track width change

Hi all and thanks for the replies.

I know of the 0 Ohm R trick to isolate nets but never thought of refining

it to a pair of small, shorted pads. It does result in physical separation

where the net size change happens, which is something I try to avoid with

bypasses on supply lines.

My 20 and 40mil track width are a hangover from days of DIY board making. I

do a quite a bit of soldering and desoldering on professionally made

prototypes and 5mil  tracks and minimal pad sizes don't stand up too well,

also a 40 mil track will cope with the occasional accidental over current

far better than 5mil. I consider the PCB fabricator's design rules a

minimum.

Rachael, "Changing your net class rules so you can finish the routing isn't

really the right way to do it", couldn't agree more, that's why I said it

wasn't very elegant and asked for the correct way.

Maybe I should explain my situation, I'm not a professional, I've made the

occasional board now then over the last forty or so years. Back in the day

it was sticky pads and black tape in assorted widths (what was it called?)

, much easier now with a bit of computing power and Eagle.

I appreciate  what you say about the auto router not being perfect but for

what I do it does a far better job than I can. My last board was a dozen 74

type packages in a random logic design along with a few opto isolators and

about 60 pins of connectors. Now, just the number of possible combinations

of package positions and track routing is far more that I can cope with. My

approach is to roughly group components together, run some power and ground

and the let the auto router have ago, followed by some close inspection and

assessment, and if I don't like the result I start adjusting.

I don't have to meet exacting standards, no GHz RF running around or sub

nano second propagation delays to worry about, it's not "mission critical"

but that doesn't mean I'll be happy with a rubbish job.

I understand there is art in a good layout as there is in any engineering

endeavour and please, please don't interpret my comments as in any way

trying to denigrate the professionals who prefer to route manually, it's

just that it's a skill I don't have. If I was to do ten hours a day for ten

years I might start to get somewhere but as I've probably done less than

200 hours in five years I'm not going to get there.

I'll take the advice offered in your last paragraph, I see it as basically

an incremental approach, let the auto router do a bit, fix/tidy up by hand,

repeat until done, as opposed to put the go button and hope it finishes.

Fortunately I recently discovered the power of Eagle's command line input

as opposed to WIMP input.

Ken.

--

EAGLE support forums at http://www.eaglecentral.ca :: Where the EAGLE community meets.

Hi all and thanks for the replies.

I know of the 0 Ohm R trick to isolate nets but never thought of refining

it to a pair of small, shorted pads. It does result in physical separation

where the net size change happens, which is something I try to avoid with

bypasses on supply lines.

My 20 and 40mil track width are a hangover from days of DIY board making. I

do a quite a bit of soldering and desoldering on professionally made

prototypes and 5mil  tracks and minimal pad sizes don't stand up too well,

also a 40 mil track will cope with the occasional accidental over current

far better than 5mil. I consider the PCB fabricator's design rules a

minimum.

Rachael, "Changing your net class rules so you can finish the routing isn't

really the right way to do it", couldn't agree more, that's why I said it

wasn't very elegant and asked for the correct way.

Maybe I should explain my situation, I'm not a professional, I've made the

occasional board now then over the last forty or so years. Back in the day

it was sticky pads and black tape in assorted widths (what was it called?)

, much easier now with a bit of computing power and Eagle.

I appreciate  what you say about the auto router not being perfect but for

what I do it does a far better job than I can. My last board was a dozen 74

type packages in a random logic design along with a few opto isolators and

about 60 pins of connectors. Now, just the number of possible combinations

of package positions and track routing is far more that I can cope with. My

approach is to roughly group components together, run some power and ground

and the let the auto router have ago, followed by some close inspection and

assessment, and if I don't like the result I start adjusting.

I don't have to meet exacting standards, no GHz RF running around or sub

nano second propagation delays to worry about, it's not "mission critical"

but that doesn't mean I'll be happy with a rubbish job.

I understand there is art in a good layout as there is in any engineering

endeavour and please, please don't interpret my comments as in any way

trying to denigrate the professionals who prefer to route manually, it's

just that it's a skill I don't have. If I was to do ten hours a day for ten

years I might start to get somewhere but as I've probably done less than

200 hours in five years I'm not going to get there.

I'll take the advice offered in your last paragraph, I see it as basically

an incremental approach, let the auto router do a bit, fix/tidy up by hand,

repeat until done, as opposed to put the go button and hope it finishes.

Fortunately I recently discovered the power of Eagle's command line input

as opposed to WIMP input.

Ken.

--

EAGLE support forums at http://www.eaglecentral.ca :: Where the EAGLE community meets.