Friday, February 25, 2011

A Visit to Some Student Engineers

A couple weeks ago I was contacted by an engineering student at Arizona State University.  He told me that he and his fellow students had been working on a Geo Metro, trying to turn it into an electric dragster.  An unusual choice I thought, but they'd bought the car after someone else had already converted it to electric drive, and they were simply going to modify it and improve it.

He asked if I'd be willing to bring the Z3 down and let them look at it, and perhaps give a short presentation on the car.  I wondered what I could possibly share with engineering students that they didn't already know, but I said I'd go. 

When I arrived, I pulled the car around to a spot outside their work shop and opened it up.  As a crowd started gathering around, they started asking questions about the car.  Soon an older gentleman joined them, clearly not a student.  His name was Jim, and he explained that he was the head of the program.  Once all the members of the team showed up, I started running through the various systems, explaining how each worked and why I built them the way I did.  I tried to be as honest and informative as I could be, showing the things I thought turned out well, and detailing the mistakes I'd made along the way.

The group was clearly interested and began asking more questions.  There were the normal question, "How long to charge it?", but also more technical questions like "How do you keep the motor from going over it's RPM limit? Do you have a sensor on the motor?"  Occasionally Jim would prod them, "See how he did that?", or "Look at that."  All the while, he took careful notes.  A couple of times, the group would say, almost in unison "Ohhh" leading me to believe that they'd seen the answer to a problem or question they'd had.  In general, it seemed to be worth their time to look it over.

After I'd shown them everything I could on the car, they offered to let me see their project.  The car had come with a Netgain 9" Impulse motor, powered by a Zilla 1K controller and a bank of AGM (absorbed glass mat) batteries; a type of flooded lead acid battery.  I don't know what the original configuration of the batteries was, but I gathered they were going to re-do the layout with all the cells in the cabin on racks along the floor of the car.  It was now going to be a single seater; after all, you don't need a passenger when you race. They've got everything out of the car and their cleaning it, prep-ing it for paint and to receive all the components again.

After wards, Jim offered to give me a tour of their facility.  That's when I found out these students weren't part of any average engineering program.  ASU has put together an Automotive Engineering program designed to address all aspects of the automotive trade.  Jim is a retired GM employee who had worked at GM proving grounds on the Corvette team.  He took me through the building showing me the CNC machines, their full machine shop, automation and electronics room, materials lab, welding shop and casting shop.  Yes, a casting shop where they cast parts in sand molds with molten metals.  How cool is that!?  I walked through the building with an increased sense of wonder and jealousy of all the students there.  Jim's been hired to build out the program and part of that is building and supplying the facility with all the tools the students will need to learn what goes on in the industry and how to fit right in to it.

All in all, it was a great visit.  I was very grateful for the opportunity to see the facility and what they do.  I hope the students learned a thing or two looking at the Z3 that might help them with their project.  They promised to invite me back when it hits the road.  I sure hope they do.

Wednesday, February 9, 2011

Balancing Act

For the past 4 months, I've been waiting for a break in the action so I could do a proper bottom balance on the battery pack.  Well, the opportunity presented it's self this last week with some time off work and fewer than normal tasks that required my attention.  But as fate is fickle, and sometimes cruel, I came down with a case of the flu which added a whole new level of unpleasantness to what was one of the most dull and tedious experiences I've had in recent memory.

A word on balancing; why do it?  Strictly speaking, I supposed it's not necessary.  However, it does provide you with a reference point about your battery's state which can be quite useful.   All BMSs balance the batteries at the top of the charge.  This, in an effort to make sure they are all fully charged and you get the most out of your batteries. But this is not without it's dangers.  The alternative would be to balance the batteries at the bottom of the charge curve.

Try as you might to get a pack of batteries that are all identical, that's not going to happen.  There will be variations in their capacity, however slight.  What this means in practical terms is one of the batteries will either be charged to 100% before the rest, or discharged 100% before the rest.  Without balancing, you can see both events could occur during one charge/discharge cycle even if one is careful.  Since the real danger to these batteries is over charging or over discharging them, you've got danger at both ends of the spectrum.  If you balance them to one end, you can make sure they all reach full at the same time, or they all reach 100% discharge at the same time, you've eliminated the danger on one side equation.

So now it becomes a question of where you want to balance the cells, top or bottom.  Pick your poison as they both have their perils.  The down side to top balancing is that you run the risk of having one or more cells hit rock bottom first during a discharge.  If you aren't aware of it, and don't do anything about it, you'll drive the cell into reversal and kill it, followed by the next low cell.  Generally considered bad.  But in addition to that, you also have to trust that the BMS isn't going to malfunction at the top of the charge and let one or more cells run too high.  Also bad, and there are multiple incidents of this occurring resulting in fires, wholesale destruction of the car and anything near it.

Bottom balancing means you make sure all the cells hit bottom at the same time.  The advantage is you're not going to kill one or more cells due to imbalance at the bottom.  The disadvantage is that you must set your charging algorithm up to accommodate the imbalance at the top.  This, as we'll see, is easy and only has to be done once.  So, we have a choice fraught with danger during every charge and every discharge, or one that needs care in setting up and then no down side.  The choice seems obvious to me. 

The goal was to take every cell down below 2.74 volts.  The CALB batteries are considered dead at 2.0 volts, but the difference between 2.0 and 2.74, in terms of the amount of power stored in the cell, is trivial. In addition, 2.74 is over the knee on the discharge curve and a hop away from dead.  Once at 2.74, I intended to charge them back up to 2.750 volts.  Once they were all even, charge them.  Seems simple right?  In concept, it is.  Executing the plan proved to be a bit more challenging.

Let's take a look at the tools of the trade:



At the bottom of the picture is the DVM, or digital volt meter.  No surprise there, any self respecting person should own one of these.  This one is accurate to 1/1000th of a volt.  Well, it displays to 1/1000th of a volt, but I have no idea how accurate it is.   I do, however, believe that it is consistent.  So even if it's off in it's readings just a bit, it's off the same amount on every reading, so that will do.  Just above that and to either side of the photo are two very sophisticated pieces of equipment used to draw power out of the batteries, otherwise known as automotive lamps.  Others may use a more elaborate and expensive piece of equipment, and they are probably very useful, but I don't have anything like that, so lamps it is.  Above those and at the top of the picture is what's commonly called a bench power supply.  This little device can put out any voltage between 0 and 18 volts and any current between 0 and 3 amps. I'll use that to charge individual cells.

I had gone for a particularly long drive and I'd used about 98 of the 120 amp/hours the batteries hold.  I decided to take a few laps around the neighborhood, and once I'd drawn out a total of 115 amp/hours, I pulled into the garage.  I then measured every battery and found the 4 lowest cells.  They were each just over 3.0 volts.  I turned the car on, turned the heater on, got my DVM ready and started measuring.  I kept my eye on all the cells, but paid particular attention to the 4 lowest ones. It became clear in a short period of time which one was dropping first so I focused my attention on that one.  By the time it got to 2.74 volts, there were still several other cells right near 3.0 volts.  I decided to add some current to that one cell and get it up to 2.9 volts.  It took about 10 to 12 minutes, with the bench power supply set at 3 volts and 3 amps.  That worked out pretty well, so I turned the heater back on to bring them all down.

By the time that cell reached 2.70 volts, a few others were just below 2.75 with the rest in the 2.8's and 2.9's.  I would need to draw the power out of those individually.  But when it was all said and done, the 120 amp/hour CALB cells had given up 127.5 amp/hours of current.  Not bad, not bad at all.  I threw the big red switch isolating the battery pack from the car and began working.

I went through all the batteries again, recording the voltage on each.  At that point, I began identifying which needed some extra power pulled out and how much I'd need to pull out.  Starting with the first battery in the pack, I clipped the lights to the terminals and watched with the DVM.  The lights issued a nice warm glow as they ate up the power.  When the DVM read 2.65 volts, I unplugged the lights.  The meter immediately began to rise and within 5 minutes it was over 2.75.  Ok, so 2.65 wouldn't do, how about 2.60?  That showed more promise, and I proceeded through about 5 batteries in this manner.

It was very slow going as it took anywhere from 15 to 30 minutes to drain each battery.  But then I realized that I could do more than one at a time if they were adjacent to each other in the series.  At that point, I started working in groups of 2, 3 or 4 batteries at a time.  The more batteries I hooked up to the lamps, the brighter they glowed.  Worked perfect.

What didn't work perfectly, or I should say, what I didn't expect was how much the batteries would recover after I got them down to the 2.74 mark.  It took me two days work, to get through all the batteries.  When I'd finally got the last one down, just below 2.74, I went back and started measuring the cells at the front of the pack, only to find that they'd recovered and were now all hovering around 2.76 to 2.82 volts.  Ok, this was going to require another pass through the whole pack taking them down perhaps a tad further.  But before I did that, I decided to run a quick test.  I worked with the first 4 batteries, and recorded the voltage of each.  I decided that I would leave the lamps on the battery for 10 seconds, plus 5 seconds for every 1/1000th of a volt over the target.  Once I was done drawing off the power I left them over night and came back to measure them in the morning.  

The next day, with the flu in full swing now, I dragged myself out to the garage and took measurements on those cells.  Each was between 2.74 and 2.75.  That did it!!  Next I simply had to work my way through the rest of the pack.  That took a few hours, but as I'd learned, I had to let them rest for a while before I could get an accurate reading off them.  Since I felt horrible and was in danger of covering every battery with plegm, I decided to call it a day and start fresh in the morning.

Next morning they all looked perfect.  Now came time to carefully charge each up to the 2.750 target I was looking for.  The short story here is that I got the first 18 or so to that target voltage, and went in for the night.  I basically charged them up to 2.83 volts each and watch them settle back to 2.750.  However the next day when I came out to continue with the remaining 30 cells, I'd found that all the batteries that had been at 2.750 had crept up and were all over the place.  The lowest was at 2.758, the highest at 2.810.  What the...

This is when the notes I'd been taking became very useful.  I found that the ones that had crept the highest were the ones that were on the charger the longest.  Even though a particular cell might have settled to 2.750, if it took more energy to get it there, while resting over night the voltage climbed higher than one that had needed less energy to get it to the target.

So what was the magic formula?  I can't really tell you because as I worked I got more in tune with the batteries and what they needed.  It really was a very touchy-feely operation.  But I can tell you with conviction that what brought me success was recording everything I did and how the batteries reacted to them.  I could start recognizing patterns and figure out how they would react to any given charge or discharge.  Plus, you simply must give the batteries time to settle once you've worked with them; 6 hours is good, 12 is better.

What I also found was that getting the cells to 2.750 volts at 11:30 at night when it's cold means that they will rise to 2.756 or so when the day warms up.  I can also tell you that getting them all to the same 1/1000th of a volt proved more of a challenge that I was willing to endure.  I settled for a spread across 2 1/1000th's of a volt.  I'd say that's acceptable in anyone's book.  So yesterday morning, I took measurements of all the cells and found, after they'd rested all night, that every cell was 2.752 or 2.753 volts.  The entire pack measured 132.1 volts.  Time to charge.

I was particularly interested to see how much current they'd accept at this point.  I threw the big red switch connecting the pack to the car and the charger, plugged in and ramped up the current.  I set the charger for the highest setting it can muster and it was putting out 29.0 amps.  That went on for three hours, when the voltage got high enough that the charger said "we're done" and shut off.  Well, I knew we weren't done and I expected that.  Those who've read my experiences from last September and the lessons I learned from Ohm's law will remember.  I hadn't touched the voltage potentiometer on the charger, so it was reading a higher voltage than the normal shutoff voltage I've selected since I'd set it for more of a lower 15 amp charge.  Increase the amps to the batteries, resistance goes up, consequently voltage to push that current goes up and the charger reads that as a full pack and shuts off early.

This was all fine and anticipated.  I turned the charger back on and turned the current down to 20 amps.  About 2.5 hours later the charger started to tell me it was near done.  I measure each of the batteries and found that they were anywhere from 3.390 and 3.43 volts.  I wrote down the four cells that were highest with the intentions of watching them carefully until the end of the charge.

I turned the charger off, turned the current knob down and started it back up again.  This time I turned it to 15 amps, which would be a full charge on the batteries before I'd done the balancing.  I wasn't sure how much time was going to pass before they were done, but I needed to stay with them, measuring the whole time to be sure no cells went over.  As the batteries approached their normal cutoff voltage of 164.7 I'd set the charger for I was measuring every cell in the pack, starting at the front, working my way through them and starting over.  What I found was that the 4 I'd originally ID'd as showing a bit higher voltage remained slightly higher of the others the whole way.  That is until just right before the charger reached it's cutoff when one cell stuck it's head up and hit 3.45 volts before any other.  Five minutes later, the charger indicated that it was going into constant voltage mode and started to ramp down the current.

No cell went over 3.46 volts, well shy of the 3.6 CALB states is the top end.  In fact the highest cell went to 3.48 volts.  That leaves me plenty of wiggle room between the charger's cutoff and the top of the batteries.  When all was said and done, the batteries had accepted 119 amp/hours.  That's 1 amp/hour shy of their reported capacity and a full 8.5 shy of their demonstrated capacity.  But I'm fine with that because that means I can plug the car in, walk away, and remain confident that batteries will charge up to near their full capacity and not burn down my car or house.  That said, I don't intend to do that for a few cycles.  Not until I'm sure I'm seeing consistent behavior from all the cells.

There you have it.  I hope it wasn't as tedious to read as it was to experience, and I hope it offers some information to others planning the same thing.  In a few months, I'll run the batteries down again and see how well they've done staying in balance relative to each other.  Should be interesting.