Has anyone heard of a Generation 2 SinkPAD with a smooth back and a cavity on top, filled with solder paste to mount the emitter on? Does this mean the emitter will not sit directly on copper? What if you want to use thermal paste instead of solder?
A quote from a SinkPAD rep:
“This Quote is for our Next Generation SinkPAD-II technology. you
probably have seen SinkPAD-I so far. SinkPAD-II will not have
depression on the back side like one you have seen instead it will
have very thin cavity on top size at each thermal pad location. We
don’t have formal data sheet yet but see attached drawing. Cavity
depth is ~0.003” which will get filled with solder paste during
assembly and you will get metal to metal direct thermal path with flat
surface on the back side”
In my opinion it's still better than traditional MCPCBs as there is no dialectric layer between the emitter thermal pad and the copper. Solder may not be as good as direct copper, but the solder layer is stupidly thin (around 0.07mm) and solder transfers heat better than the dialectric layer ever will.
They are also 20-30% cheaper than the original sinkpad.
I would rather have the best possible thermal junction at the point closest to the thing you're trying to keep cool, that being the LED itself. The farther from the LED you get in the thermal path between the LED and the ambient air the less critical those thermal junctions become, because usually each step farther has greater surface area than the previous one.
But, I'm guessing they wouldn't have made the change unless it showed some benefit - though shaving pennies off the manufacturing cost could be considered a benefit, from their point of view at least.
Looks a bit backwards to me, I’d prefer the emitter to be kissing the copper, not lead and tin. I priced some boards with SinkPAD the other day and at the 1000 level their a buck apiece. So a 20% savings amounts to 20 cents a board. The whole point in using a copper board to begin with is to amp it up, drive the expensive LED to it’s limits with the best driver you can wedge in the backside of the light engine. Dropping 20 cents worth of machining to remove copper to emitter contact seems to go backwards.
Thanks for the pic Dale. I'm obsessed with drop-box (for convenience) but it's image links don't work in forums. Something I hope they fix soon.
SinkPad told me they made the change due to complaints. Not sure who was complaining lol. They will always offer the SP1 but the SP2 is cheaper as mentioned previously. I think all they're doing is using a super thin top layer and using a 100% flat copper pad. As I said before - it's still better than normal MCPCBs, but not as good as a direct copper method.
Maybe they'll prove me wrong, but my gut tells me that their new design is crap. Having to span additional distance with solder is:
A) Dumb. Compared to Copper and Aluminum, solder is really a pretty poor conductor of heat - it makes sense to keep that junction as thin as possible
B) Dumb. Solder paste has flux in it. A lot of flux, actually. When you reflow, a good portion of the volume of solder paste disappears and you're left with much less actual solder. This means you're going to have voids in the junction between emitter and MCPCB.
C) Dumb. If there is a large enough gap (looks substantial from the picture, but who knows in the real world), there is a pretty good possibility that solder tension won't be strong enough to keep the solder adhered to the LED at all! You could pretty easily end up with a solder blob on the MCPCB that isn't even physically attached to the emitter. Thermal Conductivity goes right out the window in that case.
I've been wrong before, and I'll be wrong many more times before I pass on - but I would want some pretty concrete evidence contradicting my suspicions before I bought a Sinkpad-II over a Noctigon at this point. Noctigon has the best of both worlds - Full pillar AND no back side cut-out. I have heard rumblings of some occasional quality issues with Noctigon, but I've not personally had any that weren't perfect.
Yeah, but they're not designing their stuff to cater to flashlight hot-rodders. If you were in charge of a project involving tens of thousands of LEDs, that $0.20 each would make a pretty big difference.
In a worst case scenario what if you had thousands of boards mounted with XM-L2s driven at 3A and there were voids between the emitter and star? When the lights started going out that savings would seem a mute point.
You speak the truth! No $15 Walmart boots here buddy - try $300 Burtons! (that's Australian pricing...so halve that for an American market)
I have done more emailing and searching today that I ever remember doing on anything else. I have a few alternatives up my sleeve to this 'budget' sinkpads.
I generally agree with PPtk on his points, compared to the original design. For our use, the 20-30% savings is not significant. Heck, I’m thinking the aluminum SinkPADs could outperform this design (I have some and will be testing them). They are $1.00 each vs $2.00 - $2.50 for a copper board (SinkPAD sent a price list with the samples).
Also, that drawing cannot be to scale. Usually the traces are flush with the dielectric layer, and appear to have been exaggerated in height for some strange reason.
I for one will keep try to keep an open mind toward this design. Honestly I dont see myself purchasing these over the SP-I though. The direct bond with only a few microns worth of solder for me is a key point.
I know from experience the smallest bit of solder elevating an emitter can take you from amazing to POOF in nothing flat. Getting even a little too much solder under an LED can cause the infamous color shift and even fatality at high current.
This is actually the one reason I was skeptical of the original sinkpads when everyone first started buzzing about them. I figured they had just left a blank hole where the center pad was in the mask layer they to be filled with solder ( kinda like this one looks LOL)
LED-tech.de copper stars have been built like that for ages and they were used heavily, tested thoroughly and performed great. I dont see why these should be different.
Now we consider measuring inaccuracies, different heatsink, different setup, different measuring method and the fact that relic used an XML2 while the graph taken from CPF is an XML U2.. and I conclude that I dont mind, as long as its copper and there is no dielectric. ;)