Pick&place files: Component rotation on the bottom side

[Lukas]

Hello Everyone,

I recently added Pick&place export to Horizon EDA (see https://www.eevblog.com/forum/eda/horizon-eda-version-1-0!/new/#new), but I haven't been able to conclusively figure out how the zero rotation is defined for parts placed on the bottom side. See the attachment for what I mean: Is Q3 or Q2 at the correct angle? This boils down to whether a component is moved to the bottom side of the board by mirroring it about the X or Y axis.

So far I've seen software do both, but is there some kind of standard, or does everyone do whatever they please and it's up to the assembly house to determine the proper orientation?

Lukas

[Deni]

I am not sure that there's a standard, but I prefer Y-axis mirroring. Similar thing to Pin 1 reference location for 0-degree rotation, where I like lower-left corner. Perhaps due to fact that most screens
have better X-axis resolution, so with that kind of orientation you can display part footprint bigger. Our p&p machines (Universal GSM) can transform data any way you want, so it is not that critical.

[nctnico]

You can ask an assembler but to me the most logical would be to use the coordinates from looking at the top of the board and mark the component as 'mirrored'. This makes checking the XY file versus the design (as seen on the monitor) easy. This is what Orcad's Allegro seems to do. Another issue is the part's origin. Recently an assembler asked me for an placement file which uses the centre of the part as the origin instead of pin 1.

[Deni]

All p&p machines use centroid coordinates as a part position reference, no wonder there.

[Pseudobyte]

Here you go! Hope this Helps. Rotation is always from the top in pick and place data.

[Pseudobyte]

As to the standard question. The format is defined in ODB++ as x, y, rot, mirror.

When you provide data to a CM they will generally be using some sort of data preparation tool. They will take your gerbers, and your placement data, interpret  the rotation mirror and x/y data. They then will correct for tape rotation of parts. As long as you give x, y, rotation, and mirror you should be fine. I would not modify the output of the tool to interpret the rotation based on mirror.

[Lukas]

Thanks a lot for trying this out in Altium. KiCad does it this way as well, so I'll keep the export this way even though it's the other way round in the board editor.

[Feynman]

I think the convention for part rotation on the bottom side doesn't matter too much (I've seen both). Either way, there is a lot of manual work involved in the assembly shop, since there is no standard format for XY-files.
Regarding Horizon EDA I would focus on generation of neat assembly drawings (PDF). I think that's much more important than the actual rotation convention.

[b2che]

Hi all, I've been banging my head about this issue today, and enjoyed the nice (and I believe correct) diagram form Pseudobyte (thanks!) However, I think it looks a bit more interesting on a part that is not symmetrical about an axis, so here it is for a SOIC8. You can see the interesting Altium behavior of mirroring over the x-axis when moving a part to Bottom. That is, the footprint is mirrored over the x-axis to move it to Bottom, and then the rotation angle is added counter-clockwise as seen from Top.

[Feynman]

Here is how Pulsonix does it: