging, so it was definitely circling the drain. in its death throes. but was the the mystery packs cell 1 much worse than the others? or did the group fail together?
let's say they all dropped to 2.7V together. this is No No land for lipos... then what happens? each successive charge is like rolling the dice? the packs puff and puff and puff until one goes...?
anyway, i hope the facts in here help. there's no reason to get hysterical about these...
some changes i'm considering:
- nanos: less imbalance, better quality, hopefully. quit using them when they're so close to end of life.
- better balance wiring
- fire alarm in the garage! with that alone, i don't see need for blankets, etc
oh yeah, i meant to answer other q's:
- gopro was in the house but the case is destroyed
- wife was surprisingly cool. she'd been on work retreat, so was unusually relaxed and 'centered', and i told her over the phone when she was about to board the plane. she returned home at 9:30pm when i was out taking pics, and was amazed, if this particular expression doesn't show it.
and itemize the damage.. and they clean the inside of the garage to whatever extent they need to so that there's no residual smell. 
but sharing for the greater good.
