led, surface mount resistors.

Also, you say the resistors do need clearance. Is using heat shrink tubing not advisable then?

Again, welcome to design.... Does the heatshrink conduct as good or better... etc? It'll probably be fine. Or maybe heatshrink isn't needed. I often don't for grounded ends, or resistor junctions, or resistor-LED junctions - but that is based on those junctions not causing damage to other circuitry (ie, low or no current through resistors, LED or gnd) as opposed to OTHER conductors hitting them (and thus doing damage). But I have learned that it doesn't matter that nothing can ever hit an exposed conductor because eventually something will. (I should have learnt from telcos - even when their DC bussbars go through inaccessible areas, the are still covered with insulation. Not that their bussbars are fused, but often neither are our exposed circuits in terms of NOT damaging surrounding circuitry.) Of late, I have tended to wrap (heatshrink) whether needed or not, though often this had been to increase joint strength for flying leads etc. Besides, if your resistors get too hot, you'll probably smell the heatshrink - ie, get a warning. And if they do overheat, they should go higher resistance. If they do blow, they should blow open - ie, a circuit failure but non-destructive (other than the resistor, and the burning Aurora (Borealis?)) The 560R (1/2W, else 1/4W) should be fine. That should give max intensity and be ok for a supply up to 14.8V shown before. If your system goes higher than 14.8V (as some do) - but at the LED source - and if those LEDs thus destruct, you can increase the resistor. 680R is the next size, then 820R. That means lower intensity but it may still be sufficient for your needs. You might find you want bigger resistors anyhow if they are too bright. (And at 680R and above, 1/4W should be fine: P = iiR = i x i x R; or (VxV)/R.) Post-Edit: Most readers should stop here... The rest has little value-add. Maybe. Funny how this thread reminds me so much of another..... The other thread gave the following link for these OSRAM HyperPointLEDs (pdf; 643kB). I liked the datasheet because it showed so many aspects of LED behavior. (It also blew several of the thread's arguments out of the water. I like people that supply their own rope!) A lot of the info is batch & physical stuff, but some diagrams illustrate color, intensity, current, voltage etc shifts with temperature/current etc. Its CIE "color" diagram may be a bit confusing. Essentially it is a 3D "color map" drawn in 2 dimensions, but essentially it covers all visible colors and assigns coordinates (u,v or Cx,Cy) so that any color can be defined. (Manufacturers and Standards use them - eg, one Standard (ISO-whatever part 17?) states that for colour redundancy (discrimination), the different colors used must be separated by (u,v) > 0.2 (or something similar...) - it's all fabulous stuff except when moron managers start specifying those standards for OH&S reasons yet have no idea of the implications nor implementation, but I'm not one to whinge!) Oops - I digressed. Anyhow, flick through the pdf and look at some of the graphs. In many ways it shows why approximations are ok, or aka near enough is close enough (aka some implement extreme pedantics for no reason, but at the expense of reliability and serviceablity). But again - try/test and see. If >560R is okay, use that for more margin. (ie, 680R, 820R, 1k etc) Do they work okay under dimming? Feel the heat from the resistors. [Note that a LED-free resistor on 14.4V dissipates nearly 50% more power than at 12V - ie, P in a resistor = VxV/R, so P is proportional to voltage-squared. P@14.4 compared to 12 = (14.4 x 14.4)/(12 x 12) = 1.44 = 44% more heat. That's a principle I've tried to knock into a few local audiophiles with under-rated alternators and poor SMPS (dc-dc converters) in their amps - ie, for normal amps (45-200W per channel), the amp at 14.4V will put out 44% more power than at 12V (usually more, but it all depends...).] Isn't it good that with all the things that can go wrong, you can get a $15 replacement? That's about $985 to spend on other projects - maybe a holographic heads up display (direct projection - not crappy 3D windscreen stuff) - then you scrap all illumination etc except for the projector (and I prefer synaptic injection rather than retinal - electro-mag is still more efficient than optic for modern transducers). But let's make sure you close this project first. Yep - I definitely need a cool soak soon!

What can I say other than thanks for the complete and thorough explanation? It all makes sense now. So, I think I have a set game plan. My charging system in my vehicle operates normally between 14.8 & 15.3 depending on loads (ac, etc). So, I will use that highest value to calculate for the resistor. Overkill? As you said I may find 560R is too bright and decide to use a higher value resistor. I will keep you updated once I receive all my shipments. Thanks again!

Your welcome. Alas I may have overkilled my replies, but it seems to have been worthwhile. As to your design overkill (using the highest voltage) - that's not overkill but good design practice. IE - do worst-case calculations & design. The do "nominal" (expected) design. Usually thanks to "preferred values" and other variances and margins, they will align fairly well. It's only when there is a borderline - eg, if 560R is close to 1/4W should I use 1/2W. Or if 560R may allow >20mA should I use 680R to allow for 15.x volts? In this case they both suggest 680R as a nice solution - less power and current! But will that drop intensity too much? When I have borderline issues, I then have to decide what is to give. Stick to 560R 1/4W for luminosity and small size and risk that >14.9V doesn't occur (long enough) etc? And that reminds me - a 560R resistor will not be 560.0R etc. These days 1% tolerance is common so it's 555R to 565R. But 5% used to be the norm - ie, 532R - 588R. (Remember I calculated 550R, not 560?) That's what makes a joke of pedantics. And you may have seen the LED electrical variations? And LEDs from the same die (crystal whatever) have differing Vf (forward voltage drops). Anyhow, no such thing as overkill in terms of finding out the worst-case scenario. Better to know that it will or might blow up than not to have considered it. The overkill instead may come from using "exact" figures when rounding suffices. EG - 3.4 or 3.6V LED? 12.7V or 14.4V supply? Burger it - call it 10V for the resistor - that suits 20mA. Or a 120R resistor - lets make it 12V across it instead of 10V. That latter seems contradictory to using "worst case" etc, but there's a difference between "getting a feel" for the design (ie, an idea of how much power is needed, how hot, wire thickness etc) and getting the "whole design" with a ballpark size & component value, and then going through with the "real" figures, the real specs & temperatures or voltages etc. I think I'm stating the obvious to anyone with good experience in any art, I mean, science. Most will have their rough sizing or intuition etc. I once sized concrete costs for a floor: 6x8=~50, thickness of ? means ~1/15th so 50/15 ~60/15 = 4 cubic units at ~$100/unit ~$400 which is no problem: irrelevant. My techHead colleague returned after several minutes having calculated $345.67. They were not happy when I said "fine - it's way under $1,000" (the budget). But they insisted on the accuracy - else were unable to do simple estimates - despite knowing the budget and measurements etc. If I had have estimated closer to $1,000, I too would have run for the calculator - unless I knew that rounding & estimated were generally pessimistic - ie, bad or worst case. But I still don't know how many bags of sugar we needed though. Alas, another ramble - so boring for those log standing gifted contributors hereon. (That's my technique to get away with errors and ludicrous statements. They never catch me.)

Giving an update! The project HVAC project has been completed. I did stick with the factory SMD resistors for the 5mm 3.4v 20mA LEDs I used. Turned out very nice and I am happy with the results! Here is one snapshot.

Thanks for all the help!! The DIY twist-locks turned out really well as well.