Author Topic: Test coin batteries  (Read 5701 times)

0 Members and 1 Guest are viewing this topic.

Offline swissTopic starter

  • Newbie
  • Posts: 1
  • Country: ch
Test coin batteries
« on: November 09, 2023, 06:20:40 am »
Hello everyone,
We currently suspect that some of the cr2032 button batteries we receive do not meet the specifications. We would therefore like to buy a setup where we can check the capacity and internal resistance of individual cells.
The solution should not cost more than 2000 USD. Can you recommend something?
 

Online TERRA Operative

  • Super Contributor
  • ***
  • Posts: 3168
  • Country: jp
  • Voider of warranties
    • Near Far Media Youtube
Re: Test coin batteries
« Reply #1 on: November 09, 2023, 07:00:01 am »
Any DC load that can apply a constant load with data logging features should do it, yeah?

Hook up the battery and test using the relevant standardised test if there is one, or the parameters in the datasheet from the manufacturer.

Even an ebay or Aliexpress special would be capable of doing this sort of test. Heck, even a mosfet on a heat sink with a couple multimeters with logging would work.
Where does all this test equipment keep coming from?!?

https://www.youtube.com/NearFarMedia/
 

Online Fungus

  • Super Contributor
  • ***
  • Posts: 17518
  • Country: 00
Re: Test coin batteries
« Reply #2 on: November 09, 2023, 07:21:52 am »
Even an ebay or Aliexpress special would be capable of doing this sort of test. Heck, even a mosfet on a heat sink with a couple multimeters with logging would work.

An Arduino and a 10k load resistor...?
 

Online robert.rozee

  • Frequent Contributor
  • **
  • Posts: 326
  • Country: nz
Re: Test coin batteries
« Reply #3 on: November 09, 2023, 07:39:27 am »
We currently suspect that some of the cr2032 button batteries we receive do not meet the specifications.

i take it the devices these cells are being used in is a DM15L or similar calculator? one problem you may be encountering (if this is the case) is high intermittent current draw pulling down the cell voltage to the point of (a) resetting the calculator and/or (b) generating waste heat within the cell (waste heat = lost energy). most lithium button cells are rated for a relatively low current draw (a few mA) and with any higher current draw the usable cell capacity will be diminished. therefore in a specific user-case cr2032 brand A versus cr2032 brand B may have vastly different usable lifespan in your application, whereas on paper and under standardized testing perform identically.

your best bet would use the actual device (DM15L) as the load, running custom software to keep it 'exercising' the cell to simulate typical usage. a simple DMM performing data logging, attached to a PC, can then monitor battery voltage. something like an HP34401A set to read cell voltage continuously and outputting data via the serial port would suffice. then use almost any terminal program to log data to a file for examination in Excel / Libre Office Calc.


cheers,
rob   :-)
 

Offline Ian.M

  • Super Contributor
  • ***
  • Posts: 13216
Re: Test coin batteries
« Reply #4 on: November 09, 2023, 08:09:47 am »
You don't have the budget for pro grade equipment to do this job (e.g. a Keithly SMU would be an order of magnitude more expensive) so will almost certainly need to DIY, as hobby grade battery discharge testers capable of detailed logging are typically intended for far higher capacity batteries, so cant provide a suitable load.

Look at the battery datasheets!  If you want to reject out-of-spec cells, you need to test with the manufacturer's  specified load, not your device's actual load current.  However if your device exceeds the highest pulsed load in the datasheets of good quality branded cells, you have a problem!

Most if not all the datasheets seem to show data for resistive loads rather than constant current loads, and the lowest resistance pulsed load I have seen is 100 ohms for an Energiser CR2032. 

I agree with Fungus: use an Arduino or similar.   The higher resistance continuous loads could be directly switched by MCU I/O pins sinking their current, but anything under 1K is going to need a small MOSFET, with its gate driven by the I/O pin.  If the load resistance is accurately known, the discharge current can be calculated from the terminal voltage, and an AVR Arduino's ADC would be good enough, if buffered by a unit gain OPAMP so its sampling current doesn't load the cell under test.

If you want a standalone unit consider hacking a M328 style component tester kit.  Fit your discharge resisters in place of the kit's resistors between the MCU and the test pins, hack in an OPAMP buffer in one of the ADC inputs from the test pins, and hack the other two to drive MOSFETs for the low ohm pulsed loads, link all the test pins together and connect the cell holder between there and ground.  For firmware development purposes it can be treated as an Arduino with no bootloader.
 

Online Fungus

  • Super Contributor
  • ***
  • Posts: 17518
  • Country: 00
Re: Test coin batteries
« Reply #5 on: November 09, 2023, 08:34:19 am »
AVR Arduino's ADC would be good enough, if buffered by a unit gain OPAMP so its sampling current doesn't load the cell under test.

Datasheet says the ADC input resistance is 100MOhm so I'm guessing it already has one.

(section 28.8 )

Write a program to output a timestamp and voltage every time you see a new minimum value. Paste the results into a spreadsheet and plot a graph.

If you want to pulse different loads you could use the I/O pins to sink current via different resistors.
« Last Edit: November 09, 2023, 08:46:41 am by Fungus »
 

Offline Ian.M

  • Super Contributor
  • ***
  • Posts: 13216
Re: Test coin batteries
« Reply #6 on: November 09, 2023, 09:05:37 am »
See ATmega328P full datasheet section 28.6.1. Analog Input Circuitry and you will find the dynamic impedance of an active ADC input during sampling is under 100K (possibly by more than an order of magnitude) so could well be significant with respect to the cell load resistor. 

IMHO a low bias current unit gain OPAMP buffer would be prudent, though it is possible that a 33nF capacitor to ground at the ADC pin to provide the charge for the internal sampling capacitor without greater than 1/2 LSB change in the pin voltage (worst case) would be sufficient.
 
The following users thanked this post: CatalinaWOW

Offline Psi

  • Super Contributor
  • ***
  • Posts: 10385
  • Country: nz
Re: Test coin batteries
« Reply #7 on: November 09, 2023, 09:32:06 am »
Hello everyone,
We currently suspect that some of the cr2032 button batteries we receive do not meet the specifications. We would therefore like to buy a setup where we can check the capacity and internal resistance of individual cells.
The solution should not cost more than 2000 USD. Can you recommend something?

Sadly no, characterizing and simulating CR2032 for engineering RnD/production is an area where the market is lacking unless you have deep pockets, I have looked.
The annoying thing is that the proper equipment that can do it is way overspeced for the task. Why do I need a 500W load/powersupply to test a coincell...
There probably is an untapped market for a $1k product that can do this for coin cells and other tiny batteries. (four-quadrant power supply with accurate current/voltage datalogging at 1uA resolution ).


Any DC load that can apply a constant load with data logging features should do it, yeah?

Sadly no, normally 1mA is the lowest discharge current you can set. Accurately trying to work out capacity when your operating at the lowest digit precision of the DC load is impossible.
The resolution of the selectable discharge current is just too low.

And even 1mA is way to high, the datasheets usually give capacity for like a 200uA constant current discharge.
« Last Edit: November 09, 2023, 09:52:20 am by Psi »
Greek letter 'Psi' (not Pounds per Square Inch)
 

Online tszaboo

  • Super Contributor
  • ***
  • Posts: 8217
  • Country: nl
  • Current job: ATEX product design
Re: Test coin batteries
« Reply #8 on: November 09, 2023, 09:32:44 am »
I have a Keithley 2380 for capacity tests. It doesn't do ESR directly, but you can either calculate it with a pocket calculator, or write a Python PYVISA script to measure it. It really depends on your definition of ESR, I've seen 10 different definitions already. I think it would fit your budget, though not sure how much it costs now.
 

Offline Psi

  • Super Contributor
  • ***
  • Posts: 10385
  • Country: nz
Re: Test coin batteries
« Reply #9 on: November 09, 2023, 09:44:13 am »
I have a Keithley 2380 for capacity tests. It doesn't do ESR directly, but you can either calculate it with a pocket calculator, or write a Python PYVISA script to measure it. It really depends on your definition of ESR, I've seen 10 different definitions already. I think it would fit your budget, though not sure how much it costs now.

Keithley 2380 current resolution is like 100uA.  That isn't small enough to do capacity testing on a coin cell discharge at 200uA.
The problem with the Keithley 2380, and a lot of similar things, is that they are designed for devices that draw up to like 200W.
They just don't have the resolution down in the sub-mA scale.
« Last Edit: November 09, 2023, 10:03:15 am by Psi »
Greek letter 'Psi' (not Pounds per Square Inch)
 

Offline Psi

  • Super Contributor
  • ***
  • Posts: 10385
  • Country: nz
Re: Test coin batteries
« Reply #10 on: November 09, 2023, 10:00:12 am »
A Keithley 2450 SMU would do it.   10 nA – 1 A range
but you wont like the price.
« Last Edit: November 09, 2023, 10:05:10 am by Psi »
Greek letter 'Psi' (not Pounds per Square Inch)
 

Online tszaboo

  • Super Contributor
  • ***
  • Posts: 8217
  • Country: nl
  • Current job: ATEX product design
Re: Test coin batteries
« Reply #11 on: November 09, 2023, 11:17:33 am »
I have a Keithley 2380 for capacity tests. It doesn't do ESR directly, but you can either calculate it with a pocket calculator, or write a Python PYVISA script to measure it. It really depends on your definition of ESR, I've seen 10 different definitions already. I think it would fit your budget, though not sure how much it costs now.

Keithley 2380 current resolution is like 100uA.  That isn't small enough to do capacity testing on a coin cell discharge at 200uA.
The problem with the Keithley 2380, and a lot of similar things, is that they are designed for devices that draw up to like 200W.
They just don't have the resolution down in the sub-mA scale.
Well let's say a CR2032 coin cell is 200mAh, you don't want to test a battery for a thousand hours, you would set up like 2-10mA discharge current.
If you want to do lifetime measurements it's not the right tool, but for a generic "Is this 50mAh or 200mAh" test it's sufficient. It's definitely not as great as an SMU but it doesn't cost as much as well. If I would know there is a market for it, I could design a coin cell testing setup, since I already made a pretty extensive, ~0.03% (current) and 0.003% (voltage) accurate battery testers in my past job.
 
The following users thanked this post: BILLPOD

Online Fungus

  • Super Contributor
  • ***
  • Posts: 17518
  • Country: 00
Re: Test coin batteries
« Reply #12 on: November 09, 2023, 11:57:20 am »
See ATmega328P full datasheet section 28.6.1. Analog Input Circuitry and you will find the dynamic impedance of an active ADC input during sampling is under 100K (possibly by more than an order of magnitude) so could well be significant with respect to the cell load resistor. 

I think it's section 28.9 in the one you linked above.

Maybe if you're switching between analog input pins and have to keep charging the internal capacitor.

If you're not switching between channels the the charge on that capacitor will be steady.
 

Offline Psi

  • Super Contributor
  • ***
  • Posts: 10385
  • Country: nz
Re: Test coin batteries
« Reply #13 on: November 09, 2023, 11:58:43 am »
I have a Keithley 2380 for capacity tests. It doesn't do ESR directly, but you can either calculate it with a pocket calculator, or write a Python PYVISA script to measure it. It really depends on your definition of ESR, I've seen 10 different definitions already. I think it would fit your budget, though not sure how much it costs now.

Keithley 2380 current resolution is like 100uA.  That isn't small enough to do capacity testing on a coin cell discharge at 200uA.
The problem with the Keithley 2380, and a lot of similar things, is that they are designed for devices that draw up to like 200W.
They just don't have the resolution down in the sub-mA scale.
Well let's say a CR2032 coin cell is 200mAh, you don't want to test a battery for a thousand hours, you would set up like 2-10mA discharge current.
If you want to do lifetime measurements it's not the right tool, but for a generic "Is this 50mAh or 200mAh" test it's sufficient. It's definitely not as great as an SMU but it doesn't cost as much as well. If I would know there is a market for it, I could design a coin cell testing setup, since I already made a pretty extensive, ~0.03% (current) and 0.003% (voltage) accurate battery testers in my past job.

Depends what the goal is I suppose, if a product is using the CR2032 at 2-10mA and you want to characterize it at that discharge then its fine.
But if the goal is to confirm it meets datasheet capacity spec you really have no choice but to discharge using what the datasheet says the capacity is measured at, which is ~200uA and takes like 45 days.
Might be able to push it up to 1mA and still get rated capacity, depending on brand, but much more than that and you will definitely measure it as under-capacity when its actually not.

In any case, for the OP situation the manufacturer isnt going to accept evidence of under performing cells unless the data shows them tested at the correct discharge current.
« Last Edit: November 09, 2023, 12:07:55 pm by Psi »
Greek letter 'Psi' (not Pounds per Square Inch)
 

Offline Stray Electron

  • Super Contributor
  • ***
  • Posts: 2253
Re: Test coin batteries
« Reply #14 on: November 09, 2023, 01:51:19 pm »
Hello everyone,
We currently suspect that some of the cr2032 button batteries we receive do not meet the specifications. We would therefore like to buy a setup where we can check the capacity and internal resistance of individual cells.
The solution should not cost more than 2000 USD. Can you recommend something?

   Do you just want a load and a meter for a manual readout or are you looking to do data logging?  If logging then what interface? I have several HP 3437A voltmeters that would be great for an application such as this.  They only have three voltage ranges; 0 to .1 volt, 0 to 1 volt and 0 to 10 volts and they're only 3 1/2 digit so they're not very useful for other applications but they will take up to 5700 readings per second and output the results over HP-IB so they would be perfect for an application like this.
 

Online BILLPOD

  • Frequent Contributor
  • **
  • Posts: 385
  • Country: us
Re: Test coin batteries
« Reply #15 on: November 09, 2023, 01:52:53 pm »
Good Morning Swiss, welcome to the forum :popcorn:
 

Offline Psi

  • Super Contributor
  • ***
  • Posts: 10385
  • Country: nz
Re: Test coin batteries
« Reply #16 on: November 09, 2023, 09:14:44 pm »
These are ok for a quick test of a cell to ensure it's good before installing in a product.
https://www.ztsinc.com/mbt1.html
Greek letter 'Psi' (not Pounds per Square Inch)
 

Offline Psi

  • Super Contributor
  • ***
  • Posts: 10385
  • Country: nz
Re: Test coin batteries
« Reply #17 on: November 09, 2023, 09:16:12 pm »
We currently suspect that some of the cr2032 button batteries we receive do not meet the specifications.

One thing that cause problems in production with CR2032 is people not knowing how to handle/store them.

Production Operator: "It takes too long to pick them up from the plastic tray each time, so I dump them into a loose pile, then I can work faster"   :palm:
« Last Edit: November 09, 2023, 09:18:32 pm by Psi »
Greek letter 'Psi' (not Pounds per Square Inch)
 

Online bateau020

  • Frequent Contributor
  • **
  • Posts: 288
  • Country: fr
Re: Test coin batteries
« Reply #18 on: November 11, 2023, 06:31:25 am »
An Arduino and a 10k load resistor...?

Indeed, something like that. Although I built mine for Li-Ion cells, I also often use it on coin cells. See https://github.com/hb020/batterymeter
 

Online ebastler

  • Super Contributor
  • ***
  • Posts: 7375
  • Country: de
Re: Test coin batteries
« Reply #19 on: November 11, 2023, 07:19:28 am »
Well let's say a CR2032 coin cell is 200mAh, you don't want to test a battery for a thousand hours, you would set up like 2-10mA discharge current.

If the goal is to check whether the supplied batteries meet specifications (as stated by the OP), I would test them under the conditions the manufacturer has specified. For the Energizer CR2032, that would be a 15 kOhm load, resulting in approx. 50 days of discharge time: https://data.energizer.com/pdfs/cr2032.pdf

So I think one feature of a practical tester for incoming inspection would be the ability to test a small batch of batteries in parallel. The "Arduino ADC with load resistor" solution doesn't look too bad...

The spec I would be more worried about (if this is for an application like a pocket calculator) is self discharge, by the way. About 1% per year is the spec, and I don't see a solution to accelerate the test. On the other hand, no dedicated tester required -- just put a few batteries into the drawer and take them out once a month for a brief measurement.
 

Offline Smokey

  • Super Contributor
  • ***
  • Posts: 3044
  • Country: us
  • Not An Expert
Re: Test coin batteries
« Reply #20 on: November 11, 2023, 10:21:09 am »
 

Online tszaboo

  • Super Contributor
  • ***
  • Posts: 8217
  • Country: nl
  • Current job: ATEX product design
Re: Test coin batteries
« Reply #21 on: November 11, 2023, 10:53:09 am »
Well let's say a CR2032 coin cell is 200mAh, you don't want to test a battery for a thousand hours, you would set up like 2-10mA discharge current.

If the goal is to check whether the supplied batteries meet specifications (as stated by the OP), I would test them under the conditions the manufacturer has specified. For the Energizer CR2032, that would be a 15 kOhm load, resulting in approx. 50 days of discharge time: https://data.energizer.com/pdfs/cr2032.pdf

So I think one feature of a practical tester for incoming inspection would be the ability to test a small batch of batteries in parallel. The "Arduino ADC with load resistor" solution doesn't look too bad...

The spec I would be more worried about (if this is for an application like a pocket calculator) is self discharge, by the way. About 1% per year is the spec, and I don't see a solution to accelerate the test. On the other hand, no dedicated tester required -- just put a few batteries into the drawer and take them out once a month for a brief measurement.
True, but if you want to check if the batteries are not meeting the specifications anymore, that's a different story. You can a-b test with a sample from a previous batch. Plus I can tell you that for free, that stopping production for a 50 day long test is going to be an issue everyone involved. You would want to get answers as soon as possible.
 

Offline mariush

  • Super Contributor
  • ***
  • Posts: 5170
  • Country: ro
  • .
Re: Test coin batteries
« Reply #22 on: November 11, 2023, 10:55:46 am »
Why not just spend money on a multimeter with data logging feature, and discharge a few batteries using various resistor values. Then import the logged data into Excel and see if the battery behaves as you should expect.

If you want to test things like on-off , you could short out the resistor using a 555 or a microcontroller ...
 
The following users thanked this post: BILLPOD

Offline HKJ

  • Super Contributor
  • ***
  • Posts: 3086
  • Country: dk
    • Tests
Re: Test coin batteries
« Reply #23 on: November 11, 2023, 12:04:14 pm »
A multimeter with computer connection, a resistor and TestController, then you can do it.
TestController thread: https://www.eevblog.com/forum/testgear/program-that-can-log-from-many-multimeters
 

Online Fungus

  • Super Contributor
  • ***
  • Posts: 17518
  • Country: 00
Re: Test coin batteries
« Reply #24 on: November 11, 2023, 02:30:00 pm »
Plus I can tell you that for free, that stopping production for a 50 day long test is going to be an issue everyone involved. You would want to get answers as soon as possible.

You could do an accelerated test against a known-good battery and see if there's any obvious difference, eg. if the suspect battery dies in half the time.

(by "accelerated" I mean over 3-4 days...)
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf