One thing that would probably fix a lot of problems noticed is to switch off the "snap to grid".
are you insane ? the whole PCB world is about grids and accurate measuring. and you are starting off throwing that out of the window.
how can you make an accurate footprint if you cant snap to the grid ?
Here is how i make a footprint : for example a TQFP style package ( four rows of pads )
grab datasheet and find 2 numbers : the pin pitch and the distance between the left and right side, set the x grid to the pin pitch and the y grid to half the column distance. ( you need the center point ) plonk down the left and right coumns. now swap x and y grid and plonk down top and bottom rows. set the y grid to half the width of the body and draw two lines ( top and bottom) with the length of the body . one coordinate is frozen already due to the grid. you onlyt need to set the length of the body, the spacing is handled by the snap to grid. now set the y grid to half the courtyard that you want and draw another two line. select the last four lines you have drawn , copy , paste and rotate the paste line 90 degrees around the centroid.
Done. i can do this in under a minute
The same goes for routing. if you know your channel and track width you can set the grids so that everything snaps neatly spaced.
the same goes for parts placement : wanna evenly space parts ? set the grid , place parts alter the grid. there are rules for parts placement ! depemnding on the proces you may need grid spacing.
For example a doublesided board that will go reflow top and wave soldering bottom. the parts can be denser on top than on the bottom. the pick and place and reflow have no problems. the wave on the other hand is a 'blob' you need a certain space between parts so the wave can get to the pins. so you setup a placement grid for top and bottom that are different.
then there is testpoint injection that also needs to happen on a grid as it determines the type of pogo pin that will be used. that typically uses a staggered grid.
so now we require snapping to :
- pin grids
- body grids,
- top placement grid
- bottom placement grid
- routing grids
- testpoint grids.
some of these can be polar grids ( circular )
and you are going to make a tool that doesn't have grids ? that doesn't work ...
and the tool bloody well should understand the difference between all those grids and the 'electrical grid' whci is a gravitational pull emanating form the center of any electrical object.
if i have a board that uses both metric and imperial parts and am on one type of grid i could not 'snap' to the center of a pad of that part happens to be using the other type.
the electrical grid solves that as it excerts a gravitational pull and allows the track to jump form the defined grid to the center of the pad , or ending of a track.
i haven't gotten that far in DEX yet , but i hope it has such capabilities or the experiment will end prematurely. you can NOT design a board without those capabilities. you may as well fling parts to a sheet of paper and hope they stick.
i am getting scared now ... my gut feeling says it will be a no .. to all of it. i took another look at that demo board ... i get this icy creepy feeling that this is indeed wired gridless which would explain the mess. it also leads me to believe there is no such thing as design rules for inter-object spacing and/or realtime checking against the rules. there are some parts and tracks that are placed very 'funny' where a proper DRC would have slammed on the brakes. it could be that they just used defaults ( which were dialed in wrong)
anyway. i haven't gotten that far yet... first things first
i will try to make a schematic symbol today ... one with proper graphics. .. i hope there is a grid there...
currently there is only one obstacle blocking me ( the rest are 'annoyances' , apart from the bugs like DEX crashing while trying to show the properties panel , or the installer trying to run autotrax instead of dex ) and that is the schematic autowiring. That has to go. i can not use this program with autowiring running. it just makes the schematics look like shit and it is too much work to clean up every time it messes up. especially the dots not being where the are supposed to be annoy the hell out of me.
so get on your horse and tell me how to turn autowiring off. ( I hope it can be turned off ... if noet : hard stop. end of excercise... )
two blockers right now :
- autowiring
- incorrect gerber
those need urgent fixing.