Charlieplex (N²) ?

[3v0]

This thread has been cut back to where people were playing nice. It was a few more posts then I would like to have removed.

Play nice
3v0 as moderator

 

[Mr RB]

...
I am trying to build a very thin (0.25 inches) LED board with 4 wire connects. No onboard chip What is the maximum number of LED's I can address? What concept design do you suggest?
...

A total PCB of 0.25" thick allows lots of options.

Do you have room to fit a tiny SMD shift register IC on the PCB? The shift register can drive 8 LEDs without needing PWM, and the 3 wires needed to drive it can be bussed over many boards, so you can use the same 3 wires to drive ALL your boards for a large matrixing operation.

Or is there some reason you can't put a tiny SMD chip on the PCB (like the PCB is entirely filled with LEDs etc)?

 

[Chipwizard]

A total PCB of 0.25" thick allows lots of options.

Do you have room to fit a tiny SMD shift register IC on the PCB? The shift register can drive 8 LEDs without needing PWM, and the 3 wires needed to drive it can be bussed over many boards, so you can use the same 3 wires to drive ALL your boards for a large matrixing operation.

Or is there some reason you can't put a tiny SMD chip on the PCB (like the PCB is entirely filled with LEDs etc)?

Hello RB:
0.25 inch seems to be wide but not wide enough for fancy details. Considering your LED's (SMD or DIP) in the center of the board, LED pads will limit the number of lines across the board. To make it even more complicated there are limited parameters for production of a PCB such as minimum trace, gap, minimum hole size for vias, and gap between trace to pcb edge. Most but not all can be overcome.

It is a very noble idea to put a register chip on the LED board and I thought about it in the begining and gave up because of wiring and lack of space issue. If you place an on-board chip (SO, SOL, SMD, DIL) you need minimum of 0.3 inch width for the board with chip pads being right along side of the board. The traces from the chip (8-wires) have to come from within the center of the chip going across the board. Now to add to that you need 5-wires (not 3) across the board to run the register chip. (+,-, Data, Ck, Latch) I do not see how we can run 13 wires along side the LED pads and make it fit on 0.3 inch board! Your thoughts...
Regards,
Rom

 

[Chipwizard]

Here is one idea that I have been working on. I can place 9 connection pads on the LED board (8 positive, 1 Gnd) in the center of the board. By doing so you only need 4 wires on either side of the connector to feed 8 LED's, plus common wire for Gnd. Using both sides of the board run one set of wires in the center to the 4th and 5th LED adjecent to the connector. On top copper run one wire from above and below the LED and on bottom copper run one wire for LED on top and common line below the LEDs. All the LED's are fed.

Following your suggestion of using a register, now add a tiny interface board with 6 wires in, 9 wires out that will plug in the LED board's 9-connector. 6 wires to the interface will allow (+,-, SI, SO, ck, Latch) feeding multiple boards (with 5 wires).

 

[Chipwizard]

Update: I used the TSSOP version of register directly on LED board. I was able to narrow the board down to minimum 0.315. That said the pads for register is a hair larger than 0.01. This board needs to be special manufactured with higher than normal cost since it is outside typical normal production parameters (not a at-home project).
I either have to use a plug-in interface or widen the board further.

 

[Mr RB]

It was a communications error on my part Chipwizard. I assumed "thin 0.25 inch" was the HEIGHT of the PCB, where it seems you meant the WIDTH was 0.25 inches.

If you have to make such narrow width PCBs full of LEDs (do they go in a tube??), maybe you could make the LEDs surface mount, and TSSOP shift register (or whatever small package you like) on the rear of the PCB. A through-hole double sided board will allow some vias under the IC and a few leadouts each side but it's still going to be tight as the IC requires 3 control wires, 2 power wires and 8 output wires...

Why on earth do you need to make matrixed individually-controllable LEDs in a tiny "stick" form factor??

A lot of design problems are best fixed at the beginning... What are the LEDs displaying, how many LEDs total are required, what is the total overall size of the assembly and and what might be the best way to do that task?

 

[Chipwizard]

It was a communications error on my part Chipwizard. I assumed "thin 0.25 inch" was the HEIGHT of the PCB, where it seems you meant the WIDTH was 0.25 inches.

If you have to make such narrow width PCBs full of LEDs (do they go in a tube??), maybe you could make the LEDs surface mount, and TSSOP shift register (or whatever small package you like) on the rear of the PCB. A through-hole double sided board will allow some vias under the IC and a few leadouts each side but it's still going to be tight as the IC requires 3 control wires, 2 power wires and 8 output wires...

Why on earth do you need to make matrixed individually-controllable LEDs in a tiny "stick" form factor??

A lot of design problems are best fixed at the beginning... What are the LEDs displaying, how many LEDs total are required, what is the total overall size of the assembly and and what might be the best way to do that task?

Let me put up the specs as you suggested and we go from there:
PCB Size: 0.25" x 13"
Board Type (Thickness): FR4 Double Sided Clad 14 mil thick (0.014) for flexibility
LED Type: SMD or DIP with each LED have minimum 20,000 mcd 100-140 degree view angle. Not sure if there is such a powerful smd LED available.
Power source: is 12Vdc. I was going to use 3 LED's and a resistor to run on 12v per channel. Willing to compromise to 5V feed and reduce to one LED and resistor per channel as long as the current per channel allows use of LEDs brighter than 20,000mcd. (I found some 4.2v, 100ma, 100Lumens, 100degrees in China)
The boards to eventually become water proof. (Shrink tube, epoxy resin, or silicone)
Number of Channels per board: Would like 8, could go for more or less channels
Wiring Feed to PCB: Minimum wiring. If there is compromise of current+brightness of LED vs. number of wires to PCB, I would increase number of wires without compromising on brightness.
Hope this helps...

 

[Chipwizard]

First Prototype Pictures...Board 0.25x13. An 8 channels (8 Cathods, 1 Anode) 12V per channel, 20,000mcd per LED (60,000mcd per channel)
Pictures show the board next to 555 timer chip in a SO package, an 18 DIL PIC, and a US-dime for size reference.

 

[Mr RB]

Maybe in that form factor you coul dput a PIC or other micro on board, and then you only need 2 power wires and one communications wire to control them?

 

[Chipwizard]

Maybe in that form factor you coul dput a PIC or other micro on board, and then you only need 2 power wires and one communications wire to control them?

Point well taken. However placing any chip (PIC, LED-Driver, Register) will limit both current and voltage supply to each channel. In the prototype configuration current per channel is about 30mA per these LED specs, however voltage per set of 3 LED+resistor is 12v. With added on-board chip then additional transistors (N-MOS) is required to handle 12V. On a 0.25" wide board as you can see PCB real estate is tight.

Also adding a chip (PIC specifically) will limit the affects only to this board. With the prototype board I am using 8 cathod and one positive feed. You can run 8 wires (CAT5) parallel from LED board to LED board and the positive feeds can be multiplexed. Example using PIC16F628A: 8 wires from PortB as your sink lines (Thru LED Driver Chip) common to all boards, and use 6 lines from PortA to drive 6 different boards(LED Driver or NMOS). Multiplexing 8x6, 48 channels on 6 boards. The 8-bit-base configuration will also allow the controller board to PWM, hand shake with 8-bit registers, and so on.

The 6 drive channels can also be expanded to 12 (50% duty), 18 (33% duty), 24 (25% duty) and so on driving additional boards. (expanding thru time sharing)

I still like your original idea of an on-board latch register. I am going to toy around with that a bit more since it allows for expansion from board to board.