Components Spacing

[gnuarm]

I am dealing with part substitutions because of the shortages and one being considered is using a SSOP in place of a TSSOP (MAX4601CAE in place of ADG1411YRUZ).  The pin spacing is the same.  While the body width is larger, this was accounted for in the board layout. 

The concern is the body length.  We never tested the board with the SSOP devices and there is barely enough room for three of these parts end to end.  My calculations show 6.7 mil (0.17 mm) worst case. 

The convention I've always heard is 10 mil between parts, but when I've worked with layout designers they've applied this rule several times over.  They would make a courtyard that is 10 mil larger than any aspect of the part, then arrange the parts with 10 mil between the courtyards effectively giving 30 mil between parts. 

I've never been able to get answers out of assembly houses as to their design rules.  Seems they just focus on working with whatever they are given. 

What component spacing do you use for common surface mount parts?

[asmi]

I don't think assembly houses are concerned with distance because placement tolerances should be accounted for in the footprint design.

As for me - if there is a chance that part will need to be replaced, I try leaving enough space to get in with the soldering iron. Otherwise I pack them as close as possible, because space on multilayer boards is expensive, so no need to waste it.

[gnuarm]

The assembly house has to be concerned about the spacing if they can't assemble the board without parts getting knocked off the pickup nozzle. 

I managed to talk to the head of the shop floor at one point and he said their machines have very high accuracy, but didn't say for sure they can do it.  The more I think about it the more I'm convinced 7 mil should do the job, but it doesn't matter how convinced I am.

Unfortunately I may be caught in a chicken and egg situation.  My assembly house does not want to mess with this until I give them an order and my customer won't finalize the order with me until I can tell them it will work.

[Mangozac]

You need to look at IPCC assembly density specifications.

[SMTech]

If 0.17mm is the worst case using the max dimensions from the datasheet including the mold flash protrusions (which is the bit that going to snag if it does), you'll probably get away with it, mostly. It is a shame its 3 in a row, with two you can tweak the placement values to place them very slightly apart and then let the paste pull them back. Placement on modern equipment is as accurate as the imaging can make it, if you have some mangy old or re-tinned stock that upsets those centering algorithms the results might not be what you want. Equally if your talking about your $4m tax dodging masterplan - do a trial run.

[asmi]

The assembly house has to be concerned about the spacing if they can't assemble the board without parts getting knocked off the pickup nozzle. 

This will never happen. Like I said above, positioning tolerance is one of the reasons footprint is larger than the actual part. Besides, modern PnP machines are very accurate (which is a necessity when working with ultra-small parts like 0201 or 01005), so as long as your footprints do not encroach on each other, pnp machine should be able to place them with no problems. They always know exactly how part is positioned on a nozzle (thanks to vision system, which is installed even on a cheap Chinese machines these days), and they use than info to place part exactly where they are meant to go.