Finished PCB thickness vs finished copper weight requirement...

Apr 17, 2009 5 Replies


It's hard to get real 1 oz copper, much less 2 oz. People start with


> 1/4 or 1/2 and plate it up, and they tend to quit too soon. One of my
> fab drawings says START WITH 2 OZ COPPER.

In this case my fab drawing said (in HUGE letters), "Board shall be


2oz finished copper." The fab house claims that they started with 1oz copper and plated up to 2oz. I have asked for them to provide in writing what they did. I am now wondering how to prove whether or not they complied with the requirement. I personally dont think it "feels" like a 2-oz copper board. The raised height just isnt what I was expecting....it seems very thin. What types of quality control rules do you use for determining what was actually delivered? I dont have an accurate depth gauge, but I suppose I could use calipers to measure the thickness of an area with no copper (only soldermask) and compare that to an area with full copper to determine the copper
+plating thickness....then determine what thickness 2oz copper should be.

A colleague told me that "Copper weight is given in oz / per sq foot.


1 oz =3D a minimum of 0.0012=94- 0.0014=94 thickness". So I suppose that
2oz finished would be 0.0024" - 0.0028" thick. Also, in a 2003 usenet archive post, Rene Tschaggelar offered these numbers (and I provided the translation to mils)...
1oz 35um 1.38mils
2oz 70um 2.76mils
3oz 105um 4.13mils
6oz 0.2mm 7.87mils
9oz 0.3mm 11.81mils

For my current situation, the PCB is measuring 0.062" thick in an area with no copper (only PCB and soldermask) and 0.065" in an area with copper on both sides. This implies 3mils of copper on top and bottom, so 1.5mils of copper on each side. I dont think this would qualify as



2oz finished copper, but now I have to convince the fab house to eat
1000 pcbs due to their fault...

(PS-My application only deals with automotive 12vdc and 10amps, but I am mostly concerned with adequate copper for heatsinking my D2PAKs because of a slooooowly switching Infineon half bridge driver running as fast as it can go).



Any insightful comments? Thank you! frenchy (Steve French)


If you're mostly concerned with using copper for heatsinking your D2PAKs, the thickness doesn't matter nearly so much as the surface area. How much power are you dissipating in the D2PAKs? I'd probably only worry about the copper thickness once it's above 5-10W (although I'd start calculating thermal resistances and what-not once it's above just a watt or so).

---Joel

When it matters, I add test traces and measure the resistance on the received boards. 1 oz should be about 525 uOhms per square.

Most of the routine "1 Oz" boards I get are actually in the 0.75 oz sort of range.

Sometimes it's easier to add copper pours on more layers, stitched with vias, than to fight the board houses over this.

A pound bag of coffee used to be a pound; now it's often 12 oz. And nobody sells "Pound Cake" any more.

You could do something like this...

ftp://jjlarkin.lmi.net/Copper.jpg

John

Extra plating causes "mushrooming" of narrow traces. Try to route planes on the outsides, signals on the inside.

Cut a sample of the copper of known or measurable area from the centre of a pad and have it weighed on a sensitive balance.

suppose that

Perhaps you could use any of several "narrow" traces as a low value resistor. Measure in "the 4 terminal mode". If they are close to proper weight the "resistor" will be in tolerance, if not it will be too high. .

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required