EEVblog 1512 - Why Bypass Your PCB Like THIS?

[EEVblog]

Why add bypass capacitors on a PCB like THIS?
Answering a Twitter question:
https://twitter.com/furan/status/1586889993234038786
https://www.vgamuseum.info/index.php/cpu/item/993-s3-supersavage-ixc-sdr

[Bud]

PCB designers who do not know electronics ?? 

Did I hear that right?

[EEVblog]

PCB designers who do not know electronics ?? 
Did I hear that right?

Yes, I knew several that came from a drafting background.
PCB layout WAS drafting back in those days.

[johnh]

The small company ( 5 people)  that I worked at when I was getting work experience for the Certificate in Electronics back in 1978, had a pommy draftsman doing  the PCB layout using Bishop tape

[Kleinstein]

If they have to stick to the rather large footprint with spacing, than the simple grid is close to the best one can do.
There are still quite some 0 ohm resistors, that may or may not be required.

In the 1990s there was a time when fast chip supposed to have even more than 1 cap per pin, but a whole sereis of 1 nF 10 nf 100 nF - I never really understood and at least with modern 100 nF caps in 0402 or separate GND and VCC planes, there is no really need for this anymore.

[Fungus]

Check for unconnected capacitors.

If it was me I'd add dummy capacitors just to fill in the spaces if there were a couple of gaps left over after layout. 

[Brumby]

If it was me I'd add dummy capacitors just to fill in the spaces if there were a couple of gaps left over after layout. 

Ah ... Your artistic perfectionist is showing through.   

[EEVblog]

Check for unconnected capacitors.
If it was me I'd add dummy capacitors just to fill in the spaces if there were a couple of gaps left over after layout. 

Agreed, that would be compulsory on a grid layout like this.

[Fungus]

Ah ... Your artistic perfectionist is showing through.   

Either that or my OCD.

[free_electron]

PCB designers who do not know electronics ?? 

Did I hear that right?

Oh yes. Lots of them. The reverse is equally true... lots of engineers think they know how to do a board ...not ! dfm ? never heard of it .. what do you mean parts can crack if they are too close to the edge ? not my problem. Why cant i use smaller drills and place them closer ? WTF is CAF ? Can't you just slow down the conveyer to increase the barrell fill ? ok , ok i'll use a larger drill ... what do you mean the part now "pushes out" and why is there solder all over the topside ? can't you lower the wave pressure ? what is tombstoning ? what is head-in-pillow ? why do i have all these voids under my thermal pad ? can't you use a better paste that doesn't void ?

The routing bit is the easy part... getting all other stuff right for production is a different story. sure, 1 or 10 boards is easy. try doing 10.000 boards or more a day with a maximum of 10 failures.

[T3sl4co1l]

(Just glancing in here, may be repeating some things.)  Also familiar about that design, the via-in-pad suggestive of internal routing or planes.  In particular I recall similar design practices on some old IBM network card (ISA Ethernet) I had, and on Tektronix boards particularly of the vintage of my TDS460.  Especially for Tek I think, and, not sure about others, I think the priority for inner layer routing was a...reliability thing?  Maybe frustrates RE too, I guess.

Tim

[thm_w]

If they have to stick to the rather large footprint with spacing, than the simple grid is close to the best one can do.
There are still quite some 0 ohm resistors, that may or may not be required.

In the 1990s there was a time when fast chip supposed to have even more than 1 cap per pin, but a whole sereis of 1 nF 10 nf 100 nF - I never really understood and at least with modern 100 nF caps in 0402 or separate GND and VCC planes, there is no really need for this anymore.

There are lots of 0 ohm yeah, definitely none of them are actually required as jumpers. Probably just placeholders for series termination, current measurement, etc.