Manual soldering processes - 100% verifiable, or always requiring validation?

Stoic

Starting to get Involved
I've seen at least two firms take the position that a manual soldering process does not require validation because the only identified CTQ - electrical continuity - is either verified on 100% of devices produced or simply considered 100% verifiable. (To muddy the waters, one of those firms has also validated a soldering process on at least one occasion, possibly in response to a customer request.)

On the other side of the argument, there's a passing mention of soldering in the 1997 FDA Guide to Inspections of Medical Device Manufacturers as a type of process requiring validation, but I've seen nothing more explicit or recent. The MDSAP Audit Model (Ch. 6, §7) still just lists the big four from the QSR preamble, including welding. And while there are of course manufacturers that are validating soldering processes, I've found no warning letters yet for any that failed to validate one. Likewise, while there are several past threads on the subject in various sub-forums, I've not yet seen a detailed explanation of why "100% verification; no validation" is out of the question for any type of soldering process.

Those focused solely on electrical continuity emphasize that the solder joints in question are not considered mechanical connections, i.e. the forces they are subjected to are negligible, and destructive testing is not necessary. Is this a valid argument, or are they missing the point about reliability failure modes associated with thermal cycling, transport vibration, shock, etc? Are there any other publicly available data points to consider?
 
Last edited:

Bev D

Heretical Statistician
Leader
Super Moderator
Yes they are missing the point of reliability.
While it is true that minimal continuity can be verified (tested for) 100% and I supposed cold solder joints could inspected for 100%. There are at least two reliability things that should be ‘validated’: The effect of excess flux and weak joints that ‘break’ after stress. Having been in electronics and automotive industry, I really don’t see the difference (other than ‘scale’) between solder joints and welds.
 

outdoorsNW

Quite Involved in Discussions
I am assuming through hole joints, but please confirm.

Poor solder joints can pass initial continuity tests but fail later due to thermal cycling effects and/or shock. Most PCB materials expand more in the z axis (90 degrees to the board plane) than the other two directions.

Will the joints in question be X rayed? If yes, can the X ray machine look at the joint from the side (ie around 45 degrees)?
Can PCB be easily rotated in the X ray to check from multiple directions?

For SMT, an AOI with the ability to check from multiple angles (we have a 9 camera machine) may be a better way to inspect.

I believe that with good inspection equipment and people who are trained in what to look for and reject, solder joints can be inspected to assure quality. But if you do not have a good Xray or AOI machine (one able to see from the sides) and well trained people, then your inspection will not be adequate to ensure quality.

Also the soldering iron and its settings needs to be a good fit for the job to avoid cold joints or thermal damage.
 

optomist1

A Sea of Statistics
Super Moderator
to a degree solder joints are like ICs subjected to unwanted VOS/ESD....they are the "walking wounded"...in many cases to manifest themselves as failures at a later inopportune date...re: soldering irons....should be part of an initial certification and re-cal program...many a pad was (partially) lifted from an overtemp soldering iron....and yes to a degree this event depends on an over zealous or improperly trained operator
 

Stoic

Starting to get Involved
I am assuming through hole joints, but please confirm.

...

Will the joints in question be X rayed? If yes, can the X ray machine look at the joint from the side (ie around 45 degrees)?
Can PCB be easily rotated in the X ray to check from multiple directions?

There are both thru-hole and SMT solder joints. They are not being inspected with X-ray.
 

Stoic

Starting to get Involved
Thanks to everyone who has responded thus far. From a technical standpoint, it all aligns with my instincts and understanding, but I know that there are some differing perspectives in the organizations in question.

Can anyone point to an authoritative source that speaks to the expectations of the FDA or other regulatory bodies on this topic?

The '97 Inspection Guide is a bit weak: "Validation should be performed (as applicable) for processes such as... manufacturing operations (...soldering...)... Review first and last article test results for continuous processes such as ... automated soldering..."

Two other aspects of the issue to consider (but which may not make a difference):
  1. We're talking about single-use disposable products.

  2. Packaging validation does include some thermal and transport stress testing (although not with OQ limits on the soldering process or multiple PQ soldering lots).
 

Sidney Vianna

Post Responsibly
Leader
Admin
We're talking about single-use disposable products.
That's an important aspect of the context to be considered.

I guess the hardware does not go through a burn-in process, does it? That would minimize the chances of the initial, high-failure rate portion of the bathtub curve of component reliability. And, obviously, we would not be concerned with the high failure rates at the end of the bathtub curve, since the product is single-use purpose.
 

Bev D

Heretical Statistician
Leader
Super Moderator
The difficulties with having to point to specific language in a standard is twofold:
- because there are many processes that fit into any given category the standards tend to define the requirements and list only a few processes that fit. We are expected to honestly apply the definitions and 'special processes' are no different.
- if you have to point a specific callout in a standard to get people to 'do the right thing' you've already lost the game, since those that need to see the 'explicit shall' are not quality focused to begin with, they are trying to do the minimum needed to 'get by'...of course the opposite is also true: many people who are trying to do the right thing must ask for the 'shall' when confronted with a 'rote' person who wants to impose their opinion or misguided belief inappropriately. (I'm thinking of a parallel thread here where the auditor is trying to impose 'special process qualification' rules on a process that is not a special process and their 'requirement' is not statistically sound.)

We must not rely on standards but on logical scientific and statistically sound critical thinking.
 

Jim Wynne

Leader
Admin
I don't think it's possible to confidently valdidate a manual soldering process. Too many variables, not the least of which is operators, and other things such as the required (or expected) production rate.
 

outdoorsNW

Quite Involved in Discussions
Without good Xray inspection for through holes, a good AOI for SMT joints, and people who are trained to understand what the machines show, the soldering needs to be managed as a special process.

In 1997 I don't think there were machines capable of adequately inspecting solder joints in use. If they existed, they were extremely expensive and rarely used.

Too much soldering heat can cause internal damage to a board. A while back at a previous job I worked on a failure investigation of an intermittent open inside a PCB. The structure was a microvia directly on top of a buried via. This type of structure is no longer recommended, but at the time this was not as well understood. (The correct way is to offset the microvia from the buried via with a short trace to allow for a bit of movement.) The difference in thermal expansion between the buried via and the microvia (they expand at different rates) was enough to tear the joint internally if too much heat was applied during hand soldering. If the soldering iron was held to the board for a shorter time, everything would work. But if the soldering iron was held for just a bit longer, possibly as little as half a second to a second longer, the internal microvia to blind via joint was damaged. (The hand soldering was next to the microvia.) In addition to the time factor, the soldering iron temp was 50-100F too high. When the board was cold the joint closed enough to pass a continuity test but when warm the expansion caused an open. Due to the intermittent nature of the problem, it took a lot of time and effort to figure out.
 
Top Bottom