1.2 V Battery checker by discharge

This project was inspired by the Martin OSSMANN battery checker. I found it can be enhanced and simplified. My scheme shows these improvements. It works fine now !

This project was inspired by the Martin OSSMANN battery checker. I found it can be enhanced and simplified. My scheme shows these improvements:

1 - The ATMega328 is programmed by means of Arduino : this makes me easier for programming.

2 - I use a 4 x 20 char LCD display: The Ossmann displays are for Elektor readers and specialists ! My displays are for friends and family who are not familiar with these purposes.

3 - I add a "START" button, so the user follows the operation.

4 - The LCD circuit is simpler (no negative voltage generation).

5 - I add a resistor (pull-down) to detect when the battery is removed from the socket (10 kOhm).

6 - The time counter uses the Timer 1 (ISP socket is available for, programming the ATMega328).

7 - The discharge current is selected by a switch depending on the battery size (AA or AAA size) : 50 mA or 20 mA. The switch position must match with battery size, else the check will not start. Very easy to use !

8 - I write the program using ARDUINO language, not ATMEL Studio. It is very simple to read and understand.

The other parts of the scheme are not modified significantly. The program works correctly. The program counts (Timer1) the discharge duration; during this time, the current in the FET is regulated to drain 20 mA (AAA battery), or 50 mA (AA battery) by Timer2 used as PWM generator.

Buildt it and post here your battery load results, or send an email !

All ideas and suggestions are welcome !

2021/01/08
I add a table showing the caacity of my batteries. Because someone are very old, I didn't mention the brand. It seems the capacity depends on the loading mode, on the 2nd battery during loading, . . . .

Project Elements

Now in the box #1 ! (JPG, 1.4 MB)

View of the PCB (JPG, 1.12 MB)

View of the display, battery selector and push-button (JPG, 1.41 MB)

View of the battery socket on the top side (JPG, 813.69 KB)

batterycapacitytable.png (PNG, 13.5 KB)

This table shows the capacity of some batteries.

AccuMonitor software (ZIP)

BatteryCheckerScheme.zip (ZIP)

Schematic of the Battery Checker : final revision

Accu Monitor vs. ATmega328 connections (PNG, 69.29 KB)

Shows the connections

AccuMonitor-Software.zip (ZIP)

Software of the Accu Monitor device

BattteryCheckerScheme.pdf (PDF)

AccuMonitor-Software.zip (ZIP)

The AccuMonitor software

AccuMonitor-Software.zip (ZIP)

The AccuMonitor software

BattteryCheckerScheme.pdf (PDF)

Discussie (2 opmerking(en))

TechDoc 9 jaar geleden

I found an article in Hackable magazine about the atomicity of the code. (See github on the web). Do you think this will help ?

Regards.

Antwoord

ClemensValens 9 jaar geleden

Please post a link to the article you refer to.

To me atomicity of the code refers to uninterruptable blocks of code. Why would that be helpful for you? If you globally disable interrupts, your code becomes pretty atomic.

Antwoord

Toon meer

1 Opmerking(en)

ClemensValens 9 jaar geleden

Did you check the clock frequency? Are your timers running at the correct frequency? The MCU in the Ossman project runs at 11.0592 MHz, the Arduino MCU normally runs at 16 MHz.

BTW, we are currently reworking the Ossman project into a Platino Battery Tester [130543]. It will also be able to recharge the battery after testing it.

Regards,
Clemens

Antwoord

TechDoc 9 jaar geleden

My idea is that we need to meet an accumulator manufacturer such as VARTA, DURACELL or GP. . . I think they will hear your request, not mine. I have 2 battery loaders and now I log the values with the loader model information.

Antwoord

ClemensValens 9 jaar geleden

Software filtering is like hardware filtering but done in software. In your case it would be low-pass filtering to get rid of the PWM ripple. Averaging is a simple form of low-pass filtering, but you can get much better results by implementing a simple IIR filter. Here is a nice online tool to calculate digital filters and generate source code for it: http://www.schwietering.com/jayduino/filtuino/ About the CtrlFET variable, just checking wat values it takes on during normal operation doesn't remove the requirement to limit the value when it can run away. Imagine a battery that cannot provide the current you want. Then the variable will keep increasing without ever reaching the requested current value. This value is used to control the PWM duty-cycle, but that is an 8 bit value (maybe 16 depending on the configuration). So when the value of CtrlFET becomes higher or overflows the PWM duty-cycle will reset etc. This will result in out of control measurements. In software design you must always think about variable overflow, it is like caring about current & voltage limiting in hardware design. Regards, Clemens

Antwoord

TechDoc 9 jaar geleden

Hi ! What do you mean by "software filtering " ? Is it a calculation of an average value on 10 (or more) values ? Or something else ? Especially on new development, I check all parameters and values, and it is the case for the variable CtrlFET: this value is very important for this software (The CtrlFET values grows from 0 to 130 (step = 1 ) at the very beginning of the test , and then varies from 120 to 125) , the serial monitor is very usefull ! Regards

Antwoord

TechDoc 9 jaar geleden

The CtrlFET values grows from 0 to 130 (step = 1 ) at the test beginning , then varies from 120 to 125. It seems that 250 mA for the AAA batteries is too high : I set this current to 100 mA, and it seems to run correctly now. About the 62.5 mV and 2.5% of fulll range : I will try with an amplifier (about x 20 with a TL082 or something else in my stock), and I come back here. I want to build a "GO/NO GO to the garbage bin" set, no a laboratory set ! (Remember : a set easy to use for the friends and family!).

Antwoord

ClemensValens 9 jaar geleden

In your code there doesn't seem to be any filtering of the measured signals. The RC filter on the gate of the FET will not eliminate completely the ripple on the discharge current & voltage. Your calculations are based on a noisy signal which may explain incorrect results. You try to measure 62.5 mV on a 2.5 V scale, which is only 2.5% of full scale. Another problem may be the variable CtrlFET whose value is unbounded either way and can (and probably will) overlow, thus introducing uncontrolled discharge events. If you want a precise instrument, you also need precise software. Regards, Clemens

Antwoord

TechDoc 9 jaar geleden

Hi ! Oh yes, during the development, the ISR of Timer1 contained a command like this : digitalWrite(13, !digitalRead(13)); and I checked this pin on my oscilloscope. It is ok. I also check the voltage on the 4 x 1 Ohm resistors : the voltage is 0.062 Volts during discharge. It's the reasons why I need help ! One of my next project is to develop a battery loader . . . . because the commercial loaders are too unprecise. I think we all "kill" the batteries by bad loading conditions. Thanks to all !

Antwoord

Toon meer

6 Opmerking(en)