Wednesday, October 13, 2010

On Battery Balancing and Charging

As I've mentioned in the past, I'm not an expert.  On anything.  But in the time I've been operating the car and charging it, I've noticed a few things about these LiFePo4 batteries that I think is worth sharing.  In addition, think you can draw some simple conclusions from these experiences.

You may remember that just over a month ago I noticed that a few of the batteries were hitting their max voltage before the others.  The question became, did those cells move in their state of charge relative to the others, or do the simply have a slightly lower capacity.  My assumption right away was that they'd drifted.  Quite frankly, it never occurred to me that they might just have a lower capacity, which meant I immediately thought they'd moved relative to the other cells.  Of course the first thing I thought to do to fix that was balance them to the others.  I drew energy out of the all cells until every cell matched, down to 1/100 of a volt.  I drove the car a few miles and then charged up.

That proved to be a complete waste of time.  The very same cells that I so carefully matched to the rest exhibited the same behavior the very next time I charged them.  This time though, I was watching them. But I'd say this is a clear indication that those cells simply reach their fully charged state before the others. They had not moved relative to the other cells, they just have less capacity.  We're talking about a fraction of an amp/hour.

Early on, and for the first month I carefully watched the batteries every time I charged them.  I saw very small differences in the finished voltages between the cells; one hundredth of a volt here, 2 hundredths there.  And so it went until I had to pull the motor.  After the motor was back in, they did the same thing, but I was checking them less often.  I guess you could say I got complacent. I went from checking once a week, to letting it go nearly a month.  When I did that last check, I found those cells I mentioned running away. There's no question they'd done it before, but there's really no way for me to determine how many times.  I know it was less than 20, more than once.  I also think there's no question those cells sustained some damaged by it.  How much remains to be seen, but I don't think it's substantial.

I think it's likely that those cells were a slightly lower capacity than the others to start with, so they filled up first.  In that 20 or so charges, they'd reach their 3.6V maximum and be forced a little higher. Each time, just a little further.  In this way they accrued a little more damage each time.  Had I watched them more carefully, I would have caught it.  But, it is what it is, and lesson learned.  Since I rarely pull more than 50% capacity from these batteries, I don't think I'll ever really feel the impact of that.  Perhaps I'll get only 2900 cycles out of those cells instead of 3000.

But all of this got me thinking about battery management systems. I don't mean to imply I never thought of them before.  I have, a lot. But, the ones on the market shunt off power in the form of heat when a cell reaches capacity.  I can't stress enough how much I don't want that anywhere near my car.  Maybe that extra heat is OK when you're in Minnesota.  It's not in Arizona.  No extra heat is acceptable as far as I'm concerned.  I'd rather ruin a set of batteries than burn my house down.

I believe what would be ideal is a charging/battery monitoring system that has the capability of charging cells individually.  A monitor on each cell would watch the cell, and when it reached it's determined state of charge, the monitor would simply shut the charging for that cell off.  Not shunt the power away, but rather throw a relay that would simply stop power flowing to that cell.  Of course it would need to send signal back to the charger so the charger would know that one less battery was accepting a charge.  The charger would then ramp down the current it's putting out by the corresponding amount that the deactivated cell had been drawing.  Otherwise, you'd have a cascade failure as more cells dropped off line, the current traveling to the remaining cells would climb until you started burning wires.

This strikes me as spectacularly complex.  Doable, but due to the complexity, likely very expensive.  Not only that, but complicated which means an increase in the likelihood of failure. You could certainly design the system so that failures resulted in a system shutdown, not a runaway charger.  But could such a system be designed?  I think so.  The net result is batteries charged to the same voltage every time.  This of course would be a top balanced pack.  Jack Rickard has done quite a bit of work and demonstrated the dangers of top balancing.  So the only way this is practical is if you simply hold the bottom 5% of the batteries state of charge as off limits.  Meaning you could never run the batteries close to depleted for fear of having a cell reverse it's self. 

The next best option would be similar but much more simple.  A similar system with a monitor on each cell.  It watches the voltage for one cell.  When any cell reaches the target charge state, it sends a message to a main board that in turn trips a relay that cuts power to the charger.  Sure the batteries might be at slightly different states of charge, but who cares.  No battery is ever endangered and your pack is charged.  The advantage to this is that you could, if you chose, bottom balance you're pack with this set up.  Once balanced the charger would cut off as soon as the first battery reached it's limit and there would be no effort to move the others any further, keeping the batteries in the same positions relative to each other.

It seems to me that this would be a far simpler solution to charging and protecting your batteries and far safer than creating heat in a confined space but shunting off lots of power.  I would consider putting that in my car.  Heck I'd design it and build it myself if I were talented enough.

When it all comes down to it, I think my problem was one of charger set up.  Setting the charger and thinking once it was tuned in nothing would change was clearly a mistake.  Like I said, lesson learned.

One More Thing

Today I met with a gentleman named Alex who has been following the EV Z3 blog and the progress with the car since just before it hit the road.  Alex is traveling at the moment and had a stop over here in Arizona, so we decided to take the opportunity to meet up.  We had a great time.  We took the car through down town Chandler, (which is unfortunately under construction) and out on the freeway.  People often ask whether it can reach freeway speeds, so I thought it would be a good idea to show Alex just how well it does. 

He's been considering doing a conversion of his own for a while, although his ideal car to convert is a Porsche 911.  How great would that be?!  Hopefully the Z3 helped to inspire him to tackle the project some day.  Keep in touch Alex, and if you decide to tackle that Porsche, let me know so that I can follow the blog on your build.  You will keep a blog, right?  :) Thanks again!


Anonymous said...

Hi Tim,

The BMS idea you have sounds exactly like what the miniBMS does - modules directly mounted on the batteries, one wire starts at a little control board, daisy chains to all battery modules, and back to the control board. Any monitor module that detects either batt volts too high or too low will open circuit the wire loop.

Here's the good part - the control board gets 12 volts all the time plus the keyswitch on voltage. If the keyswitch voltage is present, there is a relay you can use to reduce motor performance (like turn on valet mode on a Zilla, or add a resistor to the potbox for other controllers) if any battery drops too low. If keyswitch voltage is not present, that means the car is not running and probably charging. An open wireloop means a battery is too high, so another relay can be used to stop the charger.

The clever thing is that what the open loop means is dependent on whether the key is on or not, and the control board has the logic to determine what relay to use.

BTW I am doing a conversion too,

Good luck! - Gene

p.s. I've decided to go lithium after all. I watch the Friday shows :)

Tim Catellier said...

Thanks for the comment Gene. I read through the literature on the miniBMS, and it does exactly what I had laid out as optimal in the post. How about that, I'm not as original as I'd thought ;)

One thing it mentions is that you can get the boards with, or without the shunting feature. The only problem is that I didn't see the non shunting variety on the products page.

If I decide that I want to have a high voltage cut off system, I'll be giving them a call to find out what's involved in getting a system with no power shunting. Thanks again!

Alex said...

Hi Tim,

I don't know how I didn't see this post before, but it was great to hang out with you and learn about your experience! I haven't given up on the 911 conversion, but it will be a while before I can drop the cash into the project. On to read your more recent posts...