Layout screwup

Grin.. I've made worse*... I'm trusting it worked after all that.

George H.

I haven't gotten that far, but it beeps out and it's not too high a frequency, so I'm pretty confident.

A great big soldering iron works well to remove ground planes -- copper doesn't stick to PCB at soldering temperatures (this is part of how pads get lifted), so if you can tease a corner up you can take the rest off with the aid of heat and a pair of pliers.

I've never taken an _entire_ ground plane off, though.

If it's any comfort we have all been there. Sometime worse than that, like when a manufacturer spec'd the pin order counterclockwise against all convention and ...

There's an ST BlueTooth module that has no indication whatsoever of the actual footprint pinout. (The footprint itself is fairly vague, too, but they have most of the dimensions present at least.)

Sure, there's a schematic symbol with pins, suspiciously in the same pattern as the footprint itself. But no numbers on the footprint, so who knows.

Turns out, using the most likely matchup, it worked. But it would've been nicer to say that (rather than thumb my nose at the datasheet) at the design review...

Tim

I saw one board (not mine!) like that. The FPGA sat on a ball of red wires. They made the layout guy do the hack, since he screwed it up.

I've had cases where the manufacturer forgot to say what the bottom pad needs to be connected to. "I just need to know whether it's ground or V-" ... "Ahm, well, ahem, oh! Good questions, we'll get back to you about this".

I've got one that's kind of related. Not actually a fail, but classic crappy data.

FT245RQ. They *call* it a QFN-32. I build a footprint that was IPC compliant to the dimensions given, for packages called "QFN". The board worked, but the solder joints looked damn suspicious. As in, solder blobbed up on the open ends of the pads. WTF?! I can see the metal on the sides of the packages, and none of it made a fillet!

On closer inspection of the solder joints, and the package itself, _the pads (on the bottom) do not actually extend to the edges_. There's a little exposed metal on the sides, like you expect to see in a QFN, but it's not continuous with the bottom metal.

What it is, is an LGA, with the arrangement of a QFN!

So ridiculous...

(And yes actually, that's the worst, recent, footprint 'accident' I have to show for myself. Which either means I'm damned good at reading datasheets... or I don't do nearly enough designs. But I prefer the former.)

Tim

That sounds like my very first layout screwup, when I was a TA for a high school electronics course (I didn't take any electronics courses in high school, but I was a TA -- it's a good way to learn stuff without having to mess with getting a grade).

I did this lovely layout, with the one chip (14 or 16 DIP, IIRC) laid out in mirror. The teacher etched a batch of boards, and then taught an entire class full of sophomores how to solder a chip on the back side of the board (and, I guess, gave them an early warning of who does a lot of extra work when a design guy screws up).

Always check packages twice, and VCC pins ;D

Cheers

That's common practice in the wonderful world of RF. Admittedly, I haven't done it very often, but often enough to know how it's done. My favored method does not require heat. Instead, I score an outline of the area to be removed with an Xacto knife. Then, I use the blade to undercut the copper ground plane and peel it off. Adhesion varies by vendor and material, but even the hardest glue cannot resist a sharp Xacto knife powered by a determined PCB butcher muttering profanity. Sometimes, the copper comes off in little pieces. Other times, it peels off nicely in one piece.

Ah, I had one board that I had to convert a SOP into a PLCC because I had the width too narrow. Good times :-)

I saw a case where the datasheet for the first silicon showed the pins in the places experienced by the silicon designers. The PCB was layed out as if the datasheet was typical for system designers.

The first PCB got very hot.

The second PCB had the ICs mounted on the other side, which was easy with through-hole components. It also made a good conversation point when demoing the system to interested parties!

Just has some extra air cooling. Nothing wrong with that.

Sylvia.

Well not the entire ground plane... just near the front end. It's when I learned that high impedance TIA's and capacitance to ground don't mix too well. (Yeah, yeah I should have known, but for me I needed to have it shoved in my face.)

George H.

There are Molex connectors with two different pin numberings on the datasheets. The cable guy gets one and the layout guy the other.

We've all done it. The nice solution is to make a little convertor PCB that mounts above the main board. Just be thankful we don't layout with paper tape any more.

NT

This one?

See? There's always someone who had it worse!

Yup, that's worse. Why not just spin the board?

That was my thought but some managers get tight-jawed about time and money. OTOH, that kludge wasn't just an hour wasted, either.