Update this project:
[Goldmine Electronics](http://goldmind-elec.com) has an interesting deal right now. $20 gets you a [Boostcap](http://www.goldmine-elec-products.com/prodinfo.asp?number=G17960), and occasionally goes on sale for half off. These supercaps were apparently pulled from large banks used for regenerative braking. So they've got some road dust caked in, use an old toothbrush to clean it up. Spec wise, 2600F is nothing to sneeze at. Previously, the largest capacitor I have ever seen was 10F. The desire to own a Boostcap just for the sake of experiencing what so much capacitance looks and feels like is pretty strong. And these supercaps feel beefy! The pictures do not do them justice. They weigh a pound. They are about the size and shape of a 16 oz energy drink can. Designed for sinking hundreds of amps, two Frankenstein-monster-esque bolts stick out of either side for attaching to bus bars. Like all supercaps, the maximum voltage is rather low. Just 2.5V. One can charge caps past their max. Doing so makes supercaps leak like a sieve and heat up. And take some lifespan-shortening damage. But entire products have been based around overcharging supercaps. Remember those tiny ZipZap remote control cars, briefly popular around five years ago? They used a small 2.5F supercap (with a 2.5V max) charged up to 3 volts. But the Boostcap is much larger and would not dissipate heat very well at all. So don't overcharge it. Now, 2.5V is not really enough to do anything with directly. And capacitors have the bad habit of dropping off voltage very quickly. Some kind of voltage boost regulator is needed. This appears as a load with a constant wattage, or variable resistance dropping proportionately with the supply voltage. At the bottom of the page you'll find source for a simple constant wattage supercap simulator. The parameters available to the sim are * `cap` the capacitace in farads * `watt` the wattage of the device * `vstart` the maximum voltage of the cap when fully charged * `vstop` the point at which the boost regulator stops functioning * `effi` the regulator efficiency The particular supercap determines `cap` and `vstart`. For the Boostcap, these are 2600 and 2.5 respectively. The limits of the regulator determines `vstop` and `effi`. Stopping at 1.2 volts and an efficiency of 80% are fairly typical for this type of boost converter. My laptop uses 6.7 watts with the screen and wifi on. Crunching the numbers returns an estimated run time of 747 seconds or just over 12.5 minutes. Not very useful. What about a theoretical ARM laptop with a transflective/eink display, consuming only one watt? Simulation says 83 minutes. Still kind of short, and 90 minutes is the absolute minimum for most laptop manufacturers. So this is almost feasible. How about 0.1 watts? That will run for 13 hours, which is useful. I've got a few devices floating around which use 100mW and will be worth hooking up. Most are much smaller than the Boostcap, but maybe the comedic appearance will compensate for the additional bulk. With minor variation the same routine can be used to estimate recharge time. One of the big features of capacitors over batteries is "instant" charging. Set the wattage to 500 (reasonable for a line-powered device), Vstop to 0 and efficiency to 1.0. Vstart stays at 2.5. Charge/discharge curves are symmetric with respect to time, so charging at 500 watts takes the same time as discharging. The charge time at 500 watts? 17 seconds. Not quite instant, but pretty impressive. Here is the code. #! /usr/bin/env python from math import expm1 cap = 2600.0 # farads watt = 7.0 # watts of device vstart = 2.5 # capacitor Vmax vstop = 1.2 # regulator Vmin effi = .80 # regulator efficiency assert vstart > vstop v = vstart t = 0 delta_t = 0.1 # seconds while v > vstop: r = v**2 * effi / watt rc = cap * r v += v * expm1(-delta_t / rc) t += delta_t print(int(round(t)), 'seconds') The biggest flaw is that it simulates timesteps. The alternative would be to solve the integral. But for the purpose of ballparking run time, this is fine. Someone asked me how much total power was in one of these caps. Assuming a 100% efficient conversion, capable of using all the charge down to zero volts. The sim says 2.6 Wh, which works out to a specific energy of 5.7 Wh/kg. Coming up with actual applications is being tricky, as anything low-current could be just as easily powered by two NiMH AA batteries for a fraction of the size and cost.