I saw a few places that it uses the G2 dies or at least dies the size of G2’s?
The spec sheet shows it is much closer to a 4x XP-G2 then E2 as well. Particularly when you take into account that the G2 has a dome and the XHP35 does not it is virtually the same on the spec sheet.
What bins are you using? I was under the impression that the E4 was only released a few months ago?
E4 1D are my latest 4.
Frying the emitter at under 3A per die is more consistent with XP-E2 performance, XP-G2’s can take over 5A apiece.
Yeah, I am learning as I go with the buck driver design, it is actually quite simple and ingenues once you wrap your head around it. The design of the actual buck driver is not a big issue as long as I can reverse engineer the MTN MAX driver to see how it is laid out to make sure I am getting the schematic right.
The only real hang up at this point that I see is the parasitic drain issue. Although like I said above I think this could be solved by powering the MCU directly from the batteries and thus not using the buck converter 5V regulator.
This allows us turn the buck converter off when the light is off by a small FET, thus eliminating any drain in the system besides the MCU. It would have a free pin to control the FET since it would only need a single output to control the buck converter. The FET would simply shut off anytime the MCU went into sleep mode.
This is the basis I am working on at the moment.
Someone like DEL would be the best to get an opinion from, I know enough to understand the basics but not how everything interacts.
The MCU does output a PWM but it is only used to tell the buck converter how much current you want. The buck converter does all the actual work.
I did a bunch of reading yesterday but can’t find some of the links that were good. Here is one that explains things:
A bit long but good.
Basically a buck converter is a normal FET circuit with an inductor placed between the FET and LED. The inductor stores energy when the FET is on and slows the voltage rise, then when the FET turns off the reverse happens. It releases it’s stored energy and causes the voltage to drop slowly (in comparison to instantly stopping).
What the buck converter does is watch all of this happen at up to 2mhz+. It then times the FET to turn on so that voltage/current increases to 5% over your target level. It then turns the FET off and lets the inductor release the stored energy until voltage/current (remember they are directly connected with LED’s) is 5% less then target.
It then repeats this process endlessly giving you a regulated and constant current with a 10% ripple (aka, 5% over and 5% under target). Better buck converters can obviously be much better then this but the one we use has a 10% ripple.
That is the basics to how it works, just an FET with an inductor, technically you could make a buck driver by simply adding an inductor to a normal FET driver. but the circuitry to actually monitor things and ensure constant current is a bit more complicated, which is where the buck converter comes in.
The ripple is not a big deal in most cases but if you add a large enough capacitor to the output from the inductor you can further clean up the signal. The old Skyray king drivers had this setup.
Far as I can tell the limiting factor when it comes to current is the inductor, small ones just can’t handle high amps and the larger the inductor the better. Since we have a lot of room in the Q8 this should not be a problem. The one I linked above is rated for 18A and would easily fit in the Q8 assuming we have at least 6-8mm of height clearance, which we should.
Honestly the biggest issue I see for a buck driver in a flashlight is size, trying to fit it on a 17mm board and make it supply high currents would be silly hard and you would need to overdrive the inductor causing a lot of heat.
I think I am going to move ahead with basically using the MTN MAX driver with an FET to turn the buck converter off when sleeping and see what the schematic looks like.
Anyone that has something to say, please do say it!
I pretty much need to finish these driver projects this week as my free time will be cut drastically after that.
Interesting, no idea what to make of it, seems odd that acebeam would be lying by that much, CPF seems to think they are making more then 2600, I saw claims of 2800 lumens.
Either way, the exact number doesn’t really matter. it can be done for sure with an XHP50, maybe the MCPCB could be designed to allow for XM pads as well as XP? I know it was brought up but it was not selected for some reason.
Another key to a properly working buck driver is a high quality, fast switching FET.
It needs to switch as fast as possible to minimize heat build up. I know the SIR400’s work well for high currents and don’t generally have any issues but has anyone looked at how they handle high frequency?
We were just discussing the die size in another thread. It seems like the lateral size of the XHP35 is more consistent with it being 4 XPG2 sized dies.
How are the XHP35 emitters frying? Is it at the bond wires? From pictures it looks like all the current flows through one bond wire instead of two like on the XPG2.
Regarding the output differences between measurements and Acebeam’s claims: could some of that be OTF lumens vs LED lumens? I would guess there could be up to 20% loss in the flashlight.
Is there somewhere that has better measurements than that graphic? It doesn’t use any unit of measurement, and it is showing “optical source size” instead of actual size anyways.
I suppose that acebeam could be using emitter lumens, although I thought they are FL-1 ratings? Or are they one of those manufactured that just slap a number on the lights? I thought the little I heard about them said they were reasonably honest about the ratings.
The bond wires is a good point though, that could very well be the case, they do seem to be going at right about half the current of normal XP-G2’s
It would be easy for someone with a HI version to just measure it, but I don’t have one. The “optical source size” seems to be consistent with the lateral size, though, because it says the XPL HD is ~1.5x the size of the XPL HI, which is consistent with our observations that dedoming reduces apparent area by ~1.5^2=2.25. Also it agrees with our observations that the XPL HI and XPG2 with dome are about the same size.
I just measured an XHP35 HI best I could without damaging it.
Comes out to between 2.65-2.75mm or there abouts, hard to get an accurate reading without messing up the silicone and risking it burning later.
Ehm a lot is way over my head here…
I assume the focus is on the xhp35 because they could be reflowed on the Q8 ledboard as hoped for now?
With the 4 18650 cells a 2S2P configuration makes more sense to me then 4 series.
A boost from 3-8.4 to 12V should be given in this case but longer runtimes and a mod I would dare start 
Yes, the XHP35 is mostly due to it being a direct replacement, if the MCPCB was changed to the other design that was purposed that had pads for both XP and XM LED’s then the XHP50 would be the one to watch.
Far as 2s2p vs 4S. 4S makes more sense in this case IMHO.
Firstly the XHP35 must have 12v so it has to be 4s.
Secondly all the XHP series can be run in 4s mode or run in series to work in 4S mode.
Lastly higher voltage and lower current = less heat in the driver/components and thus less resistance. This allows you to be more efficient and push more power.
Why do you think that all the major manufactures go with the cells in series whenever possible? It is easier and better from an electrical point of view.
There is nothing that 2s2p offers that 4s doesn’t do better if using a buck driver in a light with 4 LED’s. No matter what the setup you can make it work better (aka, more efficient) with 4s. Now sure the changes are not much but there is simply nothing that 2s2p offers that is better.
Technically the best possible driver setup for the Q8 would be a buck driver running 4s and driving all 4 XP-L’s in series. This way each LED would get exactly the same current and it would take less total current flowing through the system.
Instead of ~16+ amps it would only need ~4 amps to reach the same wattage/lumens (4 amps x 16 volts is the same as 16 amps x 4 volts).
Because there would not be as many losses in the driver/wires/springs/everwhere you would actually be able to push a noticeable amount more current as well netting more power. The gains could be as much as half an amp per LED depending on how much resistance is in the system.
This is why all the big manufactured run the batteries in series, it simply works better.
[quote=Texas_Ace]
Can sign all posted.
Guys, do not exaggerate XHP35 possibilities. They are limited by thermal conductivity and thermal pad area. If both are same, led can not be able to provide more than XP-L. Just few percent more efficincy cause less losses cause lower current rates.
selfbuilt's review has measured lumens, and I trust his measured estimated lumens more than the manufacturers. His # is 2,275 lumens, here: http://www.candlepowerforums.com/vb/showthread.php?421264. This makes more sense, plus that big 88 mm reflector - big reflectors do seem to squeeze out more lumens, like the old TN31. Maybe between the size, SMO, and good AR lens... dunno.
I actually didn’t even want to talk about XHP35’s in this thread, I wanted to design a buck driver to run them lol.
I honestly do not know what they are capable of, I never claimed I did. I simply said that if acebeam could get 2600 lumens out of them, then I think we can as well. I still do think we can match whatever acebeam did, there is no magic involved in it afterall.
I do think that is best left to another thread though.
Good review there, I would say that does sound reasonable.
Just for the sake of keeping everyone happy I took the 35’s out of the OP, I would not have mentioned them if I had known people would focus on that instead of the driver to use them lol.
I only used them as an example because they must have 4s and I wanted an example that used 4s.
So, back to the buck driver design
I have been reading everything buck driver all morning and they are really fairly simple, just a lot of components working together.
I think that by using the SIR400/800 FET we will have a pretty good FET efficiency, they have already proved themselves and the conditions will not be much different then how we use the FET now. Except it will have a full 5V and tons of currant feeding it.
The buck converter itself has already proved itself and looks good.
We need to pick an inductor but that should not be that hard, lots of options that will fit in a Q8.
Might possibly add a large cap to smooth out some of the ripple as well if space allows.
Then just a matter of picking out the parts.
Now for the fun part, reverse engineering the schematic….
If you stick with the freewheeling diode in MTN MAX, you will probably need to use 2 parallel to each other as they are only rated for 10amps. That or find a diode that can handle more current. It is important they are well heat sinked too. When the Buck converter cuts battery power to the LED’s, all the current supplied to the LED’s flows through the freewheeling diode.
So you have determined that this buck driver will have enough voltage overhead with 4S cells to drive 4S (3v) emitters? I would think that a buck driver in such a situation will provide limited High/Turbo mode performance due to battery voltage sag as current increases, increased Vf of the emitters, and losses in the buck conversion. I hope I’m wrong though.
I agree. This is why we don’t normally use buck drivers for 1S lights either, isn’t it? For 12v application and 4S, maybe we need a higher voltage linear regulator?
I don’t know any specific numbers, but I think the fact that it is 4s will help the voltage overhead issue. For example, say a single cell with buck driver driving a single 3V LED at 4A has 0.2V overhead. Now a 4s battery driving 4 3V emitters in series at the same current has 0.8V overhead.