i have had very good success with the 7135 drivers and red xpe emitter. i’m driving them with 4x7135 on copper sinkpads. and 2 or 3x7135 on aluminum stars. at 1.4A i’m getting about 2.7v.
Thanks for the input guys. I guess the 7135 drivers would work ok but there is a lot of excess voltage that just get turned to heat.
kevind43: Do you see any noticeable difference in brightness between the 3×7135 and 4×7135?
there is a fair amount of heat with the 4x7135 driver and the low Vf red emitter. i have been using aluminum stars with AA epoxy on solid brass pills, there is a measureable drop at start up then levels off, i can’t remember but it was around a 10 % drop in light output. i just finished a light driven at 1A on a copper sinkpad got a 14% gain in overall lux and no drop when compared to same emitter on aluminum board, at least i couldn’t measure the drop with the tools i have.
for example i have put the red xpe on aluminum boards in c8’s with a 3x7135 i was getting around 12k and very little heat. and with the 4x7135 driver about 15Kcd and good deal more heat, but manageable.
that’s what the diode is for, to take some of the load off the 7135 chips. Those cheap small buck drivers aren’t terribly efficient either, so I wouldn’t be too worried about the efficiency of one approach vs. the other. At least with a 7135 driver you have a huge amount of options and they’re very robust.
Yes I think I understand the principle, the schottky diode goes between battery+ and led+ ?
Might give it a try 
I don’t understand it (even though I just read the wiki page on schottky diodes), but I’d like to, since I’m sort of into red LEDs.
How does the diode lower voltage more efficiently than the driver, it just turns it into heat also, right?
it’s not a schottky diode, it’s just a plain vanilla diode, although I’m pretty hazy on the differences between the two. The important difference is that the plain diode drops more voltage than the schottky, which is what we want, even though that’s often a bad thing I’m told
The circuit is Battery>diode>LED>driver>battery. So all the diode does is increase the effective Vf of the LED, so there’s less voltage drop across the driver. It doesn’t affect efficiency in any way that I can understand, the main point is to reduce some of the burden on the 7135 chips. The higher the voltage drop and current draw per chip, the higher the power that needs to be dissipated as heat. As they get hotter they either get less efficient, shut off independently or die, none of which are good things. So all a diode does is add another place from which voltage can be burned off as heat, so there’s less to burn off in the driver.
You can do a similar thing with a resistor in series with the LED, but for my application the voltage drop across it would change too much with current, so getting the right resistance value for one extreme would make it inappropriate at the other.
These
http://www.ebay.com/itm/180847140820?ssPageName=STRK:MEWNX:IT&\_trksid=p3984.m1439.l2649
were what I bought. I have some spare if anyone wants one, although for a buck you could just get 10 of your own From what I remember these were what suited my voltage drop/ current rating/ size requirements at the time, although exactly how I came to that model I can’t quite remember. Here Driver Matching was the thread where I learned about diodes 
I took pictures (although not very informative ones) but my camera’s at home, so I can upload some tomorrow if that helps?
edit: the end with the band points towards the LED. Very easy to figure it out - wire it up the wrong way and the LED won’t light
So, ideally, the diode would be attached to a heat sink?
(Watching this because I know amber has the same issues as red — and I’m very fond of amber LED lights — been hoping for something informative about drivers. Thanks for asking)
ideally, but I don’t think it’s necessary at the kind of currents you’re likely to be driving red/amber LEDs at. Even at 1A you’re only looking at 0.5W at most. I JB welded mine to the heatsink, but that’s more because I didn’t want it rattling round and shorting against something by accident than any other reason.
i think if your going drive the red xpe at around 1A there wont be any real problem with the heat and the 7135’s. so far i have not had any(knock on wood) failures. most of the builds i’ve done with 1.4A drivers run continuously for 1 hr-three to four times a night 3 or four days a week.
looked at the data sheets for amber xpe’s. they are very sensitive to heat, so they will need very good heatsinking or not much overdriving.
that makes sense, especially the part about efficiency…thanks
I have used red XPEs with 3x7135s with no issues… I am planning on building a torch with a XPE (on Sinkpad) at 2 amps and unless I can find a more efficient driver then I will be using 7135s.
If I get around to doing the above tonight (test run with a C8 host) I will let you know how I get on.
XPE red foward voltage looks like it goes up about .25V per 500mA (from 2V / 300mA), so at 2 amps I am guessing the forward voltage will be very close to 2.9V which is only .3 or so volts difference from a XP-G2 running at the same current.
I don’t have personal experience with driver failures (not even sure I’ve even heard of any) and this is my first light with a red LED, so this was just what PilotPTK advised me to do. I had the space and the parts were cheap, so I figured, what the hell, I might was well do it
I measured the Vf of my parallel triple and it oscillated between 1.8V at 115mA and 1.9V at 460mA per LED (it’s an on-and-flashing circuit), so the voltage drop at 1.4A overall would have been ~1.7V at full charge or ~550mW per chip. The AMC7135 datasheet doesn’t say anything about max power dissipation, only that above 700mW additional heatsinking is required to maintain chip temperature at ~120C (chip auto shuts off at 150C). Single LEDs will be less of an issue as their Vf will be higher for the same current, so the heat dissipation per chip will be less. There wasn’t anything about efficiency and temperature, only that the power consumed by the chip itself (Iq) rose from 0.17mA at 40C to 0.2mA at 100C, which shouldn’t affect output much
doesn’t care about low Vfs or even a Vf of zero but to reduce power lost in the driver keep your battery voltage just high enough to run the (linear) current driver. If you have a switch mode driver you will have minimum overall power loss and the supply voltage is not so critical.
The If determines the LED brightness, the If x Vf determines the LED heat dissipated and any battery voltage above the max value for Vf is power lost somewhere.
You can even out the Vf variation for each LED by putting several in series.
For two LEDs with a Vf tolerance of /- 0.3v the Root Sum Square tolerance of the combo is sqrt(2 x [0.3^2]) =/- 0.4v (instead of +/- 0.6v). On a percentage basis this is a 1/3 reduction.
Many LEDs in series should give you a more predictable total Vf for whatever forward current you select.
To figure 7135 dissipation I’d pick ICs with similar packaging and see if the other chip makers give you the theta [junction to case] thermal resistance. The other thermal resistances are predictable and so every term in the 7135 power dissipation formula will then be known or can be calculated.
I’ve seen it again and again; the less the buyer knows, the better it is for the vendor.
a constant current driver does care about LED Vf, but indirectly through the power it needs to dissipate (which varies by dV and I obviously). Try using an Lflex to drive a single XM-L from a 2S pack at 3A and see how much it doesn’t care about Vf Even driving 3 XM-L at 3A from a 3S pack is pushing it as you’re nearing 5W heat dissipation in the driver alone.
will allow you to measure the thermal resistance, junction to case (or to tab). Fasten one to an infinite heatsink (a copper pipe passing water at some known temp) put in one watt of power and wait a sufficient interval, probably a half hour.
If it doesn’t shut down increase the power by 50% and repeat. Then the resistance is (150C -water temp)/(watts in).
The sufficient interval can also be measured. For a tiny thermistor it’s about 15 minutes and has to do with the thermal time constant, which is published for some devices.
My wall thermometer, despite having a tiny sensor with low thermal mass, takes about 5 x 15 minutes depending on how close to the final temp you want to get. If the sensor responded too fast the furnace would short-cycle.
> a constant current driver
There’s the problem — is there a way to identify a driver, to know if it is a constant current driver?
A little update, I received these drivers today https://www.fasttech.com/products/1105800
Hooked it up to a red XP-E and used a bench supply as power source, but to get full output I had to use 4.6V 
Efficiency seemed pretty good but with just one Lion it’s pretty much useless unless there is some way to modify it?
I tried two drives and same result on both.
even though the efficiencies don’t look good on paper you should give the 7135 drivers a shot. i have built many, many lights using the 7135 driver and red xpe’s with very good results. i just tailor the amperage to the heat dissipation ability of the light i’m working on.
Yes I will use 7135 instead, 1050mA or maybe 1400mA.