Voltage and Ground lines

[FalloutBoy]

Hello,

I have sunk a large sum of money into my hobby project so feel it's better to ask this now.

I am using Eagle PCB 7.10 professional and am re-entering a very large schematic in order to produce a board for a piece of old tech under a modern board design, to give you an idea of the scale of the project I am undertaking the board itself is 13 x 15 inches (330 x 381 mm), is double sided, has 2895 air wires to be routed and contains 3725 pads and 558 components.

The main 5v DC supply runs at 3 amps, this is the largest draw in the whole board.

My concern is that Eagles router although it routes the whole board 100% and optimizes it down quite happily swaps sides of the board a lot as it is routing and I do not know if it is an acceptable / advisable / safe or done practice to let supply lines and grounds do this - yes its only 15 watts we are talking about; however is all this side swapping ok or am I better to drop the auto route and route the whole thing by hand?

What possible problems could occur?

 

[ads-ee]

Well stitching the board with vias all over the signal/power/ground lines will add inductance to the traces, and that inductance will be worse if there isn't a ground return path for the signal close to the via.

But as you are talking about old tech, does that imply very low speed design with low slew rate signal transitions? If you want "modern" board design, that usually means a ground plane with a minimum of via poking holes in it and absolutely no slots (or big holes) in the ground plane. Of course that is the standard for high speed design practice with high slew rate signals, as every signal should be treated as a transmission line.

Anyway back to your question...it depends on the design. If the design is susceptible to problems caused by having ground loops and poor trace impedance matching then hand routing is advisable.

Just be forewarned that an old tech board being upgraded to use modern tech, could easily stop working reliably, as the parts are now significantly faster, and will necessitate the use of high speed design rules due to sub nanosecond edge rates.

Regards

 

[FalloutBoy]

I am keeping you busy ,
The tech in question is a re-imagining of the Amiga 2000 computer onto an HPTX board suitable for modern day cases, the base board itself is an 8Mhz cpu with nothing faster than 28Mhz running on the board ( the video IC ), when it comes to expansion subsystems the fastest CPU I am aware of that could run on it's expansion system would be a Motorola XC68060RC75 although a 66 would be more likely as I am not sure that the 75 was ever brought out or at least if it was it would be extremely rare.

I have all the necessary parts right down to extremely rare brand new pal TCXO's and it should be a just a matter of getting the board routed nicely, I am trying a different auto routing approach in routing the grounds and voltages by hand and then allowing the autorouter to route the signals but barring that I may have to go by hand unless I can get away with it at such low speeds.

 

[FvM]

Power and ground routing has to observe rules that can't be kept by an autorouter without detailed constraints, and it's probably impossible for a simple autorouter like that brought by Eagle.

A suitable compromise could be to do the power and ground routing manually, implementing a meshed topology with well considered placement of bypass capacitors. Some critical nets like analog or clock networks can be also manually routed. Then fix the routes and let the autorouter do the remaining signals.

 

[ads-ee]

The tech in question is a re-imagining of the Amiga 2000 computer onto an HPTX board suitable for modern day cases, the base board itself is an 8Mhz cpu with nothing faster than 28Mhz running on the board ( the video IC ),

Looks fun.
You do realize you could easily build this with 1 chip (FPGA), a low end video DAC, some extra components for keyboard, audio, and a bunch of passives. On a small board. Of course you'll have to be more careful with the layout and routing, but imaging you could make the Amiga 2000 have a turbo mode (100's of MHz with just a different configuration or a dynamic reconfiguration of a PLL).