[dorkbotdc-blabber] Re: LED Cube Workshop

[R. Mark Adams, Ph.D.]

I like the bipolar LED idea!  If you want to buy a set to try out, I would
be happy to go in on the buy so I can hack around too.   I will do the
holes at the 0.1 pin-spacing multiple, too.

Perhaps we could have the 'cube builders' add the driver transistors at
the same time as they build the cube.  We would need to amend the plans
and test the build in advance.

Also- in order to enable the "mega-cube" how about if each cube is wired
to shift-register(s) and expose a clock and data line- that way you could
either (1) daisy-chain them, or (2) have each cube be driven off of one
pin of the microcontroller.  I have done something similar for driving LED
matrix displays from my embedded computers in the past.

THanks!
-Mark


Tim Slagle wrote:
> Sign me up for a kit, but I would probably want it before the
> meeting so I can put it together in advance.  If other people
> want to do this it will complicate the logistics of a volume
> order I'm afraid.
>
> Also, the circuit that drives the cube is a hack, and probably
> wouldn't work too well to drive multiple cubes without
> additional parts.  The cube is driven as three multiplexed
> planes of 3x3=9 LEDs.  The LEDs in a plane are driven directly
> from the AVR uC pins, which is okay, since they can drive about
> 25mA.  But then the planes are addressed by pulling the current
> from the whole 9 LEDs into a single AVR uC pin, which is a lot
> more current than the AVR can sink.  So, the cubes will already
> be operating at ~1/27th the brightness of a directly driven LED,
> and adding more cubes in parallel would only make them dimmer.
>
> One change that could be easily done to increase brightness is
> use two AVR pins in parallel to sink current from the planes.  I
> think there are enough extra I/O pins to do this for the 3x3x3
> cube.  The firmware would then have to be changed to set two
> bits instead of one to address the planes, but this should not
> be a big deal.
>
> A better approach would be to drive the planes through a
> transistor that could sink more current.  This would require
> soldering three ~1Kohm resistors and 2N3904 or 2N2222 NPN
> transistors to the plane drive I/O pins, and inverting the bit
> values of the pins in the firmware.
>
> I would recommend making the LED cubes on a piece of perf board
> if possible, so there would be room to wire up the drive
> transistors on the board and you wouldn't have to solder them
> directly to the POV kit.  But then, I think like an engineer :-)
>
> Mark, if you make the LED grid assembly fixtures, space the LEDs
> at some multiple of 0.1" so they will fit on a perfboard.  They
> need to be less than 1" apart so I'd say make them 0.7" or so.
>
> One variation I've been thinking about that would be cool (but
> require more firmware changes) would be to use bi-color LEDs.
> As long as the drive voltage is kept around 3V and two-pin
> reverse-connected LEDs are used, they could just be wired up
> like the single-color cube.  Then in the software, the I/O pins
> for inactive planes are set to be high-impedance inputs so no
> current will flow, and the active plane is set high to address
> one color and low to address the other color.  So the matrix
> would be scanned like six planes of 3x3 with only one color on
> at a time.  The plane bit values for the high-level addressed
> color would have to be inverted since they are now active low.
>
> I will try to buy some bipolar bi-color LEDs for my own use but
> others may want them as well...
> --Tim
>
>
>
>
>
> --- adam koeppel <koeppel at gmail.com> wrote:
>
>> Yeah, if this works out well, it could be an installation
>> worthy
>> piece.  I definitely think the working towards a massive
>> assembly
>> approach is a good idea, and it will solve the shortage of POV
>> kit
>> problem.  We should probably do a little design work ahead of
>> time to
>> make sure a single POV kit can provide the power we need,
>> although I
>> don't imagine it will be a problem.  I'd just hate for it not
>> to work
>> because of a power shortage.  What do you guys think?
>>
>>
>>
>> On 12/6/07, Alberto Gaitán <alberto.gaitan at gmail.com> wrote:
>> > Yes! I like the idea of divving up the work factory-style!
>> ...some folks
>> > building cubes, and others POV thingies. And the idea for a
>> > mega-LED-cube is genius enough to stop traffic for hours in
>> Boston!
>> >
>> > A
>> >
>> >
>> >
>> > On 12/6/07 11:33 AM, R. Mark Adams, Ph.D. wrote:
>> > > I think this sounds great- if we follow the 'sewing
>> circle' approach,
>> > > people are certain to feel like they had an enjoyable
>> time, as well as
>> > > (potentially) making something interesting.  I would
>> suggest that we buy
>> > > and build at least a couple of POV kits in advance, so
>> people who finish
>> > > their cubes could hook them up and get so see what they
>> look like working.
>> > >  in fact, if we encourage the group to work together
>> collectively, we
>> > > could set it up so that folks are each working on parts
>> that can all come
>> > > together in the end to have something working.
>> mega-LED-cube, anyone? :-)
>> > >
>> > > I am happy to volunteer to buy a couple of kits/LEDs to do
>> some
>> > > pre-building if people would like.  I can also knock off
>> some wooden jigs
>> > > to aid in the LED cube construction. (I can use the robot
>> I showed last
>> > > time to make them- cool!)
>> > >
>> > > I also have some tool kits and soldering irons that
>> tool-less folks can
>> > > use at the meeting if they want.
>> > >
>> > > Thanks!
>> > > Mark
>> > >
>> > > Gareth Branwyn wrote:
>> > >> Hey Everybody,
>> > >>
>> > >> As I'm sure you all know, we need to step on it to get
>> the LED Cube
>> > >> workshop in gear.
>> > >>
>> > >> I talked to Dan Woods at the Maker Store and it's not
>> firm yet, but
>> > >> it sounds like they'lll sell us the MiniPOV kits for cost
>> in exchange
>> > >> for a "Sponsored by MAKE" and a link on their website, in
>> the
>> > >> announcements, etc. The problem is that, given the
>> season, the kits
>> > >> are in short supply. He said supply should free up after
>> Xmas, so
>> > >> hopefully they could have them to us in early January.
>> > >>
>> > >> Tim Slagle said that he could get the LEDs themselves.
>> Katie Bechtold
>> > >> and Mark Adams have also volunteered to help. (all copied
>> in here)
>> > >>
>> > >> Alberto said that he'll draft the announcement that
>> Koshland needs to
>> > >> see, sooner than later. I'll also help in editing that.
>> > >>
>> > >> So, we need to figure out ASAP how we're going to run
>> this thing.
>> > >> Personally, I'm not that concerned about whether the
>> cubes are
>> > >> finished by the end of the workshop or not. To me, these
>> things are
>> > >> more like a sewing circle than anything else, a chance to
>> socialize,
>> > >> geek out, learn a skill and a few tips from those more
>> experienced,
>> > >> etc. But others might have a different take.
>> > >>
>> > >> Given the "shortage" of POV kits, we may have trouble
>> getting a lot,
>> > >> if say, 25 or more people each wanted a kit. One thing we
>> could do is
>> > >> also approach Lady Ada, the maker, and see if she'd give
>> us a
>> > >> discount. Even if she didn't, we could buy them full
>> price from her
>> > >> (at $17.50 ea) and spread out the cost over all the kits
>> we offered
>> > >> to people.
>> > >>
>> > >> So, we need to get a show of hands ASAP for how many
>> people plan to
>> > >> purchase a kit. And for one thing, we should insist that
>> you have to
>> > >> be participating in the workshop to get the kit at the
>> discounted price.
>> > >>
>> > >> Gareth
>> > >>
>> > >>
>> > >
>> > >
>> >
>>
>
>


-- 
|                            |   "Information is light.      |
| R. Mark Adams, Ph.D.       |    Information in itself,     |
| Computational Biologist    |    about anything, is light." |
| rmadams at epotential.com     |       - Tom Stoppard          |