16 RGB LED's in parallel controlled by atmega328 PWM pins for pc case mod. (Updated)

I don't see anything obviously wrong, but this is only my opinion with NO WARRANTY.

I don't see any reason to use +12V and a regulator: I'd go with +5V from the power supply.  Be sure to keep the 10 uF bulk bypass cap and add a 0.1 uF ceramic cap for the processor.

20 mA per LED seems like a lot these days.  I recommend that you try various resistor values starting with 1K to see how bright you really want it.  Also, different color LEDs typically have different voltage drops and different efficiencies, so you may need different resistor values to get the right color mixture.

I recommend prototyping on a solderless breadboard to get it working before making your final resistor selections and soldering it up.

The only comments I 'd make is that AREF isn't tied to 5v on most other boards.

It can be set under software so they usually keep it isolated.

You also seem to have no high freq filter caps (0.1uF) showing across the supply lines

R54-57 aren't required if you set the pullups to on.

Pinmode(xx, INPUT);

DigitalWrite(xx, HIGH);

will set the internal 20k resistors and achieve the same thing.

Depending on the gain (Hfe) of transistors Q4-6 you might want to drop the R51-53 down to 4k7 or 2k7 to ensure you fully saturate the C-E at 100% PMW.

And don't forget the series resistor on the RGB leds

Mark

Made some changes based on feedback, I still need to correctly calculate the values of R1, R2 and R3

Looking good.

I'd change the 10uF to 1000uF (or even larger) since the 5v is also supplying your RGB leds which are 320mAx3 (960mA)

(A rough rule of thumb is 1000uF per Amp)

For R1-3 you may wish to use a darlington transistor that is capable of 500mA ie MPSA13 or MPSA14 which have a gain of 10-20,000, then you can use a 10k resistor.

Try to keep your current paths seperate, ie the ground for Q1-3 should go to the supply, the ground for the ATmega 328 should go to the supply and both join at the supply point.

A bit like keeping heavy traffic from residential streets but joining them at the motorway on-ramp.

Your FTDI ground should go to the ATmega.

The idea is to keep any switching currents from influencing the processor.

Most of the ccts I've seen also have a 10k to +5v from the RST pin, and use a 0.1uF to DTR

Without it it can float, and possibly reset when you don't expect, or not always reset.

(Arduino Pro shown below)

Mark

Thanks very useful points, as i said i'm totally new to electronics so this project is more of a learning experience for me than just to create something that lights up. You said that the FTDI ground should go to the ATmega, i don't fully understand how you mean, wouldn't that mean the ATmega would be disconnected from the supply ground?

Here is what i have now,

I've Tried to keep current paths separate best as possible.

Changed the filter cap from 10uf to 1000uf.

The transistors switching the LED's will be MPSA14's (however couldn't find any Darlinton transistors in fritzing) and updated their resistors accordingly.

Changed the 22pf to a 0.1uf and added a 10k resistor to +5V on the DTR to reset line.

Also noticed in my last revision that instead of sending +5V out to the LED's i had the pin connected to ground

Oops

I posted and you updated while I was constructing the reply.

Normally a drawing doesn't match the physical layout, however this is a good exercise to get you thinking about avoiding the pitfalls.

As you become more familiar with electronics these become second nature ... but they still creep up behind you and slap you if you are sloppy.

If you build it very similar to your drawing (ie laying it out in the same manner), you should avoid any potential headaches.

one minor edit ... I'd make the 10k on the reset connect to the ATmega supply line, rather than the 5v to the leds.

Well done.

Now the code ....

Cheers

mark

sorry I made an edit while you were marking it.

mark

No worries. And thank you for your help, I really do appreciate it.

I love learning new things. This project is a way for me to learn something new and make something i want in the process, Its actually a good way of learning, your motivation to carry on comes from seeing your idea turn into an actual thing and the satisfaction you get from seeing your creation working at the end of it makes you want to go further, learn more and create more complex things.

There is probably hundreds of systems pre-built to do what i'm trying to make here, but by building it myself not only do i learn something, but at the end of it i can step back and say "i made that".

It was the same when i started learning Java, started with a small simple project, and worked my way up from there, did some dabbling with android apps just to learn something else. Wrote tons of programs most of them to serve a purpose, and aid in doing tasks i do regularly, most where a success, a few failed, but the fun is in trying.

But now down to the code for this, I'm looking forward to this part.

Also amended my schematic

The above attitude to learning will hold you in good stead here .......

But now down to the code for this, I'm looking forward to this part.

Plenty of comments so that six months later you (and any others) can understand why you did that bit that way....

Have a look at the example "Blinkwithoutdelay" in the IDE to use the inbuilt millis() timer, rather than a delay.

Don't forget to enable the pullups on the three button inputs, and I use a small delay (25mS) then check again to de-bounce the switches.

Cheers

Mark