Beyond component side (and heavily stitched solder side) copper pours, is there any value to adding stitched internal planes when it comes to helping to move heat away from components?
And, when does copper thickness start to hinder other product aspects? (neglecting material costs)
Den onsdag den 9. november 2016 kl. 23.51.06 UTC+1 skrev Don Y:
everything helps
when you can not longer get the trace width/spacing you need
There are two issues to cooling a topside part:
One is that eventually you have to get rid of the heat, from the pcb surface to air (or maybe conduction cooling to some metal surface.)
The other is that you need to spread the heat laterally, away from the part, to use more surface area for cooling.
The lateral spreading is helped by internal copper layers, connected to the topside copper pour with vias. The vias should be spread out under/around the part, to get the heat out from the central hot spot. The vias inside a cluster of vias don't help much.
Often existing internal ground or power planes can be used for heat spreading, even if you can't connect to them. Maybe have a topside copper pour a bit bigger than the part, to get heat away from the center hot-spot of the part, and then a lot of vias to a plane or to a pour sandwiched by planes.
1 oz copper has a lateral thermal resistance of about 70 K/W per square. Thinner copper is proportionally worse. So a small part can easily be power limited by the spreading thermal resistance of the copper.PCB houses lately tend to prefer 1/2 oz copper for fine stuff, like 5 mil traces. You can mix copper thickness on various planes, but that can cost more.
I sometimes include test traces, and measure their resistance, to see if the board house is delivering the promised copper thickness. Generally they come in a little low.
I'm doing an 8-layer board now with a 484-ball ZYNQ chip, and we hope we won't need a fan over the BGA. We'll have three plane layers to help spread the heat.
-- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
Certainly. The more copper the better. Of course, surface copper is better than internal planes. Most parts with thermal pads show this in their EVM designs.
Heavy copper and fine-pitch devices don't mix well. At 2oz copper, I would think twice about using BGA pitches less than .8mm and QFP/QFN pitches below .050".
At 1oz thickness, 100% coverage of copper foil about doubles the lateral heat spreading capability of the PCB.
At 2oz and 4 layers, that's a factor of 8. Big win.
A dense grid of vias (typically plated to about 1oz) about doubles the through-plane conductivity of the PCB (which, for plain, unadorned FR-4, is about a tenth the lateral conductivity!). Filled with solder, tripled. Big win.
If that's still not enough, a thermal pad between PCB and mounting plate is an excellent strategy. Choose a material with reasonable conductivity (>1W/m/K) and soft durometer, so it conforms around small; chip components. (Avoid using tall components on the PCB's heatsink side.) Thermal pad tech is awesome these days -- at least as good as PCB material, but flexible and insulating (do watch out for partially conductive varieties, though).
Pro tip: cut the thermal pad into strips, leaving gaps inbetween to give it somewhere to flow. Design it to compress slightly under the PCB (by a few thousandths), taking into account the volume of any components placed on that side.
3oz boards are okay for 0.65mm pitch leaded parts, and may be questionable for 0.5mm pitch and leadless parts. I don't think I would go over 4oz for anything that needs TSSOP/SOIC and power devices on it.Board fabs cannot make vertical sidewalls. The sides of copper traces are always tapered, or lumpy in some way (depending on the exact sequence of etch and plate used). Thus, the minimum trace width/space goes up proportionally with foil thickness.
There are some "super heavy copper" processes out there, which may involve milled copper plate, by the looks of it (milled, then stuffed into a laminate, and more glued on top). The vertical edges in such a process would be good (as long as you don't expect the gaps or webs to be too thin, which would be impossible to machine). But I've never quoted, or seen quoted, such a board...
Tim
-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website: http://seventransistorlabs.com "Don Y" wrote in message news:o0097p$ejt$1@dont-email.me... > Beyond component side (and heavily stitched solder side) copper > pours, is there any value to adding stitched internal planes > when it comes to helping to move heat away from components? > > And, when does copper thickness start to hinder other > product aspects? (neglecting material costs)
John, could you say a bit more about how much variation you see? A colleague wants to use pcb for getting copper of a "known" thickness, and my response was that it varies "a alot" maybe by 25-50%, This is for the top and bottom layers. The one time I looked at this I found that the inner
1 oz layers were petty close, but that the top/ bottom layer was thicker by ~50%. (it was a while ago, memory is fuzzy.)George H.
I've seen as low as 0.7 oz (measured resistively) on boards specficied as 1 oz. Most houses start with something thin for the outer layers, etch, and plate it up to build up the outer thickness and plate the vias. The plating time can vary, more or less than 1 oz.
We're planning a team tour to Cirexx to better understand the process. I've never been inside a PCB fab. We could stop somewhere interesting on the way back, make it an excursion. Filoli, Half Moon Bay, Moss Beach Distillery, something like that.
-- John Larkin Highland Technology, Inc lunatic fringe electronics
OK Thanks, so +/- 25% is not a crazy number. I guess I should ask the board house.
totally unrelated... my brother works at Aurubis
George H.
We are looking for a "new" building to move into and just looked at an old industrial facility. The family business made sawmill blades and industrial knives, since around 1880. This building dates from around
1908.They made some really beautiful circular saw blades. Apparently there's a lot of art and science to that.
(Sorry for the terrible pics.)
There's 100 years of black oily crud everywhere. It would need some serious cleanup.
There's a sad quiet beauty to abandoned factories. You can almost see the ghosts of generations of workers sitting at those grimy benches. There is a hobby, groups of people breaking into abandoned sites and exploring. That would be fun.
-- John Larkin Highland Technology, Inc lunatic fringe electronics
As long as you don't start a fire. :^) We had a huge fire at the old Bethlehem Steel plant.
They just had to let it burn out. (It's still smoldering and stinking up the place.) The TV weather guy could see the smoke plumb on doppler radar.
George H.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.