A number of years ago I got 4 batteries that were used in alarm systems, and preventively replaced. All these batteries were identical: made by Yuasa model number REW45-12. The capacity was labeled as 12V 45W/cell (10 min).
I've always kept these batteries charged with a Deca SM1208 charger. The main use was when going in the field with radios, I'd use these batteries as a power source. A single battery would typically last me for an entire afternoon. The battery that I labelled as "Accu 3" has always been the weaker one. For a while I've been planning to measure the capacity of these batteries, but I never got to it under the 2015 summer vacation.
Last week I've build myself a constant current load using an LM358 opamp and a IRLZ44N Mosfet. Using this load to discharge my batteries, I measured the voltage of the batteries using an analog input from a Arduino Mega board that was lingering around in the home-lab. A picture of the entire setup is shown in the picture below.
The program running on the Arduino Mega was quick 'n dirty. The output caught on the PC using the default serial port of the Arduino:
int analogPin = 0;
int value = 0;
value = analogRead(analogPin);
voltage = ((float) value / 1023.0) * 15.0;
Measuring the voltage over the loaded battery as it was discharged, produced the following graph for "Accu 1". The picture clearly shows that measured voltages fluctuate around 2V for any point in time during the measurements. Some further analysis showed that this was caused by an oscillation in the Constant Current Load. Once this issue was fixed the next measurements where much cleaner.
These pictures make clear that Batteries 1 and 2 can be used to get 1A (used load) for roughly 2 hours, but Battery 3 is completely discharged in just over 1 hour. For all batteries, it has become clear that they are getting close to the end of their usable life.