|Ask the Experts|
August 6, 2020
Estimating Failure Rate During Rework
Is there a way to estimate the potential failure rate introduced by hand rework of one SMT part. The process involves de-soldering and soldering a new part using soldering iron/tweezers. We're trying to understand the benefit of the rework versus the potential defects induced.
|Expert Panel Responses|
As a general rule of thumb, there is a 10X jump in defect creation between each assembly process.
SMT Reflow will, on average, have a 50 DPMO level. Wave will, on average, have a 500 DPMO level. Hand or Rework will, on average, have a 5000 DPMO level.
Even when you go to best in class, the 10X rule still tends to apply SMT Reflow best in class can be 2 to 10 DPMO. Wave best in class can be 20 to 100 DPMO. Best in class will avoid rework at all costs.
CEO & Managing Partner
One alternative is to examine your rework area and answer the following questions:
Senior Manufacturing Engineer
I'm not sure anyone can answer this question with the information provided. Test failures were typically balanced between process defects and product defects,so to try to determine the failure rate of a component removal and replacement would required knowing the MBTF failure rate of the particular component.
Secondly this also depends upon the type of component being replaced and why it is being replaced. For example, was the component removed to fix a board condition beneath the component and the component was going to be reused, or is a new component going to be installed. I would not suggest using the hot tweezers to replace the component if it is a chip type component such as a capacitor or resistor as the heat is applied to close to the component itself and can damage the component.
There are lots of issues to address but if the operator is well trained I would say the odds are just as good as the first time the component was installed onto the product.
Vice President, Technical Director
Reworking a solder joint typically increases the thickness of the intermetallic layer. Thickening the intermetallic layer can translate to a more brittle solder connection.
In extreme situations, the intermetallic can become so thick that the solder will actually dewet and not want to "stick" to the pad.
It should also be noted that successive rework cycles can reduce the thickness of the pad metallization. This can be especially problematic with the "knee" of a through hole where a substantial amount of the metallization may have already "washed away" during wave soldering.
Technical Support Engineer
How about putting on the right part in the first place?James M. Fournier
Benefit of the rework vs the potential defects induced. I’m only seeing two options to help someone have this discussion of worth it or not. Either we guess with general statements or quantify with data.Wade Geary, Collins Aerospace
The benefit aspect could be very easy or difficult to quantify. It could be difficult in production because it is not likely that we leave large quantities of ‘things that should be reworked’ in a product then get the customer return data (or other data) to find out how big the problem was. Then make further assumptions that we can reuse this data to predict another scenario that has a significantly different root cause than this precise circumstance. It could be easy if it’s the exact same or very similar circumstance that you have already seen, which makes it easy to predict and quantify the benefit of reworking.
On the other hand, the potential failure rate may be possible to predict because defects from reworking generally have a similar root causes (if you are primarily considering shorts/opens/damaged pads or traces) and using similar repair processes. Most facilities also track their own internal repairs, so should have the data to determine this. Of course, if you want to consider every repair process and every type of defect, ending up with one generalized defect rate answer for all of them would be very misleading. One type of part/repair could be 1,000 times more prone to repair induced defects than another.
The first question to ask the person asking the question, can a general failure rate be used? Or do you need a failure rate for your specific scenario? In other words how accurate do you need to be?
If you can quantify each (with relevant data), you can then have a very good discussion of the options and risks.
|Submit A Comment|
Free Newsletter Subscription
Circuitnet is built for professionals who bear the responsibility of looking ahead, imagining the future, and preparing for it.
Insert Your Email Address