Group Actions
Engagement
  • Author Author: phoenixcomm
  • Date Created: Date Created
  • Views 357 views
  • Likes 6 likes
  • Comments 9 comments
Related
Recommended

QUESTION: I think the sky fell on me. I need some help please. The tail of too, too many wires.

phoenixcomm

Ok I have been working with the NAV Computer Display Panel, for a while now an I run up to this wall. OK, this beast as there are 4 sixteen-segment alphanumeric displays and another 13 seven-segment displays, DISCLAIMER THEY ARE NOT LEDS THERE INCANDESCENT! They run at 4 vdc @ 15ma. per segment.

First, this is a whole lot of wires drivers, cussing, etc. Second, they run off the 5vdc supply. There are 4 sixteen alphanumeric displays (that is 64 wires), plus 13 more seven-segment displays that is another 91 wires and there are 4 decimal points (4 more wires, are driven via the GPIO on an Arduino.). When I first got them years ago I thought that I could run them with a MAX part but it only supports LEDs and it uses a technique called 'Charlieplexing'. 'To make this easier you can drive the 4 alphanumeric displays on one bus but then you need 2-bits to act as selectors these can be driven via an I2C 16-bit port IC. The 2-bits are driven via the GPIO on an Arduino. Now you have left, the 13 seven-segment displays.. 

Here is the caveat, the refresh rate of the Human Eye. 

Any Ideas please post a response. 

  • in reply to shabaz

    There is also the PCA9685 which has 16 outputs which might be worth a look if you want to reduce the chip count. One per alphanumeric display.

    (There is the MCP23017 16-Bit I2C I/O Expander but are limited to 8 per I2C bus as standard whereas the PCA9685 allows 64 per bus.)

  • in reply to phoenixcomm

    Hehe I didn't miss that requirement : ) I'm aware of what's written in the datasheet, but, despite being marketed as an LED driver, that chip actually has no constant-current drivers for LEDs. It's simply open-drain (or totem-pole; it is configurable), allowing the driving of all sorts of devices, including incandescent. 

    That's why the circuit snippet has resistors in series with each output, because using that chip is effectively the same as driving the connected indicators from a constant voltage source, and it requires the series resistors to limit current, which is 100% as compatible with incandescent filaments as with LEDs.

    It will easily fit the dimensions you mention, with a 5-board or 10-board sandwich, since only two wires (I2C) plus two or three more wires for the supply voltage(s), are required to be connected between the boards (can be done with jumper wires or connectors). Each board would only be 1 or 1.5 inch square.

    The wiring requirement is therefore very small.  What's more important is how you wire from the boards to the indicators. If you have a photo of the connections, it would be interesting to see if a PCB could be soldered directly to them, to cut down your wiring between the indicators and board to zero.

    Also: suggesting 20 of the chips might sound a lot, but that example board in the earlier reply has seven of those chips on it, easily controllable, so I can confirm it is very feasible to have three times that amount driven by a microcontroller, to control all your incandescent filaments with simple code.

  •    The PCA9634 is an I2C-bus-controlled 8-bit LED driver optimized for Red/Green/Blue/Amber (RGBA) color mixing applications. Good try but you must have missed the spec.."DISCLAIMER THEY ARE NOT LEDS THERE INCANDESCENT! They run at 4 vdc @ 15ma. per segment." This is a big swing and a strike (baseball)!!  try again. LOL the total displays = 17  16segment(4) 7segment(13) also please remember that this fits into a 3" deep box 4.2" high, and 4.75" wide.  so you need 5 bits for the 17 selects +  3 more bits for the decimal points(4), so all I need in the unit is the drivers (UJTs?)  seven seg 91 + 64 more for alphanumeric.  

    the controller or FRU/LRU is one of my plastic shoe boxes that can hold all of the logic the two CPUs and whatnot. The main Idea is to reduce the interconnect wires there are 81 7segment drives + 64 alpha drives(16x4) just to toss a wrench  13 select line + 4 for the decimal points so the load 91+64+4 = 159 drives, x .015ma segment = 2.385amps. Here on ebay I found 5A Input DC 5-30V Output DC 3V 3.3V 4V 5V 6V 9V 10V 12V Synchronous DC-DC Buck so I would most likely split the load between two of them just to be safe.  This will provide the 4volts that I need for the displays.  now about dimming the beast back in the FRU/LRU I could do something with PCM.. 

    plus 2  thumbwheel switches with BCD outputs image

    plus 2 rotary switches, one 6 positions and the other 7. I can use two  8:3 bcd priority encoders which will go to a mcp2017.

    plus a 14 keypad which I can use a 74C922 keyboard 

    thoughts to ponder. I was planning on using 2 processors maybe an Arduino for the lights and maybe the switches as well.

    Then I could use an esp32 (small)  to handle the canAerospace/over ethernet traffic. and use an every this runs on interrupts. 

  • PCA9634 (PDF doc) is your friend! : ) 

    It's an excellent chip with 8 outputs, with zero multiplexing. Just plain and simple individual outputs which can even be slowly ramped up (very high speed PWM) if desired, to look after your display filaments).  There are only 8 outputs per chip, so you'll need a lot of them (20 chips to cover your 160-ish segments, so you'd need to design a board to (say) contain two of the chips each, and stack ten boards (as an example). Each chip can easily handle the power requirement of the filaments, a total of 200 mA per chip is supported. You'll need a series resistor for each filament btw.

    Another excellent feature; there are tons of address line inputs per chip (could use DIP switches per board, or solder links), so you can parallel all the I2C pins, and drive it all using just two wires into your Arduino, although it may be best to split it into two I2C buses. All of this minimizes 99% of wiring down to just the connections between the boards and the displays.

    I've used those chips with LED displays, but they will work with incandescent filaments. The benefit of those chips for me was the rock-solid display, with zero multiplexing flicker. I used it for an exhibition event, where I didn't want flickering.

    The circuit below is what could be replicated. The green boxes are the address lines, which are hard-coded to be different per chip (e.g. DIP switches as mentioned, or solder bridges) The blue box shows the resistors, Would be faster to solder in a resistor network there, rather than 8 individual resistors. Instead of the HDSP LED display, it would be the incandescent filaments. The yellow box shows the connections to the Arduino of course. 

    image

    Just in case it sparks any ideas, this was what my board looked like.. a load of large(ish) segment LED displays, all non-multiplexed. The PCA9634 chips were all soldered behind the LED displays, on the underside. The board would get a little warm, but not a lot. My board had all the I2C connections joined together, so that just two wires went to the microcontroller (on the right side of the board). The displays were remotely updated (using old-fashioned RS232! - there's a socket for that on the left side (I used a RJ45 for that, not DB-9). It was super-bright and clear, visible from many meters away, with zero flicker.

    image

    One other thought, is that wiring 160 connections is going to be a huge time-drain. If you can find a way of attaching a PCB (or PCBs) directly to the segments, then that opens up options that use zero wires. For instance, using flat flex to go from the PCB soldered to the segments, down to the display board. There is available very widely spaced flat flex that is super-easy to solder using any soldering iron. A photo of the current setup will help with more ideas hopefully.