Footprint Dimensions Accounting for Tolerances

[Sparky49]

Hi,

Recently I've moved to Altium and have started paying more attention to the footprints I create for components. However, I notice that for a large amount of smt chip packages for capacitor, resistor, etc that the tolerances on datasheets can seem huge, especially for the end cap dimensions. This often results in in footprints with pads far closer together, for example when I use the Altium IPC wizard.

Is it the case that such a wide range of tolerance should be designed for always, or is it fine to use a 'standard' footprint, acknowledging that it some very rare cases it will be incorrect? Or should it be designed for on an on-case basis? Ie, use the large footprint, unless some other factor such as the use of high voltage prohibits this?

For reference I've attached a datasheet for a resistor I'm examining, as well as what the Altium IPC wizard churns out when I punch those numbers in.

Regards,

Sparky

https://industrial.panasonic.com/cdbs/www-data/pdf/RDM0000/AOA0000C328.pdf

[jonwilhelmjr]

Which tolerances are you concerned with. Like the stop mask or copper pad size to pin size.

Sent from my SM-N950U using Tapatalk

[jonwilhelmjr]

From my understanding pad\solder mask dimensions are to help with reflow and preventing tombstone.

Sent from my SM-N950U using Tapatalk

[Sparky49]

Hi Jon,

Apologies, my primary concern is the size of the copper pads. Red in altium, and W and b in the datasheet.

[Siwastaja]

Ridiculously excess tolerances on some datasheets are a pet peeve of mine.

Their grand idea is that they want to 100% guarantee that all parts fit within tolerances, but while doing so, and being lazy/extra careful, and adding excessive safety margins, they sometimes end up with ridiculous numbers which can cause actual problems, like the part being totally useless as per specs, or footprint ending up causing increased risk of solder bridging between pins, or even tombstoning.

It's fairly typical that an industrial process where die-cut and machine-bent metal parts (lead frame) are injection molded together, they could easily guarantee 0.1mm repeatibility - or at least 0.2mm - with basically 100% certainty, but they define tolerances like 1-2mm or sometimes 20-30% of the total product dimensions, just "for fun"!

I tend to try using common sense and experience to guesstimate the actual worst-case tolerances. Sometimes, it helps a lot if you find a similar part in a similar case from another manufacturer, and look what's typical on the market. If you wanted to be pedantic, you would declare the underspecified part as "unusable" and look for another part, but this is often not possible, especially today you need to use what you can buy, and just like you sometimes need to work outside the electrical specifications of a part (due to misspecifications, assumed typos, etc.), you sometimes need to do it on the mechanical part as well.

Often, it's a good idea to look at the "recommended footprint" from the manufacturer. It's fairly typical that this footprint is not generated from their published product dimensions, but instead, is created using more sane set of values. Or just copypasted from somewhere. Of course, sometimes the recommended footprints are broken.

So, there's no easy answer. You spend some time with this problem, and might see problems in mass production with yield, and slowly gain experience to work around poor documentation.