[dorkbotdc-blabber] Re: LED Cube Workshop - use an Arduino?

[Peter S. May]

R. Mark Adams, Ph.D. wrote:
> Okay-  I think I have a way to get the Arduino to run the LEDs without
> needing anything fancy (read shift registers, etc.)  Now I need a way to
> test it, and I do not have enough LEDs.  Someone (Tim?) had mentioned
> that they buy LEDs on ebay and have good luck doing so- any suggestions
> as to a vendor, delivery time and price?  I figured it might be the best
> way to get some quickly to we can test out our ideas!  :-)
> 
> Thanks-
> Mark

I've so far had decent luck buying LEDs and certain other electronics
items (CR2032 coin cells, for example) from the various China/Hong Kong
(gray market?) sellers.  It just happens that both LED orders I've
placed came from giorgio11185, aka "Asia Engineer".  No complaints so
far, other than the 17-20 days it takes to ship since it's not domestic,
and at least in my experience 12 cents shipped (lot of 100) is a good
price for a high-brightness white LED.  Some come with free resistors;
see the auction description if you want to ask for resistors adjusted
for 5V (the default is often 12V).

Your mileage may vary, but for me it's been unproblematic and worth the
wait.

(As with any LED, take note of both spot brightness (mcd) and viewing
angle (degrees) when picking one out.)


Incidentally, what is this way you have to drive the LEDs "without
needing anything fancy"?  Is it different than the multiplexing method
used by Bre and Mitch?

Getting pedantic for the benefit of the unaware, so please skip with my
apologies if this doesn't include you. :-)

Somebody who's done the Make version of this project, please tell me if
the following description is correct--I took a look at the PDF for the
Make mod of MiniPOV for the 3x3x3 cube, and it wasn't trivial to decode.
 But from what I can tell they took the 8 LED outputs already on the
kit, added one more by sticking a resistor onto one of the other I/O
pins, and then implemented the multiplexing by using three more of the
I/Os as a ground select.  The result looks a bit like the following
ASCIImatic:

.               (X,Y,TOP)
. XY ---/\/\--+->|-------------------- /TOP
.             |    (X,Y,MID)
.             +----->|---------------- /MID
.             |        (X,Y,BOT)
.             +--------->|------------ /BOT

Where XY is _each_ of

. NW N0 NE
. 0W 00 0E
. SW S0 SE

It's wired up such that all LEDs of a given Z share a cathode and all
LEDs of a given XY share an anode.  That adds up to 12 I/O pins (9 to
select XY and 3 to select the Z layer) and 27 LEDs (one per X per Y per
Z).  A given LED on the cube will be lit if and only if its anode (XY
pin) is high and its cathode (Z pin) is low.

The firmware draws the pattern one Z layer at a time by, for example,
setting all Z pins high to clear, setting the XY pins to the desired
pattern, and then setting the desired Z pin low.  By cycling through the
layers faster than the viewer's flicker threshold, it appears that all
layers are being drawn at the same time.

So, assuming the above is true, as long as the Arduino has 12 available
outputs it can pull the same trick.

More garbage on this topic in the PS.

Was this what you had in mind?  If not, what?

Word
PSM



Post Scriptum:

This is me getting (even more) carried away...

There's even a means of addressing 3x3x3 = 27 LEDs (or as many as 30)
using only 6 pins, as long as each one is tri-state-able.  But this
method can typically only address one LED at a time (as opposed to one
layer) and it majorly complicates the firmware.  (See also
"Charlieplexing".)

An unrelated project I have going on is a little heavy duty for any sort
of multiplexing and instead uses a small fleet of 74HC595 shift
registers.  For an immense number of LEDs (over 60 in my case) it's a
bit more straightforward and can use as few as 3 pins for an arbitrarily
large number of LEDs--as long as you buy one 74HC595 for every 8 LEDs.