Cost: Realistically, I see this thing climbing into the $15-30 range for materials. That’s more than I’d pay to buy a puck from ThinkGeek, but not more than I’d pay to construct one myself that I helped design. Anyone else?
Battery: Seems like we’d get a lot better power density (translating to smaller size and longer runtimes) if we used an RC-style LiPo battery, I’m thinking about the size of a matchbook or so. This’ll be more expensive; but I’m concerned that fitting in 2-3 AAA or AA cells will make the puck bulky that it’s not cool anymore. How do you feel about having to buy a special battery for it?
Charge circuit: I can find a charge-control chip and use the datasheet’s sample implementation as well as the next guy; but that’s really just a starting point, and I understand there are often subtleties involved in getting the best real-world performance. Does anyone have a lot of charging circuit design experience, or know someone who does, who’d be willing to design that portion and contribute it to an open-source hardware, community-developed project? We might want to look at LadyAda’s WaveBubble or the Chumby design for reference.
Voltage-boost circuit: Depending on what battery, microcontroller, and LEDs are used, we may need to boost the battery voltage. Same questions as on the charge circuit.
Surface-mount components: I don’t think there’s room to use through-hole technology. I’m envisioning at least the microcontroller, LED driver, and optional real-time clock chips being SMT. LEDs need to be through-hole, in my opinion, both for rated brightness and so they can be aimed slightly away from the center for better light dispersion. Will a few SMT components be enough to keep anyone from building this who would otherwise be interested?
Enclosure: My first thought was pouring this thing in resin, but I’d really like to be able to take it apart and service / enhance it. I’m now leaning toward a two-piece milled enclosure (probably lexan) that would sandwich the PCB edges in rubber for shock control and hold together with recessed screws underneath. Add an appropriate O-ring and the whole thing could probably be made watertight — which in my book counts as another feature. 
Thoughts?
One idea for the resin potted unit is that you could use induction charging. That way no connector or external wires are needed. Add a coil of wire inside the potted puck and one in a charging base.
David — is there anything about induction charging that would work with potted resin and wouldn’t work with waterproofed lexan?
David: do you know of any good references for circuit designs for induction charging? It’s something I’d love to play with.
Resin or Lexan is fine. A non-magnetic metal box would work as well although you could have heating hysteresis and that would not be efficient.
If you place the puck on top of the charging coil so that both the external and internal coils were close to each other, you should get good coupling. The primary concern will be the distance between the coils. Magnetic energy drops of very fast so the coils need to be close.
There would be some interesting experiments to try. I am not an expert but I do know that this will work. I have seen similar technology.
If you really want to try this I can do some research? I have access to an RF engineer or two who will know what we need.
If the puck is round, perhaps the charging coil can be wrapped around the outside. Then the puck could be placed inside the charging coil.
I did a quick search on Inductive charging and found several links to expore.
Of course Wikipedia has tons.. http://en.wikipedia.org/wiki/Inductive_charging
Here is an article that had me lauging so hard I amost passed out. Yes it’s about inductive charinging.. A mouse that charges through it’s pad. Why do funny, go see for yourself..
http://www.afrotechmods.com/cheap/arnoldpad/arnold.htm
David, you offered to check with some RF engineers on inductive charging techniques if I was interested; and if you can really do that, I’d like to take you up on it.
For the past couple of weeks, I’ve carried the puck idea around in my head and thought about times I’d use it. In doing so, I’ve come up with two decisions of much higher importance to me than I had guessed at first: It needs to have a magnetic base so it can be stuck to things, and it needs to be water-resistant enough to use underwater for a few hours.
Water resistance will be far easier to accomplish if the puck doesn’t have a charging port penetrating its case, so I think inductive charging (besides being just plain cool) is a fantastic idea. Would you mind looking into that?
As a fallback plan, I’ve been thinking about the pager “coasters” you get when there’s a waiting line at chain restaurants. They have a couple of contacts on the bottom to charge when placed on a base (and corresponding pins on top to charge the next one up in a stack, but that’s not relevant here). We could probably waterproof metal pads recessed into the bottom of the case, and diode-protect the system from discharging through the pads in (impure) water.
But inductive charging would be nicer, if it can be made practicable.
http://forum.chumby.com/viewtopic.php?id=13
Here is a link to the charging circuit on the Chumby. I didn’t realize that it was designed for a rechargeable battery. I wonder why it doesn’t ship with one.
Joe, thanks for the Chumby link. If I read the forum correctly, Chumby actually doesn’t have a battery charger built in — my mistake.
The circuit shown is a hacker’s contribution, and I’d want to look it over a little more closely before using it than if it were Bunnie’s design.
One consideration in the puck is that I’d like to be able to plug it into USB and have it both drive the lights and charge the battery. 16 LEDs at 25mA is 400mA total, and some LEDs I just got can probably take 50mA, so 800mA total.
I understand that USB can supply 100mA or 500mA, which means we couldn’t run 50mA per LED off USB power. But it also complicates (I think) the charging design, to try to maintain a constant current to the battery whilst the battery is being used at the same time.
The bq24030 from Ti is perfect for this kind of charging application, but is surface mount. It is designed for in-line charging applications where the device is normally connected to mains power, and handles the balance between circuit demand and current delivery to the battery charge circuit seamlessly. When used as an off line charger it does lose a little efficiency running in stand alone, but when connected to the charging source it will balance power delivery very nicely. It also accepts a very wide range of charging input voltage, making the design of an inductive coupler much simpler (no need for regulation etc, just a zener to limit the voltage to 15V). Lithium poly foil packs are quite cheap if you are buying for OEM use. I pay under $2.50 for 280mAh packs with over charge protection circuitry built in. Less for large orders. Surface mount is cheap if you’re making a few, but possibly too hard for amateurs. Most LED development is in surface mount so the very brightest LEDs are available in SMD format