Trial & Error
Tonight I've been trying to generate enough energy out of motion to light an LED; my target is to use 'harvestable' energy from a bicycle ride to generate enough energy to light the back and/or front light. As you could have read in my previous post, I first had the plan to use vibration as the energy source to light the LED. Unfortunately, no piezo was supplied, and the energy density of piezo actuators is too low to be useful on a bike.
Instead, I tried to explore the possibilities of using electromechanic parts to harvest energy. I thought of winding a coil, and hanging a magnet in the middle that would move, suspended by springs. After pondering about this for a while, I suddenly realized that a loudspeaker is about that configuration, just with a coil suspended by a spring (the membrane)! I took two small loudspeakers, and tried several ways of hooking them up to the energy harvesting kit. No succes. I tried using rectifiers (to use a speaker safely at single ended harvesters), using a plain mains transformer to transform up the voltage (to use on the LTC3588), all to little or no avail; I was barely able to generate energy from tapping / hitting the speakers. I put enough mechanical energy in the speakers, but almost nothing came out electrically. Apparently the coils were too weakly coupled to the magnet field, and the frequencies generated too low to let the speaker generate electricity efficiently.
"It is typically not possible to combine high [acoustic, note from Victor] efficiency (especially at low frequencies) with compact enclosure size and adequate low frequency response. One can, for the most part, choose only two of the three parameters when designing a speaker system. So, for example, if extended low-frequency performance and small box size are important, one must accept low [acoustic] efficiency. This rule of thumb is sometimes called Hoffman's Iron Law (after J.A. Hoffman, the "H" in KLH)."
Then I thought of using a motor as generator. I had a motor from a robotic kit laying around, and thought it would make a nice generator; it had a gearbox, so a small movement on the output axle would rotate the 'generator' quickly. Unfortunately, it didn't quite work. I tried multiple configurations, with the LTC3108 (because it has a nice and low input impedance) and with the 3588 (because it can be used bipolar). Both needed a lot of physical effort to get the demo board running; let alone turn a LED on.
Several items on display; DC motor w. gearbox connected through transformer, speakers and rectifiers on top left
The reason that I tried using a DC motor with a transformer is that the LTC3588 starts operating at a voltage of 2.7V, and has a fairly high input impedance. The voltage coming from the motor was only a few mV to 1V, and did have enough AC (probably because of the brushes) to expect some output. again; lots of effort, no result.
Then I thought of some other motors I have lying around; stepper motors! These have lots of magnet poles, so short movements (and I'm thinking of using saddle displacement) will almost certainly energyze some coils. Also, these motors are made to have high torque at low rpm, which is kind of equal to what I want; small displacements/second with high torque generating a lot of energy!
I had a few steppers lying around and found one that worked very well with the LTC3588, without any other coil! Maybe a 1:2 coil might improve the performance a bit more, but this looks really good!
I realized that matching the input and output impedance of the generator circuit and its source is of vital importance; the motors could be completely shorted by the energy harvesting transfomers (input impedance <0.5Ohm), and thus loosing most of their energy in their own copper impedance; they did turn 'harder', but this didn't generate more energy. Very interesting to realize, as for me this put an end to looking for ever-higher transformer ratio's; those ratios alone are not enough to generate a higher voltage, the input impedance should match the source aswell for good efficiency. This may be first grade Electrical Engineering, but I still have to realize the beauty and truth of it by making errors
I found that during my first experiments the LED would flicker too much, even with the capacitor bank attached. I then tried using a supercap for storage. This is a capacitor of 1F (yes, 1F; no milli or micro or nano), with a bit of series resistance to give the power supply a chance to charge at all. That was clearly too much; charging the supercap took far too long (eventually helped with a bench supply), and when charged the LED would light for minutes; too much energy storage. In the video above you can see what it looks like with an 1mF tantalum capacitor placed over the 3V3 supply. I think it holds the charge well enough to keep the LED on.
Now I'll have to think of a way to get energy out of the bicycle. I was thinking about using the saddle motion as a way to generate the movement. Short movements are no problem, but I do need to have sufficient speed to get the input voltage high enough.... Mechanical stuff!
Edit; LTC3105 is sweet!
I wrote this article last night, and somehow it didn't get posted. Today I realized I didn't do justice to the LTC3105 by not giving it a look; I then found out it might be just as good as the LTC3588; it operates differently, and I still have to re-read the datasheet several times to get a good grasp of what is happening. The good news is: it works like a charm with the stepper motors, haven't been able to do a fair comparison.