Second system effect
November 25, 2009
I’m pretty happy with my bike light, but gosh darn it, it could be better!
Various features that I want, but demonstrably don’t need:
- Able to tolerate higher supply voltages.
The Buckpuck tops out at 32 volts, and the desired voltage for best efficiency is probably closer to 2 times mph — 40v at 20, 50v at 25, and 70 at 35 (which I have surpassed on some hills). Obviously, eventually I am more concerned about the survival of the electronics than I am about perfect efficiency, but I’d like to be moving faster than 25mph before that matters.
Furthermore, at moderate speeds, voltages above the peak efficiency level allow for lower drag at reduced power. This might be good for daytime running lights, for example.
There’s at least two ways to get this; one is to disable the voltage doubler once an adequate voltage has been reached, and the other is to choose higher-rated parts.
A real one — everything needs to be on at least half power, meaning 2.5 watts. In terms of energy storage, 150 joules per minute, so an ultracapacitor is not a solution. A capacitor holds 1/2 C V2 joules, so a 1-farad ultracapacitor run down to 10 volts (50J) must start out at 20V (200J) — such devices seem not to exist, and certainly not at a sensible price.
- Charge-back into the battery
Given that a standlight needs a battery, it would be nice to run the battery as long as possible, meaning to put some charge back into it.
Getting a battery to 100% charge is not a job for amateurs, especially on a bicycle (varying temperatures, varying voltages), but perhaps we can do an approximate job.
- USB charger
A USB charger needs to supply 500milliAmps at 5volts — 2.5 watts. This is possible either if the lights are low or at good speed. So why not, it would be fun.
- Daytime Running Lights features
These include, mostly, the ability to blink and do other things to call attention to the bicycle. Reduced power, compatible with USB and/or battery charging, is also a good thing.
- Semi-open, so other crazy people can do more than my already ridiculous set of plans. So, attempt to follow standards set by the Arduino guys, make provision for a crystal good enough to run USB, be sure that one of the connectors could support not just USB charging, but USB communication. It’s not an exact match for the Arduino pinouts, because things just didn’t fit, and because it was important to get the two AC lines to the comparator inputs.
It’s still at the do-I-dare-build-this stage. Not only is it a lot of parts, there’s some surface-mount bits, and SOFTWARE. Said software has the option of setting some (non-essential!) parts on fire, and as a control system, there’s ample room for funny oscillations. However, the big-ass capacitor (4.7 milliFarads) takes almost 5 milliseconds to charge by 1 volt, given a 1 amp input, so there is some wiggle room.
- Two ac inputs, to measure wheel rotation
- System voltage
- Rectified AC current
- Four analog lines, for controls
- Four digital lines, for controls/USB
- Programming header, and its lines
- Three enable/disable lines to LED power supply, can be PWM’d
- One voltage-doubler disable line (ON by default, so that system powers up fast at low speeds)
- One battery-enable, for charging/discharging the battery. Software has to know what the charging voltage is; current to battery can be estimated somewhat by strobing the lights and battery and monitoring the change in the system voltage; as noted above, a goal of 100% charge is not a good plan.
- Off, is off. Need to roll the bike to turn it on.
- Slow. Use batteries to keep the lights on. Boost on.
- Medium. Batteries off, boost on.
- Daytime running lights. Lights on steady-dim, pulsing bright.
- Hi/low beams. Of the three LED power supplies, one is high beam, one is low beam, one is rear. Normally the lights are high or low beam.
- USB charging. USB charging will put a load on the supply, which should cause some dimming in order to stay within the power limits.
- Fast. Riding fast produces excess power, which means that it must be dumped somewhere. So, both high and low beams will be activated, as much as necessary to cap the voltage. (Need to put a zener clamp on this somewhere…). One possibility is to dump power into the batteries; for plausible descents, for plausible amounts of battery, the excess energy would “fit” in the batteries, and could then be deliberately run down by powering the lights on full after the descent for a corresponding amount of time.
The weird slots in the PCB planes are intended (not that I know this will work) to provide some noise isolation for the microprocessor and in particular, for its analog side. The switching supplies for the LEDs each put out at least 350mA, switch in the 100s of kilohertz, and could be a tad noisy. The slots under the inductor on the right are intended to reduce eddy currents; the inductor wires are perpendicular to the cuts.
(The PDF images will display in some browsers, or can be clicked for download in others. I need to make them be SVG. Aha, trying to get that done right, that is a problem — a problem caused by misconfigured web servers).