Re: gEDA-user: Test pads in PCB

[sdb@xxxxxxxxxx (Stuart Brorson)]

> Use a manufacturer with good equipment and they can do all the testing 
> on the element connection points. i.e. where the components will be 
> soldered. 

Yeah, but this is post-assembly testing, I believe.  (Or am I wrong?)
Therefore, you can't probe the empty component pads.  You need
dedicated test pads.

> 30-50 mils is huge . . .

I just measured a test pad on one of my "day job" boards.  (The day
job flow is Viewdraw -> PADS with a dedicated layout engineer doing
the PADS work.)  The test pad is 35 mil diameter.  50 mil is probably
too large, you're right.

> far too expensive in board area to be 
> seriously considered for every net. 

A couple of jobs ago I had an Operations VP who insisted that
*every* net on *every* board required a test pad.  Even the 622Mb/sec
busses were test padded.  I *was* able to convince him to not testpad
the 2.5Gbps CML tracks, but that was only because there weren't many
of them.

In any event, correctly done, the test pads didn't affect the signal
integrity, and they were extremely useful for manufacturing board
test.  Therefore, I tend to agree that every net should be test
padded.  I realize that the choice of doing this or not is often a
"company policy" thing.

> My employer uses such a test machine 
> and there are never test pads added to the designs. We have probes 
> capable of probing a 100 um copper area (because we sometime test hybrid 
> substrates too) on our automated tester.

Interesting. Who makes the tester?

Stuart