can the arduino handle this idea???

Forum 2005-2010 (read only) General Frequently-Asked Questions
1

i have a job i want to do, that incorporates 4700 pixels. when i say pixels, i mean points on an xy grid that can be on or off. ideally, it would be on or off w/ pwm, but .... thats a LOT of points.

With that said, i can go into detail. the goal is to create a 4x8 sheet w/ walls to contain ferrofluid. every square inch under the table will have a electromagnet. By selectively turning on magnets, i can create a "liquid display", with approx 4608 points of resolution. each of these magents will be a 12v, and with the possibilty of pwm, i can not only control the pixels that "light up" but the degree to which they 'light up'. in this application, the difference will be in milimeters, but i feel most likely visible despite the small scale, over such a large surface.

Its important to know if the arduino could handle this many points of 'illumination'. i've seen pretty extensive LED cubes w/ dozens of lamps, but this is dealing with thousands.

so. ideas?

2

An important question may be at what speed do you need to control the points?

3

uhmm... great question. if i wanted to refresh the screen (all points on or off) 60 times a second, would that be possible? half that, a quarter of that?

4

You might want to get in touch with whoever build the old SnOil project, your idea seems similar but scaled up. They might have some good ideas how to approach the project: - YouTube

5

60x per second doesn't seem that fast. I think you have the wrong question, though. It should be "How many Arduinos do I need to handle the job"? Anything can be solved by throwing more processing power at it, and there's no reason you can't have communicating Arduinos, each sharing a small piece of the work.

That's not to say a single Arduino can only take on 20 digital outputs. Certainly you can do more with multiplexing. But there's no reason to limit yourself to thinking it all has to be done with a single microcontroller.

--
The Quick Shield: breakout all 28 pins to quick-connect terminals

6

heh... well... then the idea is 'how many pixels' can a single arduino handle?

i would think that multiplexing IS the solution to this problem, but 4608 positions is quite a few.... that i would assume is a bank of chips. the question is wether or not the arduino could handle it. (and the materials i need to read in order to understand how to work with it)

and i've contacted the snoil guy, or tried to. waiting to hear back. the scale of this project however seems like it might be a bit different in its end methods.

7

what i'd liek to do, in all honesty, is take the chip out of a lcd screen and hack it to turn on / off a WHOLE BUNCH of transistors, or whatever, and illuminate 'pixels' as if the pixels were still the points on a lcd screen.

8

found this chip, and i think its large enough to handle the scope. Problem is i am having trouble finding illustrations as to how the screen itself is wired/assembled. this chip obviously does not have 9 thousand leads coming off of it. lol.

EDIT:

also, if anyone has a multiplexing for dummies solution, i'd love it. lol. i wont be powering little 3v leds though, i'll be switching 12 v coils on and off. i plan on using a cap to make the image 'persist' long enough until the next cycle comes around.

9

Below is a DIY project for controlling ~400 christmas light strings from a pc parallel port using multiplexing. You could make 12 of these to individually control your 4700 pixels. Serial shift register chips probably could be used with the arduino to control a lot of latching parallel chips. You good with a soldering iron?

http://computerchristmas.com/christmas/link-how_to/HowToId-4/LowLimit-10

10

I don't have any multiplexing examples, but there is no reason a single Arduino couldn't handle this. Refreshing each of your pixels at 60 times a second would only be a total of 276,480 refreshes per second. I don't see why an Arduino running at 16,000,000 cycles per second couldn't handle this.

As far as multiplexing, it is pretty straight forward. Probably use a horizontal and vertical axis... print out one horizontal line then trigger it to the next line and so on, then reset and do it all over.

11

i am in fact pretty good with an iron... but the cool thing is there's a couple dozen folks who are probably going to be involved in this project. a number of folks over seas might be particularly interested in soldering modules for us as i imagine they'd love to be involved in this project. even if not, there's the locals.

Meanwhile, i'm starting to understand the horizontal 'refresh' business. there are 48 rows in this build, with 96 pixels accross each. What i don't quite grasp is the how behind not having 48 wires to select a row, (for ground) and then 96 wires to illuminate a said pixel on that row. Or is that what my bottom limit on wires is going to be?

12

i plan on using a cap to make the image 'persist' long enough until the next cycle comes around.

Caps are not the right thing to do this. The coils themselves are "caps for current" so to speak.
This is in fact is the problem you have to investigate in:
What kind (length) of current pulse do you need to activate a single coil, and how long will the magnetic field exist? Mechanical inertia of the medium will add to this...

When multiplexing LEDs they have virtually no latency, that is fundamentally different with coils...

You are aware that a pixel will cost around $2 (not including the work time)?

Power consumption will be immens: Just multiply the energy in each "pixel's" magnetic field by the number of pixels by the average refresh frequency.

It is not like a motor running. It is like a motor starting all the time.

However this all is extremly simple to test: just make a 4x4 array of 16 of the smallest relais you can get. To "show" something connect an LED to the secondary side of the relays. Then multiply all data...

The software of curse should behave "as if" the matrix were the original size, otherwise you would be fooled.....

13

My concern is that you haven't thought about the memory implications of what you want. If you have 4700 pixels just assuming they are on an off that is 588 bytes of memory. You have a total of 2K so you won't be able to store more than a few patterns. If you have PWM for each point then it is likely there will not be enough memory for even one pattern.

You can't multiplex electromagnets the way you can multiplex LEDs.

14

The pattern might be generated by an algorithm...
And he can't multiplex coils the way he can do it with LEDs.

15

And he can't multiplex coils the way he can do it with LEDs.

I just said that?

16

the price isn't ... an issue. i'm not the guy footing the bill.

is there a 'reason' why i can't multiplex coils? if i gotta build my own multiplexer outta something beefier, i will, instead of using a little chip.

saying i can't do it is really sad guys. I imagine you're the folks that said the personal computer was of the devil and could never work anyways.

instead, lets challenge our imaginations and come up with a way it WOULD work. power isn't a problem, cost isn't a problem. The scope is hte problem. help me solve it?

17

We did not say it will not work. We (I) said there is something like a magnetic field you have to handle. That is JUST DIFFERENT. I have no experience with this and so I advised you to make an inexpensive experiment with 16 relais.

If you do not understand the postings in reply to your requests, please ask!

18

I imagine you're the folks that said the personal computer was of the devil and could never work anyways.

No we are the guys (or at least I am) that built the first personal computers and got them going, and defended them when every one asked what they were useful for. What we are doing is saving your time, money and more importantly your enthusiasm.
There is nothing more soul destroying working on something for a long time spending lots of money and then finding out you were battling the laws of physics. You think fighting the law is tough (as the song goes) but just don't take on physics.

The reason why multiplexing LEDs works is that you power each LED for a very short time, it produces a flash of light and the persistence of vision of your eye makes it look like it is on all the time if you flash it quick enough.

Now with a coil you can't turn it on very quickly, it takes time for the current to build up in the coil, this is called the time constant. You can cut it down by driving it at a higher voltage than it can stand and never letting it reach that voltage. However when you remove that voltage the magnetic field starts to collapse, normally that would generate a back emf that would keep the field up for a short time but this generates a negative voltage that can damage electronics so this is normally shorted out with a diode. This makes the field collapse even quicker. So in effect a magnetic field is slow to come on and fast to go off. This is what buggers up multiplexing (to use a technical phrase).
This is unfortunate because it leaves you needing quite a few high current drivers.
In short there are no easy answers to this one which is why I suspect no one has done it before. Now that has got to make it interesting.

19

My subconciuos had worked on this from the early morning, I think....

Simple LED multiplexing provides energy and control by the same pulse. In your application it might be useful (necessary?) to decouple that. The multiplex pulse will just trigger a monoflop or such which will gate the current for a definite time, much longer than the multiplex pulse. That needs a connection of all 4000 nodes to the supply voltage. However that will also help to handle the current issue....

20

sigh

firefox crashed as i was completing my post. it was an epic loss.

i will rewrite it:

first, i'm sorry if i insilted anyone with my last post. it was not my intention. I consider it to be an insult and a disheartening to hear 'cant do it, move on' when there's a new idea. This mentality, incidentally, has led to my contemporaries (i'm 26) all being consumers and not inventors. The biggest thing some of my friends will ever do is invent a new skill rotation for world of warcraft that optimizes their dps. If you think what i'm doing, compare it to that and reconsider. Meanwhile, this forum embodies a pool of experience and knowledge that has already once changed the world, by stretching the bounds of imagination and understanding. It stands to reason that that skillset would be incredibly useful in inspiring more generations to invent and reinvent things to make them better and more awesome.

Regarding power. This project will be housed not in a private residence, but a factory w/ 3ph 440 power. The power requirements of the table are intimidating at best:

1,520.64 amps for 4608 of these:

http://catalog.apwcompany.com/viewitems/electromagnets/1-0-diameter-round-br-em100

(12 v, .33 amp, 19lbs pull)

if all of them were on at once. These only serve as an example, as i doubt i need much more than a pound or two of pull from the magnet, and definately not 19. We've already started talking with manufacturers to find out if they can scale down their magnets to work with our application.

The money - not really my issue. my issue is integration of hardware to a common result - to put a array of electromagets under a 4x8ft sheet (with sides) filled with a 2millimeter deep ferrofluid w/ the intention of displaying letters and or graphics for educational use.

The goal is to encapsulate display technology and combine it with a crowd favorite of magnetism in order to create a 'fun' and otherwise 'cool' learning tool that can be used in the non-profit company i ~represent~. They've done things like build the world's largest newtons cradle, and were the guys to originally come up with 'singing tesla coils' (using steve wards solid state theories and builds). They specialize in the geek, and the enormous. This build is small and pales in comparison to some of the other undertakings they have in the works. All this while they have an open door policy, and often times are a destination for fieldtrips of students for a fun and definately more upbeat learning enviroment.

Now, regarding the backlash of the magnets on the electronics. Can the electronicsw be repositioned away, or shielded in a box to protect them. can larger diodes be used to stop current 'backwashing' and causing hell on the circuit. Each coil will be recieving a 12v 1a transistor, would the transistor itself fail in such a way that the current would possible go down the signal wire and nuke the controllers?

Its not about 'thats impractical'. its about 'how do we do it?'.

I'm hoping that you folks can help me, because i'd love to be able to accredit the arduino community for more than just the logic behind the code that drives it. If folks here are disinclined to help breach the 'impracticality' of such a electronics snaffu, i dig it, and will seek another forums of folks who are more open to the 'well... thats interesting, here's how i'd approach it...' way of doing things.