Background While resistors are obviously non-polarised, from a manufacturing point of view they might not be rotationally symmetric: specifically the labelling of 0805 SMD resistors. My board was designed with all the resistor orientations aligned, and specified labelled SMD components (from Panasonic): so I was very surprised to receive a batch of boards for approval with half the resistors rotated 180. They had a very reasonable explanation: they are random in the reel from the manufacturer.

enter image description here


Can I and should I solve this?

  1. Do some pick and place machines read the writing for orientation? (If so, I could use a different manufacturer)
  2. Is it usual for SMD resistors to be randomly oriented on the tape? (If not, find another resistor manufacturer)
  3. Is there consensus that label orientation doesn't matter? (If so, I'm probably fighting a losing battle.)

In through-hole days, we certainly considered it sloppy if the resistors weren't all right-reading: it just prevented errors.

  • 1
    \$\begingroup\$ If you pick 0603 or 0402 resistors they have no markings, so visually they will all be "symmetric" . (some 0603 have IRC) \$\endgroup\$
    – Wesley Lee
    Dec 14, 2023 at 16:01
  • 3
    \$\begingroup\$ When it comes to work, I'm a fan of the phrase "pick your battles". This is not a fight I would waste any time on. \$\endgroup\$ Dec 14, 2023 at 16:24
  • 1
    \$\begingroup\$ My experience, which is not extensive, is that p&p machines don’t attempt OCR but are limited to comparing each PCB with an image of a sample board, so a resistor that’s rotated 180 degrees could be rejected as ‘doesn’t look quite right’ depending on how sensitive the comparison is set to be. \$\endgroup\$
    – Frog
    Dec 14, 2023 at 18:18
  • 1
    \$\begingroup\$ But here’s one example to the contrary: cognex.com/en-nz/industries/electronics/pcb-assembly/… \$\endgroup\$
    – Frog
    Dec 14, 2023 at 18:20
  • 1
    \$\begingroup\$ Automated optical inspection won't care, and neither should you. \$\endgroup\$ Dec 15, 2023 at 6:20

1 Answer 1

  1. Some machines can detect rotated diodes, for example. So it might be possible technically. Although, for resistors, the machine would need to detect actual text including edge cases like "0". Keep in mind this will probably slow down the assembly process and increase cost. It will definitely increase setup cost, since this just isn't day-to-day routine for 99,99 % of all assembly shops. The bottom line is: Most assembly shops would happily align the resistors for you. But they will charge a fortune. Probably because most assembly shops would do this with manual rework.

  2. That should be the norm, yes. Just do a picture search on the internet for taped SMD resistors: On most pictures you'll find that the orientation is pretty much random. Maybe some manufacturing machines will "accidentally" produce tapes with aligned orientation. But no manufacturer will guarantee this. So it just needs a change in machinery or production site and the next batch of the very same part number will have random orientation.

  3. It would be neat, if all SMD resistors where oriented the same way. But that's simply not practical in most cases, even if you could rely on the resistors being oriented the same way in the tape. By connecting both pins of a resistor in a schematic, you also define its orientation as a byproduct without any knowledge of how it's going to be oriented on the actual board. If a resistor turns out to be in wrong orientation during layout, you would need to go back to the schematic and flip/mirror the resistor with most CAD tools. That's just a total waste of time. Remember that your footprint library can't be agnostic of the orientation, either.

tl;dr: Don't bother ;)

  • 5
    \$\begingroup\$ One day audiophiles will discover resistors sound better one way around than the other XD \$\endgroup\$
    – bobflux
    Dec 14, 2023 at 21:24
  • \$\begingroup\$ Thanks, that's extremely informative. Obviously I'm new to SMD production, and it's amazing to me that the resistors are random on the reel, but image search shows it to be the case exactly as you say. Regarding 3 (having to rotate on the schematic), we have an x-y-z postprocessor which rotates the position file (for panelisation etc) and it will normalise R, C, L for us without effort. \$\endgroup\$
    – jonathanjo
    Dec 15, 2023 at 8:39
  • \$\begingroup\$ @bobflux that is easily the funniest comment I've read in a long time. \$\endgroup\$
    – jonathanjo
    Dec 15, 2023 at 8:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.