Comments on: LED-Brite http://dansdata.blogsome.com/2009/04/10/led-brite/ the blog that is not dansdata.com Sat, 11 Feb 2012 14:54:20 +0000 http://wordpress.org/?v=1.5.1-alpha by: mangoman http://dansdata.blogsome.com/2009/04/10/led-brite/#comment-4704 Sun, 12 Apr 2009 20:29:57 +0100 http://dansdata.blogsome.com/2009/04/10/led-brite/#comment-4704 Actually you can, you just can't be quite as fussy about efficiency. For example, take a red LED with a forward voltage (V_f) of ~1.5V, and a blue LED dropping ~3.0V. Wire them in parallel, each via. a separate 200Ohm resistor from 4.5V. The red is sinking (4.5-1.5)/200 = 15mA, and the blue is sinking 7.5mA, half as much. But since the forward voltage for the blue is twice as high, the power (P=VI) is the same: 22.5mW. An LED with a V_f in between (say, a green one) will dissipate slightly more (maximum 25.3mW @ V_f=2.25V). The cost is that the red LEDs will only be ~33% efficient, 2/3 of the applied power is dissipated in the dropping resistor. A board fully populated with 1000 red LEDs will be supplying 25W to produce light, and wasting 50W in the dropping resistors. For blue it'll be 25W LEDs, 12.5W resistors. Anyway, it's totally possible, just not particularly efficient. Cheap though, and for business signs, I don't imagine the boards will be more than 10-20% populated. Actually you can, you just can't be quite as fussy about efficiency. For example, take a red LED with a forward voltage (V_f) of ~1.5V, and a blue LED dropping ~3.0V. Wire them in parallel, each via. a separate 200Ohm resistor from 4.5V. The red is sinking (4.5-1.5)/200 = 15mA, and the blue is sinking 7.5mA, half as much. But since the forward voltage for the blue is twice as high, the power (P=VI) is the same: 22.5mW. An LED with a V_f in between (say, a green one) will dissipate slightly more (maximum 25.3mW @ V_f=2.25V). The cost is that the red LEDs will only be ~33% efficient, 2/3 of the applied power is dissipated in the dropping resistor. A board fully populated with 1000 red LEDs will be supplying 25W to produce light, and wasting 50W in the dropping resistors. For blue it'll be 25W LEDs, 12.5W resistors. Anyway, it's totally possible, just not particularly efficient. Cheap though, and for business signs, I don't imagine the boards will be more than 10-20% populated.

]]> by: Changes http://dansdata.blogsome.com/2009/04/10/led-brite/#comment-4703 Sun, 12 Apr 2009 19:25:08 +0100 http://dansdata.blogsome.com/2009/04/10/led-brite/#comment-4703 But you can't just have a resistor for each LED slot. LEDs of different colors use different resistors. The only way I can see this working is with a current-regulated supply for each LED. It could just be flickering them really fast in order to use one power supply for several LEDs, but just resistors on the board won't do if you have to be able to plug different colours of LEDs in the same slot and have them work fine. But you can't just have a resistor for each LED slot. LEDs of different colors use different resistors.
The only way I can see this working is with a current-regulated supply for each LED. It could just be flickering them really fast in order to use one power supply for several LEDs, but just resistors on the board won't do if you have to be able to plug different colours of LEDs in the same slot and have them work fine.

]]> by: mangoman http://dansdata.blogsome.com/2009/04/10/led-brite/#comment-4702 Sun, 12 Apr 2009 19:18:31 +0100 http://dansdata.blogsome.com/2009/04/10/led-brite/#comment-4702 I couldn't say with certainty, but for a reflow-soldered board the extra (time, and therefore money) cost would be in the pick-and-place for the extra components rather than the soldering (as the joints are all formed in one pass). There's presumably some sort of socket being placed for each LED already, so the extra cost is doubling the number of component placements. I couldn't say with certainty, but for a reflow-soldered board the extra (time, and therefore money) cost would be in the pick-and-place for the extra components rather than the soldering (as the joints are all formed in one pass). There's presumably some sort of socket being placed for each LED already, so the extra cost is doubling the number of component placements.

]]> by: TwoHedWlf http://dansdata.blogsome.com/2009/04/10/led-brite/#comment-4701 Sun, 12 Apr 2009 17:33:49 +0100 http://dansdata.blogsome.com/2009/04/10/led-brite/#comment-4701 What's the per board cost of an extra 2000 solder joints? What's the per board cost of an extra 2000 solder joints?

]]> by: mangoman http://dansdata.blogsome.com/2009/04/10/led-brite/#comment-4700 Sun, 12 Apr 2009 14:15:11 +0100 http://dansdata.blogsome.com/2009/04/10/led-brite/#comment-4700 Changes, it wouldn't be a big deal to have a resistor per LED on the board. For a 20x50 position board (1000 LEDs), and a per-resistor price of <a href="http://au.farnell.com/multicomp/mc0603saf0000t5e/thick-film-chip-resistor/dp/1631319" rel="nofollow">$AU0.004</a>, that's a per-board cost of $4. You'd simply have the top plane of the PCB V+, the bottom GND, and one resistor to to either the anode or cathode of each LED. Changes, it wouldn't be a big deal to have a resistor per LED on the board. For a 20x50 position board (1000 LEDs), and a per-resistor price of $AU0.004, that's a per-board cost of $4. You'd simply have the top plane of the PCB V+, the bottom GND, and one resistor to to either the anode or cathode of each LED.

]]> by: Changes http://dansdata.blogsome.com/2009/04/10/led-brite/#comment-4699 Sun, 12 Apr 2009 11:40:12 +0100 http://dansdata.blogsome.com/2009/04/10/led-brite/#comment-4699 I'm aware of the resistor thing, but I understand what you're plugging in that board are bare unresistored LEDs. I'm aware of the resistor thing, but I understand what you're plugging in that board are bare unresistored LEDs.

]]> by: peridot http://dansdata.blogsome.com/2009/04/10/led-brite/#comment-4697 Sat, 11 Apr 2009 14:17:07 +0100 http://dansdata.blogsome.com/2009/04/10/led-brite/#comment-4697 Changes, all you have to do to light up a bunch of LEDs is attach an appropriate resistor to each and then feed them all from the same V+ and ground. If you want different colors the V+ will have to be high enough for the bluest color (where white is actually UV) and so it'll be a little less efficient, but a resistor per LED is cheap. Alternatively, if you multiplex it so only one is lit at a time, you can have one resistor/constant-surrent supply for many LEDs. But if they're all going to be on all the time this seems unnecessarily complicated. Changes, all you have to do to light up a bunch of LEDs is attach an appropriate resistor to each and then feed them all from the same V+ and ground. If you want different colors the V+ will have to be high enough for the bluest color (where white is actually UV) and so it'll be a little less efficient, but a resistor per LED is cheap.

Alternatively, if you multiplex it so only one is lit at a time, you can have one resistor/constant-surrent supply for many LEDs. But if they're all going to be on all the time this seems unnecessarily complicated.

]]> by: solipsistnation http://dansdata.blogsome.com/2009/04/10/led-brite/#comment-4696 Sat, 11 Apr 2009 10:03:42 +0100 http://dansdata.blogsome.com/2009/04/10/led-brite/#comment-4696 Oh, you know. Peggy and Peggy2 aren't charlieplexed-- they're for-real multiplexed using a couple of LED driver ICs and a couple of 4-to-16 demultiplexers. Charlieplexing does look cool, but Peggy is a little more complex than that. Oh, you know. Peggy and Peggy2 aren't charlieplexed-- they're for-real multiplexed using a couple of LED driver ICs and a couple of 4-to-16 demultiplexers.

Charlieplexing does look cool, but Peggy is a little more complex than that.

]]> by: Changes http://dansdata.blogsome.com/2009/04/10/led-brite/#comment-4695 Sat, 11 Apr 2009 06:29:29 +0100 http://dansdata.blogsome.com/2009/04/10/led-brite/#comment-4695 How exactly does the device work? Does it actually have a tiny power supply and/or current regulator behind each LED? It seems hard to believe, but I can see no other way in which it could manage a series/parallel connection of LEDs without knowing which is which and without burning some up and underdriving others... How exactly does the device work? Does it actually have a tiny power supply and/or current regulator behind each LED? It seems hard to believe, but I can see no other way in which it could manage a series/parallel connection of LEDs without knowing which is which and without burning some up and underdriving others...

]]> by: FuzzyPlushroom http://dansdata.blogsome.com/2009/04/10/led-brite/#comment-4694 Sat, 11 Apr 2009 06:02:32 +0100 http://dansdata.blogsome.com/2009/04/10/led-brite/#comment-4694 I'm feeling some awesome <a href="http://en.wikipedia.org/wiki/Connection_Machine" rel="nofollow">Connection Machine</a> influence here. I'm feeling some awesome Connection Machine influence here.

]]>