What did you mod today?

+1 with Dale, though I've been too embarrassed to ask...

L4P's CC drivers do result in the FET's getting hot though, just seems odd to put it on the known hot MCPCB. Obviously though it must work somehow because I'm sure L4P tests all this stuff out.

Also confused that 150ºF is being called tame, if that doesn’t burn blisters on your hand then nothing is going to. If the light isn’t too hot to hold at that setting then the thermal path from the emitters is compromised.

To elaborate, I truly do know what 150º is all about, my driveway gets that hot in the summer here. It WILL burn you!

Mercury thermometer showing an actual 120ºF air temperature on my front porch…

So yeah, I’m confused as to why one would want more heat in a small light.

The FET’s in L4P’s driver run very hot, mind that in a 8x7135 driver the burnt off heat is shared by 8 chips, a comparable amount of heat is shed by 1 FET in L4P’s drivers, and more with the modern low voltage leds and triples. So it does not really matter if the MCPCB is already say 60 degC, the heat path away from the MCPCB is so much better than from the middle of a FRP board (unlike 7135 chips, these type of FET’s do not have the thermal pad connected to ground but to led-minus!) that the higher temperature of the MCPCB is more than compensated by its better heat path.

Because as soon as the MCPCB hits 65c it gets throttled. So the runtime at max is lower than I know it can handle.

Oh, I fully understand all that, but with my E07 that makes 9300+ lumens set at a mere 50ºC the light gets too hot to touch in 20 seconds. Throttling it down AFTER it gets too hot to hold leaves it too hot to hold for a while… the heat dosen’t just disappear because the light stepped down. 25A in this one definitely makes for some heat, and this is a larger light with VERY ample cooling fins… the little tube lights have no such advantage.

djozz, the FET is a direct drive switch of course, as such, it’s not shedding any heat but merely passing the current through. I’ve held the driver’s between my fingers at full power and they don’t burn me, the MCPCB is another story altogether, especially a triple! Granted, I’m talking a standard FET driver and not Neven’s. I have more than a few of his as well and never noticed the driver getting hot, I don’t use the external MOSFET set-up.

Of course 8x7135 regulation chips shed heat, they’re regulating what would be 6-8A of current into 3A, that difference has to go somewhere…

I do understand what you’re saying. Maybe I’ll try it again at the default setting. It was definitely hot, but not as hot as I am used to from my Nichia 219C triple! I could grab it without losing layers of skin, anyway.

The question was about Led4Power’s drivers, they use different type of FET’s in a different way, not unlike 7135 chips, in which they do not just switch on and off, but actually regulate current while burning excess power off. At max unregulated current they are not hot at all, they just let everything through like a switch does, but at half regulated current is when they get hottest.

Ah, ok, that makes sense then. Didn’t realize the FET was performing in that way. Thank you for that layman’s explanation. (digesting it with a good Ethiopian coffee. :smiley: )

Installed Nichia NV4W144AM under 45 degree TIR and H1-A in Thorfire VG-15S host.

Album with all pictures

Tricky points:

TIR is shorter than stock reflector:

So I built copper spacer using 16mm MCPCB and 2 pieces of copper foil. It should be thicker but it was compensated by extra oring between bezel and glass:

White small round thing was glued with the switch. Tailcap parts:

Mine H1-A looks different than kaidomain’s product picture. Mine:

Kaidomain:

Capton tape was added to isolate positive resistor (marked 01c?) from negative. Without isolation max current was only around 2A@4V. With isolation - over 5A@4V.

Linear CC drivers work by burning extra voltage into heat. This happens in MOSFET, which basically works as smart variable high power resistor. So, MOSFET is NOT in fully conductive state all the time like on DD drivers.

Math for power dissipated in FET is simple: Pfet=(Vbatt-Vled)*Iled

Djozz' statement that "FET in linear CC drivers runs hot" is not accurate in general, how hot will FET be depends on three things:

1.Amount of power disspated in FET (equation above),

2.Thermal resistance Rth (C/W) between FET case and flashlight body - this constant tells us how many C degrees will FET get hotter for each Watt of power dissipated in FET(the lower the better)

3.Flashlight temperature.

Math is again relatively simple for steady state conditions: Tfet= Tflashlight + Rth*Pfet

Why move FET to LED pcb?

Answer is Rth.

MOSFET on FR4 PCB has relatively poor value of Rth, I measured about 20C/W for 17mm LD-A4 driver without any thermal "improvements" like adding silicone fillers, so just driver in pill.

This means for each 1Watt of heat in FET, FET would get hotter by 20C. If you take worst case scenario where flashlight is already hot at 60C, to reach 100C FET temperature you need just 2Watts of heat Tfet=60C + 20C/W*2=100C.

That means max. allowed power dissipation should be declared as 2Watts@60C flashlight temp. FET can handle higher temperature, about 150C, but rest of electronic parts (caps,ICs...) which are on same pcb are rated to ~105-125C.

The reason for high Rth value is bad thermal conductivity of FR4 material combined with small cross section and relatively long distance through which heat must travel, and very small contact area between driver GND ring and pill.

Copper layers help a bit, but the main problem is that FET drain pad, which serves as cooling pad, must be isolated from GND ring - no direct copper thermal path is possible.

If you put MOSFET on LED PCB like mosX which has very high thermal conductivity, Rth between FET and flashlight becomes very small,<1C/W!

Thermal conductivity of ceramic dielectric is much higher than FR4 material, thermal path is much shorter(1.5mm), and cross section area is bigger. All this gives much lower thermal resistance.

If we do the same calculation as above for mosX PCB with Rth=1C/W, for 2Watts of heat in FET, FET temperature will be Tfet=60C+1C/W*2=62C! Temp. increase is just 2C, compared to 40C for FET on driver PCB.

I can see why some people intuitively think that it's better for FET to stay on FR4 driver PCB because it's "isolated/protected" from LED heat, but remember that LED PCB/shelf is also a great heat sucker!

As math shows us, FET on driver PCB would get much hotter (and by only its own generated heat), because it's too thermally isolated from flashlight cooling surfaces. Heat builds up as it can't go anywhere efficiently. FET on LED PCB would run much cooler because despite the heat generated by surrounding LEDs, there is still plenty of cooling ability caused by very low thermal resistance of LED PCB/shelf.

For those who still are not sure about all this, one little thought experiment: take two 20mm LED PCBs with any LED and place one on cold (20C) wooden surface, and other on hot (60C) aluminum/metal surface.

Drive LEDs with same current/power, let say 3Watts. What do you think, which LED will get hotter after few minutes?

(LED on MCPCB represent FET/heat generator, wooden surface is FR4 driver pcb, hot metal surface represent LED MCPCB)

My flashlights are liquid cooled at 98.6 degrees F. Sometimes the heat sink complains though.

Thanks Neven, that really clears it up for me. I know there are Toshiba MOSFET’s that have an additional thermal pad on top for just this sort of occurrence but adding a heat sink can be problematic as there is often very little space.

Thank you for going to the trouble. Appreciate it. :wink:

Finally got my 3x White Flats in my 18650 C8F. I had a virgin PCB from the 6 LED light I built. It was much easier to reflow on a new PCB with less solder and some flux.

With extra long 20 gauge wires (for re-flashing) and MTN-17DDm FET+1 driver. Both ends bypassed.

Get about 13.x amps on a lightly used VTC5D for 144kcd - way more satisfying than the shaved SST-20’s that were in there before.

Nice contactcr, I wondered how those would work in such a light… :wink:

[quote=DB Custom]
Nice contactcr, I wondered how those would work in such a light… ;)[/quote

+1!!

beam shot?

I went stripping again :smiley:

I just couldn’t take the “baby blue” of my EDC01 anymore (nor my wife saying “awww… that’s cute”).

I took out the internals and stuck a cotton ball coated in Vaseline inside at the business end and covered the threads with it as well. I wiped the excess off with IPA and then have it a bath in lye (a solution prepared from granulated drain cleaner crystals).

I haven’t decided yet it I’m going to polish it. I think I’ll leave it for a while and see what happens.

Oh yeah, and I swapped a 4000K Luxeon V2 in it. Makes for a great “around the house” light.

Kinda cold for stripping. I like. I use suspension clip on my SAK

Put a new MOSFET on an odd driver this evening. Been a LONG time since I’ve done this! Why? It’s the huge 70N02! :wink:

Nearly doubled the output over their FET. Still a very long way from their claim of 100,000 lumens. lol