Rectangular LEDS

128 LEDs on 2 pins?
formatting link

Multiplexing though.
Or without multiplexing, that's 32 pins worth - add another 2 mcp23017's to get you to 64 pins - still 2 pins off the Pi (I2C)
There is another board that has (I thina)k 4 x mcp23s17's on it (SPI, 4 Pi pins), but not hard to hang these together yourself. The hard part is wiring up the individual LEDs.
Gordon
Reply to
Gordon Henderson
Loading thread data ...
interface
Is there any limit to the number of 595's that can be cascaded?
I can get away with three 594s or two 595s if all four of the LEDs in each of the 12 light bars can be controlled by one pinout (i.e by switching a constant current source for each light bar, so using 594/595 chips looks pretty attractive.
I liked the specs of the I2C LED driver chips until I realised that: (a) these chips are tiny and way beyond my capabilities when it comes to soldering them onto proto boards
(b) since these pups can only drive a single LED per output pin, I'd need four of the 16-bit types.
Point (a) is the killer for me. Point (b) is just a little more programming and there's oodles of time to run that, since the lamps only change once a minute.
Fantastic. That looks good.
--
martin@   | Martin Gregorie 
gregorie. | Essex, UK 
 Click to see the full signature
Reply to
Martin Gregorie
I thought that as well with the 1-Wire 8 port IO/switch chip that is only available in surface mount. but after watching a couple of "how to hand solder surface mount chips" on YouTube decided to go for it. Bought a small board with correct pads for the chip and points to attach proper wires, bit of selotape to hold chip in place, non acid plumbing flux, solder blob on hot iron, quick run along the pins, good wash with water and scrub with old tooth brush to remove the flux, job done. Well chuffed...
--
Cheers 
Dave.
Reply to
Dave Liquorice
If you want "readable" then the whole idea of encoded time should be dropped. There is a reason why this type of display is not popular, it is hard to read, period.
It may be a cute light show but it will never be practical. 90% of people will not be able or willing to use it.
--

Rick
Reply to
rickman
Because I'm an engineer.
There are 4 "lights" in the 5 hours row. There are 4 "lights" in the 1 hours row. There are 11 "lights" in the 5 minutes row. There are 4 "lights" in the 1 minute row.
So 4 rows and 11 columns, 15.
Simples, as the meerkat would say.
Reply to
mm0fmf
I think the brightness issue is overblown. If you drive them from a 5 volt supply the current limiting resistor will dominate control of the current and they will all be fairly equal in brightness. I doubt that you need a constant current source. What will be more important is to set currents so the different colors and different LED types look the same. This will take a bit of experimentation.
Even so, there are *many* LED drivers that include current control.
You need to understand the electronics. For a given current all LEDs of a given type and construction (read that as same part number) will have the same brightness to the eye. What varies is the voltage vs. currrent curve as that depends on the details of doping which varies a bit from batch to batch. However, this voltage vs. current curve does not vary widely. If you have an adequate voltage on the current limiting resistor it will dominate the small variation in voltage and all the same type of LEDS will look the same with the same value resistor.
I don't think there is such a thing as "a standard" soldering iron. It all depends on your skill level. PCBs are the way to go. You can get them for $2 a square inch (3 each) if I remember correctly.
Are you planning to use this in your final unit? Most people will say this is a bad idea.
Yes, printed circuit boards... lol
Micro Computer Unit. MCU is an all-in-one device with RAM, program storage and I/O. The rPi processor doesn't have memory so it is just a CPU with some I/O. MCUs are cheap enough that they are often used to take the place of specialized I/O devices. Heck, drop the rPi and just use an MCU with a 32.768 kHz oscillator, 23 outputs, 59 transistors and 59 resistors. You could put the electronics on one side of the board and the LEDs on the other and have a completed clock on a single board. Three PCBs for about $10. Add an RS-232 or I2C port and it can talk to the rPi so it can be set from a network time server if you must...
--

Rick
Reply to
rickman
Did you read anything else in the data sheet? You shouldn't stop reading at the marketing blub. Look at table 8 - LIMITING VALUES, lines IDD and ISS. 100 mA max for the entire device.
--

Rick
Reply to
rickman
15 outputs
Reply to
mm0fmf
Static test??? What good is that? The spec sheet says 100 mA max for the Vdd and Vss pins. Going over that can create problems internal to the chip and often the problem is the dynamic operation. Ground bounce can cause the chip to forget it's internal state.
As long as they need to be on/off together. Also you only need 1 transistor.
--

Rick
Reply to
rickman
Bluster. You said "This part is not designed to drive LEDs" except it is. It's there in black and white. In the data sheet. On the front page.
If you meant, it will drive a few LEDS but not as many as needed for this project then you should say so. Most of the people who are reading your advice are essentially zero knowledge on hardware issues and imprecise language will confuse them further.
Reply to
mm0fmf
On Sat, 29 Nov 2014 21:20:02 +0000 (UTC), Martin Gregorie declaimed the following:
I think you need to change your viewpoint and treat the "bar graph" as your row, not your column -- especially as "bar graphs" are normally seen as growing from one end -- not something with gaps in the bar (unless one has a bar graph with a peak indicator that decays slower than the main bar).
That makes four "graphs" of 12 LEDs... And you'd cycle among then using 2 pins and some external logic (you want 00 => top row, 01 => next down, 10 => next to bottom, and 11 => bottom row)... an 1 of 8 decode with one of the three inputs tied to ground may do with some wasted outputs.
You'd need to set the 12 column pins and the 2 row select pins; then prep the next rows data, and repeat... Do this fast enough and no one should notice that only one row is active at a time.
--
	Wulfraed                 Dennis Lee Bieber         AF6VN 
    wlfraed@ix.netcom.com    HTTP://wlfraed.home.netcom.com/
Reply to
Dennis Lee Bieber
Nope. The display is a ragged array containing 23 cells. The rows contain respectively 4,4,11 and 4 'lights'. Every cell contains a 'light'. Each 'light' can be switched independently of any other 'light', so it needs its own individual output.
Hence there are 23 outputs.
--
martin@   | Martin Gregorie 
gregorie. | Essex, UK 
 Click to see the full signature
Reply to
Martin Gregorie
The OP said he wanted 23 outputs. Are you building a different clock?
--

Rick
Reply to
rickman
Thank you for the correction...
--

Rick
Reply to
rickman
Who ever said it was practical? Not me: in this case cuteness is all. I never intend, or ever intended, to make more than one.
If it has any practicality at all, its purely to show that the RPi is still running and paying attention to the time of day while it hets on with its other tasks. Besides, it will probably be sitting alongside a Logitech Touch which shows the time when its not streaming Radio 4 or FZ.
--
martin@   | Martin Gregorie 
gregorie. | Essex, UK 
 Click to see the full signature
Reply to
Martin Gregorie
Let me introduce you to what you may consider is voodoo. We engineers call it multiplexing.
4 rows of 11 columns = 15 outputs
So the 5 hour lights are in row 0. The first in column0, next in column 1 and so on. Turn on 5hours we drive row0 and column0. 10hours, row0 and column0 and column1.
Scan down each row in turn and select the correct columns for the lights in each row. When you get top the bottom you go back to the top. (Hey that would make a great line in a song!). Lather, rinse, repeat.
Scan the matrix at say 1kHz and nobody knows the LEDS are flashing, good old persistence of vision. Of course the muxing will reduce the brightness because each position is only active for 1/4 of the time. So increase the current by 4. The LEDS are happy, they have a peak current limit but it's the average that is important.
If you don't believe the LEDs will survive, look at an infra-red remote control LED. The forward current is specified for 20mA or so. But to get any range they are driven with massive currents, 1A or so. But with a very, very small duty cycle. Peak 1A, average 20mA. The duty cycle calculation is left as an exercise for the reader.
Reply to
mm0fmf
No, I can light all 23 with 15 outputs. Especially with a 700MHz CPU running the software.
Reply to
mm0fmf
Cute!
Though, remembering back to when I had The Game Of Life running on a mainframe and printing each generation, 60x60 is really the practical minimum display size, and more would be better, especially if you want to use really interesting starting patterns such as 5 or 7 in a row.
I'll certainly drink to that!
--
martin@   | Martin Gregorie 
gregorie. | Essex, UK 
 Click to see the full signature
Reply to
Martin Gregorie
I thought we were talking about the clock someone else was building? I didn't know you were building one too.
--

Rick
Reply to
rickman
Hmm, thought it was 200mA - still 12mA should be more than enough for modern LEDs - some are really bright at only 5mA!
Not if you want the transistors to do the current control. I first saw the configuration I have in mind in Practical Wireless circ. 1975 and have used it may times since.
--
W J G
Reply to
Folderol

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.