Constant current driving RGB LEDs with common cathode/anode

[amaschas]

Hey all, I'm trying to layout a board for LED light that will consist of alternating rows of red, green and blue LEDs. My current layout is strings of 8 LEDs in series, with 3 RGB groups, for 72 LEDs total. For heat dissipation reasons, I'd like the final board to be aluminum, and to make that affordable, my board needs to be 1 layer. What I've discovered is, if I need to route both a positive and negative terminal for each string of LEDs, a 1 layer board is impossible if I have more than 2 RGB groups, so I've been exploring whether I can make my design work with a shared cathode or anode, but I'm also hoping to use constant-current drivers rather than series resistors to drive the LEDs, and I'm having a hard time understanding whether parallel circuits with a shared ground or voltage source can maintain different voltages. I appreciate any input on my plan or illumination of the issue. Thanks!

[greenpossum]

Draw us proposed circuit(s). Better than a thousand words.

[amaschas]

Sure! Here's my current circuit diagram and board layout for both the shared anode and 3 separate circuit designs. For the driver I was considering something like the TI LM3409, or something similar that could support shunt-FET PWM dimming. The diagram of the typical application indicates the LED is routed to ground, rather than an negative terminal, so I'm wondering if I can just run three of them in parallel.

[greenpossum]

Run three of what in parallel? Three LEDs or three CC sources? The former is not recommended due to variation between LED forward voltages, while the latter is perfectly ok, as the current will be three times of each CC source.

[amaschas]

Ah, I see what you're asking. In the attached diagram, my current approach is on top. On the left, there are three drivers, each with positive and negative outputs for each color of LED. This necessitates the approach on the right, with positive and negative pads for each color strand of LEDs. I'm hoping to use the approach on the bottom, which would allow me to use a common ground (or common source if that's better) for the three strands.

[Zoli]

There are two ways to solve your problem:
1. Route the return traces at the side of the board, one on the top, and two on the bottom; it won't be the nice RGB(anode)-RGB(cathode) output order, but it can be done easily, with no extra parts.
2. Use 0 SMT resistors(1206 as example), and route the traces under them; if the output order is important, this is the way to go.
And of course, you can use the combination of both...

[amaschas]

Oh, interesting, I actually saw 0 ohm resistors on a few LED array boards I've been studying, and I've been wondering what they were for. Now I realize they were used to help with routing! Thanks!

[greenpossum]

Ah, I see what you're asking. In the attached diagram, my current approach is on top. On the left, there are three drivers, each with positive and negative outputs for each color of LED. This necessitates the approach on the right, with positive and negative pads for each color strand of LEDs. I'm hoping to use the approach on the bottom, which would allow me to use a common ground (or common source if that's better) for the three strands.

(Attachment Link)

Sure, as long as the driver allows the return connection to be connected to ground. It's a bit misleading to call them "in parallel" as the top connection is not common. There is nothing magical about ground. It's just a very often used connection point.

[greenpossum]

Oh, interesting, I actually saw 0 ohm resistors on a few LED array boards I've been studying, and I've been wondering what they were for. Now I realize they were used to help with routing! Thanks!

They are basically jumpers that can be inserted by machine. Manually you can use a wire.