Wall sized Tetris?

[Mosaic]

Hi guys:
I'd like some advice on creating a wall sized tetris from:

Tetris on a PIC Microcontroller | Swordfish Compiler Projects

What type of display....? Are there LED displays around 3' x 3' that can be used?
Would the best approach be to add darlington chips like the ULN 2008 to help source/sink the currents for the large display?

Or do they have to be hand assembled and then driven by a grid of MOSFETS?

thanx!!

Edit: Seems swordfish SE can't compile the code (no Randgen or Internaloscillator), would anyone be able to do a compile and share the hex with me?

 

[Wilksey]

Hi Mosaic.

Darlington's would be the best option for high current sinking / sourcing.
You may be able to find large displays, I have made some in the past, each 'pixel' contains 4 LED's for example.

Regarding the software,
Swordfish SE is a full featured version, you need the addon pack from digital DIY, nothing to do with swordfish:
Document Details | SF User Module Pack

Hope this helps.

Wilksey

 

[Sceadwian]

That's definitly a DIY project. I imagine a nice way of doing the display would be using something like automotive light bulbs (12 volts) and power transistors for your light elements, cheap simple easy off the shelf stuff. Then you just put those behind a large panel you create out of wood and paint that uses opaque plastic panels for the dots.

 

[Mosaic]

Hmmm, 128 Auto lightbulbs can be 3/4 amp each. = 96 Amps peak....I'd need more than ULN darlingtons to drive it. And a serious power supply!

Perhaps driving 4, 25 ma (100mA) hi intensity LEDs per 'dot' would work. 12.8 Amps max. A PC power supply 5V rail would happily do that.

Are there suitable higher voltage /lower current LEDs? Perhaps 12VDC etc?

 

[Wilksey]

Yes, you would need more than a darlington! lol.
You can get some high current LED drivers from Maxim, you may be able to get some free samples?

Dont forget that not all LED's are on, and if you are using column + row drivers then the LED's are only on for a short time, with visual perception.

Once you have 4x 25ma per dot, and you have 4 dots on for example, that is 400ma, that is going to be quite bright! You can increase the number to 8 for example for bigger displays.

You need to set your "canvas" size and see what looks good.

Wilksey

 

[AtomSoft]

I think some Piranha leds with something over it to defuse it would be better... there nice and bright. I would thing some tracing paper would defuse it well... never tried it would be cool..

Makes me want to try it out heh

 

[nickelflippr]

Hmmm, 128 Auto lightbulbs can be 3/4 amp each. = 96 Amps peak....I'd need more than ULN darlingtons to drive it. And a serious power supply!

Perhaps driving 4, 25 ma (100mA) hi intensity LEDs per 'dot' would work. 12.8 Amps max. A PC power supply 5V rail would happily do that.

Are there suitable higher voltage /lower current LEDs? Perhaps 12VDC etc?

I don't think 4 x leds would be required, it really depends on the overall pixel size. Before going to far down the road of power supplies, arrive at a pixel size, diffusion method, led type and acceptable brightness level.

Like Jason says, try a piranha style led (green shows best brightness), many of which will take higher drive currents like 35-50ma. Choose wide viewing angle over unrealistic mcd ratings and corresponding narrow viewing angle.

One idea for a pixel, is a baby food jar (approx. 2-1/4" dia). Diffuse by having them bead blasted.

Another idea is to use short sections of abs plumbing pipe for the pixel, along with an acrylic diffuser meant for fluorescent fixtures.

 

[Mosaic]

Well, I think the piranha suggestion is a good idea, some kinda diffuser. I recently made a diffuser with clear 1/8" acrylic and a can of 'frost' spray. Worked pretty good.

But I think I'll buy a dual 4' fluorescent fixture with a built in diffuser and build the whole thing in there. That's a 4' by 2' size and is just right.

I am considering this LED:
MCDL-1860UYC-TL: JAMECO VALUEPRO: Opto & Illumination

What do u think?

My ISIS sim shows that that the 8 LED columns are refreshed every 20mSec.

So only 1 column is active at any moment, if I overdrive the Piranhas at 80mA or so using a 5V supply & PIC, I can use 38ohms or thereabouts as the 3V dropping resistors. That means 1280 mA max load per column...which exceeds the Darlington 500mA max.

I can replace the ULN2803 with 4 of these:
https://www.newark.com/fairchild-semiconductor/fds6892a/dual-n-channel-mosfet-20v-soic/dp/82C2569

Also the Y axis drive requires 80mA drives, which could be done with a bank of 2n3906's or cheap logic mosfets to eliminate the base resistor.
Perhaps these:
FAIRCHILD SEMICONDUCTOR|FDC6312P|DUAL P MOSFET -20V 2.3A SUPER | Newark.com

 

[Sceadwian]

You wasted money on the frost spray, just sand the acrylic. The flicker it creates would probably drive me insane, and over driving LED's drops the efficiency down to at or bellow good incandescent lighting efficiency. LED light output is NOT linear with current so doubling the current will drop the efficiency dramatically and only increase the light output a small amount.

 

[Mosaic]

What flicker ? ....there's a 50Hz refresh, flicker would be minor. To compensate for the 1/8th duty cycle brief LED overcurrent is a well established method of augmenting the light output....so far as I have read in this forum and others.

 

[Sceadwian]

Mosaic, apparently you didn't read my post. light output WILL go up, but efficiency WILL go down faster. Find the lumens vs current chart for your LED's and calculate it out. You will see I'm correct. be mindful that the bulk resistance of an LED won't change much and that means greater power loss in direct heating of the materials. If you're calculating that a 100ma pulse of 500ms over 1 second is equal to a 50ma constant current draw with the same light output, you're incorrect.

I can see flicker in standard CRT monitors when my eye cavities vibrate because I'm walking towards it at upwards of 50 feet, that's vertical refresh, I see the bars with each step. I can also see the same modulation in LED clocks when I cough. Anyone that thinks that 25-30hz is 'good enough' for persistence of vision and thinks we're not capable of seeing higher order harmonics knows absolutely nothing about the workings of the visual system of a human being. It works fine for film, flicker is a totally different story.

 

[Mosaic]

Mosaic, apparently you didn't read my post. light output WILL go up, but efficiency WILL go down faster. Find the lumens vs current chart for your LED's and calculate it out. You will see I'm correct. be mindful that the bulk resistance of an LED won't change much and that means greater power loss in direct heating of the materials. If you're calculating that a 100ma pulse of 500ms over 1 second is equal to a 50ma constant current draw with the same light output, you're incorrect.

I can see flicker in standard CRT monitors when my eye cavities vibrate because I'm walking towards it at upwards of 50 feet, that's vertical refresh, I see the bars with each step. I can also see the same modulation in LED clocks when I cough. Anyone that thinks that 25-30hz is 'good enough' for persistence of vision and thinks we're not capable of seeing higher order harmonics knows absolutely nothing about the workings of the visual system of a human being. It works fine for film, flicker is a totally different story.

Hmm, I do believe u have some errors in this post:

**broken link removed**

That chart reveals for a current increase of 100mA thru 400mA, lumens increase from .18 to .62 = 344%. Net efficacy loss is from 1.35 to 1.05 = .3/1.35 = 22%.
That works out to 344% * (100-22)% = 268% lumens gain for a 400% current gain, or a ratio of: about .67:1, Lumens:current.

Thus I am realizing 2/3 of the increased current in more light. 67% efficiency in over running the Leds is not bad at all , and quite beneficial in my book.

Your claim about massive efficiency loss doesn't seem to happen until we reach 6 times the current ratings. I am merely doubling the current rating from 40mA to 80mA.

If u read carefully, I said 50Hz not 25 or 30Hz...that's a 100% difference!! 60Hz and better leaves no discernible flicker, 50Hz similar to PAL TV has acceptable and minor flicker.

 

[Sceadwian]

Notice on that graph that the lumens per amp goes down very slightly as current goes up? Notice also that the lumens per watt goes down both just as I said? What's also not in this graph is that you can't put that much current through an LED and maintain a 25C junction temperature it's physically impossible, the thermal changes will accentuate that graph in a real world device even more. That graph is pure theory not practical.


No discernible flicker at 50hz? I can see 60hz flicker on an LED at 50% duty cycle easily even without moving my head.. all you have to do is move your head to see it at higher frequencies if the modulation depth is strong, it's especially noticeable in peripheral vision because it's picked up as motion. You're missing the modulation depth portion of the equation, an LED is HARD on and HARD off, NTSC or PAL TV is one image to another there is no absolute blank phase in between each frame. Even past 100hz I can see the flicker if the modulation depth is high enough.

Even the higher frequency electronic ballasts for modern florescence lighting can be seen by sensative people, when the tubes go bad the arc can skip cycles, and I can pick the flicker up from that in the corner of my eye, typically when looking at a CRT monitor because the harmonics create beat frequencies with the differing frequency of the lights. This is on CRT's and bad flourscent lighting mind you. If I clench my jaw and hum or tap my skulll when looking at a typical LED clock I can tell if it's a fully driven element or multiplexed, if the duty cycle is less than about 75% the elements appear to shimmer for 'float' when the head is jogged, or turned sharply.

 

[Sceadwian]

By the way, what you may be talking about with pulsed LED's is the PERCEPTION of the light being brighter, it's a trick of the eye the actual amount of light output does not increase, just human perception of it's relative brightness, it won't help you see better in a dark room, especially if you're moving fast =)

 

[blueroomelectronics]

For the amount of money it'll cost to put together you could probably buy a projector and play all your games on the wall.

 

[nickelflippr]

Mosaic:
A green led is going to be perceived as much brighter than the amber one you linked too. I picked up a few these Cree green 4mm ovals for experimenting. They are really bright, and have a moderately wide view angle.

 

[Mosaic]

thx for the tip nickel.

I did some more digging and I think I found the ticket:

https://www.newark.com/jsp/search/productdetail.jsp?sku=32R7693

That baby can really crank it out! I can prob run >500mA into this using PWM as it has a 1A peak current rating.
With a 2.2 Vf, =>5V reg => 2.8V Pd across a resistor => A 5 Ohm 3W Wirewound resistor delivers 560mA!!
Heat gen in the resistor = .56 *2.8 = 1.6W.

Unfortunately this:
https://www.newark.com/vishay/vlmw711t3u2us-gs08/smd-led-little-star-white-e3/dp/46R4436

costs 2 much....otherwise it would be even better. Might make a decent flashlight though!

 

[nickelflippr]

Mosaic:
There is no end to led brightness and power. The Optek part looks good for the price. That package looks even more of a hassle to solder than a more normal plcc type though. The green version puts out twice as much luminance as the red, at rated power of 150ma. Heat could start to be a real issue, as I see they recommend some extra copper under the package.

The green Cree 4mm oval led is about twice as efficient as the Optek part.

The Optek max current figures are for less than 10us pulse width, and 1/10 duty cycle. I have to think the on time, even for a multiplex display, needs to be wider than that for any sort of bright display.

I like your idea of using the 2'x4' fluorescent fixture as the case. I still think the baby food jars have merit, cost free, and would have a neat 3d effect. The downside of the jars could be their cumulative weight?, and fragile nature.

I will try and throw up a pic of the jar, with some frosting paint, and the green 4mm oval.

 

[Mosaic]

Nickel, I'd be interested in seing a pic of the dispersion from the cree LED. I can't seem to find an data on the viewing angle.

This one might be better:
https://www.newark.com/cree/c566c-gfs-cv0z0792/led-5mm-green-5200mcd-535nm/dp/73P7898

 

[nickelflippr]

The dispersion chart is in the data sheet, and since it is an oval there are two different angles. It looks like 100 deg and 45 deg for horizontal and vertical planes respectively. At a focal distance of about 2" it seems pretty diffuse, with still some hot spot. A plcc or 5050 chip will give a closer focal point, more like 1", I would think.

In a dim lit environment, the Cree 4mm seems pretty good. In a more daylight environment, I could see wanting more lumens. Kind of a subjective thing.

It was impossible for me to get an accurate rendering of the Cree 4mm oval and the frosted baby jar. Used the Red-eye flash, macro setting, with some tissue over the flash. Using no flash, the picture comes out overexposed. Not going to spend anymore time on it for now.