Skip to main content

48v Bike battery pack investigations

As I think I mentioned a few posts ago, my 48v bike battery seems to have a problem with cutting out under full throttle.  This is the case even immediately after charging.  I haven't known much about the battery pack, since I bought it used and it came with no documentation, but after reviewing lots of eBay auctions selling similar items, I have a pretty good guess at what I'm dealing with.  I think I've even figured out what happened, and what to do if there's any chance of fixing it.

So I believe I'm dealing with a 16S LiFePo4 pack with a nominal Vcell of 3.2v, (or with all 16 cells in series, 51.2v Vpack).  I disassembled my BMS (battery management system) so I could get a look at how it was configured. The battery has a main output (2 wires), as well as 16 wires (one for each cell junction) that are all connected to the BMS.  This allows the BMS to measure and even balance the cell voltages within the overall pack.  The BMS can shut down the output or charge if any of the individual cells are out of safe range.  Incidentally, for a LiFePO4 cell, this is quoted as ~2.2v (discharge) and ~3.9v (charge).

My pack voltage settled out around 49v after full charging cycle.  This is low, about 3-5 volts lower than it usually is.  I then measured the individual cell voltages using the 16 wires.

B1 = 0.0v
B2-B16 were all somewhere between 3.1v to 3.7v, with quite a bit of variation.

As you can see, B1 is a big problem.  I didn't really know what to do.  I know that no cell should ever be lower than 1v no matter how dead it is.  It'd basically have to have been shorted.  After thinking and researching about it, I realized my mistake.  I stored the battery for the winter months with the BMS attached.  I read somewhere that a lot of BMS will use the first cell to power its circuitry.  If this is true in my case, the BMS killed the first cell down to nothing.

Further intuiting has lead me to the conclusion that the BMS can't balance the pack when its power ain't on.  So that explains 0v on the first cell, and no balance on the pack.  The BMS at least threw the pack way out of balance, and at worst, destroyed the B1 cell.

Since I don't have a 16S cell balancer, or an individual cell charger, the first thing I tried was to leave it on charge for a long time.  I noticed the charger did briefly cycle on and off about every minute for about an hour or two.  I then unplugged the BMS and left it overnight.  I then measured the following:

B1 = 2.8v
B2 = 3.3v
B3 = 3.3v
B4 = 3.3v
B5 = 3.3v
B6 = 3.3v
B7 = 3.3v
B8 = 3.3v
B9 = 3.3v
B10 = 3.3v
B11 = 3.3v
B12 = 3.3v
B13 = 3.3v
B14 = 3.3v
B15 = 3.3v
B16 = 3.3v

This is a very good sign.  My B1 cell is not destroyed (though it may be damaged, but it's still alive).  It's clearly still badly out of balance.  The solution seems to be one or a combination of the following:

-Somehow kill batteries B2-B16 until they are around the same level as B1.
-Somehow charge B1 individually to 3.3v.

Whatever happens, it needs to be done through the dinky little 20 gauge wires that are attached to the cell junctions. At this point, it's possible the BMS will have enough power to do its job and start balancing the cells, but the big question will be, if I leave the BMS plugged in, will the first cell go dead again.  As far as I know, from some basic research, the BMS doesn't charge individual cells - it only shunts the other ones in an effort to drain them to the lowest cell voltage (I believe I read it has a max of 70ma).  It would seem to me that it would have a lot of work to bleed off that much power, since the lowest cell is so drastically different than the others.

Comments

Popular posts from this blog

Timbaland rips off a Demoscene artist

I knew this day would come. The new Timbaland/Nelly Furtado song "Do It" uses a song made in 2000 by Finnish demoscene artist "Tempest" (Janne Suni). It's a 4 channel .mod (the ripoff is from a playback using the C64 SID soundchip). The song was hosted on scene.org's servers (the main repository for all everyones demos and tracked music, etc.). As you might expect, no permission or royalties were paid to Tempest. Just to clarify, we're not talking about some kind of coincidence here. There is no question that this track was used to create the song "Do It". In an interview, Timbaland tries to downplay it, saying things like "he sampled it from a video game". (This track was not written for a video game- it was actually written for the 2000 demoscene music competition, in which it won 1st place). Regardless, he basically claims he has no legal obligations because it's just like all the other pop artists that sample other m

Reaper, Linux, and the Behringer X-Air - Complete Studio Solution, Part 1

Introduction and Rationale This is part one of a major effort to document my experiences with recreating my home studio, entirely using Linux.  Without getting into too many of the specifics, a few months ago I decided that I was unhappy with Windows' shenanigans - to the point that I was ready to make a serious attempt to leave it behind.  For most in this situation, the obvious choice is to switch to Mac OS.  With its proven track record, support, and options for multimedia production, it is naturally the first alternative to consider if your goal is to simply use something other than Windows. For me the choice was not so simple. I despise Mac OS and, in general, the goals and philosophies put forth by Apple in an effort to ostensibly provide users with an "easy" working environment.  It does not help that I have also failed to find any aspect of the Mac OS UI intuitive, but I realize that this is a subjective matter. With my IT background and user-control* favori

Windows 8 audio clicks and glitches narrowed down to Malwarebytes

Ever since I got my Windows 8 PC, I have been having serious problems with audio.  Basically all sound playback on my system experiences a brief  but frequent click, skip, glitch, stutter, whatever you prefer.  I can reproduce the issue on any sound card or firewire sound interface (devices tested include the onboard Conexant SmartAudio HD, my external Phonic Helix 12, and my Edirol FA-101).  All of them seem to have audio clicks, with the firewire interfaces' clicks seeming more harsh for whatever reason.