Thursday, July 19, 2012

A Look at the Bottom

Juvenile jokes aside, at the end of the last post I alluded to the fact that I was interested in seeing how the batteries were doing since I bottom balanced them in February of 2011.  Quite a lot has happened with the car since then.  The batteries have been through 353 cycles.  They've put out 2209 kWhs of electricity and then had it stuffed back in.  I've driven a total of 6324 miles.  The car has been out of commission twice for motor problems, and off the road for a grand total of 7 months.  During those occasions, the battery pack was partially disassembled with half of the cells out and laying on the floor of my garage, while the other half remained connected together in the car.   That last point has concerned me a bit.  I've wondered all along if having them apart might have introduced some variable that may have caused cell drift.  I kind of doubt it, but I simply wasn't sure.  So, I set out to find out. 

Last Sunday I had a chance to drive down to the brand new Tesla store in Scottsdale and take a first hand look at the Model S.  I even passed someone that was out for a test drive in one of the demo cars.  I don't believe I've seen a bigger grin on someone's face while they were driving.  It was a great trip, but I'll write about that later.  At any rate, with a round trip to Scottsdale, and a couple errands thrown in, I'd used about 95 amp/hours out of the pack's 120 amp/hours available.  I decided to make a quick trip out to run the batteries down a bit more.  I figured I'd get up to 115 amp/hours or so and then run the rest down by running the heater in the car.

I set off for a quick 10 mile lap that would do it, when I had a second thought.  I remembered that the last time I'd charged the batteries, the charge had cut off a bit early with the top cell being about 3.40 volts.  That equates to the pack starting off about 7 to 10 amp/hours down, so I decided to cut my trip short.  Turns out that was a good thing.

As I was driving adjacent to my neighborhood heading for a specific entrance, I noticed the car was not really accelerating any more.  A half mile before that I'd accelerated to 40 mph with no problem but now, it was acting dead.  I turned into the neighborhood quickly and headed for home nursing the car the whole way.  I'd brought my multi-meter along but I was afraid if I stopped, I wouldn't get going, so I coasted (running a couple stop signs along the way) and turned the final corner to my house.  As I was heading up to the garage, I was hoping the door made it open in time because if I had to stop, there was no way I was going to get it up the hill of my driveway into my garage.   I made it in, but the car was dead.  It would barely move the 6 more inches I wanted to go.  I quickly jumped out and measured the cell I know to have the lowest capacity and it was at 2.043 volts.  I started measuring others and they were in the 2.5 to 2.7 volt neighborhood.  Well, that doesn't seem balanced to me!  I decided to let the batteries rest for an hour or so and come back to measure them.

By the way, when I pulled into the garage, I'd used 113 amp/hours (for those of you keeping score at home.)

Now, I'll elaborate on this more in a moment, but take note.  The lowest cell was 2.043 volts right after it had had a load on it, which is just above what CALB considers dead.  The car would barely move.  Yet no cell was below 2.0 volts and no cell was ruined.

I came back an hour later and measured the cells and found that the gaps, or differences I'd seen in the voltages had closed up dramatically.  The lowest cell that was 2.043 volts had bounced back and was now 2.684 volts; the highest cell was 2.937 volts.   I should let the numbers speak for themselves.

You can see that the cells were mostly between the 2.700 and 2.800 range, with a few just 1/100 off in either direction.  But there were 4 that were more than 5/100's of a volt off, with the spread from the lowest to the highest cell at 0.253 volts.  Three things come to mind looking at this data.  First, they aren't balanced.  Second, the amount by which the are out of balance is quite small.  At that end of the discharge curve, the difference between 2.684 volts and 2.937 volts is a fraction of an amp hour.  The third thing is that I think I simply wasn't patient enough when I performed the bottom balance.  You may have heard this elsewhere, or experienced it yourself if you've ever bottom balanced a pack of batteries, but it is an extremely boring, tedious, and lengthy endeavor.  Or to put it another way, it sucks big time.

On that first attempt at bottom balancing, it wasn't until after I was sick of the whole process and charged the batteries back up that I realized the proper thing to do would have been to let them rest for several hours to be sure they remained balanced.  I had them all within 1/200ths of a volt when I charged them, but I now know that if I'd waited, I would have seen them settle, and found they were likely a bit further off.  I think that inaccuracy is reflected in the variations in this data. 

In spite of my ineptness demonstrated here, I must have balanced them well enough to be, what I consider, successful.  The car would not have moved another 10 feet if I needed it to, yet no cell went below 2.00 volts let alone reversed itself and died a horrible death.  Something a top balanced pack simply can't do.

Now, I know what you're thinking.  "But Tim, you brought on this situation yourself.  This was completely and utterly self inflicted!  There was no need what-so-ever to discharge the pack this much.  I never intend to take my pack that low and expose them to this peril.  Consequently I'll never face the jagged, rocky bottom of the discharge curve, risking one or any cells in the process."   In part, you're right.   But consider this.  These cells, like any other, lose capacity over time.  How much and how fast is determined by how you treat them.  The problem is, the dangerous, jagged bottom of the curve sneaks up with every charge.  In other words, a pack that started out as a 120 amp/hour pack eventually becomes a 110 amp/hour pack, and then 100 amp/hour pack.  If you don't know where the bottom is, you risk hitting it and running a cell or 10 into reversal.  Since mine are bottom balanced, I "see" that imbalance at the top.  The charger cuts off at a preset voltage and the batteries will eventually reach that voltage regardless of how many amp hours they can actually hold.  The difference is, if I hit bottom the car stops moving and the batteries are fine.

So what's a fella to do at this point?  Best try to balance them again and do it properly.  This was monumentally difficult.  Not because the job is hard, and not because the batteries put up a fight or anything.  Rather because it's miserably hot and humid in AZ at the moment and spending 3 days in the garage balancing the batteries was not my idea of a good time.  One of those days was 18 hours long!  Suffice it to say, I got them all between 2.757 and 2.761, 4/1000ths of a volt, and that was with letting them rest for 4 hours at the end, before I put the charger on them.

One of the key pieces of information I wanted to get and was eager to share with you was the total number of amp/hours that went back into the pack.  That really is a measure of how the batteries have held up to the 545 cycles they've seen.  Sadly because of another, yet different stupid mistake, I'm not able to share that with you.  In may haste to get the pack balanced and the car back on the road, and my zeal to get it done right, I forgot something very important.  When ever you disconnect the main battery pack from an e-xpert pro meter, you MUST remove power from the meter.  I forgot to do this.  I opened the car and saw the following...

Notice the haze in the lower corners of the display?  The sharp eyed ones among you may also notice that the meter is not actually displaying any data.  I opened the door and immediately smelled the distinct aroma of a fried circuit board.  NOOOOOOOOOOO!   It gets better.  I took it out this morning, hoping to send it back to Evolve Electrics for repair, and this is what I discovered:

I'm no doctor, but that does not look good.  This meter was a few oxygen atoms away from catching fire.  I'm not sure if the deformation of the cylinder is clear in the photo, but it is not healthy looking.  Don't be like Tim.  Disconnect power from your meter before you work on the battery pack.  Incidentally, it fried the 1:10 prescaler as well.  *Sigh*

So, the end result of that is that I don't have a precise number to give you regarding how much power the batteries were able to accept when I finally charged them.   I can tell you that I turned the charger's dial to what I believe to be the position where it delivers 20 amps, and it took almost exactly 6 hours to charge.  That works out to 120 amp/hours, but it's really no better than a guess at this point. 


Gizmo said...

Wow, I never would have thought that the e-xpert pro nearly catch fire like that! That was close. I do like the fact that the CycleAnalyst uses the battery pack it self for power so it wouldn't have had the issue that the e-xpert pro had.

Does your charger cut off rather quickly at the end? I ask because 20 Amps for 5 hours is only 100 Amp-hours. Maybe it was 6 hours of charging. As you said, however, it is a rather crude measurement of capacity.

Your comment about the "bottom" coming "up" as the batteries age is a definite concern of mine with my pack. I top balanced my pack which makes it easier to see if there is any "drift" between the cells when they are fully charged. That leaves the ragged bottom which is undoubtedly creeping "up" over time.

I was curious if a Lee Hart type Batt-Bridge like I have in my Gizmo would have told you about a low cell so I put in the voltages you recorded and find that the difference in voltage between the two halves of your pack was 0.045V. It would have been nice to know if it would have pointed out your low cell while under power, or what power was left any way.

Tim Catellier said...


I was as surprised as you. Shocked, if you will (no pun intended). When I removed the e-xpert, not in my wildest dreams did I expect to see such damage. I do wish it were a bit more robust. I think I'll look into the CycleAnalyst you mention.

The Manzanita charger does cut off automatically at the end, and it's pretty reliable. It is no easy task to set it up though. It uses a little trim pot, and you have to dial it in to the voltage you want it to cut off at. The trouble is, it takes about a dozen cycles for it to settle down and start being consistent at the setting you've dialed in. So that means if you've got it wrong, you adjust it and start the process all over again. Once it is dialed in then it's dead on consistent. That is as long as you don't change the current you're pushing to the batteries by too much. It then falls victim to Ohm's law.

If you set the charger up while it was charging the pack at 22 amps, then you need to charge the pack at or near 22 amps for it to work right. If you turn the little dial and raise the current to 26 amps, the voltage rises just a bit in order to push that extra current (Ohm's law), and it's enough to cause the charger to cut off before the batteries are fully charged. Conversely, if you lower the current to say 18 amps, it does exactly the opposite and you risk over charging the batteries. But if you get it dialed in and follow the routine with military precision, it works great.

Am I happy about that? No. It sucks, but it's the charger I have.

Thanks for catching my typo. You'd think I'd get better at proof reading over time. But like I said, it's a crude guess. I might have been charging at 18 amps, or 22, I simply don't know.

I've heard of Lee Hart's Batt-Bridge, and I understand in principal how it works, but I've never seen one and I don't know how sensitive it is. I would have loved to have had some way of capturing a voltage reading on each cell that last time I depressed the throttle pedal. I think that would have been very interesting data.

Any way, thanks for the comment. I hope you found all this useful in some way.


Brad Woodcock said...

You let out the magic brown genie! Now it'll never work!

Tim Catellier said...

Right you are. Stupid brown smoke! I got a new one yesterday. I'm going to try and NOT destroy it.