Guru
Joined: 21/06/2018
Location: AustraliaPosts: 1286
| Posted: 04:36pm 21 Apr 2021 |
Warp once again you have posted something I consider derogatory to a post I have made and it has annoyed me enough I feel the need to take you to task over it. If I (or others) post something that is just plain wrong or totally misleading by all means post your own experience or opinion, but be careful & respectful of others points of view which may differ from yours - it does not make you right or others wrong that they have had different experiences.
I can think of many instances in my designs where just because something has the same net doesn't mean I want them connected together wherever the PCB fabricator feels like joining them with crappy etching and inspection. I run sometimes long traces for instance where a kelvin connection is routed back to circuitry for precision or noise free measurement - yes the same net as the ground plane with its multiple instances of connections surrounded by a ground plane and very narrow annular & linear clearance but there is no way I want them joined anywhere except where I specifically join them.
Ultra high speed memory can have multiple buses that are critical to have the same track length, this involves purposely weaving a track back and forth at times to use up length with minimum clearances involved, again the same net but a bad etch can undo your hard work and cause issues that are hard to find or fix.
I was even really clever on a big project and placed hard wired addresses on a back plane with multiple charging units that plug in, all with serial comms, This was done by selectively having plated through holes in a dual grid of pads for address line selection, but I even put annular pads with a centre void but no hole in case I ever wanted to drill through to a like pad underneath and change an address/bank addressing. The PCB fabricator was so helpful he drilled and plated through all these holes for me where he assumed they were meant to be without consultation !
There are also PCB track antennas & loops with stub connections, pcb baluns that might have the same net but they dont share the same function, planar coil windings etc.
I simply posted that unless it is necessary to have tight clearances it is not best practice to rely on the fabricator to get right what you perhaps could have avoided with a little extra care. I was not trying to big note myself or criticise Peters creation, I still consider my advice as valid despite your negative response overtones & claims that PCB fabricators dont make mistakes or pick them up before delivery - that would be a false assumption in my experience.
My RM6 track windings were 6/6 mil, their capability is supposedly 5/5 mil for general 1oz boards and I have found more than 1 issue with the 10 returned arrays of boards.