Page 1 of 3

RGB dashboard backlight, PCB's are in!

Posted: Mon Oct 01, 2012 10:07 pm
by ivan141
Here's a little project I've been working on for the last few months. This is quite likely the
only forum I visit where people have enough off an understanding of the subject to point
out any flaws.

It started out by trial fitting a couple 12mm RGB led's in my ae86 cluster and trying to see
if that would fly (it didn't, hot spots and not enough light).
Image
Image
Image

Disappointed with that result I figured I might as well dabble a bit in PCB design and just slap
some rgb led's on a pcb with some led-driver IC's. Not exactly rocket science, now is it?

Reality bites, as they say, and it ended up making my GF very cross with me some evenings,
sweating over some PCB routing conundrum or other. Anyway, after a few weeks of mulling
over what more I could do to improve the design, I decided that I wasnt going to come up with new
ideas. If you don't take action, nothing ever gets finished, so the gerbers have been sent off to china
(iteadstudio) for production last week (should be in by the end of the month I hope).

Here's the 'panelized' board I sent out.
Image

Here's holding my fingers crossed that I will actually get it to do something. The firmware shouldnt be
too challenging, apart from some elegant way to handle animated sequences.

Some general specs:
-Atmel ATMEGA328 in smd package for control (overkill, but I like some room for my not too optimised code).
-LM2576 5v 3A switching supply (I know you guys love linear, but I dont feel like melting my plastic dash).
-WS2803/2801 24 bit / rgb channel led drivers.
-Pixel precision on the tach side, slaved configuration on the speedo.
-Really small RGB led's (4) on a special needle board that slots in the place of the old light guides.
-Small PLCC RGB led boards for auxiliary lighting of pressure, temp and fuel gauge.
-FTDI based USB micro interface (in an impossible position, I know, it couldnt be helped).
-Optocoupler separated input circuit for tach signal (coil-), connected to an interrupt for a simple counting rpm measurement scheme.

Re: RGB dashboard backlight (disco disco huh huh)

Posted: Mon Oct 01, 2012 10:22 pm
by Fred
"-LM2576 5v 3A switching supply (I know you guys love linear, but I dont feel like melting my plastic dash)."

That's just me! Trying to be compact, and have switcher, and low noise analogues, in the same place = challenging, a challenge we don't have resources or need for right now. Fine for anything/everything else, though :-)

What's the goal? Just to look cool? Or to change colour for ambient temp? or for RPM like a giant shift light? or what? :-)

Fred.

Re: RGB dashboard backlight (disco disco huh huh)

Posted: Tue Oct 02, 2012 7:31 am
by DaWaN
Cool idea, I had the same idea some time ago and then basically thought: "It will take too much time"
I think you experienced that ;) Nice you kept going.

For the design I have some questions:
Why a 5V power supply ? A 3.3V power supply seems much more logical to me as the dissipation in the LED driver is much less
What LEDs and how much current are you pumping through them ?

Also: filling 32KB with code for the LEDs....I think it would easily fit within 2KB :) But it does not matter as these mcus are cheap anyway.

Mind sharing the EAGLE design files on GitHub ?

Re: RGB dashboard backlight (disco disco huh huh)

Posted: Tue Oct 02, 2012 8:02 am
by ivan141
The goal is to have a showcase of personal electronics prowess on board the car ;-).
Really a pointless thing to have, but it's educational and fun to do (so far).

Functionality I had in mind:
-Any combination of backlight with any colour of needle.
-RPM dependent animation:
Needle can change colour with rpm.
Possibly flash a colour on the needle for shiftlight purposes.
Theoretically I could leave all tach backlights off and turn them
on based on RPM, basically a swipe on behind the needle. Think it will be a bit gaudy though.
Adjustable redline, just turn on an extra red led if you want to lower it (not much room to go up on a 4age anyway).

@DaWan:
5V because: the forward voltage on these RGB leds is around the 3v range. 3.3v would be cutting it close methinks.
I'm using some generic PLCC RGB leds (no common anode/cathode), the design allows for 20mA, but I intend to
PWM drive everything since full brightness is likely not required (and the WS280x drivers are built for PWM).
About the 32kb: I've messed around with a data logger where my code size got a bit out of hand, so I thought
it best to give myself a wide berth there. Then again I am the first idiot I've seen so far to create a quaternion
based library for 3-axis accelerometer calibration.

I dont mind sharing the files (dunno about github yet), but the brd files are rather pointless if you dont have an
ae86 (just look at the awkward shapes).

Re: RGB dashboard backlight (disco disco huh huh)

Posted: Thu Oct 25, 2012 9:28 pm
by ivan141
My first prototype boards are in! I'm proud, excited and anxious at the same time because they still have to be soldered together.
What's even worse is that they have to work as intended, which is where my biggest fear lies. It's all straightforward electronics,
but I wouldn't look surprised if I made some bloopers in trace design, especially the parallel switched grounds for my speedo LED's
seem a prime candidate.

Front & Back:
Image

Detail of needle boards. The cobbled together logo came out nice in silkscreen. The holes line up perfectly like I hoped. Just looks like I will have to file out the middle hole a little so the needle stem can pass through it if necessary.
Image

I am on the lookout for a snackoven to do some reflow work on the bottom. I was foolish enough to copy a reference arduino design for the microcontroller, and now I'm stuck with ceramic resonators which cannot be soldered by hand (actually, they can, but it will look like shit). If all else fails I could fall back on the internal oscillator, no time critical business here.

Re: RGB dashboard backlight, PCB's are in!

Posted: Sun Nov 04, 2012 12:45 am
by ivan141
Well, after a lot of cursing and swearing getting anything to work on my first prototype, I finally figured out that the MCU is good, and have got some test code in there finally.

Image
Image
Image

A couple of observations:

-The power draw is quite big, I can see it has a slight dip when the LED's are at full brightness. I haven't even connected all boards yet so I'm going to have to measure how much the voltage drops and make sure it's not too much. I'm curious what's using all the power since all LED's together shouldn't use more than 1A (and the supply is supposedly good up to 3A).

-As a result of the power usage, the power regulator gets quite hot. Will need a heat sink attached to it when installed.

-I'm bummed about the 3 dollar FTDI chip being toasted, because I haven't got any external inputs at my disposal to change settings with while the board is running. The idea was to be able to do this using a USB cable, so I'm hoping I'll get the next one to work better. I think I killed the FT232RL by soldering too long, tough puppy to get on there with an unfit soldering tip.

http://www.youtube.com/watch?feature=pl ... ML40qJub9c

Re: RGB dashboard backlight, PCB's are in!

Posted: Sun Nov 04, 2012 2:19 am
by Fred
Cool! Don't feel bad about the FT232RL, I killed one too, and I didn't feel as though I'd been rough on it.

Re: RGB dashboard backlight, PCB's are in!

Posted: Sun Nov 04, 2012 7:46 pm
by ivan141
How'd you kill your FTDI? Curious to know what to look out for.
I had difficulty getting rid of some solder bridges on a few unconnected pins, and ended up cutting them off completely.
Think I did it in with that + way to much heat applied too long. Next one will have to pass through a pizza oven.
Also discovered a small design flaw of the arduino nano design I used as a base, the test pin is left floating while the datasheet
clearly specifies to tie it to ground for normal operation. Fixed that with a deliberate solder bridge, but no luck.
Mine lights up both the Tx and Rx debug leds the moment a cable is inserted. Apparently this is an indicator of premature death.

Not my only layout cockup, a couple of silkscreen texts ended up the wrong way round (not a big issue), but my needle lighting
boards had mirrored LED footprints, a result of not having tested the parts and working off the wrong datasheet, which I assumed
matched the ebay ordered (and datasheet-less) 1210 RGB leds. Luckily that's only a small board, which is cheap enough to make
(10 usd for 20 of the little fuckers).

For the rest, it's more things I can work around, but could be done better:
-Should have an external activation signal pin to detect when the lights are actualy turned on. Can work around this by having it always on,
or just use an external relay to do the same thing.
-Above-mentioned pin should be able to sense the dimming value of OEM rheostat (which is broken on my car anyway).
-An extra connector to expose a few more IO pins on the MCU wouldnt hurt (alternative methods to switch settings would be nice).
-Need to test my supply circuit to it's 3A limits on breadboard. With the LED's at full chat the supply voltage plummets to 3v!
I now use the LED's at about 40% PWM, and it's still bright enough, the supply seems to cope with that a lot better, but I'd like
to know where the weak link is so I can fix it (might be a must-have once all boards are running off the supply).

Re: RGB dashboard backlight, PCB's are in!

Posted: Sun Nov 04, 2012 8:11 pm
by Fred
Probably the same way. I didn't feel as though it should have died, though... seemed fragile? :-)

Re: RGB dashboard backlight, PCB's are in!

Posted: Tue Nov 06, 2012 12:42 pm
by masterkorp
This is really sweet, making it changing color with the rpm would be really cool.