VanIsleDSM,
I really think that what you put together is really great. Good job with the construction and the execution!
However, I have checked through this thread several times before and I am a bit in disbelief of some of the numbers that you presented. When I see two of your figures, something goes into red alert in my brain. It bugged me for some time, but I did not know if I would sound like a jerk by bringing it up.
First of all, I believe that the total lumen output of this cluster is too high. Secondly, I feel the same about the efficiency numbers that you came up with (which gives you the high lumen figures). I believe that you just used the lumen figure that you got from the eBay seller that you got the 5-die Super Flux emitters from. First of all, it is very hard to trust some of the information presented on eBay about popular imported electronic goods. Typically, the unknown brand of through-hole LEDs that you get there (the typical 5mm, super-flux phirana style, and "super"10mm LEDs) are sold by salesmen or saleswomen, and not by engineers. They are certinly not sold on eBay by the original designers and electrical engineers. How credible are the eBay sellers? Sure most figures are based from factory specsheets, but still how do those figures compare in accuracy to those from major manufacturers like Cree, Seoul Semiconductor, Lumileds, etc?
As a eBay buyer, you often have to believe the specs that they give you, since it is usually the only source that it can be found. Some products can be researched for more info, but if you do not know the brand name of the LED, then you have to take what you see on eBay.
Anyway, the main reason that red flags went up was the lumen figure on the eBay LEDs. I think that you had it listed as 25lm on your original post at one time before you may have cleaned things up in a update. Still, 1000 / 38 = 26.3lm each. That seem rather high for a low-power, epoxy encased LED emitter. From memory, the through hole LEDs that you can get now do not have the highest lm/W efficiency compared to the latest generation XR-Es, Seoul P4s, Rebels, etc. Due to the lack of heatsinking, they degrade due to higher temperatures easily. Through hole LEDs used to be pretty efficient, but that was when comparing them to yesterday's high-power emitters (such as the Luxeon I, V, and then III) that operated at under 30lm/W at their rated currents.
Anyway, I also remembered reading through a thread called "White LED Lumen Testing" (created by jtr1962) a while back, and I read about many of the greatest through hole LEDs that were available (and still are, of course). Here is the link:
https://www.candlepowerforums.com/threads/89607
I end up running back in to that thread every now and then, and I re-read the old posts, as well as read the latest ones. When seeing your special 25lm super-flux LEDs from eBay, I immediately thought of this thread. Go ahead and have a look. You will see a few mediocre emitters, some not so good ones (the standard quality ones, I believe), then you see the cream of the crop LEDs. Read through the efficiency comparisons and see how lumen output of each compare. It is quite interesting!
--------------------------------------------------------------
Okay... I created a few tables to make my point clearer. I show pictures of the tables below.
What I did was look at the few best LEDs presented on the graphs found on jtr1962's thread, then I looked at the very worse (which you may say are your standard emitters, maybe). For the "best", I picked out the best 4-die LED, the best (and only) 2-die LED, and the best single die LED emitter (which was a premium part). For the "worst", I picked the worst single die LED and the worst 4-die LED on the graph. I found the luminous efficiencies each emitter when the die, or each die, was at 20mA. Your LEDs were rated for 100mA (20mA per parrallel configured die). Just for comparison, I did take the figures for when the multi-die LEDs were operating at a total of 20mA. This will show you how good the efficiency could be if every die was underdriven below specs. I also found the luminous flux of each (the lumen output) from one of the graphs.
Check it out in this picture. Download them if you would like:
Make sure everything looks fine. If something looks wrong or out of place, please comment on that. Thanks. Also, check out the notes at the bottom of the above picture.
With this data, we can clearly see that for a typical emitter part (like yours), even 75lm/W would be a stretch, unless you are really underdriving your emitters. Once things heat up, and as the LEDs get old (which the through hole emitters do at a faster rate than power LEDs), the luminous flux will decrease a bit.
Below, you will see where the luminous efficiency of the above parts will have a role in the total lumen output. You will also see, if you would be using these premium or typical LED dies, that the total wattage consumed by the LEDs, when expelled as heat, is pretty high! Let us construct a five-die LED from the above figures, then build a single 38 LED cluster like yours...
It seems like the 25lm LEDs you got are either some of the highest quality parts available, or the original documentation was not the most correct. I believe that a 600 to 700 lumen estimate for each of your clusters would be more realistic, but you should test them some how (or compare them with a known source) so there isn't any pure guessing. 1200-1400lm in a wide 100 degree beam is still very bright, but I have a feeling like someone is being mislead (you from the eBay sellers, or even the viewers of this thread). I hate to see anyone use information that is not the most accurate. I do not want to bring out the b and the s word out since you probably didn't know that the lumen figure couldn't 100% correct.
By the way, look at the output wattages (as heat) from the above table. 6W to 10W to dissipate in tiny leads and copper foil, acrylic plastic, and fiberglass circuit board seems risky to me! With enough surface area, and some slight airflow, I bet that the whole package can stay just "warm" and not hot. I worry if you will have these confined in a tiny plastic housing. If you have any questions about the formula I used, just asked. It is a common formula that many use here and there on the CPF to see what the overall efficiency of their white LED is. The 320lm/W was a general number thrown around by many that, I believe, represented the maximum theoretical efficiency of a cool white LED.
Oh yeah, if these LEDs are the "Best Hong Kong" LEDs, or like them, then look them up on jtr1962's thread. I seen some there, even if they are not the 5-die phirana versions. They should be the same low power dies used in yours.
I also wanted to say that "Power LEDs" would work as well, if not better. They are often more efficient than the through hole LEDs, even at 350mA, which can consume less power (overall) depending on the number of emitters used. Prices are pretty cheap from DX and KD (same range as buying 60 or so of the super-flux LEDs). Plus, color and lumen maintenance is much better with these power LEDs; they can handle higher temperatures safely, too! With more room for a thin aluminum plate instead of a circuit board, you can get enough material to spread out the heat and keep it cool enough for how long it will stay on. If you only need 600lm to 700lm to maybe compare to your current 38 LED cluster, then you will need even fewer, meaning far less heat than when having to produce 1000lm.
So, I do not want to sound like an *** by trying to raise a warning about your figures. I hope it doesn't seem that way by this long post and tables. I just wanted to make sure that I had all my facts semi-straight before presenting it (in as clear of fashion as possible -evident by the length of this post). Again, if anyone sees any error in my numbers or logic, or something is flat out wrong, please tell me. If you did have a little error in your numbers, and it is corrected, just ask and I can even clean this post up so it doe not look like an ugly scar.
--------------------------------------------------------------
Finally, with all of that in the open, I do have some questions. I know this thing is killer bright, even if it is mostly flood, but how does it compare to you existing reverse lights, or even your headlights with a diffusing paper over the fronts? Have you tried to compare the output to anything else, just to show its great magnitude of brightness? I do like the electonics that you used to make it fade and stuff, but is it a PWM circuit alone, or is there any current regulation?
Everything looks great, and keep up the good work. I hope that you post more pictures of the finished product, or when they are in use in a parking lot. Thanks for sharing!
-Tony
Note: I accidentally posted this into the wrong thread, so I corrected everything. I guess the thread names did look alike. It didn't help that I had several windows open.