24x12 matrix bicolor LEDs

[tEaM]

LT8500: 48 leds, SPI communication, US $5.04/piece
vs
HT1632C: 24x16=384 leds, SPI communication, US $0.65/piece

I'm starting to like the HT1632C...

The information of the switches will be read by the PIC and sent to the computer.
I am thinking to use a CMOS 8-Stage Static Shift Register (CD4014B) in cascade for read all switch.

[temtronic]

OK..please report where you found them for sale AND is there a pcb for them? I haven't found any Holtek distis up here in Canada,eh !

Might be a fun project over the winter...

Also, the 'project' has now doubled in size as now you want to READ 576 switches. Are these SPST switches ? It makes a big difference, so we nned to know what the switches are!

jay

[tEaM]

Yes, the switch is SPST.

The project did not increase . It already exists in my mind. Initiated by the control LEDs, after read the switchs. Finally communication with the computer.
Everything has to be thought out, step by step.

If I choose the HT1632C , I buy it in china.
Is there any code to control this chip?

[temtronic]

There isn't a 'driver' for the HT1632C yet. that's why I asked where you found the chip for sale. If you had them premounted on a breakout board, I'd code a driver for you in exchange for a few units.


In 'reading the switches', do you need to read more than one at a time?

Also just how big is the 'display/control' board?

hth
jay

[temtronic]

over coffee, been thinking about the switches ...

The 'display panel' appears to me as two sections ,each with 24 columns by 12 rows. The 'easy' way would be to have one PIC scan each 'section'. You'd need a PIC with 36 I/O pins, like my goto 18F46K22. Using the CCS supplied kpd example, it's real easy to expand it to read 288 switches(24 x 12).
The obvious details like key debounce and noise have to be dealt with, but for about $10 ( 2 PICs) it can be done in a day ! While it could be done with a smaller PIC and additional decoders, doing it my way allows for duplication of 'sections',simpler pcb and perhaps easier coding.

just one way to do it...

Jay

[tEaM]

Jay, you want this chip (HT1632C) to test? When I make the order, we send you a sample.

Yes, I'll have to always be reading all switches individually. Also was thinking make a matrix using 36 IOs (24 x 12 colunm row) to read all the switches.

What is the "kpd example" supplied by CCS?

tEaM

[Mike Walne]

[temtronic]

You'll also need to make a 'breakout PCB for the HT1632C. If you design it 'right', I'm sure you might be able to sell a few.
Old guys like me can't solder SMT devices....heck DIPs are bad enough these dayze !!


Jay

[dyeatman]

Breakout already available on EBay
_________________
Google and Forum Search are some of your best tools!!!!

[temtronic]

hmm. interesting though I'd rather have a 'user friendly' breakout of the signals....micro pins on 1 side, outputs sequentially arranged on opposite sides.

given the off shore sourcing for the part and the time needed to create the driver the LT8500 is looking like a better option. Easy SPI interface vs. bit banging means the driver can be cut a lot faster.

guess it all comes down to time vs money.


jay

[Jerry I]

The Status panel shown in your pictures looks quite large, What is the pitch of the LED's. Horizontally looks like 1", but Vertically looks like 2-3" ?.

24 LED's Across = 24", Vertically ??.

I would create multiple boards that would inter-connect 24 x 4. Already this board size is large.

3 boards vertically per side.

[temtronic]

Hay Jerry, I asked the OP how big it is a few msgs ago... to me it's got to be 4 feet wide maybe 6 feet tall? Compare the floor tiles which are usually 12" by 12" these days. Also the structure looks like 2by3s in the back braces ??

Size really does matter when it comes to wiring 'display panels'.

jay

[tEaM]

The pitch of the LED's is 1" horizontally and 3" vertically.
Each board has 8 LEDs and 8 switches. Each panel has 36 boards. 3 horizontally * 12 vertically.
The boards are connected with flat cable.

[temtronic]

I've been rethinking this project and have another idea.

Since each 'board' consists of 8 switches and 16 LEDS( 8 red, 8 grn) it probably could be done using a single 16F884 per 'board'. Doing it this way allows for easy switch reading( 1 switch per PIC I/O pin) and direct drive for the LEDS. PIC is about $3, add an RS-485 for communications and it's 'electronically' a very simple solution. While you will need 72 'boards' for the entire project costs are reduced due to qty. As well wiring is simple RS-485, use IDC connectors for 'daisy chaining', makes installation a snap. Testing is also easy !!
There are huge benefits in doing this. Time to code is very short as same program is in every PIC. You could use extra pins for 'board address' or hard code in flash.
You will need a 'master' PIC to control the 'action', again rather easy, simple read 'boards', decide what to do, send 'control' data as required.

This 'distributed process control' is how I've done remote energy control systems for 2 decades over leased datalines.
Using this approach R&D time and costs are minimal, parts are inexpensive and easily obtained. NO SMT needed either.


jay

[Jerry I]

Hi Jay

I see that we are almost neighbors.

Your idea is very interesting, and would work very well for this type of display of status information.

I come from having worked at an LED display type company and we would do it in a different way. Because we needed speed and constant refresh of the display with text and graphics, scrolling, and animations, your option would be slow for that. But in this case none of that is needed we think!.

We would use single PIC to drive the whole display. and shift data to each row of LED's to LED drivers like Toshiba TC62D748CFNAG. This is a 16 bit driver, the first 8 outputs would represent the RED LED, the next 8 would be GREEN LED. Or you can use 8 bit drivers, but to get the cost down 16 bit is better value. Then have glue logic to to latch the data to the LED for each row. Similarly using what tEaM suggested CD4014 shift register to read the switch inputs. Now the clock that shifts the data the the LED drivers can be used to also drive the CD4014 the data return path would be switch data. I have used the SPI Port SDO and CLK to shift data to LED Drivers, then use the SDI input for the switch data. I know 16bits are required for LED data and only 8 for switch data, but that can be distinguished in the programming.

As you write out LED data for that row, you would read in switch data or vise versa. Then do the next 11 rows, then back to the top. This can be done very fast. At this point you can have a message/graphic board.

Whats needed is the decoding of the rows to write LED data, and read switch data.

As per cost, I think this can be less expensive than having a PIC per each board. I would make a PCB that are 4 rows x 24 LED. Only require 6 Boards for whole display. Other glue logic like row decoding, data buffers between boards, I would also add.

Just my thoughts.