Great little 1W luxeon driver circuit

[andrewwynn]

after about 20 hrs of scouring the web for a pre-fab decent value tiny driver for a luxeon i found this driver

It took me a long time to get it because i ordered it within a couple days of them posting it to their site, and there was some quality issues they had to re-build some.. then.. much to my shagrin.. when they mailed them to me, they never came.. a month later, when i contacted them they were amazing.. they 2-dayed me a replacement set.

Well i didn't have any unused luxeons at the time so the drivers sat for about two months.. then i got my new luxeons and whatever 'very safe place' i chose to keep the drivers had slipped my mind.

'fast forward to today'... found the drivers tested the drivers love the drivers.

holy smoke i love 'em... i have one complaint.. they took the worlds smallest '1W luxeon driver' and made it about the 4th smallest by potting it in a pretty large block of epoxy.. i need to shave it down to fit it in my device.

In any event.. testing so far showed this:

mind you.. this is a '1W 3V' driver... it was also the only one i found that can handle MORE than 3V.. i.e. 3.7 of a LiON battery or 4.5 of 3 alkalines.

with the 3.7V LiON cell (not fresh charge.. oops).. I got 1.9W output from the 1W driver.. efficiency wasn't much to write home about : 67%

with a single AA.. it output about 1/2W... beeautiful... efficiency probably similar...

with TWO AAS however damn this thing shines... literally.. 1.5W and 90% efficient! it's the perfect driver for 2-up AAs

with TWO Ns... it output about 6 tenths of a watt.. amazing what sucking that much juice out of poor N battery (not even alkaline) will do to the voltage.

I haven't had a chance to test with AAAs but i have one of the new lithium energizers in my arc aaa and will test it soon and post..

I should mention.. i have a 'j' binned lux3.. so this may be why i'm getting more power out than they advertise.. it's advertised to output 350mA at 3.75 which is more than a watt, but have to say, extremely satisfied with this driver.

All that said... i'd have to pair up two to get full-power into a lux3, but that's still only $20.. it's quite a nice driver for the smaller light builds like mag2AA lights and such.

awr

 

[shiftd]

BTW, micropuxes can also accept input above 3V (i ran it with 4 rechargeable once and it survived, though heaven knows what the efficiency is).
If you want to run lux3 at full power, may i suggest dat's BB's? they can go up to 1000 mA with no problem, though eff might suffer at high currents. However, if you are going with 700 mA, it is still cheaper than the $20 of the mini-circuits.

i have a q though. If it is given 4.5V input, and lux Vf is about 3.7V, does it act as a step down or more like linear regulator? or even, perhaps it is a simple step up that the current will go up more than 350 mA?

 

[LED_ASAP]

I am curious if they have other components beside the inductor on the back side of the PCB. The front side is missing the input cap. If there is no input cap on the back side, then you will need to re-do the efficiency measurement with a giant bypass cap, or use a sens resistor plus a low-pass RC network. The problem of measuring pulsed current with a DMM is that your reading could be way off.

A ZLT unit is impossible to get to 95% efficiency range---there will always be 0.3V loss from the Shottky diode, which is 10% off already. What about the ESR of the inductor and the capacitor? What about the Vsat of the transistor? You can pick the best components on the world, and still won't get the claimed 95% at 350mA output. All confirmed >90% efficiency boost ICs use synchronous rectifier with NMOS/PMOS pairs, which has milliohm-range R(on). The components on the front side of their PCB really look like the old FMM617/ZHSC1000 combo, and ~70% efficiency is just about what one would expect from them.

The 4.5V limit is also an exaggerated claim. If a 3.7V Li-Ion is already giving off 1.9W, you can guess what kind of power the poor LS will get from a 4.5V source. It is going DD under those conditions---it's not regulating any more.

 

[andrewwynn]

first off... some more power measurements with different batteries:

2N -> .539W (crappy 'heavy duty' batts.. not even alkaline)
1AA -> .491W
LiON -> 1.9W (camera battery.. 1AH)
2AA -> 1.43W
1D -> .443W
1AAAA -> .335W (the most intriguing to me)
1AAA -> .419 (also intriguing).. about 130mA to emitter.
1AAA -> .403W (this was the new energizer lithium... the rayovac alk out did it)
2AAA... oops i forgot to write that one down.. but it was about 1.4W I remember it being almost identical to the 2AA..

Ok.. now on to answering questions:

I'm working with a space that is roughly 8mm diameter, 7mm deep.. so the 'off the shelf' typical drivers that are 14mm diameter just won't cut it.. the other light i'm working on has a space about 9mm x 30mm.. much more promise for the bigger fancier driver, but still obviously a no-go on any of the other drivers out there.. the 4x 8mm of this driver is a key element to it's charm.

This driver for me is to run 100-400mA from a 3V source.. or possibly a LiON cell.. however wasn't fond of using 600mA at 3.7V to operate a 3.15V Vf at 400 or 500mA... i'd rather DD and get a blinding bright light than just heat up some ckt with something on the order of a WATT of wasted power.

I'm going to remove the potting on one of my drivers today and should be able to answer more questions about what's on the back of the board, etc.

I'm fairly confident in my meters ability to deal with the voltage ripples.. it's supposed to be good for 250Mhz true RMS... i have noticed that the value definitely wobbles around a mean.. and it's quite handy that the meter does averaging.

The best reading i've gotten so far is on the order of 90% which is what i get with my buck puck so that doesn't surprise me.. as far as the 95% claim.. no idea where they got that.. maybe they goofed in how they determined that, but i'm pretty confident that 90% is accurate... of course that is just at 'just the right' voltage etc.. all the voltages not near 3V were much lower like 70% efficiency for the likes of a single cell or a LiON cell.

Oh.. this driver obviously does not limit the power to 1W... with a LiON batt i was getting almost 2W out... good think i'm using a lux 3 for testing... it wasn't DD though.. it's doing some form of inefficient DC conversion (efficiencies were in the high 60s from a LiON batt).. however.. it did hold the power back to a mere 1.9W where DD was 5 with a fresh battery... and 'only' sucking 3W from the battery...so it is obviously doing some regulation... the efficiency would go up a lot as the battery drained.. so i could get a seriously higher runtime... it will be worthy of a test i think.. but 2W vs 5 to 3 is a lot dimmer, so my current test bed light will stay DD 'til the emitter is noticeably damaged then i'll replace it and put in a resistor or something.

I initially bought the converters to run a test light like the micromightys i built.. was planning to run 3AAAs and pair up the two i bought for 2W output.. but obviously i could get by with one for that power level.. or pair them up for 3W from a 3V source... hmmm /ubbthreads/images/graemlins/wink.gif

me thinks i will be trying a parallel-serial of aaas sometime soon... the best efficiency seemed to be around 3V.. and doubling up some AAAAs might give me an incredibly small light.

I did run it from a single AAAA and it was amazing.. i think that's my next toy.. i'll be making a single AAAA light.. it should be about the size of an AAA battery.. in-fact that could make a very nice case.. especially if i can manage to keep the label intact. it outputs about 1/3 a watt from a lux3 and it was absolutely gorgeous.. even with a cut-down S017XA reflector.

I guess some runtimes with some known batteries will show if the efficiency is being measured correctly or not.

 

[andrewwynn]

Ok.. I killed one of my drivers trying to remove the potting.. but i managed to do the second one a little more carefully... i did discover 'the hard way' there is a rather tall inductor on the back of the board.. 'the hard way' being one second i'm looking at a solid black piece of plastic.. and the next second i have a spool of wire falling out on the counter.

The good news... this little gem did exactly what it's supposed to do... i gutted my brother's infinity ultra and made and "infinity plus 3 (don't forget to check out the beamshots as well...

here's a pic:

and a beamshot:

The beamshots are of the 'infinity plus 3' and an arc aaa for comparison.

at the moment it's only outputting about 400mW (something in the order of 140mA from a single AA).. but we will be putting in a 3V lithium tomorrow and that'll boost output up to just about 1.5W.

Well.. other than the huge BOX they put this little gem in i'm very happy with the driver.. I don't know if i'd recommend using it other than with 1.5 through 3.0V solutions, but it works awesomely well with these...

I will be putting my last lux3 emitter into my peak2n with a lithium 3V battery very soon. it's solid brass so it'll hold a lot of heat.. no concern about it pumping out 1.5W other than it'll probably suck the battery dead in a couple hours.. though that will take months and months of intermittent use to have happen.. my intent for that light is to keep in the glovebox.

 

[shiftd]

Andrew, that circuit looks very similar to a micropuck, in term of size, that is.

 

[andrewwynn]

they are pretty similar in size.. but when i ordered the one i got.. i was 'buying' the one without the potting so it was suppose to be 4 x 8 x 6mm..that volume is 192mm cubed.. compared to the micro puck at just about 575... 12 x 8 x 7 mm or so... it's a really inky dinky ckt inside that relatively big blob of epoxy.

i shaved mine down to about 9mm diameter and 6mm tall before putting it into the cmg infinity mod i just did.. that is only 380mm cubed... which puts it at roughly 2/3 the size of the buck puck even inside what is left of the resin blob around it.

The blob it came in was about 12mm round and 9mm tall.. or about 1017mm cubed... you can imagine my frustration when i saw it!

in fact.. i just sent a rather direct note to them about it:

[ QUOTE ]
Hey... some positive and negative feedback for the product i bought from you... 1W LED driver.

1) it kicks butt... i can't wait to get some more
2) the potting around it sucks ***... what the hell

here's the thing.. i scoured the web for dozens of hours to find 'the perfect driver' and i find 'the worlds smallest'.. because the other 'small ones' like the micro puck (at about 600cubic millimeters was way too big for my project.

Well after a shipping oops lost my first shipment, you guys were very solid in sending replacements 2-day (i tell plenty of people.. including a post on CPF about that)..

Anyhow.. imagine the shock and horror when what is supposed to be 4 x 8 mm according to your ad.. and presumably maybe 6mm deep... ends up coming and being more like 12mm diameter and 9mm tall...

Well that's a huuuge problem... now what is claimed to be 192mm cubed is over 1000... that's a factor of 5 times as large as advertised.. and it's just about twice the size of the mircopuck (which in itself is too big for most small projects).

So.. in the process of trying to go by the inky dinky picture on the web to know where the circuit board is inside that huge mass of goo.. managed to cut into the ckt board on one of my drivers making it defunct.. fortunately with the 'mapping' learned from attempt one.. managed to avoid that on round two.. you can check out the project: http://rouse.com/flash just click on the 'infinity plus three' link at the top.

In any event i would absoutely LOVE to have more of these drivers but i think i'd rather build my own than have to dremel them back to the size they belong.. who do i have to sleep with to get the driver advertised on the website.. no potting just the ckt board.. i can pot it to fit in my design.. the huggarse plug is just no good.

The insides of the driver are top-notch.. i started a thread on cpf to talk about it and am quite awed and said so.. i warned people about the relatively large size, and mentioned that it cost me twice as much because one of the drivers was destroyed in attempting to get it remotely down to the size it's claimed to be.

So.. tell me who, when and where /ubbthreads/images/graemlins/wink.gif

I jest..

but seriously... a little help here.. make me some drivers without the crap on them and i'll buy a good handful, and i'm pretty sure i could mention the fix on cpf and get more people buying them.

ps.. make a 3W version !

pps. don't POT THAT ONE EITHER.

:-D

hey love that mag 3W emitter upgrade.. too bad no regulation... we lux freaks love regulation

thanks for your attention and for the previous great customer support.

[/ QUOTE ]

hope it gets some attention i want some more of these drivers they will be perfect for my AAA and AAAA mods.. but not if i have to cut off that ludicrous blob of resin first.

 

[shiftd]

LOL

i love your direct note to them /ubbthreads/images/graemlins/thumbsup.gif
especially this part:
"pps. don't POT THAT ONE EITHER."

yep, potting is the one thing i hate from buying the micropuck also. why did they increase the already small size? /ubbthreads/images/graemlins/rolleyes.gif

 

[andrewwynn]

the ans. to the question is two-fold..

1) makes it like 1000x more reliable and indestructible.

2) hides any 'top secret' designs in their driver.

I didn't see any warranty at all with the driver i bought, so #1 shouldn't applyl.

thanks for the note on 'don't pot that one either'.. i laughed out loud as i wrote it.. such a 'duh'.

-awr

 

[RobD]

I'm beginning to think the only way to get any efficiency is with an 'H' bridge when using Li-Ion batts.
A fully charged Li-ion is 4.2 volts if charged correctly. I'm working on a circuit to run constant current (330 ma) from 4.2 to 2.7 volts. At 2.7 volts the light shuts off to protect the batt.
I'm not using and diodes in the secondary side to keep efficiency up. (I'll try diodes to see how much it drops. The BAT46 and BAT49 have decent drops but I haven't pushed them under any load)
I'm not interested in any drive circuits with less than 88%, personally. I figure , might as well use a bulb. I should be able to approach that but we'll see. I haven't seen anything on the market that can run Li-Ions at high efficiency to a Star. So if some found something I'd love to hear about it.
The down side is the larger component count but I'll use SMT to keep size down.

 

[rayearth]

I'm also interested in building a circuit to drive a luxeon 3w with a li-ion battery. As you said, the problem is that the luxeon wants to be driven at about 3.5 v at 800 ma but the li-ion gives about 4.2 to about 2.5 v.
According to my limited research, a SEPIC circuit can function in a step up/down fashion. However, I can find no published numbers on efficiency. In addition, SMT dual winding inductors are a bit hard to find affordably.
I'm not an electronics expert ... I could use some input too.

 

[RobD]

You can get away with SEPIC on buck or boost but not both in the same circuit. The problem with up/down SEPIC is efficiency losses. Most cases expect about 75%, give or take.
The 'H' bridge will run boost with current sensing and limit so it always has high enough voltage to create the 3.4 volts or so for the 330ma.(4.2 volts= buck, below 3.4 boost) to drive the LED.
The key is current limiting which is the way to control the LED, not voltage. (I've found voltage will give different currents for different LED temps).
I run 77 cores from Amidon, not SMT and wind my own. There's considerable math to get the cores wound with just enough wire for max efficiency but you get a better circuit in the end.
If you just want to run a single 'AA' then run the Zetex 310 chips. You can get away with high eff. especially if you leave off the diode and cap.
I'll publish the circuit here or on my web site when it's done. I have a few configurations to go through but it should be up relatively soon.
Rob

 

[RobD]

330ma is for 1 watt. You can run your 800ma too.
Also don't run the Li-Ion down to 2.5 volts it's to low for good battery recovery. I run it down to 2.7 and then switch the circuit off with a 339 comp.
Rob