Shop OBEX P1 Docs P2 Docs Learn Events
Word clock - Driving multiple LEDS - Whats the current best practice? — Parallax Forums

Word clock - Driving multiple LEDS - Whats the current best practice?

pacmanpacman Posts: 327
edited 2012-09-13 06:00 in Propeller 1
So as part of our first project together, my daughter wants to help build a word clock (http://www.instructables.com/id/Sleek-word-clock/#).

I intend to do it with a prop.

What is the current 'best thinking' on how to drive multiple leds with the ability to PWM? (a search of the site gives a host of methods)

Given the prop has a bucket of I/O pins (man I love the prop) I don't see then need to use shift registers, but current 'best practice' might mean I should...

suggestions please....

Thanks in advance

Comments

  • tonyp12tonyp12 Posts: 1,951
    edited 2012-05-07 10:33
    I would make a 11x10 matrix, so each letter can be on/off (10mA led), to allow centipede animations with ghost trail when it updates time etc.

    So you need 21 prop pins, attached to a IC of some sort that will give11 high side that could supply 10mA on each output (100mA total for IC)
    The prop may handle that by it self.

    and you need also 10pin low side IC that can handle 110mA sink for that pin that is on for the whole column
  • pacmanpacman Posts: 327
    edited 2012-05-07 13:58
    Thanks for that

    I was thinking that seeing as I always want the complete 'word' on' why not just combine all the LEDs in a word and drive each word with a transistor (or mosfet or.....) then I could drive each word separately.. Far less mucking about with shift registers...

    Fade in/out when changing time

    one drive pin for each 'hour' = 12
    one drive pin for each of 'about/nearly/almost/just" = 4
    one drive pin for time increments (five, ten..half) {twenty five would use two pins (20 and 5) = 5
    one drive pin for "to/Past" = 2

    total of 23 drive pins (with a couple left over for miscounts in my calculations).

    Words like "it is" and o'clock may or may not actually appear on my version - who would actually say "it is 5 minutes to 5 o'clock" rather than "nearly five to five"?


    For a base board - I'm thinking the quickstart - but very long term a custom board _might_ get developed. With the QS version perhaps a few more converts to the prop would be generated?


    Or am I missing a deeper truth?
  • HShankoHShanko Posts: 402
    edited 2012-05-07 14:23
    I'm wondering what sort of "LEDs" you are planning to use.

    Your text sounds like multi-segment LEDs might be your thinking to get each letter (7-segment LEDs can't display many letters). If so, there will be many lines needed to drive the segments and common of each. Are you thinking of common Anode or Cathode types? Either way lots of (shift) registers will be needed; plus many common drivers for each letter's LED.

    Your reference to 'instructables' appears as if a photo frame display was used for the letters and words. (Only wish interfacing with them was simpler.) Lots to consider with your 'clock'.
  • jmgjmg Posts: 15,173
    edited 2012-05-07 14:25
    pacman wrote: »
    I was thinking that seeing as I always want the complete 'word' on' why not just combine all the LEDs in a word and drive each word with a transistor (or mosfet or.....) then I could drive each word separately.. Far less mucking about with shift registers...

    Interesting idea - would use a shipload of LEDs!.
    The only detail would be physical - are these words separate on a large PCB, or do you expect them to overlay and share LEDs where possible ?

    If you want to share, some transparent graph paper could let you know which ones need 'diode or' and which do not.
    A manual nudge to the fonts and spacing could save on total LED count, giving a sparse grid appearance.
  • pacmanpacman Posts: 327
    edited 2012-05-07 15:05
    JMG and Hshanko - you are both correct

    Yes a (metric) shipload of leds - the original build uses 98 (hopefully mine with some of the 'excess' words removed would use less).
    Not sure of the _exact_ type of LED yet, but I just ordered 100 'hi-brite white' off ebay for $5 (including shipping) so for the prototype I think that will do.

    No PCB yet, the concept is to is to use a template for each word and a single led for each letter - the anodes and cathodes for each word are tied together (single resistor for each led) and then each word 'sunk' through a transistor/mosfet/opto coupler?.

    Certainly there will be a mile or two of wire 'behind the scenes'..

    More details as the concept formalises...
  • HShankoHShanko Posts: 402
    edited 2012-05-07 15:12
    I don't understand! 98 LEDs. OK, are these 'discrete' LEDs?

    If so, how will you be able to form very many letters, yet words, with that low count? Are you forming letters with many discrete LEDs? Think 14-segment LEDs to form each letter. That is 14 segments and one common (anode or cathode) for each letter.

    If not, you must have some new tricky scheme in mind. Give us a hint for more help.
  • kwinnkwinn Posts: 8,697
    edited 2012-05-07 15:17
    I generally use an npn transistor with a current limiting resistor between the emitter and ground. With the prop providing 3.3V to the base of the transistor it becomes a constant current driver so you can use an unregulated supply for the leds and still have consistent light output from the display.
  • jmgjmg Posts: 15,173
    edited 2012-05-07 15:43
    pacman wrote: »
    No PCB yet, the concept is to is to use a template for each word and a single led for each letter - the anodes and cathodes for each word are tied together (single resistor for each led) and then each word 'sunk' through a transistor/mosfet/opto coupler?.

    Ah, so the LEDS are backlights, to separate Stencil/Mask words.

    You may not even need one led per letter, just even illumination. ( Backlights use surprisingly few LEDs )
    Some type of frosting helps improve the viewing angle, and allows more distant LED placement, for fewer (brighter) leds
  • kilokilo Posts: 1
    edited 2012-09-13 03:26
    Hi Pacman,

    I too am trying to build a word clock using a propeller chip, was searching the web for a bit of inspiration and stumbled across your thread.

    I have the same idea as you in respect to using one output of the prop to power a goup of LED's (with a common -ve) which will backlight a single word. I don't see why it wouldn't work instead of using shift counters, just wondering if you've got any further along than what's already been said. I've penciled the circuitry for what would be the LED plate, and also just recieved a propeller proto board 32212. Just need to figure out the rest.

    Programing the thing is going to be my downfall as this will be my first time, also can't quite figure out how to set the time on it.

    Hoping to hear from you.

    Kilo
  • Mark_TMark_T Posts: 1,981
    edited 2012-09-13 06:00
    pacman wrote: »
    So as part of our first project together, my daughter wants to help build a word clock (http://www.instructables.com/id/Sleek-word-clock/#).

    I intend to do it with a prop.

    What is the current 'best thinking' on how to drive multiple leds with the ability to PWM? (a search of the site gives a host of methods)

    Given the prop has a bucket of I/O pins (man I love the prop) I don't see then need to use shift registers, but current 'best practice' might mean I should...

    suggestions please....

    Thanks in advance

    The advantage to using shift registers (more particularly LED driver chips with serial input) is that each LED can be run at full output current - multiplexing schemes reduce the average current per LED which can lead to rather dim display. The STP16DP05 and similar 16-channel LED driver chips have programmable constant current outputs (set by one resistor). eliminates the need for current-limiting resistors per LED too. And the outputs can be commoned to drive power LEDs too (100mA per channel), provided you pay attention to power dissipation.

    So I'd go that route for large bright displays.
Sign In or Register to comment.