New Page 1

Capacity testing cells.

Under construction .............

Introduction

Most projects start with a single thought. A few years ago I was quite shocked to find that my digital camera actually took 2 amps from two AA cells ... which explained why they never lasted very long !. I thought part of the problem was the old Radio Shack Ni-Cad charger I had been using with the more modern Ni-MH cells, coupled with the use of relatively low capacity cells. So when a leading supermarket has a special offer on an all singing and dancing multi-charger I bought one. I also bought two dozen AA Ni-MH cells plus some AAA and P3's. The AA's are 2500mA/h .... so I was somewhat disappointed to find that they were failing almost as quickly as my old ones. The question was why ?. The first possible reason is that Ni-MH cells do not reach full capacity until after several cycles of use. Mine have had plenty of use so it couldn't be that. I then started thinking about the internal resistance of each cell and wondered how uniform this was across a batch. The truth was I had no way of measuring the capacity of a cell, so I had no idea what was happening.

Ni-cad Cell capacity.

Yes I know that the useable capacity of a re-chargeable cell is usually printed on the can. By 'useable capacity' we mean that it is the amount of energy we can take out of the cell before the cell voltage drops to one volt. At that point there is still energy left in the cell and you may ask ... "so what ?".

Ni-cad cells discharge themselves with time. New ones may come completely discharged. In most cases the only way to know how much charge is in a ni-cad cell is to completely discharge it at a safe discharge rate and then short it out for awhile. Not only will it do no harm to a ni-cad cell but it can do it a lot of good. Let us assume that we leave the cell shorted out for a year .... what would you expect the cell voltage to be when you removed the short circuit connection ?. If you wait a few minutes you will find it rises up to close to a volt. It may surprise you even more that some Ni-cad chargers are designed not to begin a charge until the shorted cell voltage has risen to such values.

Now that really fast chargers are available it makes good sense to store ni-cad cells shorted out and only charge them up just before use. The question is how much charge do we have to put into an empty 2500 mA/h cell to get it fully charged ?. Remember that the 2500 mA/H is only the useable capacity. If we are charging a completely empty cell then we also have to put in the un-useable amount of charge. Usually for a ni-cad this is about 40% more, or in other words when charging a completely empty cell we have to charge it to 140% of it's rated capacity. Most modern 'smart' chargers will automatically do this for you. Again, always follow the manufacturers instructions.

Safe discharge rates. Discharge the cell using the recommended rate ( 5 - 10 hours). When the cell voltage has dropped to 0.5v the cell can be shorted out with a one ohm resistor.

Constant Current discharge

Capacity is the product of rate and time. If we discharge at a constant rate, capacity measurement becomes a fairly simple matter. Constant current in amps x Time in Hours = Capacity in A/h. Each cell under test must have it's own constant current discharge circuit. A typical example of a constant current discharge circuit can be found below.

A BUZ11 has been used to discharge the cell via one or both of the 1 ohm current sense resistors. With the link in the parallel resistance of the two resistors is 0.5 ohms. The voltage developed across them, at a discharge rate of 500 mA is 0.25 volts !. To make this possible we have to use an operational amp that can handle rail to rail input voltages and the CA3140 is capable of working down to 0.25 volts BELOW the ground rail. Here it is being used as a voltage comparator against the reference voltage set by the 2K potentiometer. This 18 turn pot effectively sets the the value of the constant current. For an experiment I needed to be able to cap test at two different rates and the link allows the circuit to also be set up to give a stabilised output of 250 mA.

The other link selects where we get our reference voltage from. As shown it is connected to the + 15 volts supply line. This may sound a little high but the BUZ11 requires about + 9 volts on the gate to fully switch it on. If I had used a TTL compatible MOSFET then I could probably have used 9 volts instead. As usual .... what gets into a circuit depends upon what you have on the shelf. This link has two positions and the other, selects an input from a PIC and a simple Digital to analogue circuit. By varying the PWM duty cycle in software one can remotely vary the output of the constant current circuit.

The PWM output from a PIC is a five volts square wave with a duty cycle that can be varied in software between 0 - 100%. When fed into the above circuit it results in the output being a DC voltage that can be varied between 0 - 5 volts. The voltage across the capacitor = (duty cycle/255) * 5. For example if in software we set the duty cycle to (100/255), then the output voltage will be 1.96 volts. The op amp voltage follower is included as a buffer to stop subsequent circuits loading the capacitor and dragging it's voltage down.

Control.

During a cap test, the voltage of each cell must be monitored, to determine the time it takes for the cell under test to drop to the 'end of life' voltage ... typically one volt. The 'easy' way of doing this is to use a PIC microprocessor. I decided to use my general purpose PIC microcontroller board that takes either a 18F452 or the 16F877. This board also contains a DS1307 real time clock, an RS232 output and one megabyte of flash eprom for data logging purposes. The 18F452 and the 16F877 both have 8 analogue inputs, so theoretically we could use up to eight constant current discharge boards and cap test eight cells at a time. The RS232 facility allows us to port out the data in real time to a PC, that could for example, plot each cells discharge curve.

My general purpose PIC micro-controller board with RTC and Flash eprom.

Software.

My version tests four cells at a time. That is the way I buy them and that is the way I use them. My PIC board uses AN.0 on pin 2, to monitor the raw supply voltage. I did it this way because many of my applications are battery powered. I therefore have used AN. 1 - 4 to monitor the cell voltages and print these up on a 40 x 2 LCD that is connected to port B. When a cell is loaded onto the tester, the software senses this and notes the time of the Real Time Clock. At the end of the cap test, when any cell reaches 1.0 volts , the software checks the time on the RTC and calculates the cells capacity as a percentage. This Percentage replaces the cells voltage on the LCD and the cell continues to discharge until it is completely empty. At this stage we can either store the cell shorted out .... or recharge the cell at 140% ready for re-use.

The process of cap testing .... discharging a Ni-cad to zero volts and subsequent recharging at 140% is called "Deep Cycling". Cells that have previously failed the 80% cap test can be rejuvinated by a series of deep cycles. Taking this to the extreme is a process called "Reflex Charging" where pulses of current are fed to the cell and inbetween each charging pulse there is a short discharge. The frequency of the pulse train is typically 50 - 60 Hz. An interesting side effect is that when a cell is reflex charged at the one hour rate ... it stays cold to the touch !. A trial was carried out by a government department using 100 cells that had previously failed to reach 80% on cap test. After one reflex deep cycle, over 95% of these cells passed the cap test at 100% !. It should be noted that these were vented aircraft battery cells.

To be continued ........