[sdiy] Design Process

[thx1138]

On 2/14/10 9:40 AM, "David G. Dixon" <dixon at interchange.ubc.ca> wrote:

>> Hardware is completely different that software. If you have a
>> short or open trace, that is easier, but if you make serious design
>> mistakes, it is much more difficult to correct if it is not a protoboard.
>> But you guys already know this I am sure.
> 
> (Not to drone on about it ad nauseam, but) that's one of the big advantages
> of rolling yer own PCBs.  Circuits are infinitely easier to assemble on
> proper PCBs than on protoboard.  In fact, I rarely breadboard circuits
> anymore, because I find that laying out and making a PCB is actually much
> faster and less frustrating than trying to figure out a breadboard layout.
> With homemade PCBs, if you do need to change the layout for whatever reason,
> you don't feel like you've thrown good money down the toilet or lost a bunch
> of time.  You just make a new board, desolder the old one, and you're back
> in business in a couple of hours.  In fact, I always assume that the first
> board I make is a throwaway, and I'm almost never wrong.  The second board
> is almost always a keeper.  Only very rarely do I have to make a second
> revision, and then usually because I overlooked some subtle issue of circuit
> design of which I was not aware; never because of layout error.
> 
Hi David,

In my case , I design a circuit and simulate it because it is usually in
45nm or 90 nm geometries.

I use tools to measure trace impedence and I also have tools that allow me
to do the same on schematic capture and layout on pcb.

I use Cadence Allegro tools and Altium DesignWorks for most of my layout.

Doggone Arm processors are real touchy about mDDR and DDR layout. Especially
when designing DDR2/DDR3 circuits.

The other issues is Mask charges are so high at 45/90nanometer that we
really try to simulate all sensitive areas first.

When I say prototype, I usually build a CPU daughterboard and a mother board
as 2 separate items. This way I can control a bit more of the layout in
sections.

I have been doing this a while and I still feel like it all comes down to
Analog design anyway with these high speed circuits. I wish my students I
mentor would choose more analog design and Material physics classes before
trying to build exotic circuit layouts.

Controlled impedence circuits are a must when working with much of the newer
CPU systems.

We even have to do landing zone examples for customers because of the 6-10
layer designs.

But in the case of PCB versus prototyping, I agree with you, it is much
quicker to build the circuit up on a board than trying to get a proto
breadboard fully functional.

In old school analog filter design, I have a breadboard I still use just for
one off's and I get pretty good results, but it usally get messy past four
pole filters.

Regards,

Terry