[mattthemuppet]

Anyone thinking about building a triple XM-L using one of Zemike's forthcoming 35mm housings might want to check out this:

As requested, 3 XM-L Driver/Emitter module DIY | BudgetLightForum.com

early days yet, but going by that guys other work (24 XM-L 4x4 light, self-designed, iPhone controllable!) it should be excellent. It would certainly make building a triple XM-L light about as easy as it gets!

 

[kwarwick]

Anyone thinking about building a triple XM-L using one of Zemike's forthcoming 35mm housings might want to check out this:

As requested, 3 XM-L Driver/Emitter module DIY | BudgetLightForum.com

early days yet, but going by that guys other work (24 XM-L 4x4 light, self-designed, iPhone controllable!) it should be excellent. It would certainly make building a triple XM-L light about as easy as it gets!

Hmmmm.... it does look quite interesting, but I'd be concerned about the thermal transfer given this doesn't use a MCPCB and instead depends on vias to get the heat to the backside of the PCB. Is that a viable technique for a board that will need to heat sink 30W or more of LEDs?

Personally I'd go with a 3up XML board with separate b3flex driver... much more robust to not have the driver and LEDs sharing the same board with LEDs that will invariably run hot putting additional strain on the driver electronics.

 

[mattthemuppet]

well, MCPCBs are actually pretty crap at transferring heat from the LED die to the housing as most of them have a thin layer of insulating material under the traces to isolate them from the aluminium star. This is are far more direct and efficient method. He's also used the same technique to build a considerably bigger light and has done pretty extensive thermal modelling, so the guy knows what he's doing.

3up board + b3flex vs. this depends on cost and flexibility in my mind as all the components are designed to handle the heat (plus any thermal monitoring will be implicitly more sensitive).

 

[rschultz101]

good find! interesting !
thought my SMT soldering skills not pretty .

 

[zemike]

Never saw an XM-L on FR4 substrate.
Matt, can you ask several questions on that forum?
1) what is the LED/heatsink temp on full power ?
2) minimum order quantity for a machine assembled module ?

 

[mattthemuppet]

just done it Zemike

 

[Itess]

well, MCPCBs are actually pretty crap at transferring heat from the LED die to the housing as most of them have a thin layer of insulating material under the traces to isolate them from the aluminium star. This is are far more direct and efficient method.

I wouldn't say that "any mcpcb is pretty crap" at transferring heat from LED. Even Cree made some research on that.

 

[mrradlos]

I'm not sure if I am able to solder the LEDs and the tiny bits on that board myself? :skep:

 

[mattthemuppet]

I wouldn't say that "any mcpcb is pretty crap" at transferring heat from LED. Even Cree made some research on that.

well, it was a relative term. Compared to directly soldered onto a circuit board with a direct path to the housing, they transfer heat less efficiently. Some MCPCBs are better than others, but the majority of the ones we get are pretty pedestrian. After all, the MCPCB format was designed for convenience (no reflowing!) not performance

 

[mattthemuppet]

huh, interesting info in that link itess, learnt a lot about FR4 and MCPCBs Looks like the dielectric layer under the Cu trace on MCPCBs doesn't have as much of an effect as I thought and that board design for FR4 substrates is much more critical. At least I now know what those little holes are for under the LEDs!

 

[mattthemuppet]

cross posting for Zemike from BLF:

"He'll be plesantly surprised by the performance of an XM-L on FR-4. I promise.

1) Completely 100% dependent on the heat-sink. MCPCB, FR-4, Platinum - doesn't matter what the emitters are mounted on - it's the heat sink that has to dissipate 30 watts.. I can't answer the question, unfortunately.

2) Unsure just yet - I haven't even got the kits shipping yet. Details will be provided if/when I'm prepared to offer a fully assembled module.


PPtk"

 

[KBARK]

Now I just need to find a square 3 up lens...

Probably not gonna happen.

 

[phburns]

This is the development I've been waiting for. I'll be watching this carefully. I've been toying with this idea in my head, but have no design experience, and don't want to climb that steep knowledge hill.

It just make so much sense to me to have the driver and Led's on the same board. Would be nice to have Buck/Boost to be able to drive it with two 18650 cells, but I guess you can't have everything.

I hope the board will be economical. In any case, PilotPTK should sell a bucket of these. :thumbsup:

 

[mattthemuppet]

This is the development I've been waiting for. I'll be watching this carefully. I've been toying with this idea in my head, but have no design experience, and don't want to climb that steep knowledge hill.

It just make so much sense to me to have the driver and Led's on the same board. Would be nice to have Buck/Boost to be able to drive it with two 18650 cells, but I guess you can't have everything.

I hope the board will be economical. In any case, PilotPTK should sell a bucket of these. :thumbsup:

it certainly would make building a triple a lot easier, especially with Zemike's housing once it's done, although some of that convenience may come at the expense of the customisation we're used to with the taskled drivers (hopefully more info on that will come soon). You should go and read Just off the CNC - Update: It's Alve! | BudgetLightForum.com - the guy takes OCD to a whole new level! He reminds me a lot of George actually in his attention to detail.

Fingers crossed on the price, I'm guessing ~$70-90 roughly, but we'll find out soon enough

 

[zemike]

Small remarks:
1) The FR4 board is thinner than an MCPCB so there will be a neeed to file down the legs on the optic, designed for 1.5mm thick boards.
2) I can get a quote from my PCB manufacturer for FR4 and MCPCB substrates but I think that the impact on total cost is negligible. (my PCB fab is the same as lux-rc, so quality is good). So why use FR4 ?

 

[Itess]

well, it was a relative term. Compared to directly soldered onto a circuit board with a direct path to the housing, they transfer heat less efficiently. Some MCPCBs are better than others, but the majority of the ones we get are pretty pedestrian. After all, the MCPCB format was designed for convenience (no reflowing!) not performance

Did you read the document that I pointed to?

 

[mrradlos]

Small remarks:
2) I can get a quote from my PCB manufacturer for FR4 and MCPCB substrates but I think that the impact on total cost is negligible. (my PCB fab is the same as lux-rc, so quality is good). So why use FR4 ?

PilotPTK on budgetlightsforum:
The answer is: He's forgetting or not realizing that this isn't a simple single layer "star" board. It has all of the electronics on it, and pretty much demands at least two layers of copper. Not only are multi-layer MCPCB's EXPENSIVE, but they're actually quite bad at thermal conduction. The board I'm using is very thin (half the thickness of the thinnest in the CREE document) meaning that it will outperform a standard thickness single layer MCPCB and FAR outperform a multi-layer MCPCB. COULD I route this in a single layer? Yeah, I could - but the module would have to get a bit larger, and I didn't really want to make it bigger in diameter than the CUTE-3 Optic. That was the design goal - keep the PCB equal to or smaller than the diameter of the lens.

 

[mattthemuppet]

Did you read the document that I pointed to?

I did, hence post #10. It was quite eyeopening, thanks!

 

[phburns]

From the article on MCPCB vx FR-4 heat transfer, apparently the FR-4 nearly approaches the coeficient of the MCPCB used in the test. The thinnest FR-4 used in the test .8mm is twice as thick as the one Pilot is using for the latest tripple-driver solution .4mm. Also, the number of vias used in the test (14) is not even close to the 55 vias being used in Pilot's module. I think he's got the heat transfer isssue covered .

 

[phburns]

I thought this might interest a few folks here

Announcing Pricing:
DIY Module which will include Circuit board, Components and LEDIL CUTE-3-SS-XM Optic
The only thing NOT Included that you'll need are the actual XM-L emitters. For now, I'm not going to include them so that each person can procure their emitters in the tint and bin that they want.
$35.00 plus shipping
I will have T6, U2 and LT5 emitters available for purchase separately. T6 and LT5 will be $9.00 and U2 will be $11.00 each.
Fully assembled Module will include a professionally reflow-soldered module complete with XM-L Emitters and CUTE-3-SS-XM Optic:
Neutral White T5 Flux Bin E3 Tint Bin: $85.00 plus shipping
Cool White T6 Flux 53 Tint Bin: $85.00 plus shipping
Cool white U2 Flux 1C Tint: $93.00 plus shipping
Your Choice Emitter - I buy it from any source you specify. $60.00 + Cost of Emitters + $10 Custom Fee + shipping
These will be the initial costs. If volume ramps up and I can buy components in larger quantities, pricing may come down a bit.
Expected Availability for the kit is 3 weeks from today. Fully assembled module 4-5 weeks from today.
PPtk