SIG Sauer P320

Opinions are like rectal orifices. Everybody has one.  So, for what it's worth, read on. 

My personal opinion (spit-balling here) is one of the supplied parts is out of tolerance at a very low frequency of occurrence.  Perhaps as low as one in 10,000.

They clearly have done a tolerance and stack analysis on all the parts and looked at the failure, modes and effects.  

The mechanical CAD model is theoretical.  What everyone forgets is that you actually actually have to go out and look at the goddamn parts.  

All of them.  

My money is on Supplier Controls.  A formal Fault Tree Analysis (FTA) looking specifically at minimum and maximum tolerances would be my next big move.

I would start with a review of engineering changes over time, and get the MCAD STEP files for review.  In parallel I would look at the sheet prints for proper expression of Geometric Dimensioning and Tolerancing (GD&T), the training records for design engineers to show who has been trained on GD&T, and most importantly specific call out on the sheet prints of those component level AND assembly level characteristics that are 'Critical To Quality' (CTQ).  

Suppliers really like it when the design engineers actually disclose which dimensions and callouts out of the hundreds on a drawing will cause the design to, you know,  fail, so they can apply proper diligence and ongoing surveillance on those CTQ's BEFORE Production Part Approval.  Often times design engineering be forgets to denote these features, and it remains a State Secret, leaving suppliers in the dark.  

"Here's your parts.  Enjoy."  

I would also look at the number of Supplier Quality Engineers (SQE’s) SIG has, and make sure they’re the right pedigree AND temperament.  

I say this because my career has been plagued with supplied part conformance issues.  They are like herpes, and recur frequently.  

This is only why I would start here.  

The FTA would also yield the threshold margins for not just one component that is out of specification, but also which combinations of parts that would result in an “uncommanded discharge” or ‘UC’ of the pistol.  I would look at the totality of specifications for critical parts, not just dimensional (material composition, etc.).  

I then would personally inspect the records associated with the inspection plan for each critical component and assess whether the plan was followed, and in turn, if the inspection plan was decidedly deficient (AQL sampling levels, measurement method(s)), I would look to see who signed off on it; if it was approved by leadership, then I’d start a watch of the cognizant QA leader(s) for evidence of competence (or, incompetence) and any telltale CYA behaviors.  

If there was only a paper CoC supplied with material / part receipts (and not material certifications, test and inspection results / associated control charting, etc.), then I would pull the PPAP document and review the overall Quality Plan.  

If the Quality Plan was decidedly deficient, I would place the VP QA/RA on “special projects” and tell leadership they need to do a better job of selecting the HMFIC for QA.  

I would ask for any process deviations approved for any nonconforming critical parts, review all records of rework, verify the associated engineering analysis and risk assessments performed, and assess the overall signature approval authority of this process by departmental function, to assure the person approving these deviations / rework were granted authority to do so by the defined Design Control Authority (DCA) and Quality Authority.  

I saw in the FBI report for the pistol under evaluation that had an uncommanded discharge in Michigan (and wounded the trooper) that there appeared to be wear on the sear / striker assembly noted, and the photographs had a telltale hook on the contact area of the sear (as I recall) [The FBI did not conclude this was causal or a contributing factor in the report].  

This may possibly implicate the stamping process and the related supplier controls for the sear and related components.  Possibly even impact damage during material handling, shipment, receipt, or manufacturing assembly.  I would audit that full process for the critical parts.  

I would have my best steely-eyed Supplier QE (who only accepts evidence and records as inputs, but only listens to the blathering talking heads and suits to discern who actually knows what the hell they are talking about) fly out to the stamping house to inspect the stamping tool, recent tool maintenance, current production samples, retention samples (if any) and associated records.  I would look at the competence of their metrology, and the traceability of measurement reference standards to a National Standard, training of their inspection personnel and their calculation of measurement uncertainty to the equipment performing the dimensional and metallurgical evaluations.  

I would tell my SQE to be professional, but if he detects any smoke or pushback bullshit, to then toggle into full asshole mode and get me on speakerphone.  

I would also tell the SQE to look assiduously at the material certifications and certified third party assay of the raw stock to see if the raw material met specification.   If there is not solid evidence, I would have additional tests of hardness, surface finish, dimensional performance over a sample of 300 parts, and metallurgy against the material specification.  

Footnote: Japan’s Kobe Steel was a world class supplier of specialty metals, and one day, a new SQE asked the question “have we ever audited them?”  The answer was no, because they were ‘Kobe Steel’ dude.  He pointed out that even Tier 1 Class A suppliers needed an on-site audit.  When he did his investigation, he found the records were falsified and management was asleep at the wheel.  This resulted in recalls from Nuclear Power Plants, Aircraft, US Navy ships, medical devices, and all manner of HI REL products.  Buzzkill.  

You have to look in the horse’s mouth to know how many teeth it has.  

In parallel, with the fault tree analysis (FTA) in hand, I would have a widely advertised 'Come to Jesus' with cognizant [Design] QA and Design Engineering to look at the analysis and testing behind the aging and reliability studies that assess the expected wear patterns over time of the critical components, and assess whether consideration was afforded design margin against each use case, and see if onset of uncommanded discharges (UC) was “out of box” or after a period of use, and if so, how much wear and abuse the produced pistols experienced.  

I would also see if there was a staff position for ‘Reliability Engineer’ at SIG and see if they were a refugee from Pratt & Whitney, or a newb freshly minted from an associate’s degree program from the Sam Houston Institute of Technology (or SHIT).  

I would assess the statistical validity of the sample sizes used to establish confidence levels.  This is the most significant contributor to Design and Process Validation efficacy; management does not want to spend the money to run sufficient quantities of product through testing.  

For safety critical (or Mission Critical — think ‘Nuclear Weapons’) the testing is 'the reveal' to weakness in design or process.  In the Semiconductor Segment, the process IS the product (the chips produced are only the accounting units to allocate cost and tally scrap).    

Note: The product that ships new out of the factory is not the product that degrades with use over time.  

The ‘use case’ anecdotally appears to influence UC.  The credible videos I have seen comport with heavier use, in the MIL / Police / and competitive civilian shooting use cases.

Occasional use, apparently not at all.  

There is a video circulating where some dude sticks a screw in the trigger assembly to "take up the slack" on the trigger and repeatedly forces an UC.  This is what you get on the interwebs these days.  

This 'test' is not a valid use case.  Pleeease.  

I would look at the validation plan, looking at drop shock, thermal shock, and vibration testing.  I would want to assess the “vibe profile” used for testing including the PSD factor(s) and over what frequency domain(s), and the test setup used.  I would look at a design FEA (finite element analysis) to see what stress gradients and torsional effects are at play (assuming a FEA was performed -- likely not).  

LEO’s are renowned for not lubricating their duty pistols, even though a majority of them are good “gun guys”.  

I would want to see the process validation test results for the population that was run to failure “bone dry”.  

Note: Trying to tell busy LEO’s to lube their Roscoe weekly is like telling someone attending a P-diddy “freak out” not to get any bodily fluids or baby oil on them.  

SIG is a nimble company.  It is run by an Israeli CEO.  They have entered the markets early for all manner of products, are quick to act, and it has yielded steller results as a company (financially, organizationally, and market timing), but I do not know their level of mastery with Operational Test & Evaluation as it relates to devising Validation Requirements.  H&K is slow, stogy, methodical, and enters the market windows late... but their shit works.  

This would be my approach to bird dogging this issue with the SIG P320 to root cause.  

The rest of this online chatter is just arm waiving, posturing, and narrative control in advance of lawsuits.