🚀 go-pugleaf

On Tue, 10 Jan 2012 21:10:15 -0600, "Tim Williams"
<tmoranwms@charter.net> wrote:

>"John Larkin" <jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in 
>message news:a6qog71oeiacboqnku38gr0qq7tnf4njfc@4ax.com...
>> He says that a transmission-line pcb trace must always be "referenced"
>> to the same "return current" plane it started with and shouldn't pass
>> through vias to become referenced to a different plane, or else the
>> return currents will get confused or go on strike or something. If
>> that were true, 98% or the multilayer boards on Earth would quit
>> working.
>>
>> He also says that you should add bypass caps and stitching vias to
>> conduct these return currents whenever a trace does via to a different
>> "reference plane."
>>
>> All silly.
>
>You're just spoiled...
>
>8-layer boards aren't cheap.  Heck, four layer boards aren't cheap.  If I 
>can avoid it, I'll pick 2 layer whenever possible.  Great proto prices and 
>turn arounds.  With two layers to choose from, you have no choice but to 
>stitch like a carpet.  You only have room for one ground plane as such. 
>Heck, if you think you have room for two (maybe VCC and GND), you're being 
>lazy and could get by with 1 layer. :)
>
>Tim


8's do get expensive, and you can't do serious stuff on 4, so we try
to use 6 layers when we can. Routing a BGA FPGA to a BGA ARM and then
to a lot of ADCs and such can be a challenge on 6 layers. 

Having a solid ground layer is great for signal integrity, mandatory
for fast stuff. And the power pours make nice heat sinks for voltage
regulators and power opamps and things.

I haven't done a single-sided board in decades. We do do some doubles
for adapters and breakouts, things like that.

But a bare board, 6 or 8 layers, is a per cent or two of our selling
price. It makes more sense to pack on features and sell a few more.

John