I have used 40% of replacement cost as time to call it quits.
Odds of success vs total cost to assemble, especially if it fails test the
3rd time.
At times no rework. 1 returned assembly would generate more work than a
month of scrap. Applies to high yielding product with demanding customers.
Limit site to 2. If the 2nd doesn't do it t may be the wrong part called
out at TS or the technique needs to change.

Favorite limit was when index cards were used. Stopped them after the
first  card filled up than after 1 side full. Scrap did not go up do the
cost of components used.
High yield after several inventions to completely change design and process.
Do not recommend hand lap soldering (18) 0.0009 in wires to a 10 mil thick
PCB with untested 1,800 turn transformers.

On Wed, Mar 28, 2018 at 5:06 PM, Yuan-chia Joyce Koo <[log in to unmask]>
wrote:

> you mean the 10x I usually disagree with MFG chaps?  it usually either a
> MEMS or RF shield can that need touch up... not full blow processor BGA
> (MEMS gyro or accelerometer that required absolute flatness, or large funny
> shape RF metal shield can that warp during reflow because the vendor forgot
> to stress relief after cold work)... or new parts that wrap  like potato
> chip ;-). parts replaced are usually throw away - almost feel like someone
> practice their skills for so many rework... ;-(
>  or your CM paid as extra for rework rather than paid by achieve minimum
> 1st pass yield (good old days, without minimum 1st pass yield you don't go
> to production... but I guess since the design/MFG decoupled, it not apply
> now). - logistic issue regarding contractual agreement... - CM on the
> forum, don't get any idea... ;-(... ;-(
>
> On Mar 28, 2018, at 3:58 PM, BEV CHRISTIAN wrote:
>
> Phil,
>> BlackBerry’s engineers (all gone now  ☹  ) would agree with Bhanu and
>> Dave Hillman about the # of reworks in one spot.  However, if we are
>> talking about DIFFERENT components I believe, so  tof remember that BB
>> allowed 10.  Why anybody would do that many for something as cheap as a
>> cell phone, I don’t know, but there it is.  Joyce, can you remember if I am
>> correct?
>>
>> Now if we are talking jumper cables, I can remember a 12” x 10” board at
>> Nortel that had 32 jumper wires on it – and that was for a long life
>> telephone switch..
>>
>> Regards,
>> Bev
>>
>> Sent from Mail for Windows 10
>>
>> From: Nutting, Phil
>> Sent: Wednesday, March 28, 2018 3:53 PM
>> To: [log in to unmask]
>> Subject: Re: [TN] when is there too much rework on a circuit board
>>
>> Thanks Wayne.  I am more concerned with thru-hole or cuts and jumps.
>>
>> -----Original Message-----
>> From: TechNet [mailto:[log in to unmask]] On Behalf Of Wayne Showers
>> Sent: Wednesday, March 28, 2018 3:34 PM
>> To: [log in to unmask]
>> Subject: Re: [TN] when is there too much rework on a circuit board
>>
>> To contrast slightly with Dave, 3 reworks was established by procedure
>> and was a direct result of preparing for NADCAP certification.
>> For BGA's, we established a 1 and done guideline for rework so that the
>> assembly was evaluated if a second rework was to be attempted.  This was
>> done mostly as a function of the component cost, but also to minimize
>> potential pad damage.
>>
>> Rework limits are mostly driven by the diminishing bond strength of the
>> functional epoxy.
>>
>> Here are the general guidelines:
>> Note, Prepreg is chemically almost identical to DEVCON 2-ton epoxy.  This
>> has a bond force of 15.5N / mm2, but each successive heat cycle weakens the
>> bond by hardening (crystallizing) the epoxy.
>>
>> Hardening (crystallization) at temperatures above 300C, decomposition at
>> temperatures above 350C.  For my non-metrically inclined, 572F and 662F.
>>
>