Hi, I would disagree that it varies "Much" but you do need to do a little math. Knowing your PCB material you know the thermal conductivity, that is your key parameter. This lavers of copper on internal layer don't help much though for a small part, like an SOT-23 they will help spread the heat. Spreading the heat gives a greater area cross section with which you apply to your thermal conductivity formula. My experience with an Al backed board is that those thin copper foils do not conduct out from the part very well, many .25 inch or so. While thermal conductivity of FR4 is about .3 w/m k and copper is 1000 time that, the cross section of a copper foil is only 1% that of a 10 x 10 mm square through the board. So the thermal resistance through a PCB is in the same order as the thermal resistance through a solid copper plane! (Someone please check my math!) You can take two VERY important design steps: 1. Use a thin PCB laminate. 2. Use a thermally conductive laminate. Then, if your device is small (sot-23 type size copper on top AND internal layers will help spread heat. Use LOTS of tiny vias to tie your top side copper (directly under the part, not around the sides) to both internal planes and bottom side copper. I am assuming you have a very thin insulating layer between bottom side copper and the Al plate so bottom copper to Al thermal resistance will be very low.. The many thin walls on the vias will suck down heat well. Avoid the idea of big holes filed with solder. Solder has about 10% or the thermal conductivity of copper. You want a lot of little vias each with a copper wall running top to bottom. With a DPAK sized device on .5mm thick thermally conductive laminate and with a lot of tiny vias I was able to get about 1C/Watt from DPAK tap to Aluminum backing. Now think about that. That means I can dump 40 Watts from a DPAK without any type of mounting hardware, screws, clamps, or epoxies. Just regular old SMT reflow of the part. Bob K -----Original Message----- From: TechNet [mailto:[log in to unmask]] On Behalf Of Wayne Thayer Sent: Tuesday, April 22, 2014 9:12 AM To: [log in to unmask] Subject: Re: [TN] Seeking : Thermal Resistance onto PCB surface Hi Roland- It can vary a bunch, depending on how much metal is in the pcb and how that metal is arranged. Most designs for thermal transfer depend on thermal vias to do most of the work. You can get fast ballpark answers with "back of the envelope" thermal transfer coefficients for PCB laminate, or even better, grab a board that's similar and just measure the performance. If you need answers within a few degrees, then the thermal analysis can get considerably more complex. Wayne Thayer -----Original Message----- From: TechNet [mailto:[log in to unmask]] On Behalf Of Roland Jaquet Sent: Tuesday, April 22, 2014 4:06 AM To: [log in to unmask] Subject: [TN] Seeking : Thermal Resistance onto PCB surface Dear Technetters, I am confronted to an unusual challenge I am not able to solve. Can someone help, if you please We are seeking : Thermal Resistance onto PCB surface (14 layers) standard approx.. 1.8 mm thick Between External Cu of approximately 30 to 35 metric microns covered with ENIG And Aluminum "Alodinized" 1200 or Anodized in black Thank you for your help Meilleures salutations Best Regards Roland ______________________________________________________________________ This email has been scanned by the Symantec Email Security.cloud service. For more information please contact helpdesk at x2960 or [log in to unmask] ______________________________________________________________________ ______________________________________________________________________ This email has been scanned by the Symantec Email Security.cloud service. For more information please contact helpdesk at x2960 or [log in to unmask] ______________________________________________________________________ ______________________________________________________________________ This email has been scanned by the Symantec Email Security.cloud service. For more information please contact helpdesk at x2960 or [log in to unmask] ______________________________________________________________________