Anyone own a Mercedes car or SUV? 354,000 of them have been recalled because a starter part can overheat and catch on fire. BAD QC?

It could be either poor QC, or the part responsible was made well but to a fundamentally weak design (e.g. not capable of carrying the full operating current) ; or indeed made well to a good design but affected by some external factor.

The problem with a recall is that you never know the actual number of failures, nor the proportion of failures in the total number of the parts in service.

Like ANY system, no Quality Control method can be 100% safe - but they are certainly not guesses!

The Standards used in science and industry are nationally and internationally agreed codes of "best practice" and of repeatability; and are based on, among things, accumulated experience. They are denoted typically by a code-abbreviation, e.g. ANSI, DIN, BS and ISO, followed by the standard's own indexing number. ANSI, DIN and BS are respectively the USA's, Germany's and Britain's own national standard systems; ISO means International Standards Organisation.

"Best practice" is essentially the safest and most effective method for the task, implicitly following formal standards.

By "repeatability", I mean in general, matching some agreed standard as closely as possible every time, such as the use of standard units of measure by having the measuring equipment traceably calibrated to the official references.

Now, manufacturers of items like cars are compelled legally to produce work that meets assorted national or international standards of design and workmanship, primarily to ensure as far as reasonably practicable, their products are intrinsically safe. The operator may be unsafe, but he or she is outside of the manufacturer's control! That italicised phrase carries legal weight in British safety law, reflecting the fact that although you may strive to do everything demanded or expected, and more, to prevent an accident, you can never remove all risk. 

It is also obviously commercially foolish to cut corners to the extent that the products are of poor quality.


In order to comply with the standards, manufacturers may have to do two things.

The first is to present samples to an independent test-laboratory who will test the them rigorously to ensure they have been deigned appropriately for the intended use, and made to a satisfactory quality. This may be legally required, such as for the Type Approval of vehicles; or come under the heading of "Best Practice", commercially valuable but not necessarily obligatory. The laws governing vehicle manufacture and Type Approval in the EU, especially in Mercedes' home country Germany, are very strict.

The second is the manufacturer's internal quality-control system, and this essentially covers making the items of the correct materials, to the correct sizes, finishes, etc.  It can be as simple as someone using a gauge to make sure that an important dimension is maintained within the tolerances designated on the drawing.

To give a simple example - suppose you are making precision-engineering components. You cannot make 10 000 copies of something exactly 1.0000 inches diameter, so the designer gives an allowance - the tolerance - of, say for this item,  plus or minus 0.001 inch.; so one made 1.0005" in diameter would be useable, one made 0.9985 inch diameter would be too small, so would have be scrapped because the error would cause it or a related part to fail prematurely in service. Incidentally, if made to any ISO quality-assurance scheme, those dimensions are more likely to be in millimetres, not inches, in normal practice. 

Both co-operate: in order to ensure the product still keeps to the approvals, the manufacturer must monitor the ensuing production closely.

So what happened to make Mercedes recall so many cars? I do not know the original fault, nor how many of the cars actually suffered from the fault, but clearly it was happening often enough to alert the makers - because their own QC system monitors breakdowns as a last resort verification of product design and construction. Not knowing what actually happened, I would hazard a guess that the faults were not in something made by Mercedes-Benz, but by a component supplier. There the component maker is responsible for his own QC system, although the user (Mercedes-Benz, not the eventual car owner) may well inspect batches of the parts to ensure they meet his specifications in turn.

Now note that new point - inspect batches. It is clearly not feasible to inspect every last nut and bolt intended for a car that itself is made in thousands, so batch controls are used. These do rely on good faith, and a lot of advanced statistics, to work, but they are normal practice, and normally work very well. So they meet the Standards, they meet Best Practice, and meet the test of Reasonableness. 

So it still possible for faulty items - by either inherent design weakness or by physical error -  to slip through, but everyone takes all reasonable steps to ensure the fault-rate is as low as reasonably practicable.


Historically, there is nothing new in this. Standardising nuts and bolts started in the 19C., so manufacturers could rely on buying ordinary nuts and bolts that actually fitted together not just on Machine serial-number 1, but on Machine No. 1001 as well. So did the first attempts at quality-control of metals as their physical and chemical properties became understood. That alone is a vast field with any number of standards of its own thanks to the vast range of metals needed to cover modern life's needs. 


As a tail-piece, consider the Tay Bridge Disaster in Scotland, 1869. The bridge collapsed under a train during a storm of severity common in that area. The spans already carrying a few hundred tons of train, were also loaded sideways by the gale; and could not withstand the complex resultant stresses. Over 60, possibly >70, people died. When the wreckage had been examined and witnesses summoned, the Official Inquiry discovered two serious failings:
1) Inherent, major weaknesses in the design, by both method of building and choice of materials. These could be said to show collective inexperience rather than negligence, as such massive engineering and materials-science were both still in their infancies.
2) Appallingly poor workmanship by the manufacturers of key components and by the contractors who assembled them into the bridge. This was inexcusable: some of the "errors" were attempts to cut corners by choosing to ignore the designer's instructions; others were carelessness, such as leaving out some of the rivets or bolts, as local fishermen had noticed. 

Incidentally, the locomotive was salvaged, repaired and returned to service, but the company's railwaymen - who had lost three colleagues among that 60 or 70 dead - refused to drive it over the more Northerly routes. They nick-named it "The Northern Diver", and it was relegated to minor duties in the South of Scotland.

It was disasters like this, and numerous smaller accidents, as well as commercial pressure for standardising, that led to the QC systems we have now. Quality Assurance plans are made as well as possible in genuine good faith with the best of current knowledge. Nothing man-made can be infallible, but above the very cheapest of trivial items intended for very short, unimportant lives, everything is done to minimise the chances of failure, whether the result of failure is likely to be merely an irritating breakdown or a terrible disaster.