I would find Phil's process perfectly acceptable if your qualification X-ray showed 0 voids down in the barrel after the ball attach process, or at least some acceptable hole-fill level.
A void would not necessarily be considered a defect; unless it creates a hole-fill issue of less than 75% for class 2 and 3 product. That is highly unlikely if there is evidence of visible solder already on the target side prior to ball reflow.

-----Original Message-----
From: TechNet [mailto:[log in to unmask]] On Behalf Of Nutting, Phil
Sent: Monday, October 17, 2016 12:40 PM
To: [log in to unmask]
Subject: Re: [TN] Solder source side

For our high voltage multiplier boards we hand solder, then "touch up" to add a solder ball on the non-component side.  This process assumes that there is proper flow through, regardless of "source side".  The addition of liquid flux greatly improves the results when generating the solder ball.

We do have a new vendor that has an automatic solder balling machine.

Phil

-----Original Message-----
From: TechNet [mailto:[log in to unmask]] On Behalf Of Larry Dzaugis
Sent: Monday, October 17, 2016 12:10 PM
To: [log in to unmask]
Subject: Re: [TN] Solder source side

It moves around based on process.

Pin in Paste is the topside/component side.

Selective solder is the side opposite the component.

Rare hand soldering for components only accessible from the top side, not the preferred situation for DFM.
Select components in module assemblies come to mind.
Some designers do not accept sockets.

Inspection callout should include the situation so that the Inspector does not reject valid joints.

Had large problem when selective solder connector went to pin in paste with Inspectors rejecting good joints.

Larry Dzaugis