Sorry to post a semi-nonrelated question but does anyone have a link to
info/datasheets/manufacturers/resellers of these LED matrices?

___________________________________________
Wesley Moore
RMIT - BEng/BApp.Sc. 2nd Year

wmoore@cs.rmit.edu.au
http://wmoore.tsx.org/

On Sun, 2 Apr 2000, Paul B. Webster VK2BZC wrote:

> Sumiec Jerry wrote:
>
> >   So after that I decided to make a 7x60 matrix with the same
> > principle of decoding each column and having the PIC supply the rows
> > again.
>
>   Right principle for five columns, wrong for 60.
>
> >  I used (4) 4x16 decoders and an 8 bit counter to cycle through each
> > column, that way I only needed 1 pin from the PIC to send a pulse from
> > the PIC to increment to the next column.
>
> >   Now the problem - when I try to turn on all the LED's they're all
> > really dim.  From reading some of the previous messages on the list
> > and actually reading the specs on the Optrex LED matrix's (<cough> for
> > the first time) I found that I need to supply 100mA to each dot on the
> > matrix with 1/10 duty cycle with a pulse of 0.1mS...  Now each pin of
> > the pic can only source 20mA, and the decoders only sink 25mA.....
>
> > Originally I wanted to keep the number of chips to a minimal, so I
> > didn't really pursue a huge shift register.
>
>   .. and now you are beginning to understand the true nature of the
> situation.
>
> >  Are their LED matrix modules that could use a smaller Ipeak?
>
>   Yes, but you'll really pay for them.  It's simple mathematics.  You
> want each LED to light adequately. 10 mA would do this, but it'd be 10
> mA continuously.  One PIC pin to drive every LED would do it, but you'd
> use a *lot* of PICS, so you multiplex.  Multiplexing means that instead
> of feeding the LED 10 mA continuously (or 20 mA for the old low-
> efficiency ones; I discriminate here between "efficient" LEDS of current
> design and "low efficiency" LEDS of older design), you feed it 100 mA
> for one tenth of the time.
>
>   If you were (head-banging) multiplexing some 25 mA into the LEDs on a
> 1 in 64 cycle, as you imply with  four one-of-16 decoders, then you'd be
> averaging about 0.4 mA and even with efficient LEDs, that is pretty dim.
>
>   So, the design of one of these displays is actually based on a 1-in-7
> or 1-in-8 multiplex, that is, one *row* at a time, all columns
> simultaneously driven (according to the data for that row) by shift
> registers and drivers.  This turns out to be the simplest solution and
> few implementations will vary.
>
>   The column drivers must sink 100 to 200 mA, depending on whether you
> want to average 12 or 25 mA per LED, and one ULN2803 driven by a 74HC595
> shift register per 8 columns, or the two functions integrated into a
> TPIC6B595, is the usual, with the current limit resistors, one per
> column.
>
>   The row transistors must source current for all the columns at once,
> so are usually replicated every 32 columns or so (four sets of 8
> columns) as a module, on which basis they must source between 3.2 and
> 6.4 amps.
>
>   You will soon see why these units often use switchmode power supplies
> (though I refer to older ones, the more recent ones with flimsy wall-
> warts must use very good LEDs indeed!) and why Jinx pointed out the heat
> problems.
> --
>   Cheers,
>         Paul B.
>