Glad I asked. :laughing: :open_mouth:

What if the led can survive whatever amount of overcurrent? When you do that you're pumping more power into the led it can fruitfully handle.

Wrong. The only reason a regulated linear setup generates more heat than plain direct drive is the price of regulation. The LD25 driver, for example, uses an AON7520 MOSFET and achieves regulation by sensing current through a 10mΩ resistor and using this feedback to precisely tune the VGS MOSFET gate voltage for it to let go no more than the preset driving current, thus no :THUMBS-UP: PWM. The net dissipated power difference versus the same driver without regulation is the additional power spent in the sensing current path, Psense = I² × Rsense. At 5A that is 0.25W, this new LD-25 is particularly efficient in this regard. And of course bear in mind the regulator has to take and bear with any additional voltage delta times current power, thus the increased heat in the driver. This increased heat in the driver nets reduced heat everywhere else in the current path: springs, switch, cell, emitter, etc.

A MOSFET misdriver is just a PWM micro-controlled switch. As I said above, you pay a regulation price in linears. However, this price is not always a hurdle. You complain about heat generation, but do you understand where and how is the heat generated? Any time you “adjust” a plain MOSFET setup via unbypassed springs, using higher internal resistance cells, etc. what you're doing is redistributing where the heat is generated. Which of course can be good.

A linear driver has to burn all excess power to avoid it reaching the emitter, which of course will make them hotter than unregulated misdrivers which of course do nothing.

Sure as FLICK a misdriver isn't going to be very hot, how could it be? It is doing nothing besides quickly switching on and off when required. However, this does not mean the heat isn't going to be generated elsewhere. It's the emitter who is going to take the penalty among other parts like the switch, my dear. Pumping 7A in the white flat isn't going to insta-kill it, but it sure increases emitter stress unnecessarily and nets you lower performance.


Sat, 11/03/2018 - 01:36; Sat, 11/03/2018 - 03:30

Ya know this guy noferek9 prolly makes the most practical real world sense outta all of us here. :laughing:

“the problem is that most of us are going to change the xpg2 on our flashlights which was with DD drive,i prefer to just change the led keep the drivers and just find better batteries for that setup.”

Cuz I can tell ya right now I’m not gonna rip out a bunch of XPG2’s either much less their (gasp….heresy!) DD drivers.


“Sure as FLICK a misdriver isn’t going to be very hot, how could it be? It is doing nothing besides quickly switching on and off when required. However, this does not mean the heat isn’t going to be generated elsewhere, you know? It’s the emitter who is going to take the penalty among other parts like the switch, my dear. Pumping 7A in the white flat isn’t going to insta-kill it, but it sure increases emitter stress unnecessarily and nets you lower performance.”

Man, I almost can’t wait to read Luminarium’s reply to this one.

:laughing: :sunglasses: :heart_eyes:

Sure as FLICK I can wait. In fact I don't really care. No one's paying me for this and I believe it is unlikely he's going to get off his high donkey. He may already be aware of the provided information, except he may not have tried a proper linear driver like the ones from led4power or an LD-25. Or he just doesn't care.

By the way, I love Donkey Kong. ;-)


Well all I can say is I’m being entertained. I was starting to get bored around here. :laughing: :+1:


You said it. And it is true.

About pumping that 7A…

Don’t be like Notthawhackjob. I explainded to people with FET DD that low current cell is must since it will not pull more than 5.5A from emitter and becasue of that it will be safe in usage.

So once again for people still using FET DD setup. Use low current cell (Sanyo GA, LGBD1, and similar) in FET DD setup.

That configuration will run from 5.5A (at 4.2V) > 3.3A(at 3.4V). So nothing bad will happen to emitter and emitter will perform at its maximum performance (look at L4P test and Djozz graph).

Don’t use high current cells in FET DD setup (samsung inr 30Q, Sony VTC and similar) cause they will pull more than 7A of current and could potentially kill your emitter.

Yes all of that is understood. Thanks anyway. :THUMBS-UP:

A lower current cell is also of benefit for a linear driver, reducing the driver's dissipated heat.

There are some duper low price deals for high capacity low current cells in AliExpress. Asked Henrik to get some reviewed, by the way.

Quite low Vf, so low it could also work very well with an efficient buck driver like an LD-29 fed with single cell. Stack an R050 over the stock R025 for 4.5A maximum drive current.

Sofirn sells the C8S host with 20+mm driver hole and smooth reflector, I wonder what sort of throw performance would that net with the white flat. 200KCd?

Cheers :-)

For people got confusing about different performance L4P and Djozz got of their tests.

Djozz got best performance at 4.8- 5A while L4P got it at 5.75A.

L4P explained why they got different readings: “Excess solder could cause that when power density is high like with this LED. Also, I’m using lead-free solder which has a little bit better thermal conductivity.”

I did my own re flow and I used very thin layer of solder paste we mostly use here (that cheap budget mechanic solder paste in syringe) and I got similar performance to L4P test.

Yes I also find that confusing but I know that my test subjects Osram white flat performs as in L4P test. And since I am testing it for more than month in fet dd setup I know it is robust emitter that will work without any issue with low current 18650 cell like LGBD1 and that it can even survive abusing with high current cell like Samsung 30Q but I don’t recommend that to anyone. Why would anyone want to run emitter on 7A if it has best performance on 5.5A?

So yes. Barkuti is right. If you want to be worry free get regulated drivers, but if you insist on FET DD be very careful and use low current cell or you could fry something out :slight_smile:

New tests:

52KCd in Godmes T01 (14500)
118KCd in Olight M22 (30Q)

Original driver in both. The main difficulty of this led is propperly focusing. It was impossible to focus the TK61 (600KCd max) and Warsung MX900 (83KCd max)


FYI. For the record, I’m not trolling anyone here, Jack, nor am I trolling anytime, period.

Other readers didn’t exactly agree with you 1:1 on every single technical point, observation, or at times confusing or contradictory comments you’ve made until you further clarified your comments but that doesn’t mean they obviously don’t understand or didn’t read. That includes me.

Look at your Post #122 as an example. Even you admit there are seemingly confusing elements involved.

All I poised was that can the Osram KW take relatively short bursts from high current cells on highest draw mode without harm if for whatever reason I need/wanted to do so. The point being this led isn’t apparently going to immediately fry with a short or even long burst from a high current cell. Is it optimum? No. Is it recommended by you? No. But again to be clear I don’t disagree with your initial comments about this led performing best with a low current cell. Totally understand.

Congruently I also surmise I can still put a high current cell behind it if I keep it away from operating on essentially turbo mode. But it’s not absolutely essential because it can take the abuse as you said. So it’s really not a safety issue according to you, right? This led apparently isn’t going to melt, fry, or cause a fire with a high current cell, right? It’s really a better performance issue, right? No problem there. I fully understand. Ergo, I’m not trolling.

You see there are times and situations when and where I foresee I may not happen to have a low current cell on me when this led happens to be in an empty light that I need to use for whatever reason.

It’s not going to definitely harm anything according to the long term results of your high current testing other than it’s just not needed for best performance so why do it? Agreed.

And again, fully understood. So am I a troll?


PS. Am I correct that English is not your first language at Location ‘X’? I guess I have trouble deciphering exactly what you’re saying at times. Maybe moreso the syntax.


Well you are wrong. Again, I’m not trolling. I kinda suggest you quit going there with that deal. It’s better that you understand you gain nothing by name calling or trying to paint someone negatively so that you try to look superior to others.

Now reread edited post #124 to better understand where I’m coming from.

Performance discussions about leds used in flashlights, however heated, however annoying, however one seems to not understand the other, are not trolling.

A troll has no opinion, has no conviction, does want to convince someone else, he just wants to mess up. This not going on here.

Well, at any rate, the BLF Rules prohibit fighting, and regardless of a user’s opinion or motives, this is always the best course of action:

I’ll lock this thread for now. If there’s something else to add the technical discussion please PM me and I can re-open it after tempers calm down.

Thread re-opened. To those who participated in the argument that broke out, please go back and delete those posts. Thanks.

I read the Texas Ace test that showed that a cheap thermal paste is just as good as a top one for flashlight use.
But Osram LEDs are quite vulnerable to bad thermal transfer. I have GD900 and I wonder if it’s worthwhile to invest in something better?

Arctic MX-4 won’t let you down.

Cheap thermal compounds tends to dry very fast.

Very worthwhile to invest cause you don’t want weak link in your modd or build.

If you have two copper surfaces, liquid thermal paste is the best way to go (it’s much, much better than any standard paste). Otherwise something like Arctic Silver 5.

The flatness of the surfaces will make the biggest difference though.