Battery Testing Data - Reports from the Road

[Russell]

Scoot,

You didn't check the individual cell voltages before you bulk charged the packs? :|

-R

 

[Russell]

Wookey,

Have you checked the individual cell voltages in a discharged state? You really do need to have data points for the cells discharged and charged to determine what's what. For example my cell #11 (of 15) goes the lowest on discharge but also is one of the highest when charged which means it has the lowest capacity but it also means there's nothing I can do about it. If that cell had the lowest voltage on discharge and exhibited a lower voltage than the rest when charged it would be a different matter and I could simply bring it up a bit with a single-cell charger so that it more closely matched the others.

I have not been running my 15-cell 10Ah LiFePO4 battery pack without a BMS for long now, just 9 cycles so far, but I found my pouch type cells remain tightly grouped when discharged. I'm actually surprised how closely matched they are considering they are lower-end generic cells but the manufacturer must have their process well under control and having cells very close in capacity is the real key to a good battery. Once charged (I use a charger which averages 3.62V/cell) the spread is wider (0.10-0.15) however no cell goes higher than about 3.70V. Oh yeah and I slow charge using a 2A charger too which might help. I have a 4A charger but I have that set to a higher 3.70V/cell for when I do wish to top-balance the cell voltages using the original BMS.

-R

 

[scoot]

Scoot,

You didn't check the individual cell voltages before you bulk charged the packs? :|

-R

Actually, yes I did, and they were all at 3.32 to 3.33 volts... not surprising to me actually as my other pack usually settles to 3.35 volts per cell after resting over night... even when they are out of top balance. i am getting the impression that this settling phenomena is pretty normal for Headways, but have heard others proclaim otherwise :?

 

[Russell]

Actually, yes I did, and they were all at 3.32 to 3.33 volts... not surprising to me actually as my other pack usually settles to 3.35 volts per cell after resting over night... even when they are out of top balance. i am getting the impression that this settling phenomena is pretty normal for Headways, but have heard others proclaim otherwise :?

I've been contemplating building a battery pack (sans BMS) using the Headway cells however if they are showing that kind of variance from cell to cell, something Wookey's data shows too, I'll pass on them. My present pack uses these cells http://www.alibaba.com/product-gs/255031097/ATL_3_2V_4950mAh_55A5D0_LiFePO4.html and they are very close in capacity but I wouldn't want to build my own pack using them because of the delicate and hard to solder tabs, plus they are basic 2C max cells.

For reference I charged my pack yesterday evening to 3.62V/cell and this afternoon it was at 3.46V/cell. Then this afternoon I wanted to see how much capacity that amounted to (I knew it would be small) so I charged it back up with a WU meter in-line and it took 0.012Ah or 0.6Wh to bring the pack back up (about the energy I use over 250 ft :wink: ).

-R

 

[scoot]

I fear you may be mis-interpreting my data Russel:? I wouldn't conclude anything particularly negative about the Headways quite yet.

I think one of the points of this thread's recent discussion has been to address "naked packs" and the hidden impacts of bulk charging (in this case). The cells I charged today were about as naked as naked can be. Fresh out of the box with no flexing or pre-balancing performed whatsover.... just basically slapped together at the factory. I thought it would be an excellent opportunity to further illustate Andy's teaching points. And boy, did it ever I think

I think the only conclusion one can walk away with in this case, is just that these cells were not equally filled up from the factory, and the consequences were very evident when charging without any high voltage control on the individual cells and/or balancing.

In regards to the headway cells themselves, I think my first pack kicks ass . I am running 8s2p with a powerfull Torkinator motor pulling 2000 watts (90+ amp bursts) and 50 amp continuous runs. Like some folks, i am not running a BMS yet, and until I do I am learning how to use these cells without trashing them... mostly through close monitoring.

 

[AndyH]

Beautiful, Gentlemen!

One of the 'beliefs' that's been making the rounds is that only 'low quality' cells need management. The cell manufacturers - even the folks that make the 'high quality' cells require management and say that not two cells are alike from the factory. It doesn't appear to matter if they're A123-Systems cells or Ping.

Good Job!

 

[Russell]

I fear you may be mis-interpreting my data Russel:? I wouldn't conclude anything particularly negative about the Headways quite yet.

.

No I don't think I am. Take a look at Wookey's data above for a good example of the spread between the cells in a Headway pack without a BMS. By comparison my pack of pouch type cells without a BMS is showing excellent results so far. Granted my results are preliminary but the cell voltages I'm getting is what I want to see from any pack I construct.

-R

 

[AndyH]

I fear you may be mis-interpreting my data Russel:? I wouldn't conclude anything particularly negative about the Headways quite yet.

.

No I don't think I am. Take a look at Wookey's data above for a good example of the spread between the cells in a Headway pack without a BMS. By comparison my pack of pouch type cells without a BMS is showing excellent results so far. Granted my results are preliminary but the cell voltages I'm getting is what I want to see from any pack I construct.

-R

Can we really compare numbers until we know exactly how both of you prepared your cells before charging them?

One of the problems folks in the 'scooter world' have had with Thunder Sky cells is that they scooter manufacturers don't appear to be following Thunder Sky's guidance of balancing the pack on initial installation. Connecting a bulk charger to a pack of cells in states of charge between 30 and 80% is a recipe for early problems at both ends of the charge cycle. Cook some cells on initial charge, then destroy one or two on the first long ride...lather, rinse, repeat...then park the bike and wait for warranty to replace cells...

Russell - how did you charge and/or balance your cells when you first built the pack? How about you Scoot?

 

[scoot]

well toady's test on the new virgin packs... nada

My first pack (2 months ago), I parallel charged and discharged 16 cells gently (1 amp per cell) a half dozen times or more then re-constructed them into 8s pairs for the pack.

 

[AndyH]

I went back and looked at Russell's info as well - sorry, missed it the first time. He charged the cells separately then assembled the pack. You showed an initial charge with an unbalanced pack. I'm not surprised you each got different results!

 

[recumbent]

I found that the best way to monitor cells is to check all cells at a moderately drained state, Then check again fully charged.

The cells stay balanced for many weeks of use, if one is too high, I just drain it with an old car headlight with alligator clips attached as my resistor.

Now i use two CellLog monitors for my 16 cell pack, and never get into trouble that way. I've had two different BMS's destroy couple cells over the years. BMS's are not quite perfected either.

 

[Russell]

Russell - how did you charge and/or balance your cells when you first built the pack? How about you Scoot?

I did not build my pack, it is an Emma/Lau 48V/10Ah pack made with 32 5Ah pouch type cells. I used it for 38 cycles without a single hic-up then it mysteriously failed over the winter. It looks like the fragile tab on the last group, the one with the red main power lead attached, broke and killed the last cell group. I removed the last group and made it into a 15S pack.



I always wanted to try a pack "bareback" as people are calling running without a BMS so I turned the failure into an opportunity. The BMS didn't work with one cell group removed anyway so out it came. I then figured out if I jumpered the pins on the sense lead connector I could still use it as a balancer. I have one charger set to 3.70V/cell (55.5V) which I used in conjunction with the "BMS turned balancer" to initially balance the cell voltages to 3.59 +/- 0.01V. Thereafter I recharged the pack with a 54.3V (3.62V/cell) 2A smart charger.

So far this is working well as you can see from the cell voltages I posted on page 2. After 6 cycles I stopped checking the individual cell voltages since everything looks good. I did re-balance the pack after the eighth cycle just for the hell of it. I installed power-poles on the original BMS so it's easy to connect and disconnect it from the battery.

View attachment 1

I don't think with my cells it will be necessary to balance them often but I will do it every once in a while because it's simpler than checking the individual voltages. I did eventually want to construct a higher power pack and not use a BMS so I view this as a trial run to see how well it works.

-R

 

[olaf-lampe]

Hi folks,
I read the whole thread today and there is still a question from AndyH lingering in the air: What is wrong with the pack he initially introduced?
My guess is, that he mixed old and new cells together. The cells going sky-high at the end of the charge cycle are the old ones that already lost some capacity.

Another note I want to share is what I found out about low current charging: I initially charged a 100s1p 60Ah TS pack in small groups of 16s and 4s with two different chargers. The 4s charger had the same initial current of 30A ( like the 16s charger too) but current drops down pretty fast to 3-4A because of a lower voltage.
My observation was, that the slow charged cells never got above 3.5x V and they didn't show the steep bend at the end of charge. I left them on the charger for hours but voltages didn´t climb anymore. ( they only got warmer )
But when I took these cells and charged them with an additional lab-PSU ( 10A extra juice ) they almost immidiately climbed above the magic 3.65 V.

-Olaf

 

[dnmun]

russell wrote> "I did not build my pack, it is an Emma/Lau 48V/10Ah pack made with 32 5Ah pouch type cells. I used it for 38 cycles without a single hic-up then it mysteriously failed over the winter. It looks like the fragile tab on the last group, the one with the red main power lead attached, broke and killed the last cell group. I removed the last group and made it into a 15S pack.

I always wanted to try a pack "bareback" as people are calling running without a BMS so I turned the failure into an opportunity. The BMS didn't work with one cell group removed anyway so out it came. I then figured out if I jumpered the pins on the sense lead connector I could still use it as a balancer. I have one charger set to 3.70V/cell (55.5V) which I used in conjunction with the "BMS turned balancer" to initially balance the cell voltages to 3.59 +/- 0.01V. Thereafter I recharged the pack with a 54.3V (3.62V/cell) 2A smart charger. "

you should be able to repair your pack if there is any amount of tab left sticking out of the pouches. if you can solder to it, then do that, but if it is the aluminum tab, you can make a tiny clamp out of two little tiny metal bars and use screws at the ends to squeeze the ends of the broken tab between to make the connection.

if you decide to give up and use just 15S then how can you charge it? you are stuck with a big hassle.

in spite of that, if you wanna run on the 15S you can still use the BMS for discharge so you still retain some cell level protection by putting a jumper across the opto for the LVC signal and the HVC signal so it will not respond to their out of range condition.

 

[Russell]

you should be able to repair your pack if there is any amount of tab left sticking out of the pouches. if you can solder to it, then do that, but if it is the aluminum tab, you can make a tiny clamp out of two little tiny metal bars and use screws at the ends to squeeze the ends of the broken tab between to make the connection.

if you decide to give up and use just 15S then how can you charge it? you are stuck with a big hassle.

in spite of that, if you wanna run on the 15S you can still use the BMS for discharge so you still retain some cell level protection by putting a jumper across the opto for the LVC signal and the HVC signal so it will not respond to their out of range condition.

Yes perhaps I could have repaired the pack if I purchased two identical cells however I didn't want to have to buy the cells, I didn't want to wait for them to arrive from China and I wasn't certain I could solder the cells into the pack given the very fragile (aluminum?) tabs used on them. Anyway I chose to remove the last cell group and make it a 15S pack and it has been working well for me in this configuration.

I found on this forum how to make a Signalab BMS function with fewer cells however my BMS is from a different manufacturer. My BMS also uses tiny surface mount components the legs of which I can't even see to modify even if I knew how. I never had the BMS trip for any reason in the 38 cycles which preceded the cell failure (even at 93% DOD) which was a good indicator I might be able to do away with it completely. I did figure out how to jumper the BMS so it could still be used to top-balance the cell voltages so I would not have to charge each cell individually and that’s what I used to achieve 3.59V+/- 0.01V on each cell prior to the first ride of the season. After that I used a 15S charger I picked up from All-Battery.com to recharge the pack. For the first 6 cycles I checked the individual cell voltages after each ride and after recharging the pack. I found no large deviation in my cells even when discharged to 87% so it’s looking good. I did re-balance the pack after the eighth cycle and will do it periodically not because I think it’s needed but because it’s easy to do.

For me running without a BMS should present no problems however I can see why having a BMS is a good idea for the majority of batteries in the hands of the masses. Still the absolute #1 attribute for a solid performing battery is well matched cells and not a BMS.

-R

 

[wookey]

Here is my data, updated to add charge after bleeding (to 54.5V)

cell 16cycles  bled to  2 cycles  bled to   
id   charged            charged                   charged               
     to 56.6V           to 56.6V                 to 54.5V
A1    3.335              3.338                           3.338
A2    3.727     3.65     3.794    3.362    - too high    3.430
A3    3.335              3.339                           3.338
A4    3.335              3.337                           3.338
B1    3.430              3.561    3.367    - high        3.408
B2    3.679     3.65     3.718    3.368    - too high    3.441
B3    3.357              3.378                           3.386
B4    3.753     3.65     3.821    3.367    - too high    3.505
C1    3.336              3.340                           3.339
C2    3.341              3.352                           3.358
C3    3.336              3.338                           3.338
C4    3.335              3.338                           3.338
D1    3.493              3.607     3.363    - high       3.399
D2    3.334              3.338                           3.338
D3    3.335              3.338                           3.338
D4    3.608              3.634     3.364   - high        3.419

So that seems to have worked nicely in that no cells got too high. High cell-sets are consistently highest. Charging was much slower (less than 2A after first hour) due to charger backing off much sooner. I wonder if I am seeing the effect Olaf reports of lower charge rates much reducing the tendency for the voltage to get too high? It did seem to take much lkonger to terminate than I expected. This gets back to the old thorny question of to what degree voltage represents SOC - if different charge rates change the shape of the curves that makes life more complicated. (like spotting 100%SOC with dV/dT on NiMHs which is much easier at high current rates).

Russell, no I haven't really done detailed checking in discharged state. I did have a check of battery pairs and they were all 6.70 (i.e the same within 0.1V). Checking individuals is fiddlier with current wiring. I'll check that next time I go to work by moped. The matching voltages may simply show that I am on the really flat part of the discharge curve where SOC variantions simply won't show up as voltage differences the way they do near 100% SOC.

I will run for a while with this reduced-rate/voltage charging and see how my cell divergence goes.

 

[Russell]

On today's ride I used 7.98Ah from my 15S LiFePO4 pack. I stopped the charger three times on the recharge to see what the cell voltages were and to see at what point the cell divergence starts in earnest.

The first point at which I stopped the charger was after 7.60Ah had been put back into the battery or 95% of what had been used. I waited 2 minutes after halting the charger for the voltage to stabilize then measured each cell.

1 cell at 3.36V
13 cells at 3.37V
1 cell at 3.39V (My low capacity cell#11)

I reconnected the charger and stopped it again when the charger current started to fall. Another 0.168Ah had been replaced for a total of 7.77Ah or 97.5%. Again I waited 2 minutes to take the measurements.

3 cells at 3.37V
10 cells at 3.38V
1 cell at 3.39V
1 cell at 3.43V (#11)

Reconnected charger, put in another 0.088Ah so that 98.4% of energy used had been replaced.

1 cell at 3.39
4 cells at 3.40V
5 cells at 3.41V
4 cells at 3.42V
1 cell at 3.49V (#11)

Then I let the charger finish at 54.24V with a total of 7.95Ah. I checked the voltages while the charger was still connected and found they were between 3.55 and 3.67. Cell #11 btw was no longer the highest at 3.62V. I then removed the charger, waited 15 minutes before recording the voltages and they were nicely grouped in the range of 3.44V-3.51V (#11 was 3.47V).

One thing I had already concluded is that if you are not using a BMS with shunts and cell over-voltage protection (~3.90V) you should keep the charger voltage on the low side (~3.60V/cell) since the voltages will diverge as the cells get plump full. It appears you can lessen the risk further by going with an even lower charger voltage and as long as the cells stabilize over about 3.40V after charging they will be at least 98% full, or at least that's what my data today shows.

-R

 

[GGoodrum]

One thing I haven't seen mentioned here is having two cells that have the same capacity, but just happen to be at different states of charge. One thing periodic balancing does is equalize all the cells to the same voltage level. If you don't, and the cells keep getting farther apart, you end up with less and less capacity. If you at least have cell-level low voltage protection, you won't kill cells, but every time you run down to LVC, the cells will get farther apart.

One feature of our 4.0 BMS is that you don't have to do balancing, but the charge controller will still monitor each cell and moderate the charge current so that no cell goes over the set point. In effect it is a cell-level CV mode. If you let the charge go until the current drops, you will at least get the low capacity cell full, and all to the same energy level. Doing this, along with periodic balancing, and cell-level LV protection, will protect the cells on the high end, as well as the low end, and will maximize the available capacity.

-- Gary

 

[docnjoj]

Hey Russell
I have been doing something similar to what you have. I have a 12 S rebuilt Ping and an 8S set of Fatpacks set up in series. I am using a Ping charger reset to 65 volts and it seems to work well. The Fatpacks take about 21 volts and the Ping the rest. I periodically charge the Ping separately with the BMS but only about 1 time out of 10. Works great and really lowers power consumption using the higher voltage (54 nominal) and easing off the throttle.
otherDoc
Actually, I guess it would be a 5s 8P set of fatpacks. Hmmmmm.......it still works great. Maybe I will lower that charger a bit as it could be 3.8 volts per cell.

 

[AndyH]

I found that the best way to monitor cells is to check all cells at a moderately drained state, match up the lower cells with a single cell charger (i use a cellphone charger) Then check again fully charged.

The cells stay ballanced for many weeks of use, and if one is too high, I just drain it with an old car headlight with alligator clips attached as my resistor.

Now i use two CellLog monitors for my 16 cell pack, and never get into trouble that way. I've had two different BMS's destroy couple cells over the years. BMS's are not quite perfected either.

You're right - a bad BMS channel can take a cell with it. Hopefully one of the things we can come up with here is a quick way to validate BMS function without waiting for a cell to gag.

The wee CellLog beasties seem to be a great way to keep an eye on things. I know that I feel naked without the PakTrakr running. Unfortunately the cell monitors pull power from the pack and can easily be an 'un-balancing influence' that can make management a bit harder.

Maybe managing a 'bareback' or 'naked' pack is easier overall -- but maybe those of us using monitors get extra points from the 'Russian Judge'? :wink:

Are we having fun yet?

 

[scoot]

While I haven't performed the remedy myself, there is one for the CellLog involving the user soldering in a couple micro jumper wires or resistors. I actually only connect mine when I am logging or getting readings. i think the CellLog8 rocks 8) i have learned so much and proven/dis-proven so much to myself using that litll' gem.

and hell yeh I'm having fun :wink:

I am charging sequentially one cell at a time on one of those bareback virgins of mine... just to see roughly how far out of balance the new pack is/was. My charger is giving me ah readings

Should be done with this 8s pack in 8 hours LOL

 

[scoot]

This is the summary from the sequential single cell charging on one of my new naked bareback virgin packs (8s - 38140se cells):

1 - 0.77 ah accepted
2 - 0.34
3 - 0.49
4 - 0.01
5 - 0.07
6 - 0.28
7 - 0.54
8 - 0.56

Cell #4 took no time at all as it was already full. #4 was the one that spiked up so high (4.34 volts )the other day during the initial 8s LFP bulk charging. Tonight I'll see how the cells respond to the bulk charging again. :wink:

 

[AndyH]

And now for a look at a pack with a shunt-type management solution. The unit is a 21-channel variation of one of Gary's LiPo shunt units tailored to scooters. The BMS has 640mA shunts. The shunt set point is 3.65-3.72V - basically the V2.5 G/F BMS setting. This BMS does not have charge control - the charger is connected directly to the pack positive and negative.

We'll look how the pack responds to the Thunder Sky charger and a 5A power supply. [edit] A reminder - the Thunder Sky charger isn't stock -I've added about 3V. This is a worst-case demonstration for the TSL charger and a shunt-only BMS. [/edit]

View attachment scooter_bms.jpg

Here's a look at end of charge from the Thunder Sky charger. The early part of the charge looks exactly like the charts early in the thread, so we'll jump to the CC/'pseudo-CV' phase and the cell bleed-down.



At the red arrow, the charger has shut itself down. The combination of the shunts and Paktrakr sensors are pulling .8A out of the pack. The red circle highlights two cells that were close to full and responded to the extra energy. Eight cells still haven't turned the corner.

I pulled the TS charger and connected a 5A power supply to see how the low cells would respond.

Here's the first 30 minutes:



The low cells are responding to the additional energy. The power supply started at a bit over 3A but quickly dropped down to 400-500mA. The 3A starting current pushed the high cells well over the shunts - maxing out at 4.01V.

I ran this top-off for data. When I use a power supply to bring the stragglers up (once a month or so) I limit it to 600mA. This keeps the high cells at or below 3.7V and still allows the low cells to catch up.

It took just under 2 hours of low-current to bring all the cells into the group, while the shunts continued to pull the high cells down.

 

[scoot]

I had to read it and study it a few times Andy to get it, plus catch up with all my other stuff goin' on, but I am with ya' now.... popcorn in hand :lol:

 

[AndyH]

One thing I haven't seen mentioned here is having two cells that have the same capacity, but just happen to be at different states of charge. One thing periodic balancing does is equalize all the cells to the same voltage level. If you don't, and the cells keep getting farther apart, you end up with less and less capacity. If you at least have cell-level low voltage protection, you won't kill cells, but every time you run down to LVC, the cells will get farther apart.
-- Gary

Hi Gary,

Thanks very much! This is part of what I'm trying to understand.

At this point, I'm pretty sure my pack has a range of capacities. A couple of cells could have been damaged thru over-discharge last summer (but seem to be slowly improving). And 8 cells have a slight but constant load from the PakTrakr remotes.

I think I've been seeing what you describe here - that a bit of imbalance (from at least the sensor loads) causes cells to diverge.

I keep the PakTrakr remotes connected 24/7 and recharge after each ride. Sometimes the bike will sit for a couple of days between rides. I'm getting a noticeable unbalance after 24 hours from a fully top-balanced pack that the Thunder Sky 'pulse and glide' mode cannot correct.

Disconnecting the PakTrakr remotes and keeping the shunt/LVC-only BMS connected changes the picture completely - the TS charger's pulses are enough to allow the pack to balance over multiple cycles. When running this configuration, I'll leave the charger conencted 24/7 and just let it pulse. There may be a small shunt overrun for the full cells but it's not much considering the 2-minute on/18 minute off nature of the pulses.

fun! Andy