I've never used LEDs before, and now I might be in trouble.
I want four LEDs controlled together, yes, they are colons. I expected to use the upper arrangement with at least 150ohms, thinking it was the most obvious. I checked on a website, Amplified Parts, and they came up with the bottom diagram and taking 80mA.
Why would they do that? Is the upper diagram legit?
.... but wrong...
When connecting LEDs in parallel, each branch MUST has it's own resistor.
Therefore the lower diagram is better than upper, but resistor values are incorrect. If you calculated R value for upper schematic as 150 Ohm, each resistors in the second scheme must be 300 Ohm
It can be. Many LED displays have that configuration internally. But in that case, the factory can utilize parts all from the same batch, have closely similar characteristics. That permits a fairly acceptable impedance mismatch, so pretty even current in both branches.
Trying to do it with random parts, you would be lucky to obtain a match.
Disclaimer:
not quite my area of expertise.
You will also have to check the datasheet. If your LEDs are e.g. blue, the forward voltage is around 3.3V for each; for two in series, that would make 6.6V and that is something that an Arduino can not provide. If your Arduino is a 3.3V model, you will probably even battle with a single blue LED.
Thank you, Gentlemen,
The LEDs are plain-vanilla red ones. I didn't see the mismatch issue coming, I guess I will have to spend a bit of time with a breadboard rig before I get a new PCB. This project is for use in the field at night and all the LEDs need to be kept pretty dim, so maybe that will help.
Why can't you just use the second (lower) circuit, which doesn't have this problem at all?
At what voltage ?
This is one of those times when you realise you might have a problem after you receive the PCB but, while the board is as densely populated as you can get with the components to hand, I have actually found I can get one last resistor in, so I am now looking to order a new one.
There are also (inevitably) other cosmetic reasons for revising the board.
Unfortunately the current versus voltage (diode) curve exponentially increases, so small mismatches can have catastrophic consequences, unless suppressed by a current limiting resistor.
Don't see why if the LEDs were all the same, one resistor should suffice.
Just need to be a higher wattage.
see post #3
That´s very simple.
Test it by choising different resitors from high values to lower values and learn.
5 volts
Yup, like I said, don't see why there would be a problem with one resistor.
That's your issue.
Take a 1k resistor and check their forward voltage individually. Then take them in pairs or 4 in parallel and check that they all have the same brightness.
plain vanilla red leds, pretty dim
Go for 5 .. 8mA or even less current on each branch, then they might all survive.
Else, if you run them at their spec limit and one branch fails, the other one will fail immediately too.
(Not sure it that makes the problem worse, though)
In my experience, LEDs are not all that fussy.....give them 20mA and they'll work much the same......give them 5mA and your eyesight probably won't see the difference and the LEDs don't suffer in either case.
All a bit"over thinking" if you ask me.
Or in laymans terms, a load of bollocks.
In parallel, you're giving them a voltage not a current, the current limiting is aggregate. It's not quite the same. I don't think it's overthinking, it's a well known problem that is typically solved by matching components. But it occurred to me, that running them well below nominal rated current would make it safer. Maybe to a point where it's completely safe, as in this case it is thermal runaway that is the risk.
OK I have now found a 220 Ohm trimpot and some extra-long header pins to breadboard this. I want the display big but dim, for use outside at night without glasses. Battery power, so I'm glad to hear I should be able to get down to 5mA, and maybe less. So what I'm looking at for starters is four LEDs in series/parallel with a single 100 Ohm resistor, and Arduino's 5v will cover it. If a LED burns out, that leaves me with two in series @ 10mA, i.e. 25% of the max current for the Arduino pin and still half the current typical for a red LED.
FWIW, I have found the site that started me on this in the first place.
Could be a bad choice....... trimpots are only rated at possibly 1/8w..... 
LOL, it's big and ancient with a broad track. Its pins are thick. Probably made before 1/8w was heard of. And every ckt I have seen says 1/8w suffices.