Small BGA pad size - likelihood of success?

[zigenz]

Hi guys,

I'm trying fan out a 36 pin BGA for prototyping - I had a look at this video by Dave:

He's managed to fan out a 36 pin 0.4mm pitch BGA on the top layer; this is what I'm trying to do.  I tried the same thing with a random 36 ball BGA I had in my library assuming a 3/3 mil process and it wouldn't work - the pad size was 200 micron as recommended on its datasheet.  Makes sense since at 200 micron, assuming an X/X mil process capability yields X = 66.7 micron => 2.7 mil < 3 mil.

A closer analysis of the video (i.e. snapshot + scaling pad size off grid) indicated that Dave had used a 6 mil (152 micron) pad with 10 mil (254 micron) mask (i.e. NSMD).  I looked up the datasheet for the IGLOO Nano UC36 package; they recommend NSMD, pad diameter of 230 micron with a 330 micron mask, suggesting that the pads are 34% smaller than recommended!

I'm just wondering based on others' experience, how this is likely to go assuming I use 6 mil pads with a 10 mil mask?  I'm just prototyping, so trying to keep the NRE cost down.

Thanks in advance for any guidance. 

Cheers,
Z

[EEVblog]

Not sure I'd even have the board file any more in order to check.
Are you actually using the same Igloo nano part?

[zigenz]

Nah, it's an entirely different part altogether, but still 0.4 mm pitch.  Recommended pad dimensions are approx. the same however.

Ignoring that specific part, any pointers on successful reflows for 0.4mm pitch BGAs avoiding blind vias would be most appreciated!  It's just for prototyping, so not THAT concerned with long-term performance, heat-cycling stress resistance etc.

Cheers,
Z

[zigenz]

Hey guys,

Hope everyone had a great Christmas and all the best for 2020!

Just FYI, I managed to get it done using a 2.5/2.5 mil process capability which quite a few board houses are competent with.  My actual target device was a 52-pin package which didn't require a "tradtional" BGA breakout in a sense.  I only used the narrow trace in the vicinity of the BGA, and expanded to a minimum of 3 mil as soon as possible.  To fit in the board envelope, I also needed to go down to a 0.2mm drill with a 6 mil annular ring - again, most board houses that can do 2.5/2.5 will be able to accommodate this too.  See below:



If you are using compatible packages (i.e. where single-layer breakout is feasible), it saves a lot of the cost associated with using laser-drilled micro-vias.

For the less experienced, I recommend the following guideline: http://www.ti.com/lit/an/spraav1b/spraav1b.pdf.  You can largely ignore the package-on-package stuff, but the 0.4mm BGA layout recommendations are invaluable IMHO.

Just thought I'd feed back in case it helps someone else. 

Cheers,
Z