DIY Electric Car Forums banner
1 - 20 of 56 Posts

·
Registered
Joined
·
40 Posts
Discussion Starter · #1 ·
14 months (6K miles) using 36 REAL FORCE 120 AH LiFEPO4 cells. Generally good results, but I'm having serious range problems now. I'm using Dimitri's mini bms so it's difficult to determine which cell is causing LV alarms but they now start at a little less than 20 miles, I'm doing most of my driving very gently at only about 100 Amps (sometimes less). Mostly at 35 MPH or less, only short excursions above 45 and this is Florida -- almost no hills. The pack has been top balanced again and again to the point that static (after driving) voltages are all within .03V of average cell voltage and highest voltage - lowest voltage is around 0.1 V after a day of driving. The lowest cell seems to change every time I measure cell voltages. For a while I was suspecting cell #6, then it was #30, then #5, and most recently it was #21. It's not at all easy to observe the LED's on the cells while driving so all I can really do is static voltage measurements after I get home. Obviously, something is amiss with my batteries but I need some sort of test procedure to determine what the solution is. Do I have to replace all 36 after only 14 months? I have never charged any cells above about 3.8V and only once have I measured a cell voltage below about 2.5V. If one is damaged, it doesn't consistently show up as way different from the rest. When I first completed the project, I could go 62 miles on a charge. Now the audible alarms start at around 19 miles. Anybody have an idea of what the fix is?
 

·
Registered
Joined
·
66 Posts
Sorry to hear about your pack problems. To start, you need to cycle test a few individual cells for capacity. If they all test low, say at 60AH then that is why you are getting half the range. If they test at 130AH, then you might have a few bad ones. Until you start testing cells, you don't know where you are at.
Joe
 

·
Registered
Joined
·
1,693 Posts
Drive your car down to where the "squealer" is going off on acceleration. Then place your cell phone on video and duct tape it to "watch" your pack to see where the Red LED is And on what cell. You might do this several times to get the varmint.

Or have an assistant watch them if possible.

Make a couple of small jumpers to go over the bad cell BMS board to isolate it.

Drive some more to see if the squealer quit.

You can fine it with patience.

I did.

Miz
 

·
Registered
Joined
·
40 Posts
Discussion Starter · #4 ·
Joe G, can you please share a little about testing individual cells? Do you make up a low resistance load (say with a hundred feet of 12 GA solid wire, or some such) and then purposely load the cells down one at a time, observing cell voltage drop, accounting for internal resistance? How do you go about measuring the fairly high current that will surely result? I don't have a professional load box like they use at the corner garage when you are suspecting your 12V battery in a ICE powered car. Maybe you know of a relatively inexpensive source of a device like that?
 

·
Registered
Joined
·
40 Posts
Discussion Starter · #5 ·
Mzlplx,
I tried to follow your suggestion. I perched my wife over the open bay of batteries (not an ideal position to be in!) as I accelerated again and again. Asked her to look for green LED's extinguishing. I learned that #21 and #6 were the first to go out, and after that it was 23 , 24, 25, 26, 27, & 29. But some runs had some of those (23-29) not going out at all), It's really pretty inexact because she said the LED's got dim but wasn't sure they were all the way out in some cases. There are two battery boxes, one under the back seat (where the fuel tank used to be) and the other under the luggage area (it's a hatchback) where the spare tire used to be. In the rear battery box, the LED's are really hard to see at all, and some are obscured by the front of the battery box so I'm not all that confident that we didn't miss one or two bad performers. I'm thinking that the other suggestion, measure the AH capacity with a load tester, would certainly be in order before I just go out and willy nilly start replacing between 2 and 8 cells (as you know, they're really expensive). Does anyone have a suggestion for where I could purchase a commercially available battery load tester (measures AH) suitable for measuring only 2-4 V? I'm sure there are lots of them available for typical lead acid batteries (6 or 12 V, or even higher) but I don't even know where to go looking for one suitable for lithium. The remaining question in my mind is that if I were to replace, say, six of them, will they ever behave "close to the same" as the remaining 31 cells that have been deteriorating for 14 months?
 

·
Registered
Joined
·
2,122 Posts
Do you have a voltmeter showing pack voltage as you drive? Can you see how much overall voltage sag you get under 100A-120A load, which represents 1C for your cells?
If you can record pack voltage sag under 1C load on a fully charged pack and then also under same load once BMS alarms start coming in, then you will have some idea of how bad internal resistance is on your cells.

I have a feeling that IR on these cells is rising too much over time, which speaks of poor quality of these cells.

Also, you can drive around the neighborhood gently until BMS alarm is constant, even when you let go of the pedal. This way you can certainly see which LED is out, this will point out weakest cell. While this cell is still beeping, measure other cells and see if any are below 3.1V , which means they are also getting close to the low knee of the curve.

If many or most cells are resting below 3.1V when lowest one is in alarm state ( below 2.7V ), then your pack is deteriorating more or less evenly, i.e. its not just one cell causing troubles. This is worst case, because replacing only a few cells is a waste of money.
 

·
Registered
Joined
·
40 Posts
Discussion Starter · #7 ·
Dimitri, Yes, I am using the Android dashboard. It gives me a reasonably accurate pack voltage reading. I'll try that (testing at 1C with a full pack, and later when it's nearly depleted). I'm going on travel tomorrow for the next week, but I'll do that ASAP upon my return. I did try today running with a nearly full pack (only about 10 miles on it) with an observer (my wife, who was none too thrilled about this assignment!) watching for green LED's going out. When I was drawing around 3C, I got alarms to ring and she noted around 7 LED's going out (that means 29 cells never got to LV Alarm state). So now I am reasonably sure that more than one or two cells are troublesome. Interestingly, after stopping and taking another round of static voltage measurements, one cell (#30) which never dropped to LV alarm state, was about half a volt lower than average voltage of the pack (3.202V) which was far and away the biggest deviation from average of any of the 36. But its green LED never came off, indicating it wasn't one of my problem cells under load. The entire pack was very close to perfectly balanced when I started that little exercise.
 

·
Registered
Joined
·
903 Posts
I'm wondering if you might have been overcharging your cells, especially if you are balancing them on every charge to 3.6/3.65V. With a healthy LiFePO4 cell at room temperature at 100%SOC will rest at 3.38V. Maybe after a full charge with balancing disconnect the BMS on each cell and all loads and let it sit untouched for 12-24 hours. Then go around and measure the voltage of the individual cells. If any are resting over 3.38V then it was overcharged. It might be that the quality of the cells is such that they can't handle the overcharge as well as some other cells can.

I have over 16k miles and over 3.5 years on my 200Ah pack and it shows no signs of degredation. I only charge to 3.455vpc and when I do the charge and rest test all my cells rest between 3.356V and 3.363V. Right at the end of charge they were between 3.450V and 3.487V and have not been top balanced for 2 years. These are TS-LFP100AHA cells made in November 2009 and are in a 2p20s arrangement. (Note that I do not have BMS boards on them any more. If you have BMS boards on a LiFePO4 pack you are likely to need to balance your pack more often since it is nearly impossible to make sure the parasitic drain on each cell by the BMS boards is exactly the same.)

As for a capacity tester you might consider something like the PowerLab 8. A description is here: http://store.evtv.me/proddetail.php?prod=revolextrix
 

·
Registered
Joined
·
541 Posts
<snip>The pack has been top balanced again and again to the point that static (after driving) voltages are all within .03V of average cell voltage and highest voltage - lowest voltage is around 0.1 V after a day of driving. The lowest cell seems to change every time I measure cell voltages. <snip>
I think there may be issues of resistance...either internally or in connection resistance--have you verified connection torque and maybe connector crimps?.
But also, after you have been signaled for low voltage, measure the resting voltage after 15 minutes or so. If not less than 3 volts, drive on for a while. Resting voltage is the only sign of actual over-discharge.
Gerhard
 

·
Registered
Joined
·
40 Posts
Discussion Starter · #10 ·
Gerhard,

I have checked with my fingers to look for a loose or cockeyed connection but didn't find any. I haven't checked with a toque wrench; think I'll try that. What torque do you suggest? Also, since I live in Florida (95% humidity and lots of salt in the air), I used No-alox when I installed all the terminal connections. Do you think that might be a problem? No corrosion but maybe some high resistance due to the grease?
 

·
Registered
Joined
·
2,047 Posts
I used No-alox when I installed all the terminal connections. Do you think that might be a problem? No corrosion but maybe some high resistance due to the grease?
Correctly used the greases are not an issue. And you can really glop the grease on and as long as you get the connection tight all the excess will be squeezed out. It looks terrible when you do this but is isn't an issue.
 

·
Registered
Joined
·
1,526 Posts
Testing

14 months (6K miles) using 36 REAL FORCE 120 AH LiFEPO4 cells. ... I have never charged any cells above about 3.8V ... I could go 62 miles on a charge. Now the audible alarms start at around 19 miles. Anybody have an idea of what the fix is?
You never mentioned what pack or per-cell charging voltage you have been using, but it sounds like reduced capacity due to overcharging.

You don't need to buy an expensive cell tester, you already have one--your car. Just buy some good heavy-duty jack stands and a clamp-on current meter if you don't have a current shunt.

Charge your pack and let it sit 24 hours and measure pack voltage at the start, then jack the wheels up and run it on the stands at a constant current and you don't have to worry about getting stranded on the road.

Measure pack voltage and current every 6 minutes (0.1 hr) and you can monitor pack capacity in real time until the pack voltage drops to whatever level you like to call drained, let sit 24 hours and measure voltage again to see the bounce-back.

For example if you are drawing 50 amps on the stands, then every 6 minutes is 5 AHr--5,10,15,20 you can do the math in your head and see what sort of reduced capacity or if are close to getting 100 Ahr out of your pack.
 

·
Registered
Joined
·
40 Posts
Discussion Starter · #14 ·
OK, just got back from a week on the road and finally got a chance to try Dimitri's suggestion: checking for internal resistance (and/or loose, corroded, or otherwise high resistance terminal connections). I made about 18 runs from 0 Amps to 400 Amps pack current, measuring the pack voltage for every 100 Amps step). Since I have 120 AH cells (made by REAL FORCE), this roughly corresponds to 1C, 2C, 3C, and 4C. LV Alarms (mini BMS head board) started sounding at 3C and above right from the get go so I am quite sure something is wrong with the pack. (Maybe it's just one or two cells?) Total resistance ranged from 0.54 Ohms to 0.61 Ohms with an Average Thevenin Resistance of 0.57 Ohms. I calculated it on a simple spread sheet by applying the following formula [V (rest) - V (1C)]/100 or V(rest) - V (2C)]/200 . . . and so on. It's pouring rain outside (has been for about 8 days now) and I haven't run the battery charge down much yet so I haven't continued the experiment again with the pack almost depleted as Dimitri suggested. Now, my question is: Is this [R(Thevenin) = 0.57 Ohm] considered excessive resistance? In the next few days, I plan to remove all the terminal connections, wipe them clean, polish with fine grit sandpaper if necessary, check all crimp connections and reassemble. Then, I'll repeat the experiment both with a full charge (as today) and with a depleted charge. My problem is that I don't really yet know how to interpret the data. Is this considered excessive resistance? Can anyone suggest additional testing I can do before just giving up on my 14 month old battery pack?
Thanks for anyones' ideas!
Dave
 

·
Registered
Joined
·
40 Posts
Discussion Starter · #15 ·
In reply to KennyBobby -- I charge routinely with an ELCON HF/PFC 2500W battery charger set on an algorithm to provide around 123VDC. I then use the Mini BMS cell measurements to shut down the charger when one or more cells reaches the pre-set cell voltage that turns on their RED LED (I believe that's around 3.65V). In any case, after a complete charge (starting around midnight for 3 or 4 hours) first thing in the morning, the total pack voltage always shows 120V or 121V, which is around 3.36 V per cell, after settling for a couple of hours at least. For today's testing, I had charged the pack and then let the car sit for about 8 days. Again, rest voltage started just slightly under 120V. Even when top balancing (which I've been doing a lot), I use the mini BMS to shunt cells, never driving any single cell higher than about 3.75 V (and that's pretty rare, I usually top them out at about 3.65V). I have never knowingly driven any cell much above that. I do have a clamp on ammeter and I guess I could use it while the car is on jack stands, but the Android dashboard app is pretty convenient to measure pack voltage and current and that's how I usually do it. I'll try the jack stand suggestion next week. Thanks.
 

·
Registered
Joined
·
3,141 Posts
I think your 0.57 ohm resistance is incorrect. That would be 57 volts at 100 amps, which is half your pack voltage. Your formula seems to be correct, and it is a measure of internal cell resistance as well as connections. I would expect the voltage under 1C load to drop from 120V to maybe 110V, which is 10/120 = 83 mOhms. You probably mean 0.057 ohms, which would drop your pack voltage 22.8V at 400A to just under 100V.

Sometimes it helps to use a figure of percent impedance, which in this case would be 0.057 / (120/100) = 4.75%. This type of characterization is used for power transformers, which also gives you the short circuit current as a multiple of rated current, which in this case would be 120/0.0475 = 2526 amps.
 

·
Registered
Joined
·
286 Posts
In any case, after a complete charge (starting around midnight for 3 or 4 hours) first thing in the morning, the total pack voltage always shows 120V or 121V, which is around 3.36 V per cell, after settling for a couple of hours at least.
Might be slow death by over charging.

Turn down your charging voltage to the range of 3.45 to 3.5 and (or) reduce the CC phase. Aim for a static resting voltage of 3.33 which is 119 to 120 V.

A BMS shouldn't be used as a trigger to terminate charge, should be used as a fail safe.

Bet way to maximize cell life is to stay away from either end. Under charge and not over discharge.

Unanswered questions are - how far from the end, and which is more pointy?

80% DOD, 95% SOC or is 85% DOD, 90% SOC better?
 

·
Registered
Joined
·
2,122 Posts
Might be slow death by over charging.

Turn down your charging voltage to the range of 3.45 to 3.5 and (or) reduce the CC phase. Aim for a static resting voltage of 3.33 which is 119 to 120 V.
I never understood the persistent notion on this forum that LiFePO4 cell is overcharged by taking it to 3.65V or even to 3.85V using basic CC/CV chargers. Is this one of those gems where one genius came up with this theory and everyone else repeating it?

Most common CC/CV chargers do not hold CV voltage for more than few minutes, in fact entire CV phase is really short, so I don't see how any damage can be done. I suppose there is some potential danger in holding CV voltage for many hours at a time, and even that idea needs some hard data to prove such effect.

I have been driving my pack for 4.5 years now, charging every night to 3.8V per cell, yet the pack is still going strong, so how can this be explained?

All commercial LiFePO4 chargers and charge controller ICs are made to 3.65V per cell, so I guess those billion dollar companies with their R&D labs are all wrong too?

So, if we call things what they really are, some people choose to undercharge their packs, and that is their choice, but it does not mean that people who chose to charge their cells more fully are overcharging them.
 

·
Registered
Joined
·
3,141 Posts
Is there a temperature coefficient for LiFePO4? The float voltage for a lead-acid cell is 2.4 Vpc at 25 C, and the temperature coefficient is (-)0.003 Vpc / DegC, or (-)0.125%/C. If it is the same for LiFePO4, then the 3.65 Vpc would change to 3.76V at 0C and 3.54V at 50C.

A discussion from some time ago says that the tempco for LiFePO4 for charging is (-)2.5 mV per DegC.
http://www.diyelectriccar.com/forum...ure-compensation-lifepo4-chargersi-64518.html

So in that case it is (-)0.07%/C and the range would be 3.71V to 3.59V. Apparently the open circuit voltage does not change appreciably with temperature. But the usable capacity changes strongly with temperature:

Also, the *usable* amphour capacity changes strongly with temperature.
It delivers about half the amphours at -25 deg.C, and 10% more amphours at 45 deg.C. The OCV might say a battery is at 50% SOC, but you can't
get more than 25% out if it's cold, for example.
Something else to consider is the effect of the resistance external to the cells (as well as their internal resistance), especially for higher charge rates. If the resistance is 0.05 ohms for 36 cells, then each cell has a resistance of 1.4 mOhm, which corresponds to 0.14 Vpc at 100A. So if the charge voltage is measured at the charger, it could be as high as 3.65 + 0.14 = 3.79 volts. This is where a distributed BMS with connections right on the battery posts may be preferable.
 
1 - 20 of 56 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top