Overlapping SOT-23-3 courtyards?

[I wanted a rude username]

A row of SOT-23-3 packages can be placed with overlapping courtyards, the central pin of one in between the two pins of the next. Under some conditions this could save about 30% of the board area they use.

What do assembly houses think of this unorthodox practice? Would most accept or reject it? Is there a rule of thumb about the degree of permissible overlap?

[thm_w]

I don't see how its any different than placing two 0603's beside each other, with the same clearance you are going to use on the SOT23's. So you'd have to check with the manufacturer what sort of clearances they allow.
The courtyard was drawn by someone, and whoever did might have just arbitrarily drawn a rectangle as it was easiest.

I usually don't use them and rely on 3D clearance checks.

https://www.hi5electronics.co.uk/manufacturing-defects-caused-by-pcb-design/
https://www.eurocircuits.com/blog/good-courtyards-make-good-neighbours/
https://www.pcblibraries.com/forum/how-is-a-courtyard-excess-defined_topic3060.html

[Infraviolet]

From a perspective of hand assembly with an iron for very small batches, or hand assembly with pasting and an oven:

it is definitely doable, and in the paste and oven scenario not even much more difficult than doing SOT-23s with wider spacing. If an automated production line places paste first for all pads then lowers the SOT-23 parts in to place (and has sufficient accuracy and understanding of the packaage shape not to cause collisions) it should be possible. For hand assembly with a soldering iron it would be unpleasantly fiddly to avoid the iron hitting the package bodies and other pins around the joint you were trying to solder, but this could be mitigated by placing and soldering in the correct order.

For low production assembly of your own I'd go ahead and see what it is like, for automaed assembly I'd ask the assembly house you're thinking of using, send them a picture of the semi-overlapped footprints, or betetr a picture of two actual sot-23 chips soldered to that style of footprints.

[T3sl4co1l]

I mean, you can... beware that placement needs to be that much more precise and a misalignment will now cause a short circuit.  Usual design rules I believe are to support rework -- you'll definitely have a harder time doing it yourself, let alone anyone else (I mean, the pros are pro, they don't care, it'll just cost you more ).  Whether auto place will have as much trouble, who knows, but you'll be demanding higher precision.

Tim

[I wanted a rude username]

Thank you, beautiful people! Ideally I'll avoid it, or ask the assembly house what they prefer.

[Kleinstein]

If space is a problem there are also smaller cases and dual transistors (e.g. SOT23-6 like).  The courtjards are often not stringent defined for the cases, more like suggested spacing for the parts.
for handsoldering one often would like even more space.

[Kasper]

It's nice to leave room for flipping them over and maybe standing them up after you accidentally mirror the polarity or all you can source are ones with mirrored polarity.

[Siwastaja]

I don't think there's anything unorthodox there. The idea of courtyards having to be rectangles is completely arbitrary made-up "rule". It's a lazy way of drawing, really.

In physical reality, there is nothing rectangular around the components. Placement accuracy (risk of components touching each other during P&P) counts from the real, physical edges of the part, not imaginary rectangle around it. During placement, component is rotated to the final rotation high up in the air, and pressed to the board vertically.

Component height must be taken into account; you can't place a small component right next to a very tall component because the P&P nozzle is small only at the tip and widens higher up. But this has nothing to do with the courtyard shape, you need to make the exact same considerations with rectangular courtyards.

In other words, I do that all the time with absolutely no issues. For significantly non-rectangular components, I draw courtyards that follow the shapes, because that allows me more flexibility during layout. One can always place components more loosely when possible, but don't limit yourself arbitrarily right from the start.

[voltsandjolts]

IPC7351 provides guidance on smd courtyards, and for sot23 it's a square box.
So, placing central pin of one in between the two pins of the next is non-standard.
But at the end of the day, it's up to you and your assembler.

[Siwastaja]

IPC7351 provides guidance on smd courtyards, and for sot23 it's a square box.
So, placing central pin of one in between the two pins of the next is non-standard.
But at the end of the day, it's up to you and your assembler.

These are examples, there is nothing normative in these "standards", and saying not following these suggestions is "non-standard" is quite a stretch.

Similarly, IPC tries to standardize how orientation is communicated in P&P files (standardization of pin #1 direction). Even though such standardization is completely non-limiting and simply helpful, no one follows this standard, most of the component manufacturers and CMs have never heard of it.

[voltsandjolts]

Sure, but for PCB work IPC standards are as close as you're gonna get to a 'normative' standard.
Of course, when there is no applicable legislation in your target markets, you get to choose what is 'normative', according to your corporate procedures and standards.

[Mangozac]

most of the component manufacturers and CMs have never heard of it.

That's not true. Any half decent CM is aware of these standards; how much weight they place in them is another matter.