Greetings,
Does anyone know where to acquire small (8mm ish) round MCPCBs for an E21A?
When I need a small MCPCB in this sort of footprint, I just grab a generic 2835 board like this and cut off a piece with a sidecutter. Thermals and power handling will be crap though.
I assume this is for your Maglite mod–why E21A? There are so many LEDs that are easier to work with and perform better in every way. Try dedomed 519A: standard footprint, much better thermals, better power handling, much-reduced angular tint shift in a reflector…
1 Thank
Unfortunately it needs to be about 2mm square in order to work with the stock optic.
Also, the 519a does not get dim enough (according to datasheet) for my needs. I’m hoping for somewhere around 60lm when the battery is at 3.9V.
I was originally going to use the 519.
Could you walk me through how you arrived at this conclusion? Something doesn’t quite make sense.
I’m assuming a direct-drive setup, since with a regulated driver you can probably get a mode far dimmer than 60lm. Under direct drive, the E21A datasheet indicates 2000mA at 3.3V and stops there. This means at 3.9V, you will instantly fry the emitter.
So with direct-drive, no emitter can really get you 60lm at 3.9V. If you want to put in a resistor, you can simply calculate the right value to use given any LED, so 519A won’t be a problem.
I see. Remind me, are you modding an incan Maglite? If so, the stock reflector will be completely unfocused with an LED, since the base of the reflector does not coincide with the focal plane. Once you ream it for correct focus, I suspect an 519A might do, Plus, the emitting surface of the 519A is indeed 2mm by 2mm, and you can have the reflector rest on top of that–the reflector is not obligated to envelop the entire footprint.
EDIT: I spotted your other thread. Could you measure the dimensions of the stock LED? It doesn’t look like the usual rectangular Luxeon in other solitaires, and vaguely resembles a 3535 emitter, which is the same footprint of the 519A.
Also, if your goal is still improving runtime, I still suggest NiMH chemistry as a simple solution.
1 Thank
519a 80CRI nominal light output is 624lm.
Using this graph, we can conclude that the lowest forward current we can achieve with a bit of margin is about 250ma.
From this graph, we can conclude that the emitter will produce about 20% output at 250ma. Thus, 624*0.2=124lm. We can conclude that 124lm is the lowest output we can reasonably achieve with a resistor “driven” 519a.
On the other hand, the nominal light output of an 80CRI E21A is 292lm
From this graph, we can conclude that the lowest forward current we can achieve with some spare margin is about 125ma
From this graph, we can conclude that that the brightness is about 25% of nominal at 125ma, so .25*292=73lm
2mm*2mm (ish - there’s a glob of dome material that spills over the edges of the chip).
I believe it is a Luxeon C.
Just doing some quick math:
Luxeon C claims 116lm nominal output. Maglite claims 47lm output, so a bit less than half power.
On this graph, 100ma corresponds to a bit less than half power.
At 100ma, voltage is about 2.67v. So 0.1*2.67=0.267w, which is assuming a perfectly efficient driver. Maglite claims 1:45 (let’s be generous and say 2hr) of runtime on a aaa alkaleak, which my testing shows is able to deliver about 0.9wh of power. So the actual power consumption is about 0.45w.
If we assume a battery voltage of 3.7v, the E21A, which draws 0.125a will consume 0.46w, which let’s just say is 0.5w.
Eneloop claims 800mah capacity, so 1.2*0.8=0.96 (let’s just say 1) wh, which coresponds to just over 2 hours of runtime. Just over 3.5hr assuming the driver is (it’s not) perfectly efficient.
Vapcell claims 450mah of capacity, so 3.6*.45=1.62wh. With the resistor driver, the light should run for a bit over 3 hours.
I did all this math a while ago with data from a Luxeon Rebel (the emitter that earlier Mag Solitaires used ??), and found the efficiency was really dissapointing. After doing the math with the Luxeon rebel (150ma vs 100ma on Luxeon C), I decided I should use a E21a, but doing the math this time, it seems a resistor driven Luxeon C would get better runtime whith less hassle.
However, I still (because I’m crazy and stupid) want to make one with a Nichia E21A. Now my justification is better light quality (everything looks kind of washed out and monochrome with the stock emmitter in the Mag Solitaire. I also wouldn’t mind the light being a bit warmer. Maybe 4000k. Although it might be easier to reflow a different Luxeon C (available on Digikey) instead of moving to an E21A.
1 Thank
That’s not what the graph says. LEDs don’t have a “minimum” current–you can drive it as dimly as you please. Almost every modern driver (e.g., Convoy’s entire lineup) has a moonlight mode that delivers only a few mA to a 519A, and outputs on the order of 1-5lm.
You can try this out for yourself. Put, say, a 1kOhm resistor in series with an LED, and direct-drive it with a Li-ion cell. The LED receives a current no greater than 4.2V/1000Ohm = 4.2mA.
I would suggest redoing all the math with an unspecified resistor value, and then plug in various resistances to see what gives you the drive current you want.
Thanks, that emitter sounds right. I would check the reflector bottom hole just to see if a larger 3535 emitter still might clear the opening. Not the entire footprint, but just the light-emitting surface, which is 2mmx2mm for a dedomed 519A.
I see, wanting better color quality makes sense.
A dedomed 519a should fit then… And supposedly 519a’s can be dedomed with q-tip.
I’m going to redo the math.
1 Thank
I just discovered that constant current LED drivers exist, are dirt cheap, and can be really tiny.
Challenge is, I can’t seem to find one that cuts off at 3v and puts out 100ma.
Does anyone know where to look? I’ve tried digikey and mouser.n
Not an expert on drivers, but some drivers can be modified for more/less output current by swapping a sense resistor, so perhaps it suffices to look for drivers with 3V cutoff and try to understand its composition to see if the output can be adjusted.
Ideally you should look for a buck driver. Technically linear drivers are also “constant current”, but are about as efficient as driving with a simple resistor.
If the goal is long runtimes, I still maintain that NiMH chemistry offers an improvement of 2-3x over alkalines, and that there is no need to complicate things going to Li-ion. If the goal is higher output, I doubt the solitaire has a good enough thermal path for it.