The frustrating intersection between smart battery charger, smart BMS, and an unbalanced lithium battery pack

7 posts / 0 new
Last post
reikiman
reikiman's picture
Offline
Last seen: 12 months 3 days ago
Joined: Sunday, November 19, 2006 - 17:52
Points: 8447
The frustrating intersection between smart battery charger, smart BMS, and an unbalanced lithium battery pack

My electric car is running and almost ready for routine use. It's a 1971 Karmann Ghia, a 9" Netgain Impulse9 motor, a Kelly KDH14800D controller, and a 144v 100ah Thundersky battery pack, with a BMS from Alliance Renewable Energy (a.k.a. Lightning Motorcycles).

The issue at the moment is getting the battery pack fully charged and balanced. What we're observing in the process has really educated me on the value of BMS's. There are some in the conversion community saying you can go with no BMS, that the charger will take care of it okay and not to worry. Eeeep is what I have to say to that, especially with what I'm seeing right now. I'm more convinced than ever that a BMS is necessary with lithium batteries.

Lead-acid batteries for all their flaws are somewhat forgiving of abuse. You can run with an unbalanced lithium pack, overcharge, overdischarge, and while that might damage the batteries a bit you'll still be able to use them just with a shortened life-span. With lithium batteries on the other hand they can quickly turn into a useless brick, or if you're using a flammable battery chemistry they can catch fire and destroy your vehicle. My Karmann Ghia is so beautiful (and cost quite a bit) that I don't want it to be destroyed, and the battery pack cost enough to put together that it's best to protect it for long life.

Anyway...

The charger is an Elcon PFC3000. It runs at either 240 volts or 120 volts, and will put up to 3 kw charge rate into the battery pack. I was trading off between cost and charging speed.

It is a somewhat smart charger - when it thinks the battery pack is nearly full, it lowers the charge rate to match. I can measure this using a clamp-on AC power meter and during full charge it will draw 11 amps or more at 240 volts, but in the more modest charge period when it thinks the pack is full it will draw 1 amp or so at 240 volts.

The BMS has an on-off signal it can use to tell the charger to go on, or off.

The BMS algorithm is to keep the charger on until any cells reach a maximum voltage (set at 3.9 volts; this is set a bit below the maximum cell voltage to lengthen the lifetime of the pack). When a cell is at max, the charger is told to turn off. Shunt resistors bleed off voltage from the highest cells. The charger is told to turn back on when cells fall below another voltage level. It means that as the pack nears full charge, the charger will be cycling on and off until all the cells are fully charged and have almost identical voltages.

This is probably a standard algorithm for BMS's ... it's certainly what it looks like the pingping packs I have are doing, because the pingping charger keeps cycling on and off when it's at the end of the charge cycle.

The problem we're having at the moment is a) the pack voltage is high enough the charger is only putting out a mild charge current, b) the cells in the pack are way out of balance.

Many of the cells in my pack at this moment are in the region between 3.7-3.9 volts (during charge) but a few are near 3.5 volts (during charge). The problem is that you can't really deduce the state of charge in a cell that's at 3.5 volts. When the cell goes above 3.6 or 3.7 volts it's real easy to deduce the cell is almost fully charged. But the voltage characteristics on lithium cells are such that a 3.5 volts cell is indistinguishable from almost charged or almost discharged.

Basically - if I were to drive the car right now there would be a danger of damaging the weaker cells because those cells might actually be almost discharged right now, and taking the car for a spin could send the cells below the low-voltage threshold that causes damage.

The charger sees a nearly full pack however, and is putting out a modest charge current. If there were no BMS a vehicle owner trusting the charger to do the right thing would have no clue how out-of-balance their pack is, drive away with an unbalanced pack and end up with a dead cell or two because of cells going below the low voltage threshold. Also the unbalanced pack could cause the high cells to be over-charged, damaging those cells because of going over the high voltage threshold.

Johnny J
Johnny J's picture
Offline
Last seen: 2 years 8 months ago
Joined: Tuesday, July 8, 2008 - 17:02
Points: 377
Re: The frustrating intersection between smart battery ...

In my experience it´s almost impossible to get all cells to the same levels with a "dumb" BMS.
There will always be a few that´s not in the prefered range.
But, if you try to "top off" a cell with a lower voltage or shunt off a high one, you most likely will find that it will only take a couple of Ah to make it reach the voltage of the other cells.
It might be that the lower cells are 95% charged, average 97% and the "high" ones 99%.
Without an active BMS you might have to live with this....blame internal resistance...:-)

antiscab
Offline
Last seen: 10 months 2 days ago
Joined: Saturday, July 7, 2007 - 23:55
Points: 1686
Re: The frustrating intersection between smart battery ...

Many of the cells in my pack at this moment are in the region between 3.7-3.9 volts (during charge) but a few are near 3.5 volts (during charge). The problem is that you can't really deduce the state of charge in a cell that's at 3.5 volts. When the cell goes above 3.6 or 3.7 volts it's real easy to deduce the cell is almost fully charged. But the voltage characteristics on lithium cells are such that a 3.5 volts cell is indistinguishable from almost charged or almost discharged.

We're talking about LiFePO4 right?

at the standard 0.3C charge rate (which you are well under), the difference in SOC between 3.5V and 3.9V is ~0.5%
If these voltages are measured when the charger is up to the CV stage (constant voltage), the difference is even lower.

3.9V is quite a high cut-off, what is the cell average voltage for the charger CV stage?
for 45cells it should be between 156V and 165V.

Generally, if running without BMS,
The best way is to bottom balance, then set the chargers CV voltage so that the highest (smallest) cell ends at 3.65v (the lowest may be as low as 3.3x V)

Matt

Daily Ride:
2007 Vectrix, modified with 42 x Thundersky 60Ah in July 2010. Done 194'000km

PzlPete
Offline
Last seen: 10 years 9 months ago
Joined: Monday, June 22, 2009 - 08:38
Points: 157
Re: The frustrating intersection between smart battery ...

Will put in my 1.5 cents here. I have owned my 4000li since January of this year, and at 3,980 miles, the ignition switch went out, so I had to take her apart to replace it and did a lot of testing on my batteries while I had access to them. The 4000li does not have BMS and relies on the smart charger to keep all 20 cells balanced. She had about 20 miles since last charge at the time I took her apart, so was at about 3/4 capacity based on the range I see.

I measured the voltage of all 20 cells, and all 20 measured exactly 3.28 volts.

So I plugged in the charger and let it run until green on both lights, and measured again. My highest cells were around 3.95 with my lowest at 3.52 and 3.66.

Then I let the smart charger go into pulse mode (charge beyond full for an hour or two). Measured again, my lowest cells had come up to the 3.7 range, and my highest cells were around 4.05, but they didn't remain there very long and were quickly back to 3.80. So at full charge, the smart charger managed to "balance" the cells around 3.7 to 3.8. And keep in mind this is after the batteries had 4000 miles on them.

How did all 20 cells end up at exactly 3.28 after running for awhile?? Does the controller do some sort of balancing?? I load tested her again after full charge for another few miles, and again the cells were exact after running for awhile.

Full charge = unbalanced. Let smart charger go into pulse mode = nearly balanced. Run for 20 miles = perfect balance. Does the controller have some sort of BMS capability??? I don't see any individual wiring going to each battery, and they appear to be one large pack just wired in sequence.

Regardless, it appears my cell are staying fairly balanced to this point without BMS. BMS is available on the 5000li now I believe.

reikiman
reikiman's picture
Offline
Last seen: 12 months 3 days ago
Joined: Sunday, November 19, 2006 - 17:52
Points: 8447
Re: The frustrating intersection between smart battery ...

PzlPete:- I don't think you can tell much from the resting voltage of a cell. The voltage you quote is in the gray middle area of lithium cell voltages where you can't tell between 30% and 70% SoC. Lithium cells behave differently from lead-acid cells, with lead-acid the voltage drops off smoothly and you can make an educated guess about SoC from the voltage, but lithium has a flat discharge curve and you can't tell much. BUT you can tell more about SoC from the voltage under load ... though, it's rather difficult to measure per-cell voltage under load while riding the scooter eh?

I don't know how balancing can be done when you are simply applying pack voltage to the pack as a whole. Don't you need to do something at the per-cell level to manipulate the voltage of each cell? For example for lead-acid pack charging we've used bank charging (one charger per battery), various kinds of balancers like the Rudmann Regulators, and monitoring gizmos like the PakTracker.

Johnny J:- What are you calling a "dumb BMS"? The BMS in question has an embedded computer, I can fiddle with settings by connecting my laptop to it via a USB port, and there are a zillion settings so e.g. you can use the BMS with a variety of chemistries.

What I think is going on is a) charger & BMS are both somewhat "smart"; b) the communication between them is very weak; c) hence they're not able to cooperate and working at cross purposes

The charger is seeing a nearly full pack (going by the pack voltage) and adjusting the charge rate really low. But the charge rate isn't high enough for the BMS to be convinced that the charger is actually on.

antiscab:- Yes, thundersky LiFePO4. The BMS as configured stops charging when a cell reaches 3.90 volts, and by that point it had already started bleeding voltage from that cell, and it keeps bleeding until cells reach 3.70 volts. Hence it's target is to get all cells to 3.70 volts.

I should mention -- by the way -- that this BMS is, I believe, still considered by Alliance Renewable to be a "beta test" unit, and I'm helping them diagnose behavior and consider what (if any) design changes are needed. Or settings tweaks.

Johnny J
Johnny J's picture
Offline
Last seen: 2 years 8 months ago
Joined: Tuesday, July 8, 2008 - 17:02
Points: 377
Re: The frustrating intersection between smart battery ...

Ohh´, what I mean by "dumb" BMS is that it only shunts energy away from the high cells, cuts the charger and warns if there are low ones, right?
An "active" BMS one the other hand is constantly balancing the cells (even under load) by taking energy from high ones and distributing it to rest.

By the way, I really envy your Karmann... :-)

Regards

Johnny

PzlPete
Offline
Last seen: 10 years 9 months ago
Joined: Monday, June 22, 2009 - 08:38
Points: 157
Re: The frustrating intersection between smart battery ...

Reikiman, I actually did test the voltage of the cells under load by putting all the lights on and turning on the throttle with the cycle basically in treadmill mode. (Not easy, but possible). My cells read between 3.41 and 3.52 under load immediately after full charge....so all twenty were still very close. I was trying to determine if after 4000 miles any of them needed replacing. But more than anything, I was suprised to see all 20 cells measuring exactly 3.28 at rest, partially discharged, and that was with no BMS on my bike.

Log in or register to post comments


Who's online

There are currently 0 users online.

Who's new

  • Juli76
  • xovacharging
  • stuuno
  • marce002
  • Heiwarsot

Support V is for Voltage