copper thickness test

Winfield Hill wrote in news: snipped-for-privacy@drn.newsguy.com:

Gotta keep those dirty fab houses in line!

I sometimes measure trace resistance to estimate copper thickness. "1 oz" copper usually turns out to be less, maybe 3/4 oz.

Possibly plated copper doesn't conduct as well as the pure stuff, which would distort my resistance:thickness calculation, but my bottom line is that I seldom get the expected conductivity of 1 oz copper.

Maybe if a pcb house puts various boards on a panel, and different boards have different requirements, not all can be right.

I would assume this was a low quantity prototype run? It may have been com bined with other designs. The plating is not small, I think it is typicall y half an ounce or so. Any particular specs that would make the plating th icker, like small diameter with extra plating time to assure plating in the small holes. Generally something like copper thickness is a minimum, not a max.

Then there is the accuracy of the trace width. Did you check that?

Hmm, looks a little on the wide side. At any rate, it's easy to see it was indeed made with 2oz copper.

I take a small section of the board and etch half of it in ferric chloride. Then I compare the etched and non-etched sections with a ten-thou micrometer. Other methods are available.

Argghh! I'll go for one of the other methods.

Double ARRGGHH! Not that method!

Wikipedia says that 1 oz copper is about 35 um thick, and that the density is 8.96 g/cm^3, so 1 oz copper should weigh 0.031 g/cm^2. Grab a scale that can weigh to 0.01 g, cut a 10 cm^2 piece of single sided pcb, and weigh it before and after etching it bare and you should have an answer good to about

No need to etch, peel is easy. Can also measure the peeled off material directly -- remember to hit it with a cigarette lighter to burn off the soldermask (the copper oxides are negligible thickness, as you can tell from their rainbow colors when this is done quickly enough).

Though if you're going to etch, you can also do some analytical chemistry once you're there. For example, cut a coupon of known size (PCB with bare or plated copper), etch it, then titrate the solution. Uh, neutralization of the etchant followed by precipitation (as whatever will precipitate, CuO or Cu metal perhaps) might be okay, in which case you get a gravimetric answer (i.e., you weigh the precipitate). There might be a good redox titrant for copper, I don't know offhand (exposure to air complicates that, too). Spectrophotometric can be good, making sure the copper is in a known complex (e.g. diluted enough for tetraaquo, or paired with say EDA to get a stronger color?), impurities being a major downside. Can always make a dilutions and perform fancier analyses like atomic absorption (your sample gets aspirated into an oxyacetylene flame, then light from a hollow-cathode lamp is shined through it), or various mass spectrometry methods (of which ICP-MS is the most sensitive?).

Or sand and polish a section, and use a dimensional microscope to measure the plating thickness edge-on. Nice way to measure trace width and etch shape, too.

Or ultrasound I guess, if you absolutely need nondestructive testing. Or eddy current testing, if you have large enough contiguous (pour/plane) areas. Or x-rays, as long as you know the transmittance of bare board versus 1-4 layers of copper. Or neutron activation and gamma spectroscopy. Or...

Note that these all have various and worse errors than direct measurements; you can make some assumptions about the properties of the materials involved (e.g., acoustic velocity and impedance for ultrasound; resistivity for eddy currents; absorbance of copper versus glass for x-ray; activation and decay of copper, silicon and other elements for neutron; etc.), some of which aren't all that obvious (of these, the resistivity of stressed (as-plated, non annealed) copper is probably the furthest out there).

Can also just feel it, but your sense of touch is easily fooled by the sharpness of the edge, and a nice soft soldermask edge feels different from a sharp cut metal edge (with or without burr as the case may be!). So it may take some practice^Hcalibration to get there. For modestly sharp corners, a touch threshold under 1 mil seems pretty typical to me, so >=

Tim

Etch a known area and performa a titration on the solution.

Nah... encapsulate a fragment in resin grind a flat edge and use a microscope...

OK, now there's a good idea, almost easier than getting out the milli-ohm-meter. This method might benefit from a ready-to-strip exposed trace, without solder mask.

Tim's titration method? Are you serious?

What, no? Use an Xacto knife to peel away a layer of PCB copper, and mic it.

How do you do that? It's bonded to the substrate.

I just connect a power supply (it's always on my bench) to a trace and let it current limit. Then measure the voltage drop with my Fluke. Just takes a minute. I do that to a trace that's long enough and wide enough to make for good math. It's often my TDR test trace too; SMA on each end for the current, and intermediate vias for voltage probing on various layers. You can learn a lot doing that.

It's cool to include test traces on boards, and measure resistances, impedances, and plane capacitances.

So would you say it's the best fun you've ever had in your entire life, John? I'm looking for a new, challenging hobby.

I have done a lot more enjoyable things, in and out of bed.

Skiing at Northstar last week, with a family of crazy Italians, was probably more fun than measuring trace resistances. I taught three of them to spin, and two to attempt some features in the terrain parks.

Think I'd rather measure the trace resistances, but each to their own.

Find somebody at the fab and say, "I'll give you a barometer if you tell me...."

But seriously, you can't etch it anyway after it's coated. Can we assume the solder mask is a constant thickness across the board? If so just use a micrometer on a section with foil and a section with no foil.

Or you could drop it off a building....