how to test LOTS and LOTS of batteries?

We have 420 panasonic 2800mAH 3.7V 18650 batteries, which we plan to use in a solar car. I was wondering if anyone is aware of how to many of these at a time. Many of the commercial "capacity tester 18650" online appear to only test a cell at a time. This would take way too long.

We need to ensure the cells we are connecting together are well-matched in terms of ESR, capacity, leakage, etc. So is there any solution that exists to accomplish this task?

sort by: active | newest | oldest

HOW long does it take to test a cell ? 1 second ? 10 ? a minute ? an hour ?

How is the test accomplished ? ESR is easy, capacity and leakage both sound time consuming.

-max- (author)  steveastrouk1 month ago

We are considering to do an AC analysis on the batteries as well. It's not something I have ever heard of, but supposedly it gives much better data. And yes, we need a way to test all 420 batteries and parameterize them individually, so that we can carefully match them in the final design.

The batteries are a bit old, sitting in storage for over a year, I would expect that they may have drifted in terms of performance. We are not sure of the variance from cell to cell.

As I said, how long do these analysers take per cell ? If its seconds, you need to create an erector set automatic feeder. If its hours, you need to make LOTS of electronic hardware, which I suspect is pretty nasty

-max- (author)  steveastrouk1 month ago

My hope was that someone may already know of a solution to do this. Given how many products are being developed with packs of batteries, I would expect at least some sort of automated test equipment to exist commercially.

I'm not sure if we want to develop custom hardware to test the cells, but at the same time I'm not so sure I can trust the cheap madea-in-china testers on amazon or ebay.

Buy one, and characterise it.

You STILL haven't put a handle on how long these tests will take.

-max- (author)  steveastrouk1 month ago

So then what is your advice for ensuring that cells of simalar performance characteristics are connected together?? Like I said we already have the batteries.

-max- (author)  steveastrouk1 month ago

We also need a BPS / BMS (Battery Protection (or Management) System) We do have a voltage monitor as per required but as far as I know, nothing that will actively balance the cells.

Back in the day the company I worked for had a lot of battery powered pagers and they were about the first things to really utilise these 18650 style Li/Ion batteries.
The style was different and size just over a AA cell.
Problem was that the supplied wall wart chargers for them had no controls of any kind.
Plug them in over night and take them out in the morning.....
Needless to say that we suffered from bad batteries a lot.
Our radio guy then created a bank for 10 batteries a time to charge and discharge them to set parameters.
AFAIK he just just 10 electronic chargers for this and packed them into a nice looking case and added some automation LOL
Anyway, the main reason for this was that the batteries needed to be "refurbished" before they could be tested for performance.
So our batteries were charged and discharged six times with 4 hours of rest between cycle.
After that the batteries rested for a full day and the no load voltage was checked.
Everything below a certain level was discarded and went into the junk bin.
For the rest and after a full charge the time to discharge at a certain load was measured - simple timer that was stopped when the specified voltage was reached.
Again all batteries performing below set parameters went into the bin.

All this took a lot of time but back then it was justified as every replacement battery was quted at over 20 bucks.
I know that basically every company dealing with big, multiple battery setups have some way of testing them in bulk but the only "machines" for that I ever saw were back in the 80's and for Ni/Cd types.
One problem I often encounter with my 18650's is that in devices requiring more than one cell they always drain at different rates.
Even when I mark them to later on match the ones that still had more juice left the problem was (or better still is) there with the next cycle.
I admit that my cells are not the most exensive but I tend to think the integrated protection circuits play a vital role in this weird behavior.
The only set of two I have that always performs nice and drains almost equally came with a cheap flashlight and is refused by my electronic charger stating no protection found.

With your topic in mind I wasted a few hours of sleep last night thinking about what exactly might affect the different discharge.
Four sets of batteries in four flashlights later it hit me hard: Temperature!
The battery always draining faster was the one down the handle.
A check after 3 hours of continous running showed the battery near the LED got much warmer due to the radiated heat from the housing.
But the flashlight with the unprotected batteries was still not really affected at all by this and the top battery was more than hand warm while the LED area was not nice to touch at all.
I think that even if you manage to overcome the problem of testing all your required parameters in bulk the end result might be quite different to the expectations.
If temperature is affecting the protection more than the actual battery than heat transfer will be your greatest issue.
And from what I could come up for testing you would need to automate the process for the battery feeding one by one or make a testing platform for multiple batteries at a time - both will be quite hard to accomplish in a reasonable time.
Manual labour might be faster than developing, testing and using a dedicated machine.
No clue how good your connections are or if you know people that can help with hiring but one of these might be of interest:


If that is too far out of budget, maybe this is better:


Wildcat Discovery Technologies in San Diego is develoding a new type of battery, maybe they are willing to sponsor your team!?

Can you test them in banks of 10 or 100 and if the numbers don't work out, then test each of them individually? It sounds like you need to know all the info of each battery, in which case, I would probably test each one. My experience is VERY limited though

-max- (author)  cerberustugowar1 month ago

Lithium ion batteries are pretty fragile and dangerous when mishandled, overcharged, undercharged, overtemp, overcurrent, etc. When you wire them in series and parallel, small differences in the ESR (Equivalent Series Resistance) and capacity can cause them to fall "out of balance" and the state of charge of each cell is no longer matched. This means that when charging one cell may charge up to 4.25V while another may only reach 4.15V, and upon discharge the undercharged cell may go below 3V while the others remain at 3.5V with some energy left. This can lead to really bad things.

So in order to make sure we can use the full capacity of each and every cell, it's imperative to test each individual cell to know its characteristics so we can wire similarly performing cells each module.

iceng1 month ago

Tesla manufactures a lot more batteries, then you can imagine, in the canyon approach to my town, so I know something about how they prep the batteries. Thousands of new batteries must be first charged at specific schedule and then discharged. The power cost would be astronomical if Tesla did not use discharging cells to Charge new cells.

Each cell has a built in fuse and is assembled in a monitored and switched parallel pack for inclusion in a buss structure as a removable block and has a forced liquid cooling.

The factory parking lot is as big as a mall for its employees and certainly has machines to automate the battery production process...

A Pi could be a sorting tester and a gravity roll alignment with after press contacts to process 10 to 40 batteries at a time...

-max- (author)  iceng1 month ago

They are definitely a "for profit" so it makes sense for them to slosh power between a bank of charged and discharged cells to wear them in, given that they need to reduce costs and perpetually maintain profit margins. I really like that idea and might present it as a suggestion for testing our cells. However we do not necessarily care about electricity costs,my guess is that UVA supplies us with the power we need for development, we have access to "free" power.

We do have some equipment like the NI VirtualBench, and the university is good friends with NI, we can get stuff for reduced costs or even donated from them as needed. (This is my a "myRIO" FPGA development platform is the embedded solution we have choose to use.) as well as a secondary raspberry pi model B for the driver interface GUI and backup camera.