Another Hybrid / Electric Car Thread So Start Screaming

Hey Everybody,

I've been all over the internets looking for grassroots ways to load test individual cells and balance hybrid / electric battery packs. So far, I've seen some high-dollar solutions except for one 50-dollar method for balancing / conditioning but it takes an entire month. Yes, you read that right. That's about 29 1/2 days too long.

Just point me in the right direction and I'll take it from there.

Thanks in advance,

Jerry

Bumped

Disclaimer: I've never done this on a hybrid battery, or any battery for that matter. In fact I didn't even stay in a Holliday Inn Express last night, but we have a machine that load tests our aircraft batteries and I think I can break down it's automated procedures into something we can do in our garage:

So it looks like at the lowest level your cells look something like this:

Where as our aircraft batteries look like this:

Physically it looks different, but if you look at the top picture it looks like 1+ and 1- are right next to each other just like in our aircraft battery cells and the rest follow suit. I bet there's a drawing on google that will tell us which is which.

So when we do our scheduled checks we first hook up a machine to the + and - of every cell to read the voltage. We have 20 cells that total 28vdc, so that's 1.4v per cell. Our job guide says that any difference of .1v needs to be "looked at". Since you don't have access to the same machine I might suggest finding the best multimeter you can and doing the same check one cell at a time (maybe write down your results for comparison at the end?)

If this goes well and all your cells read exactly the same minus one or two that read significantly less then I would say "that's that" and replace the bad cells. If your readings are all over the place then something's different (either the info doesn't relate, or your meter sucks, or there's something I'm missing).

But that's not a "load test". That's just a "voltage check" at the cell level. For a "load test" we need a "load" (really just a low resistor). Something like this:

If we use 12v car battery load testers as an example (they are usually rated at 500amps for a 12v battery) then we're looking at a load resistance of .024 for a 12v source.

I think we can safely assume the load tester would be rated ABOVE the max discharge rate of a car battery then at a 70% duty cycle we're looking at 350 amps, 12v (still), and a resistance of .34 ohms. We can keep using ohm's law to find out what resistance we'd need for your cells, but honestly it sounds like we just need a wire that will handle the current.

Expected current = voltage of cell / resistance of wire

I would suggest rigging up a clip system that allowed you to connect your "load wire" and your multimeter to the battery at the same time. The multimeter should already be on and set for DC voltage. Maybe a push-button switch could be used on the wire to control when the battery was put on a load?

The procedure would be to connect the load and meter to each cell's +/- for about 5 seconds. Write down the voltage, repeat, compare, and look for any deviation from the "norm". Replace the suspected cells as necessary. Of course this is again assuming you get a consistent reading from all but a few cells.

You might want someone with a stopwatch to keep your times consistent as you're looking for deviations in voltage from cell to cell. Any deviation in time (longer or shorter) could give you false readings. Our machine just says "test complete" and lets us know which (if any) need to be replaced.

There's also a capacity check that'd tell ya more, but we let that run overnight and it involves completely discharging the batteries (8+ hours?)... Not sure if we want to do that with yours.

Couple things to note:

• armchair amateur here. I've never done this manually, and I've definitely never done this to a hybrid (and I are teh scared).

• our battery cells are NOT the same. We use lead acid, you're probably nicad, or lithium ion, or unobtanium exotic, or something else that we're not. The tests may or may not translate and may or may not have catastrophic effects on your cells.

For the "balance check" we generally have a "balance charge" option on our machine. Before we go replacing battery cells we have the option of charging at a higher than normal voltage level for longer than normal to bring up the low cells. We then do another voltage check and load check to make sure they were just "out of balance" and not "weak". I'd be a bit nervous to do the same thing on your batteries without some sort of written guide.

Good luck!

I'm pretty interested as well, both because I've been shopping off-lease Leafs (24 month old EVs with 20K miles on the clock starting around $12,000!) and because this seems like a completely untapped market.

EVs and Hybrids tend to lose 5-10% of their battery capacity each year. That means that the already limited range of EVs shrinks year over year until the battery pack must be replaced or serviced to keep the vehicle useful.

Right now car makers don't want owners or independent technicians touching the battery packs at all, so owners are stuck with whatever their dealership wants to charge them for battery servicing. An effective method for isolating and replacing problem cells, or reconditioning entire battery packs, could be a fantastic alternative for consumers and would go a long way toward making EVs a more compelling used car option.

Here's my $.02: (I have not done this, but I do have a physics degree and an itch to do something similar, take my advice for what you paid for it - ha)

On the cheap, you can voltage test, load test, and balance charge a battery pack like this.

1. Voltage test - Like Hungary mentioned, look up the schematic and use a good multimeter on each cell.
2. Load test - Same as above, but get a power resistor (Volts = Resistance * Current, size your resistor to get a similar current to what your motor pulls) and measure the current with your multimeter. Might need a 'shunt' resistor to measure voltage drop across the shunt rather than current because multimeters can't handle huge currents.
3. Balance charge - go by one small AC/DC converter (like a cell phone charger) for each cell or small group of cells. They output a specific voltage and don't have large current outputs. As long as you size the charger to be slightly higher voltage than the cell, it will charge it to full capacity given enough time. Since the cell phone chargers are low current, you won't overheat a big cell. You will have to watch the voltages on each cell periodically and remove the charger when the cell is charged. But at least you can charge them all independently to a 'balanced' state. If it takes too long, charge the cell with your main charger first, and then balance manually as necessary.

For bonus points, you can probably do all this in an automated way with a few arduinos (dependent on how many inputs you can have on an arduino and how many cells you have) and a bunch of relays to connect and disconnect the chargers. Just need a connector from the battery that has one lead for each cell.

Edit, maybe I should add some on what to look for! haha.

1. Voltage test - tells you when the pack is charged to a balanced level. Do step 3 until step one reads all cells similar.
2. Load test - you will be looking for the battery to sustain more than a specified current for a specific amount of time (at a certian temp). I.E. automotive batteries have 'cold cranking amps' which is the current they can output for at least 30 seconds at 32 degrees or something like that. All batteries will have some sort of spec of current and time to pass the load test. Bad batteries could show good voltage, but not be able to sustain high current for the right amount of time.

Where's Nashco? I remember he had a setup for balancing the batteries in the hybrid Fiero that involved bleeding energy into a hair dryer. Took just a few seconds.

Great stuff, guys. Keep it coming. I'm thinking about a side job fixing / replacing battery packs. Balancing, conditioning, individual cell replacement, etc. When warranted, complete pack replacement from the increasingly ripening aftermarket.

Hungary, that looks like a Honda battery pack. Each tube contains 6 D-type cells soldered together and encased in plastic. Testing and replacement is done by the tube, providing one has access to good used parts. Toyota has a rectangular cell with prominent terminals.

I'm not a hybrid / electric fan but I am a pragmatist. I see a niche forming with people who want to sell hybrids with bad packs or want to keep their cars rolling. I was inspired by Ashyukun's build / flip of an '05 Prius.

Besides, repair rather than replacement when warranted will benefit both the consumer and the environment.

http://priuschat.com/threads/gen-ii-prius-individual-battery-module-replacement.125588/

Not sure if that's the particular thread I remember, but it looks a lot like it. Much faster than a month, but longer than a day. Pretty good explanation and some interesting links to other threads and such. Seems quite DIY, but a bit time consuming with regards to get all the cells to play nice together.

I wasn't at all interested in Hybrids until this thread. Now I kind of want to grab an Insight with a bad pack and see what I can do in a weekend

Damn you Jerry.

Hungary Bill wrote: Damn you Jerry.

Have you been speaking to my wife?

Ixnay on using the Ireway.
500 Amps at 12 V = 6,000 W of power.
Unless you want to deal with fire and molten metal, use a load capable of safely handling that much power, or choose a lower test current.

A lower test current, say 25 Amps or even 10 Amps is enough to identify a bad cell. Ideally all the cells will show about the same voltage.
Any that are substantially different are suspect, and can then be individually tested.

Also, I'm wondering if a conventional battery load tester could be adapted to testing individual cells. By "adapted," I mean converting the meter scale to measure accurately a string of Honda D-cells (as in Hungary's example) that might measure 8.4 volts fully charged instead of 12.75 volts for a conventional lead-acid car battery.

Really got to love a forum where folks can give detailed how-tos that include the phrase "may have catastrophic effects..."!!

In reply to Jerry From LA:

I agree with your business assessment. I have been scheming for a while....

Here's a lead on a guy with good info:

Tampa Prius. The owner's name is Todd Thompson, and his cell is

813-241-5307.

He is in the business of selling cars, so he is not going to want to give away too many of his trade secrets, but he is very knowledgable and helpful.

He spent many years as a car dealer, and has now decided to scale back to spend time with his family. His business model is now selling only 2nd gen Priuses. He has 15-20 on his lot at a time, and doesn't finance.

He knows a LOT. He can tell you just about everything there is to know about Priuses, including how he rebuilds single cells of the battery packs.

If you call him and ask questions as a potential buyer willing to ship, he will share a lot.

Jerry From LA wrote: Also, I'm wondering if a conventional battery load tester could be adapted to testing individual cells. By "adapted," I mean converting the meter scale to measure accurately a string of Honda D-cells (as in Hungary's example) that might measure 8.4 volts fully charged instead of 12.75 volts for a conventional lead-acid car battery.

A conventional load tester can work OK.
1) Battery voltages are nominal, depends on the condition of each cell, and you should never expect to see those precise values.
2) Exact cell voltage changes with load, due to the internal resistance, which varies by how the cell is constructed.
3) The load tester itself varies with amount of current, as resistance of most materials decreases with temperature, and temperature of the resistance elements will be lower with lower current developed by a lower source voltage. So using it with a 8.4V battery instead of 12.75V battery will pull less than 8.4/12.75 Amps.

It's still a valid test scheme, since you are attempting to identify weak or suspect cells, because it will be dissipating a smaller load, and running at a lower temperature.

If you plan on load testing lithium based batteries, make damn sure your load is lower than the cell's max current rating. Remember, each of those cells is probably about a hand grenade's worth of energy. As mentioned, you should only need a few amps/fraction of max load (lightbulb?) to see the difference in voltage drop. You'll need a good voltmeter, you can get a older Fluke off ebay for around $50 that will be plenty accurate/precise.

Kenny, I saw a good video where a tech uses a conventional V-O-M to load test a battery by using the meter's "min-max" function. After insuring the battery was fully charged and switching the meter to min-max, the tech connected the meter leads to the battery and observed the voltage change while starting the car. If the voltage drops below 9.6 volts (as observed by checking the min-max log), the battery is deficient.

I'm sure there's a chart of known NiMH and Li characteristics that would allow someone to observe battery condition without using expensive test gear other than a good V-O-M.

Sigh.

LS-x swap!

In reply to Jerry From LA:

For a cranking battery that works, you don't even need min-max/logging, just a helper to turn the key while you watch for a number less than 9. I wouldn't want to be doing the same in the trunk of a hybrid with an opened up battery pack, driving down the road going over bumps and such though, that's a recipe for disaster. Unless the car will let you brake torque it to load the traction battery stationary, and I don't think any of them do.

By good voltmeter I mean a typical auto ranging digital VOM. In all honesty a HF freebie would probably do the trick, but the fluke is a much nicer piece of gear to use if you're taking measurements all day.

The only reason I like the min-max function is the numbers go by really fast. I'm old now so it's nice if they get captured so I can read them.