[sdiy] *facepalm*

[Magnus Danielson]

Tim Ressel skrev:
> Experience is the best teacher:
> 
> I've had boards fab'd, only to discover a near-fatal mistake. It is no fun to shell out money for boards only to find out they don't work. Nowadays when a layout seems complete I let it sit for at least a day without looking at it. This allows the neurons to reset. Then I look over the design (schematic and layout) for errors. If errors are found they are fixed, and the cycle repeated until no more errors. 

We had one design where a number of pins burried deeply into the board 
was swapped. The engineer was doing a much deeper facepalm than Aaron, 
but we just said "Let Sten have a look at it" and a day later the board 
was thoroughly patched but working. Sure, we had to do another run, but 
we did not loose critical time as much as you would expect to. Mind you, 
it was fairly high speed digital memory interface.

> Another handy tool when making new library parts is to print out the PCB symbols 1:1 and hold the actual part to the paper to check the footprint is correct. Nothing worse than using the 0.65mm QFP symbol then the part is 0.5mm.

In another incident the footprint and the ordered components did not 
match. We got the board fully populated except for that chip (it was 
just 208 pins and critical component to the design). Another Sten moment 
later and he had handsoldered in 208 silver leads, handling pin swapping 
and everything. 8 full hours it took him. We could actually start 
debugging the board. He did a few more boards...

> Lastly, download one of the free gerber viewer programs. ViewMate seems to work okay once you understand its interface. Check all layers including the silkscreen layer.

Always assume there is an error. Always assume you have been a lazy bum 
and just not found it before.

Use a varity of techniques to test the design. They have different merits.

Cheers,
Magnus