Shop OBEX P1 Docs P2 Docs Learn Events
Hexagonal dot-matrix LED display — Parallax Forums

Hexagonal dot-matrix LED display

m00tykinsm00tykins Posts: 73
edited 2014-10-23 17:37 in General Discussion
Hello all,

I'm interested in making a custom RGB LED dot-matrix display for a project I'm working on. However, since the LEDs I'm getting will be circular, the best way to pack them closest together to make a display isn't as a grid like most dot-matrices, but as a honeycomb pattern. The advantages I can foresee with using a honeycomb pattern for LED dot-matrix would be higher resolutions simply since the LEDs are closer together, as well as having three axes that can be used to form straight lines without need for anti-aliasing instead of just two. The biggest disadvantage though would be the software needed to drive such a display. Has anyone thought of something like this before? Is it feasible? Thanks!

PS: Here's a link showing in pictures what I mean: https://loomsci.wordpress.com/tag/hexagon-tiles/

PPS: Here's the best guide I could find on the internet to writing code for hex grids: http://www.redblobgames.com/grids/hexagons/

Comments

  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2014-10-22 01:41
    Essentially, you are referring to a topic that is referred to as tiling.

    Mathemtatics has long recognized three alternatives of 'regular tiling' -- squares, triangles, and hexagons. Those are the only solutions that will cover a flat surface with one regular tile shape

    I am also fascinated by a two tile shape solution called the Penrose tiles, but it isn't important to your solution. I just mention that there are other tiling alternatives if you get away from single regular shapes.

    Hexagons will work nicely for tiles and you even show some links to graphics that are very interesting. But in the background, you might find that the actual wiring of a square grid is optimal. After all you are dealing with lighting pixels or diodes with a positive and negative connection.

    Wiring hexagons confounds me. If it can be done, it is going to be more complex to light and refresh the images displayed. I would love to hear if you have solved this.

    My main point here is that we use square grids because the wiring works extremely well in digital display. It also works extremely well in digital memory. A triangular grid or hexagonal grid doesn't seem to offer a simple wiring solution and may actually create a huge disadvantage.

    Also, you will have to create your own circuit boards for display. Everything that is ready-made is based on a square grid.

    It is quite remarkable how much of our man-made world is square grids and cubic spaces. I have been trying to find ways to break out of this for 50 or more years; but humans seem hardwired to the square and the cube, or the circle and the sphere. Hexagons are represented in organic chemistry. Triangles to come into play in inorganic molecular structures.
  • m00tykinsm00tykins Posts: 73
    edited 2014-10-22 08:24
    Thanks for the feedback loopy. But I'm afraid I don't understand, why would wiring be an issue? Since these are discrete LEDs, I should just need to arrange them in a hexagonal array on a protoboard, attach all (+) pins to the same source, and have a mess of wires out the back equal to the number of pixels, which can each just be attached to a single I/O pin on a MCU in the simplest case, right?
  • Heater.Heater. Posts: 21,230
    edited 2014-10-22 09:08
    m00tykins,
    ... attach all (+) pins to the same source, and have a mess of wires out the back equal to the number of pixels, which can each just be attached to a single I/O pin on a MCU...
    No. Wiring is a big issue.

    Doing what you say requires a lot of pins from your MCU and a lot or wiring to build. The Prop only has 32 IO pins remember. That does not get you very far.

    It is a lot of pins if you have a lot of LEDs so mostly we don't do that. Rather we use multiplexing. Where not all the LEDs are actually on on the same time, but they are lit up long enough and fast enough that the human eye perceives continuous illumination. Google for "led multiplexing" or some such to read more. For example a square grid of 16 LEDs can be controlled by 8 pins from an MCU. 4 pins switch on the power to a row, and four pins switch on power to a column. In this scheme 32 LEDs can be controlled by 12 pins, or 64 LEDs by 16 pins.

    Now. The easiest way to do such multiplexing is basically to assume a square grid. At least electrically. But of course the actual physical layout of the LEDs can be whatever you like. A row of seven segment displays for example. Or a hexagonal layout of LEDs.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2014-10-22 09:10
    The wiring should be no more complicated than it is for a rectangular grid. In fact, a hexagonal grid is equivalent to a rectangular grid with every other row offset. No big deal.

    One nice thing about a hexagonal grid comes into play if you're using discrete red, green, and blue LEDs: clusters of three repeat naturally and work much better than in a rectangular grid, where a Bayer pattern is more common.

    -Phil
  • m00tykinsm00tykins Posts: 73
    edited 2014-10-22 14:23
    Ah, ok. Thanks for the explanation Heater, and yes, multiplexing does make much more sense.

    If you look at the PPS link I posted above, this is just a problem of which coordinate system we use. Phil is referring to using the offset coordinate system, though I think the axial coordinate system is a more elegant solution. I could be wrong, but since axial and offset coordinate systems can refer to any single point with a set of two numbers, the wiring should be very similar to conventional dot-matrix displays, just with weird angles XD
  • Mark_TMark_T Posts: 1,981
    edited 2014-10-22 17:46
    You gain 15.5% increase in packing density at the expense of having to rescale every graphic and lose easy portrait/landscape reorienting.
    60deg reorienting is easy though.
  • PJAllenPJAllen Banned Posts: 5,065
    edited 2014-10-22 18:26
    "hexagonal" matrix display -- not a new idea

    Anybody care to come up with 0-9 and A-Z ?
  • kwinnkwinn Posts: 8,697
    edited 2014-10-22 19:35
    The wiring should be no more complicated than it is for a rectangular grid. In fact, a hexagonal grid is equivalent to a rectangular grid with every other row offset. No big deal.

    One nice thing about a hexagonal grid comes into play if you're using discrete red, green, and blue LEDs: clusters of three repeat naturally and work much better than in a rectangular grid, where a Bayer pattern is more common.

    -Phil

    I agree. A gamma camera captures images using a hexagonal array of photomultiplier tubes so there is no reason a hexagonal array of leds cannot be used to display images.
  • tomcrawfordtomcrawford Posts: 1,126
    edited 2014-10-23 07:17
    PJ Allen wrote: »
    "hexagonal" matrix display -- not a new idea

    Anybody care to come up with 0-9 and A-Z ?

    I wonder just what system these guys want to license. Surely not the idea of putting pixels in a hex array.
  • BeanBean Posts: 8,129
    edited 2014-10-23 07:31
    PJ Allen wrote: »
    "hexagonal" matrix display -- not a new idea

    Anybody care to come up with 0-9 and A-Z ?

    I'm not impressed by that layout at all. Look at the "B" layout. Looks terrible. How is it different from "8", by maybe 1 pixel ?

    Bean
  • PJAllenPJAllen Banned Posts: 5,065
    edited 2014-10-23 17:37
    I wonder just what system these guys want to license.

    The article is from a long time ago. Whatever it was did not set the world on fire.
    Maybe it was "suppressed" by the vested interests.
    Maybe they wanted interested parties to buy it (to "pass the bill") to find out what's inside.
    I'm not impressed by that layout at all. Look at the "B" layout. Looks terrible.

    Agreed. The straight 'A' doesn't go with the slanting 'B', but maybe they're from different fonts.

    I've googled and wikied the heck out of this, but found nothing that filled in the blanks.
    No leads at all (a lawsuit against NEC for some printing technique or sim.)
Sign In or Register to comment.