(If you don’t like reading technical information with detailed explanations, I suggest you exit this page now and go to a website where you can buy a ready-made battery replacement.)
I never have liked devices that are powered by rechargeable batteries. Sooner or later (usually sooner for me) the batteries will either lose capacity or short out, leaving me with a device that is useless unless I buy a new battery. Sometimes replacement batteries cost as much as the device did originally! As a result, I avoid battery operated tools, but I am stuck when it comes to cordless phones, cell phones, portable computers, electric shavers and many other devices.
MTH Trains used a Nickel Cadmium or Nickel Metal Hydride 9V (nominal voltage) batteries to provide power to their PS-1 locomotives. Battery failure in these devices can lead to serious problems that can scramble the train’s microcontroller or cause corrosion due to leaking chemicals. In my opinion, these batteries are ticking time bombs.
To avoid these problems, I began in mid-2002 to design my own circuit to replace the battery with a module that contains ultracapacitors. The ultracapacitors have a very long operating life permitting thousands of recharging cycles.
Ultracapacitors are a relatively new form of capacitor that can have vastly more capacitance than conventional electrolytic capacitors. Unlike conventional capacitors which are rated in hundreds or thousands of microFarads, small ultracapacitors are available with ratings of a few Farads.
The Achilles heel of present day ultracapacitors is that the voltage rating is very low, coming in at 3 volts or less. To get the 9 or 10 volts, we will need to connect four ultracapacitors in series. For series connected capacitors we add the voltage ratings of the capacitors, but we must also divide the capacitance value by the number of capacitors. In one example below we use four 3.3 Farad capacitors in series, yielding an effective capacitance of 3.3 divided by 4 or .823 Farad. This is sufficient capacitance to keep a PS-1 engine alive long enough for a complete the 20 second shutdown cycle including the shutdown sounds.
Conventional batteries provide a nearly constant voltage as they discharge, but capacitors begin losing voltage as soon as discharging begins. If the current drain is constant, the voltage will drop linearly with time. For our application, we must be sure that the terminal voltage stays high enough at the end of the shutdown cycle to sustain proper operation.
The UP Turbine that I used for my testing has a 3-stage shutdown that draws 65 mA for the first 10 seconds, then it ramps down to 50 mA for 5 seconds, and finally down to 35 mA for the final 2 seconds. This totals up to just about one Coulomb of discharge. (A Coulomb is the number of electrons that will flow in one second at one ampere.)
My first unit was built with a pair of PC5-5 ultracapacitors from Maxwell Technologies in San Diego. These units are flat modules that can be stacked to fit into the recycled shell of a 9V battery. Later on I tried some cylindrical capacitors that were also small enough to fit into a battery shell. Both types work well.
The PC5-5 is actually a module containing two 3.6 Farad capacitors in series to yield a 5V rating at 1.8 Farad. Two of these modules in series ups the rating to 10V, but drops the capacitance in half again to .9 Farad. Discharging a .9 Farad capacitor 1 Coulomb will cause a voltage drop of 1.1 volt. This means that if we start at 10V with a fully charged unit, we will still be at 8.9 V at the end of the shutdown cycle. (MTH trains do not mind a voltage somewhat higher than the original 8.4V NiCad battery because the battery voltage is regulated down to 5V.)
A major problem with series strings of capacitors is that the capacitors don’t necessarily share the charging voltage evenly. To illustrate the point, we will assume a series string of 4 capacitors, with three of them being 1 F and the fourth being a weak unit with only .5 F capacitance. (This is an extreme variation, but it makes the math simple!)
When we pump a charge of 2 coulombs (equal to one ampere for 2 second) through the series string, each capacitor will get a voltage boost of Q/C (capacitance divided by charge.) The three 1 F capacitors will charge up 2 V, but the .5 F capacitor will charge up 4 V! This may overcharge the smaller capacitor, leading to degradation or failure. Real capacitors don’t vary quite that much in value, but their typical tolerance of +/-20% can easily provide a 40% spread.
There are a few methods for trying to maintain equal voltage across a series string of capacitors. Most of these schemes use resistor ladders across the capacitors to force the capacitors to share the voltage, but this doesn’t work very well. We would like to have a high resistance in the ladder to avoid wasting a lot of power in standby mode, but we also want low resistance so that the voltages will balance out quickly. Even a very low resistance ladder with multiple 10 ohms resistors will take 180 seconds to balance out. (Capacitor/resistor circuits take about 5 Time Constants to reach 99% of the final charging value. The Time Constant value for our example circuit is R x C, or 10 ohms x 3.6 F = 36 seconds.)
A better solution for equalizing the voltages is to use Zener diodes across each of the capacitors. If one of the capacitors starts to overcharge, the excess voltage is bypassed around the fully charged capacitor so that the other units in the string can continue charging to full voltage. If the fully charged unit is removed from the charger (as when the power to the PS-1 engine is turned off), the Zener will not bleed off the charge on the capacitors the way shunt resistors would.
I have chosen 1N5223B Zener diodes for my design. These diodes have a voltage drop of 2.7 V at 20 mA, which is about the maximum charging current at 10 volts. (Mouser has these parts for $.05 each.)
The schematic for my second battery replacement unit is shown below. Nichicon JUMT1335MPD capacitors found at http://www.mouser.com/catalog/catalogUSD/644/840.pdf are suitable. If you find less expensive devices elsewhere, please notify me.
There are a few more pertinent comments that are appropriate.
- Ultracapacitors, like most other electronics components, don’t like heat. If the capacitor’s ambient temperature is raised 10 degrees Celsius, the life of the capacitor is cut in half. Mount the unit away from heat sources.
- Ultracapacitors should not be used in applications that require rapid charge and/or discharge. The capacitors have more internal resistance than normal capacitors, resulting in internal heating when large currents flow frequently. This isn’t a problem in the PS-1 application.
- The internal resistance will cause an abrupt drop in voltage when switching from the charging state to the discharge state. The size of the step depends upon the amount of current being drawn.
- Allow time for the capacitor to charge fully when first turning on the track power. I wait 15-20 seconds.
- My measurements with the UP Turbine indicate that any track voltage above 6 volts will fully charge the module.