Finding additional component libraries/footprints?

[Rene_Poschl]

@SGK

Can a pin or pad number be a letter?

Yes but i would advice against it. (But stay away from special characters like $ or similar.)
I would advice you to keep within the kicad library convention. (Or keep your footprints at least kind of similar.) This makes your work a bit easier.

[Rene_Poschl]

@SGK

In this case I wanted all the schematic’s caps to be renumbered as I had added a decent number of new ones. Hence it didn’t seem correct to “keep existing annotation”. The whole point was to rename the schematic. I was not expecting this to undo any layout work. I was expecting the placed items to be renamed according to the new net list but not moved.

Look at the netlist import options. Maybe try timestamp instead of reference for footprint selection.

[SGK]

[quote=“SGK, post:40, topic:4991, full:true”]I was expecting the placed items to be renamed according to the new net list but not moved.
[/quote]

Before

After rereading the netlist:

Not too big of a deal as clearly hadn’t done that much work and I’m still only finding my feet. But could have been very painful had I actually done a decent amount of layout work.

[SGK]

@dchisholm

This sounds like a problem with precedence in the library search paths. If a library name (or component name) is used in two different locations, KiCAD grabs the first instance in finds in the search order, which may not be the one you have carefully edited to fit your needs.

I’m not sure I follow you. Here’s how weird it is. I can see the footprints in Pcbnew (no rats nest connections as pads not understood) and I can even hover over one, hit E, then hit Footprint Editor. Up comes the footprint with the correct library designation but with the old, erroneous pad numbering. Staying in footprint editor I can hit “select active library” choose the very same library and bring up the same footprint - it has the correct pad numbering.

Using timestamp did not help.

[SGK]

@Andy_P

It should not do that.

I agree with you. But it does…

[SGK]

@Andy_P

Two things to check:

1. in your schematic, make sure that the footprint field in the symbol is correct. This ensures that what’s in the netlist is correct.
2. in PCB’s Netlist Import, make sure that “Exchange Footprint” is set to “Change.”

That said, I can see why the netlist import might not actually change the footprint if the name is the same. In that case, delete the old footprint and let the new one come in. Yes, that’s a problem if parts are placed and routed, but it seems like that’s not the case here.

It was correct and exchange footprint was set to change. There was no old footprint as I merely changed the pad number designation and saved it. I could have deleted the footprint and started again…

Quite by chance, I’ve managed to fix the issue (as far as I am concerned it is a bug).

I hovered over one of the footprints on the board, hit E and then hit Change Footprint(s). I left the new footprint name alone i.e. exactly the same as the existing footprint name, and hit “update all footprints of the board”, and then apply. Finally they updated to the correct footprint and the ratsnest connections updated. Hooray, after a day I’ve managed to take an existing symbol, an existing footprint and edit it very slightly, and have it work on the board. Time for a beer!

[SGK]

@Andy_P

@SGK

I agree with you. But it does…

I just investigated a bit … You said that you re-ran “annotate schematic components” on the schematic before generating a netlist. Did you make sure that you chose “keep existing annotation?” If you re-annotate the entire design, then when you import the netlist into pcbnew, you have to choose the “Timestamp” option for “footprint selection.”

A general rule of thumb is to not change previous annotations, just let them be, and only annotate new parts added to the schematic. Sure, your annotations will be a jumble. When it’s all done, run a script and re-annotate the board in a geometric fashion (so your assembly tech won’t hate you), and back-annotate those changes to the schematic.

And I thought this was a cool feature of KiCad. Update the annotation to a logical sequence at the click of a button. Perhaps I did not on that occasion use timestamp. Trap for newcomers.

[Rene_Poschl]

@SGK

I hovered over one of the footprints on the board, hit E and then hit Change Footprint(s). I left the new footprint name alone i.e. exactly the same as the existing footprint name, and hit “update all footprints of the board”, and then apply. Finally they updated to the correct footprint and the ratsnest connections updated. Hooray, after a day I’ve managed to take an existing symbol, an existing footprint and edit it very slightly, and have it work on the board.

Oh now i understand. This is normal behavior.
Until now i assumed you made a new footprint that was named differently (or at least in a separate library.)

This behavior ensures that footprints don’t just change in a finished pcb when you change them in the library.

Yes it would be nice to have a button that says update all footprints from library. (Otherwise one needs to search for all the footprints that have been changed and update them manually.)

A short warning: The placement of text-fields is reset when update the footprints.

The only thing that does not trigger an update in this manner is when you change something in the 3d settings. (This is a bug.)

[SGK]

Thanks guys 'n gals. I’ll get there. Not having the default libraries working properly was a big issue. I’m used to being able to call footprint pads whatever I like, ignoring symbol pin designations because I’m used to creating these linkages when assigning a footprint to a symbol to create a device. Symbol pin “K” for cathode can map to footprint pad 1 or whatever. Hopefully I won’t make the mistake again and if I do I now know how to update a board’s footprints if required. As for annotation, it’s a shame as I thought this a cool feature. Annotation - aka part numbering - should not affect part placement under any circumstances. Onwards…

[Rene_Poschl]

@SGK

Annotation - aka part numbering - should not affect part placement under any circumstances.

As i (and @Andy_P) said above. You can control that when importing from netlist. you need to select the option by time-stamp instead of by reference.

[SGK]

I should explain my comment that the lack of footprint updating was a bug. I understand, of course, that a board should not update automatically for an updated footprint library. However I repeatedly ran Cvpcb and reassigned the updated diode footprint to the parts in question. In my mind, I was asking KiCad to assign the new footprint to the parts in this schematic and, so doing, I expected Pcbnew to read the new footprints when it read the netlist again.

(I’m used to placing devices on a schematic - a symbol and a package/footprint that have been combined via symbol pin to footprint pad mapping. This segregated approach is fine except when updating or redoing the mapping doesn’t work as expected. Had I not re-run Cvpcb and redone the package association of the diodes I would understand that Pcbnew should not have read the new package/footprint.)

[SGK]

@Andy_P

You will quickly learn that the whole idea of CvPCB is silly and you’ll start to build your own libraries with what we like to call “atomic parts,” that is, a symbol which calls out a footprint. You build the part once and then you never have to worry about footprints, pads and pins matching, and all of that.

That sounds more like what I am used to - building libraries which contain symbols, packages/footprints and devices. Then one places devices onto a schematic.

[SGK]

Hi. I’ve made some progress and have created a number of footprints now. I’ve also started to layout parts roughly. But…

Any idea why Pcbnew would decide to all of a sudden duplicate every single one of my connectors? The schematic has just one instance of each. I have re-run the Netlist generator in Eeschema and tried a number of combinations of reading the netlist in Pcbnew: timestamp, reference, delete extra footprints etc. The duplicate connectors have airwires to their counterparts and have the same references and values as their counterparts.