One of the coolest parts of working on the Yahoo Purple Pedals project was figuring out how we could get the bikes as far off the grid as possible. Once we decided on using the N95 as the brain for the bikes, we had to figure out how much juice it takes to run the code, how often the code was going to run, and how we could prevent our users from having to plug-in the bike everyday like a cell phone. After thinking through and testing out a few different ideas, we settled on solar power as the simple, inexpensive, green, and aesthetically pleasing solution. In the following walk-through I am going to explain the design process and how we got from the idea of using solar to creating a solar replenishing backup power system that can run off the grid for weeks on end.
1. Hack N95 charger
The first step was figuring out how the phone charged. Power can be a tricky business and the phones weren’t cheap so we didn’t want to accidentally fry one in testing. Luckily, this step turned out to be a lot easier than we expected. The factory charger was a regular 5V AC adapter which output about 7.2 volts DC at 880 milliamps. The next step was to figure out the minimum and the maximum power levels the N95 could take in at the charge port and still charge correctly. Our idea was to cut the tip off the power adapter, test the polarity of the wires inside, mate it to a variable power supply, and see what happened to the N95 as we cranked up and down the juice. The first challenge came when we went to strip the wires on our newly cut adapter wire. The N95 charger uses a flat cable to connect from the tip to the body, which means you can’t use regular wire strippers to get at the copper inside. We devised a technique of splitting the wire in half with an exacto knife to create two independent wires which were then possible to strip with regular tools.
We discovered that inside the flat cable are two conductors, one bare and one coated in red enamel. On the first hacked adapter the red wire was ground, but as we later found out it was completely random which wire color was mapped to which power source inside the charger. Once we got at the wire and got our polarity figured out we figured it would be easy enough to just clip onto the bench supply and start testing. Wrong. Even with the bench supply set at 7.5 volts and current limiting set to 880 milliamps the phone refused to make its little beep and display the charging animation. On a hunch we tried putting a resistor between the charger input and the bench output, and somehow, even though the bench supply was supposedly already current limiting, for some reason this worked. Hurray! We got the phone charging on anything from about 4.1 to 10.5 volts DC.
2. Get solar panels
Now that we knew the acceptable range for charging the phone we had to find some solar panels that would fall within this range, fall within our budget, and look cool on a bike. We decided on the “large solar cell” available from sparkfun.com (PRT-07840). For $35 it outputs over 8 volts and 300 milliamps in direct sunlight, beating all others that we found in its price class. To top it off they look cool and come all sealed and soldered up with long leads. Later water tests proved that they were water resistant which really “sealed” the deal for us.
3. Connect and test
We ordered one to try it out, expecting that we would have to build a circuit between the panel and the phone to actually get it to charge. With low expectations, we soldered the already hacked charger tip to the solar panel (double checking the polarity of course) and took a trip to the roof on a sunny day. We found, much to our surprise and delight, that the phone charged directly! Further tests showed that the solar panel would charge the phone in very low light conditions and even indoors under an incandescent bulb. This was very exciting.
4. Measurements
Now we wanted to know how long it would take a single solar panel to charge a depleted N95 on a sunny day. To figure this out we needed to know how much energy we were getting from the solar panels over an extended period of time. We found a neat little product called the “Watts up” which tracks current and voltage consumption and displays both instantaneous values as well as a summation of the total power used over time. Perfect. We hooked the Watts up in between the solar panel and the charge port on the phone and let it sit for an hour in full sunlight on the roof. We got 219 milliamps of current over the course of an hour with a 310 milliamp peak and a 1.2 Watt peak. Since the N95 battery is 1.2 Amps this means we would need 1.2A / .219Ah = 5.479 hours to fully charge a completely depleted phone in direct sunlight. Not bad.
5. Backup power
We could have stopped there, connected a solar panel directly to the phone and, as the phone ran, the panel could top off its charge. But we weren’t quite satisfied with that. For one thing, the phone might not run long enough if the weather was cloudy, or if the bike was parked in the shade, and we didn’t want to make our riders have to be constantly thinking about charging their bikes and parking or riding in the sun. We were also worried that after a few cloudy days in a row the phone might die, and when the sun came back it might not be strong enough to charge the battery up from a completely dead state. Another concern was that the phone and panel would get stuck in an infinite loop where the N95 would have just enough charge to turn on, start itself up, and then exhaust its small charge in that process and die again. In this scenario, even with the bike in full sunlight for hours, it might never completely start itself up! We decided we really needed a back up power system–a bigger battery that we could charge using a solar panel (or a few solar panels) that would in turn charge the N95 in small doses, as needed. This battery would be big enough that we were bound to get some sunny days before it ran out of juice, and if by some fluke we didn’t–we could include a simple charger and plug on the bike so that one overnight charge would last for a couple weeks rather than the couple days the N95 provided on its own. Easy right? Well, sort of.
As we found out, different types of batteries require different charging methods. Originally, we thought we could just buy a 10 amp hour lithium ion battery and use that to feed the N95, but as it turns out they are not only ridiculously expensive but also require specific charging circuitry which we couldn’t find for batteries large enough to suit our needs. Time was running out and we needed an elegant solution.
6. The final circuit
After extensive battery research, we decided on an old standby, friend of cars and boats alike, the trusty sealed lead acid battery. We picked a 6 volt, 12 Amp hour battery from batterymart.com for $17. Yes, they are big, and bulky, and heavy. Yes, they take a long time to charge. But they are a tenth of the price of lithium ion and they’re easy to charge. We could have hooked a single solar panel up straight to the battery with nothing but a diode indefinitely. But we wanted to do more, faster. We wanted three solar panels to give us triple the current and reduce our charging time by 2/3. However, three solar panels in direct sunlight could give us as much as 900 milliamps, more then the battery’s trickle charge rating. (A trickle charge rating is how much current you can supply the battery constantly without monitoring it or worrying about overcharging). This meant that we needed a simple current-limiting charging circuit that would sense when the battery was getting to capacity and gradually reduce the amount of current supplied. The circuit we used was based on an old classic, the LM317 adjustable voltage regulator. The design we needed was right in the datasheet, a “current limited 6V charger”. The circuit required just 4 resistors, a transistor and a capacitor in addition to the LM317 and it worked like a charm right off the breadboard. The circuit was originally designed for charging a lead acid battery from a power supply, and the only modification we made was to reduce the capacitor across Vin by an order of magnitude since we were coming from solar panels rather than a transformer and didn’t have any ripple to worry about.
[JR holds up the l2 pound lead acid battery used in the project]
This solar charging circuit makes up the majority of our final circuit. We added a switch so the user could choose between solar charging and wall charging, we added a couple diodes to keep the power flowing the right way, a fuse just in case, and a zener diode with a threshold of 4.1V connected to an LED to indicate when the battery is run down past the point where it will charge the N95.
And that’s it! the circuit is real simple once all the annoyances are out of the way, and hopefully will be of use to others who are interested in turning solar energy into cell energy. Next on the list is to see if the same circuit will charge my Samsung and build a general purpose “anything in” to “anything out” charging device to take wind, solar, a dynamo–you name it–and make it whatever I need for the device of the day…
[The whole system]
