search

matthewlai / JLCKicadTools

Tool for using JLCPCB assembly service with KiCad
GNU General Public License v3.0
346 stars 62 forks source link

QFN-48 incorrect rotation in CPL file #5

Closed atanisoft closed 2 years ago

atanisoft commented 4 years ago

When placing a QFN-48 chip (ESP32-PICO-D4 as example) with pin 1 as right side bottom most pin the CPL puts it as bottom edge left most pin.

The KiCad all-pos csv shows it as:

"U6","ESP32-PICO-D4","QFN-48-1EP_7x7mm_P0.5mm_EP5.3x5.3mm_ThermalVias",163.576000,-91.440000,180.000000,top

generated CPL file shows it as:

U6,ESP32-PICO-D4,QFN-48-1EP_7x7mm_P0.5mm_EP5.3x5.3mm_ThermalVias,163.576000,-91.440000,90.0,top
matthewlai commented 4 years ago

Can you verify and see if this is consistent with other QFN sizes? I used QFN-64 (I think) on a design, and the rotation was correct.

twam commented 4 years ago

My SOP-18 was also incorrect. Suggest to include to pin count in the database file as well.

matthewlai commented 4 years ago

It's a bit strange that different pin counts would have different canonical orientations. But definitely, we should include pin counts (that's why I used regex for pattern matching) if they actually have different orientations.

TobleMiner commented 4 years ago

Ack, just had that happen to me, too. See #23

matthewlai commented 4 years ago

I wonder if they just changed the canonical orientations for a bunch of footprints, and we should just nuke the rotations file and start new.

TobleMiner commented 4 years ago

Idk, even EasyEDA does not get it right for all parts Bildschirmfoto_2020-05-09_21-42-44

We were trying to find a method for automating generation of the rotation database, but there does not seem to be a single source with consistent data.

Manawyrm commented 4 years ago

Hm, I've looked over a couple of different ways to automatically get rotation information for all footprints now, but unfortunately wasn't successful (yet).

I've looked at the EasyEDA component API ( https://easyeda.com/api/products/C7171/components ) but it doesn't correspond to the SMT assembly rotation in any way.

Automatically generating JLC orders and looking for the location of the red dot would work (in the json data), but that's really an evil hack and I'd prefer not to do that :smile:

I've sent JLCPCB an e-mail asking for more information about the rotations and asked for a list of parts and their rotation or an export of all footprint preview images (including the red dot) to hopefully be able to generate a full list of parts (and their rotation).

matthewlai commented 4 years ago

Yeah it would be great if we can get an authoritative reference!