A light difusser would look nice in there...

 

Actually it would be less wires since you could hook all of the LED legs together, but the main point is they would NOT be the same brightness.

Each LED has a "typpical" forward voltage (V_fw), but the true V_fw can vary by several tenths of a volt between individual packages. When you place several LEDs in parallel and run current through them, the LEDs will essentially "fight" to be the winning V_fw (there can only be one voltage value across a parallel circuit). This has the effect of sending more current through some LEDs and less through others.

In an extreme example (and entirely possible, as it happens to me...this is why I sort ALL of my LEDs by hand), let's say we have two LEDs in parallel, one with a V_fw=3.0V and the other with a V_fw=3.5V. Run 20 mA through them and they're both going to drag each other to an average of 3.25V (not exactly, but for these purposes let's just say that's what it does). The LED with the V_fw=3.0V is now being pulled to 3.25V, so it's going to shine quite brightly, but the LED with V_fd=3.5V is being dragged down to 3.25V, so it's going to give off a dull glow (IF it even start to shine at all).

RULE: Whenever putting LEDs in parallel, EVERY parallel string of LEDs should have its own resistor.

EXCEPTION: The LEDs are matched up perfectly, or you just don't care what they look like.

 

If the LEDs are of good quality (and similar spec), and manufactured to low tolerences there should not be an issue. The only issue i see is withe the gauge/length(s) of wire used on each parallel circuit, as this will add a small resistance to the circuit. The voltage drops you mention seem rather extreme.

 

Sorry to be the bearer of bad news, but even high quality LEDs just won't match up that well, especially between different batches. I'm not talking about just no-name LEDs, I'm talking about stuff from Phillips, Sharp, etc. These guys know how to make LEDs, but the processes just aren't accurate enough...take a look at just about any spec sheet for LEDs and you'll notice a significant voltage difference between the "Typical" and "Max." forward voltage...for red LEDs the difference may only be 0.1-0.2V, but for blue and wite LEDs the difference becomes much larger, say 0.5-0.6V. I've designed enough LED-heavy products around them to know you have to design your circuits with the wide range in mind.

For example, everyone like to pick up LEDs from superbrightleds.com. The RGB LEDs will have a wide range of forward voltages, so much so that 25% of them have a V_fw so drastically different from the other 75% as to drag the entire LED string down...one shines normally while the other 3 in a string are quite dim (at low current levels...things even out quite a bit as the current goes up). 25% is significant...things get better with the more reputable manufacturers, but 10% is more typical.

I'm not sure where you got the info about wire, but it's pretty far off base. If we were to assume 20 AWG wire (pretty common guage for wiring up LEDs), the resistance in a foot of it would be 10 milliohms (0.010 ohms). If your dropping resistor was 100 ohms, the 5% error given to most resistors would make a difference of +/- 5 ohms, about a 500 times SMALLER difference. So wire resistance is NOT an issue.

 

I can't seem to edit my previous post, so I'll have to clarify in this one...

What I meant to say was the WIRE resistance is about 500 times smaller than the error found in a typical resistor, so I would be more concerned about getting an accurate resistor before I worried about resistance of the wire.

 

I stand corrected.