LiFePO4 battery failure - what can I do?

[elastictrickery]

Hi everybody,
One of my LiFePO4 batteries has gone down.
In my LWB recumbent I have two Valence Saphion batteries of 60 cells each configured 5s 12p - 16V 14Ah. I used to run one after the other, until I discovered I could get quite a few more miles by using them together (16V 28Ah) they are separate until I switch the two 70amp relays. This has worked well until yesterday. As it’s been really cold and wet I haven’t used the machine for several weeks, so I decided to ‘exercise’ it. Both batteries were reading 16.56V and 16.54V when I started them off spinning the back wheel at 30 mph in fourteenth. I put a clamp on the Magura twist grip to hold it open, and checked it every 15 minutes. The Hacker controller was set to cut at 11.5V
After about two and a half hours I was surprised to find the revs had dropped and the data-logger showing 9.8V!!!. I closed the throttle immediately and the (combined) voltage crept up to 12.1 V. I have no idea why the ESC hadn’t cut. And I was somewhat annoyed that the logger was showing that only 21Ah had been used – when I’ve done this in the past It has shown nearly 30Ah but that was in the summer - It’s very cold in my workshop now – could this be part of the problem?
In retrospect it is unfortunate that I didn’t check the voltage of each battery individually before I started recharging. Number 1 battery came up to full capacity in 1hour 56 minutes. I put Number 2 battery on charge – but after 36 minutes the charger cut out at only13.3V. Some thing doesn’t add up here. Having taken both batteries out of the case I checked each 12 cell parallel string - four of them are reading between 3.34 and 3.32 - one is dead. There is no swelling or discoloration on any cell
I suspect that Number 1 battery was only down to 14 volts or so when I switched off, as it would normally take about three hours to recharge if it was down to 11.5V. I haven’t got the batteries wired for a balancer –I was told by the supplier it wasn’t necessary with this chemistry. And five years ago when I made this machine I couldn’t find a balance charger for LiFePO4 batteries as big as these. I had just decided to fit balance leads - in fact I was looking at leads, plugs and adaptors in a catalogue while all this was going on! Ironic or what!
What do I do now? Will the whole string of 12-cells be irrevocably dead or just one cell? If it’s just one cell - can I fit a new cell? I suppose I’ll have to tear the whole battery apart to do anything. I would be very grateful for suggestions.

Thanks.

 

[monster]

if you can get replacement cells you should just beable to swap them over and it will work. you should get a bms though they are good at protecting your batteries. i have no experience with this pack you understand.

 

[elastictrickery]

Thanks Monster,
I'm not sure if I can find replacements after this length of time. I suppose there is no alternative to breaking down the faulty 12 cell string - do I try to charge each cell in the hope of finding the damaged cell or more to the point the undamaged cells? If two cells have survived I could shorten the pack to 10p.
But what I'm really bothered about is why this has happened. Reading all the posts on this subject seems to suggest that Below 2v per cell it's like dropping off a cliff - and perhaps that tipped this string over the edge.
I certainly will fit a BMS and I'll never trust the ESC voltage cut-off ever again.

Thanks

 

[dnmun]

had you ever tested the controller LVC? why did you think it would shut down? cell level LVC through a BMS does help prevent this from happening. and then you could rewire into something like 36V or 48V with all your remaining batteries, and just buy a signalab BMS from li ping.

 

[dogman dan]

Best bet might be to make one big pack out of the remainig good cells, like 20p 5s or whatever you have enough left for.

Charge the pack and look for weak cells , then discharge the pack and look again. The bad paralell group should show up.

 

[elastictrickery]

had you ever tested the controller LVC? why did you think it would shut down?

No, my grasp of electronics is minimal - mechanical engineering is my forte - so I wouldnt know how to test the LVC or even think I would need to, and I assumed it would shut down at 11.5V because it says so in the manual. Why do you suggest I should rewire to 36V or 48V? I need 16V for this machine.

 

[elastictrickery]

Best bet might be to make one big pack out of the remainig good cells, like 20p 5s or whatever you have enough left for.

Charge the pack and look for weak cells , then discharge the pack and look again. The bad paralell group should show up.

Why make an even bigger pack? I was told that the longer the paralell string the more trouble I would have. Not sure what you are suggesting - how do I look for individual weak cells? I know where the bad paralell group is.

 

[dnmun]

when you said dead, what was the voltage? if you pushed current through them when they were lower than 2V there is not much hope, but you can still cut it apart and see if you can charge any of the individual cells will hold a charge.

you test the LVC by reducing the input voltage to the controller until the controller shuts off the output power.

your controller should be able to be used up to 36-48V if you add some resistance to the input power resistor. or use a DC converter to deliver power to the circuitry and use 36 or 48V on the output FETs.

if you can do some engineering, you can do this stuff. good excuse to study electrical engineering stuff like circuits and how they work, and how to understand power.

 

[dogman dan]

You could also make two packs of 11p then. or whatever you have enough cells left for not sure what you mean by the benefits of a smaller pack though. whether its 12p 5s twice or 24p 5s, either way its 24 cells paralelled to me.

You do need a bms though, and for two packs you need two. If you run a bms, then more of them on smaller packs would be a greater level of protection I suppose.

I did the same thing, I read 12p 5s, but I just saw it to be what we see lots of, a 12s, 36v pack. Not many bikes running 16v out there. had to delete a post when I read your first one again better.

 

[elastictrickery]

Hi dnmun,
When I said dead I really mean dead! Nothing, it's as though that string isnt there at all! The rest of the battery checks out as a fully charged 4 series pack - 13.32V. I'll slice it up tomorrow see what's what .Thanks for the tip on how to check out the LVC - I'll do that before I rely on it again.

Hi dogman,
For the time being I will have to stick to two separate batteries as the custom battery box is an integral part of the frame and has a divider for stiffness.
It must seem odd to use 16v so I'll briefly expain. When I was first designing this machine in 2004 I was advised I would need a big outrunner motor. I wanted a Torqueman but just couldn't get one. So, I went for the only other big outrunner on offer - the Flyware TorquRex. I wanted 250Kv to run on 32V but I was told that was too difficult to wind (?). So I changed the belt reduction box to 19.2:1 and halved the battery.
I'm not to impressed with the motor - I dont think it's very efficient. The magnets are flat - have not been machined so the air gap is bigger than it needs to be and it's noisy. Been wondering if I shouldn't put in a 200Kv Sporpion motor, double the size of the primary pulley to halve the reduction to 9.6:1 and reconfigure the battery to 32V (10s 11p) From what bit I've gleaned this last few years this just might be a win-win. What do you think?
Thanks now.