# filling vias

• posted

Has anyone filled vias with solder to improve thermal and electrical conduction?

I figure that if a via has an unmasked pad, and it's solder pasted and reflowed, the solder will melt and flow into the via, partially or completely plugging it with solder.

Maybe arrange for 2x the paste that would fill the via volume?

I found this online:

which seems to contain a lot of nonsense.

Seems to me that plugging a via with solder will radically reduce its resistances, and that lots of small vias are better than one big one.

• posted

The devil is in the details. The size of the via will determine the ratio of copper to solder in the via. If you favor smaller vias, that will lower the relative amount of solder and reduce its impact.

The raw material, as the article you mention, indicates, solder is much less conductive than copper, they say about 10x. I found references

the same solder is 55 W/m?K while copper is 401 W/m?K. In both cases the ratio is closer to 7 or 8 but close enough. The point is you need

*lots* of solder to appreciably change the electrical or thermal resistance of a via.

So how large are your vias?

```--

Rick```
• posted

I agree; no more than 23mil hole as the probability of a complete fill is not too good. Beware of an array of vias can result in a solder "pool"-like blob, requiring squeegee-type removal if the surface must be flat. ..and hot-air "blast" may remove solder in some of the vias...

• posted

I wrote a spreadsheet I use for evaluating those trade offs. Here's a sample:

25uM hole plating, 1.6mm board thickness, lead-free solder

(K/W per via) Plated hole via solder- copper- diameter alone plugged plugged

----------- ----- ------- ------ .010" (.254) 190 K/W 139 K/W 57.3 K/W .025" (.635) 80.2 41 11.3 .039" (1.0) 51.6 20.5 4.8

For Cu plugs the sneaky way is to mount a dummy through-hole part with copper leads, e.g. a SIP, a .025" header row, or a 2W resistor.

Cheers, James Arthur

• posted

All that will do is make the board really heavy. Press a piece of adhesive pyrolytic graphite over the component if you need to hack a low profile heat spreader.

```--
I will not see posts from astraweb, theremailer, dizum, or google
because they host Usenet flooders.```
• posted

Nice.

25 mil vias might be about right, so I get about half the unfilled theta if I fill them with solder.

I'm thinking about paving over a PCB with DPAKs and dumping a lot of heat to the back side. Maybe half a square foot and hundreds of watts, maybe even a kilowatt? 20 or so vias per dpak is a couple of K/w.

The PCB could have one of those Bergquist gap-pads on the bottom, for insulation, then a cold plate. A thinner board would help.

Just thinking.

• posted

If you are going to use Bergquist, why not put the chip side against the cold plate and get rid of the board as an insulator?

```--

Rick```
• posted

What is the thermal conductivity of air? ISTM that even a poorly conductive metal would be better than nothing.

• posted

Do you get that much copper in a via? I heard that a typical 1 oz pcb starts as 1/2 oz and they plate up to

1 oz. so the via would have 1/2 oz... and then does it fill with as much as the copper surface. I could imagine that a small via could get "starved" for copper near the middle.. but I have no idea.

(And not to change the subject too much, but how do the holes plate through? does the copper grow down from each side... and from middle layers?)

For prototypes I've make big holes (not a via anymore) under the thermal pad and then fill the thing with solder.

George H.

• posted

The assembly folks gripe about that many holes sucking up all the D-PAK's solder, but since you are the assembly folks, you could work out what works and just slather on enough paste.

My little spreadsheet calculates the total solder volume.

A thinner board helps a lot. Quartering the thickness quarters the K/W.

Another possibility is using one of the new packages that lets you tap heat top and bottom, then sandwiching a heat sink collector plate on top. I don't know if that works for you electrically, though.

Cheers, James Arthur

• posted

25um is sort of a typical spec., rather standard AIUI.

It makes approximate sense. The plated-through hole process itself has to seed, then plate a cylinder through prior to plating up the rest of the board. If they add 1/2oz after starting the PTHs, 25um would seem a reasonable end result. (1 oz. is 35um).

Magic :-)

They seed the holes with conductive pixie dust, then electroless-plate 'em to make a solid cylinder.

One option for John is ultra-thick plating. Twice the copper in each hole halves the resistances, thermal and electrical.

Cheers, James Arthur

• posted

Most board houses will fill the vias with copper for you. Might be better than using solder.

Cheers

Phil Hobbs

```--
Dr Philip C D Hobbs
Principal Consultant ```
• posted

Paulo, my production manager, will get consulted. He' s also a really good mechanical designer and machinist, and he comes up with great ideas. So whatever we wind up doing, he can't complain.

That would be hard to troubleshoot! All the parts would be hid.

I'm thinking about using our pick-and-place to pave over the board with a lot of cheap dpaks. Spread out the heat.

IR makes those DirectFets, which let their heat out the top. I could put them on the bottom, and press the board into the gap-pad, but the contact footprint area would be tiny, so theta would go up. And I need to cool other parts, not all fets.

• posted

Google heavy copper pcb and extreme copper pcb.

How about 20 oz boards? 200 oz?

• posted

Definitely better. And pretty.

Cheers, James Arthur

• posted

Yeah, that theta reduction is radical.

Copper-filled vias and 10 or 20 Oz copper gets interesting. The boards would be heavy.

The copper-filled vias that I see online are sort of granular looking, so probably don't conduct as well as pure copper. But probably still better than solder.

• posted

Sure, but that doesn't mean it is worth all the effort required to implement. James Arthur has some numbers. Interestingly it shows smaller vias are worse. That is expected since the area to circumference decreases with smaller diameters. John seems to like smaller vias, but in this case it makes more sense to have larger ones. I expect the reality is that more small vias *without* filling with solder are better than fewer large ones *with* solder. Look at how much the numbers change for copper plugged vias! A 25 mil via goes from 80 >> 40 >> 11 for air, solder and copper filling. Even a 10 mil hole drops more than 3 fold.

Better yet is to get the board out of the way and put the cold plate on the devices directly.

```--

Rick```
• posted

Or get rid of the board entirely. Flip it over and put the devices against the cold plate.

```--

Rick```
• posted

Yes, but then the devices still need to be wired. The PCB fixes that.

20 x .025 filled holes per device gets him to 2K/W even with solder-fill, and 0.5K/W with copper. A .031" thick board halves those figures.

The Berquist pad-thing is a surprise. I'd have thought John would go for straight mounting on anodized Al. Maybe he wants copper, instead.

Cheers, James Arthur

• posted

That's actually right, but instead of magic pixie dust, it is electro-less copper plating. That means they use chemicals that plate copper *everywhere* in a thin layer. They only make this plating thin because it is everywhere. No point in plating the parts you are going to etch.

Or just get the board out of the way.

```--

Rick```

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.