Here’s where things start to get pretty interesting–and where money really starts leaving never to be seen again. I’ve talked briefly about the battery pack itself, and still plan to write more about that decision process, but it’s in-hand and looks well suited to do the job.
That’s the easy part. Now we’re into figuring out how to charge the battery, how to protect the battery, and how to use energy from it. Before we go too far, let’s look at a couple of things about the battery pack.
We’re dealing with a module that’s 12 cell packs in series. They’re lithium ion, which means that 100% charged is at 4.2V/cell. That means 50.4V is as high as we ever want to go, and 48V is really more in line with what we’re shooting for.
This introduces both some opportunities and some challenges. Challenges primarily in the sense that 48V chargers for lead-acid batteries go too high (~56V) for these batteries. Most off-the-shelf battery managements systems can’t handle this–they’re designed only to deal with slight cell imbalances and shunt across fully charged cells to balance them. If all cells are being shunted, you’re going to get a good bit of heat. We also don’t want to use a normal charger set to float at a lower voltage either–that would do away with most of the fast charging capability we’re going after with the lithium batteries.
An opportunity arises though in trying to use one of these chargers. If we had a battery management system that would simply turn the charger on and off at preset battery voltages, we’d be able to bulk charge at the charger’s maximum rate until we shut it off, and save a bundle over a specialized charger. At 48V, there are tons of golf cart chargers available, including a number of brands that are American made. The going rate for a 1200W charger is about $300.
If that can be made to work, just one will let us charge the entire 12kW Volt battery on a 15A circuit in less than 10 hours, which isn’t too bad. Unlike with a lead acid battery bank, we’ll actually gain linearly by adding a second one–cutting total charge time to less than 5 hours, and we could still do that on a 30A hookup.
So what we need is a way to control the charger externally. With my background in nuclear engineering, there’s a tendency to overdo things with redundancy, but in this case I think it’s justified. We simply can’t tolerate overcharging the battery bank. There’s not a commercial product available that satisfies my requirements, so I’m into the territory of building up my own controller. Here’s what I want it to do:
- Turn on the charger only when voltage readings and temperature are within acceptable ranges.
- Shut off the charger at 49.2V (4.1V/cell)
- Keep the charger off until voltage drops to 46.3V, to keep the battery reasonably well charged but not have a ton of on/off operations
- Have two independent and redundant voltage measurements
- Have two redundant relays for turning off the charger
- Detect whether the charger is powered, to confirm that it has been successfully disabled.
- Detect when either relay or voltage sensing circuit is faulty
- Have wireless connectivity to log battery charge history and performance.
It sounds kind of complicated, but that system can be built with less than $100 in parts, and a couple of pages of code. We’ll get in to how to do that next time.
2 Comments Add yours
I have a working programmable BMS for Chevy Volt Packs that were converted to 12 Volts. I have been working on it for last two months, and just need to build the battery box, plumbing, etc. I finally got around to posting what have done so far, on my blog, with pictures today. Checkout my blog for more info. http://montana3611rl.blogspot.com/2016/02/chevy-volt-lithium-battery-packs-for-my.html
Awesome! I’ve been tinkering with my monitoring system today, and was actually writing up the next post when I saw your comment. My approach is a little different, with a Particle Photon running the show, and monitoring pack voltage and current in/out. Since voltage varies quite a bit with state of charge, I’m looking for a deviation from an expected voltage to detect a problem with an individual cell–which cuts way down on the number of wires involved (which would be 4x your setup since I’m sticking with 48V). We should definitely compare notes! –Dave