Simple question about Beau-Plexing LEDs

rjo_ · 2007-09-09 03:21

Guys,

¿Qué es esto?
Was ist dies?
Что является этим?

The story is this...and as always the story is completely true:

I tentatively decided to make a Propeller based birthday present for daughter dearest. I don't think she would appreciate a barely functional, servo-based coil winder, and I have hundreds of LEDs laying around from a project that got abandoned by my (now defunct) tech team.

So, I've decided to Beau-Plex something together... I'm thinking of an LED wall clock that spells out... "Take Your Pill" every 24 hours, in a variety of hypnotic patterns.

Anyway, with almost half a clue as to what I am actually doing... I believe that I have implemented a one LED "Beau-Plexed" test circuit. To my delight, I found that if two Prop pins are connected to the anode of an LED and set to output... and one of those pins is set to high and the other is set to low... and the cathode of the LED is connected to ground... very little happens.

... just like Beau promised[noparse]:)[/noparse] Amazing.

I then went to the app note posted by Graham and the explanation centered around a reverse bias across the LED, and I'm not buying it.

I don't see how my arrangement can reverse bias anything...(and therefore, I suspect that I am misunderstanding the discussion).

I put an amp meter across the legs of my Beau-Plexed LED... and the current dropped by 3.6 millivolts between the LED ON (53.2 milliamps) and the LED OFF (49.6 milliamps) corresponding to 2.3 volts and 1.8 volts.

I've concluded that either Beau-Plexing and CharliePlexing are not the same or that CharliePlexing has nothing to do with reverse bias across the LED.

Which if either of these alternatives is correct?

How would I apply and control a reverse bias across my LED?
.
Thanks,

Rich

p.s. I don't want to pressure you guys, but my daughter's happiness is in your hands[noparse]:)[/noparse]

By the way... about my current project... "the worlds slowest coil winder." Things are progressing nicely. I figured out the spindle design and I got the Bank to sign off on a purchase order for 50 meters of wire. I'll be posting methodology and results soon.

rjo_ · 2007-09-09 04:53

Beau,

I would bet dollars to donuts that in the mode that produces the 1.8 volts the wire heats up... which would mean that resistance could be cast as a function of information statistics applied to the current.

Rich

Skogsgurra · 2007-09-09 05:41

Hello again rjo_! Upo early? Or is it late?

This is where you misunderstood things: "if two Prop pins are connected to the anode of an LED and set to output... and one of those pins is set to high and the other is set to low... and the cathode of the LED is connected to ground"

If you connect two output pins and set one high and one low, you are probably very close to overloading the device because one pin is feeding directly into the other, which is sinking current with very little resistance. Do it like this instead:

Connect LED cathode to one pin and LED anode to the other pin.

Output high to the anode and low to the cathode. LED lights.

Output high to both, or low to both. LED is dark.

Output low to anode and high to cathode. LED is dark, because it is reverse biased (wrong polarity).

Connect another LED (with cathode to first LED's anode) - and it should com on.

So, you are changing polarity from the LED's point of view. Even if you are not outputting anything but 0 V and + 3.3 V.

BTW, it is not recommended to run outputs short-circuited. Put at least a 47 or 100 ohms tresistor in series with the LED.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔

deSilva · 2007-09-09 06:30

I am sure Rich was pulling our legs, but a litle drawing might still further the understanding.....
(1) Understand how a matrix works? 1 of N going "plus" and 1 of N going "minus" ? So you kann address NxN LEDs
(2) Now connect JUST THE SAME N lines as row AND as column drivers. Don't ask - Do it!
(3) Of course you now can no longer use the diagonals as they are connected to one line only..leaves NxN - N possibilities
(4) Light them as you would have done in case(1)

I upload a 2 and 3 silvaplex diagramm...

Post Edited (deSilva) : 9/9/2007 8:44:32 PM GMT

Beau Schwabe · 2007-09-09 07:54

Whew!!! ... Ok, here is what I was trying to describe.· With 12 I/O lines you can control 6 stacks of 3x3 LED's (54 LEDS)

0ZZ_111111111 - ALL LED's "on" ;  FIRST LEVEL
Z0Z_111111111 - ALL LED's "on" ; SECOND LEVEL
ZZ0_111111111 - ALL LED's "on" ;  THIRD LEVEL
1ZZ_000000000 - ALL LED's "on" ; FOURTH LEVEL
Z1Z_000000000 - ALL LED's "on" ;  FIFTH LEVEL
ZZ1_000000000 - ALL LED's "on" ;  SIXTH LEVEL  
1 = HIGH (3.3V)
0 = LOW (0V)
Z = INPUT

Note1: for simplicity, I just gave an example with ALL leds on each level "on"... ANY 9-Bit pattern for the LED's would be valid as well.

Note2: When a desired LED needs to be OFF, the corresponding BIT needs to be made an INPUT

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

Beau Schwabe · 2007-09-09 08:13

Charlie-Plexing would actually be better in this case... the way it is above you're only saving 3 I/O's. With Charlie-Plexing you could control 9 levels of 3x3 LEDs (81 LEDS) with only 10 I/O's.
Later today if I get a chance, I'll show a schematic of what that would look like.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

deSilva · 2007-09-09 09:07

I think silvaplexing shows all that is needed - but I am certainly "biased"

Graham Stabler · 2007-09-09 10:42

Rich, If you look at the app note again you will see that when you adress a given segment you will also be sticking positive on one of the common cathodes so will be reverse biasing a diode in another segment.

Graham

Beau Schwabe · 2007-09-09 19:55

Here is how you would Charlie-Plex 81 LED's with 10 I/O's....· You could actually go one more and have 10 groups of 9 LEDs for a total of 90 LEDs.

The formula is

Total LEDs = N*(N-1)

where N is the number of IO's used.

deSilva,

I can see how your method works well with a 2d-matrix, but I'm not sure how it would lend itself to a 3d-matrix.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

Post Edited (Beau Schwabe (Parallax)) : 9/9/2007 8:09:29 PM GMT

inservi · 2007-09-09 20:23

Hello,
here is a post with nice sample (see post 9 from OakGraphics) i think it's the 'Beau-Plexing' with clear schematic.
http://forums.parallax.com/showthread.php?p=667572

Here is my proper schematic witch show that the principe is to connect a couple of led 'flipped' between two pin. And repeat for each possible combinations with each others pins.

dro.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
in medio virtus

Post Edited (inservi) : 9/9/2007 8:30:17 PM GMT

deSilva · 2007-09-09 20:56

Beau Schwabe (Parallax) said...
deSilva: I can see how your method works well with a 2d-matrix, but I'm not sure how it would lend itself to a 3d-matrix.

All schematics given (Beau's, Inservi's, mine) are TOPOLOGICALLY identical.
I was only interested in the question how it works, and gave the answer that seems most obvious to me.

Another question is, how to ARRANGE this GEOMETRICALLY. Beau's schematic will help to form a 10 story tower with 9 LEDs each.

I think the question how to CONSTRUCT it MECHANICALLY is the moste tricky one...

Drone · 2007-09-10 18:31

Whoa Beau... Nice schematic of the Charlieplexed arrangement - but a 700+ MB .JPG attachment! Most outside the 1st-World have to deal with dial-up modem connections. I attach a version of the file that is 10 times smaller is but still screen readable (won't print to paper as well though).

Regards, Dave

Simple question about Beau-Plexing LEDs

Comments