Joining SMD component pads together on the PCB

[rokspy]

So recently I have seen a practice to join SMD pads together, which seems like a super unnatural thing to do. I can't figure any major reasonable reason to do it from the electrical standpoint besides a bit smaller impedance for the signal/return maybe. On the other hand, i see that the solder flow is probably gonna be uneven and there is not a huge heat relief, thus harder to solder. Are there any other significant reasons to doit/notdoit?

[nvmR]

Hard to understand exactly what was planned here, but maybe it is preparation for different footprints? both 0402 and 0603 or something of the sort?
Could be to make it easier to bridge it using solder maybe.

It is common practice to make footprints larger for hand soldering, but this probably isn't that either.

[thm_w]

Limited space, on a project you are going to hand solder, is the only reason I can think of.

Probably an oversight

[T3sl4co1l]

I guess it's a thing you could do, but there are good reasons not to:
1. EDA usually prohibits it (component collision).  (Assuming these are separate components.)  So you either ignore the warnings, or add rules to exempt those parts specifically?
2. Chip components butted together, will probably form a large solder fillet between them, potentially increasing stress.  Especially if mismatched types, I suppose?  Or kinda not, like: a resistor and capacitor together, the resistor doesn't have side metallization so at worst it makes a thin fillet around the edge of its electrode?  Maybe this would only apply to capacitors and ferrite beads/inductors (having wraparound end cap metallization)?
3. Makes some ballsy assumptions about the accuracy of the pick-and-place machine.  I suppose you'd at least nudge the centers a few 5 mils apart, so the one is unlikely to crush the other when it comes in.  And they probably stick back together due to surface tension when soldering.  Not sure if that's also a tombstoning risk?  Seems like they'd mostly tend to do things together, it'd be hard for just one to go tits-up I'd think.
4. Likewise, it's a lot harder to service.  You're probably melting both off, picking them apart with tweezers, and replacing new parts.  Which inevitably glom back together again, of course.

Electrically, there's nothing wrong with that.  If they're identical components, it's like it's a single double-wide component.  They act in parallel, ESL is lower, etc.  Likewise there's very little inductance between them, which might be of value say for very compact filters (possibly useful at GHz?).  Power rating isn't quite double because they heat each other; there's a missing side for heat dissipation plus they're in each others' heat islands (i.e. where the heat is spreading out over the board / traces).

Tim

[Feynman]

Doesn't make much sense to me. Looks like a special solution for a special problem. Whatever that problem might have been in this case.

[KrudyZ]

There is one case where I sometimes share a pad and that is to partially overlay a pull-up and pull-down resistor. That eliminates the possibility of stuffing both.
You could create a special footprint for this which removes the violation that will otherwise be flagged by the EDA tool, but the overlaid footprints are better for pick and place.
This one is different though, in that both pads are shared.