CNC Machining, QA/QC, And the World of Good Enough

[SomeOtherGuy]

It has everything to do with attention to detail. Everyone puts out a bad product once in a while, but not every company has a "where is my order" thread; not every company puts out canted FSBs; not every company sends parts kits out missing parts.

This. Quality control costs money and requires good people (who also cost money). Anyone with a Mazak and a half-decent operator can machine approximately the same parts. So who will catch the parts that are out of spec or otherwise defective?

The farther you go in any profession, the more you will discover that common sense isn't, and simply getting the basics right puts you at a minimum of the 50th percentile, and often towards the 80th or 90th percentile.

[Hayseed]

Companies like DD, Colt, KAC etc are much much better at stacking tolerances. Others are not. The problem is not in one part of the manufacturing but the entire process as a whole.

Sent from my SM-G930T using Tapatalk

[MistWolf]

This. Quality control costs money and requires good people (who also cost money). Anyone with a Mazak and a half-decent operator can machine approximately the same parts. So who will catch the parts that are out of spec or otherwise defective?

The farther you go in any profession, the more you will discover that common sense isn't, and simply getting the basics right puts you at a minimum of the 50th percentile, and often towards the 80th or 90th percentile.

The person doing the work controls quality. the inspector can only assure the quality is there

Companies like DD, Colt, KAC etc are much much better at stacking tolerances. Others are not. The problem is not in one part of the manufacturing but the entire process as a whole.

Sent from my SM-G930T using Tapatalk

The AR is designed to to eliminate "tolerance stacking". It's designed so that when everything is made within tolerance, the parts will fit. If the parts of an AR don't fit or work together, something is out of tolerance

[MegademiC]

As a man who has spent quite some time assembling / building in the aerospace industry, and then performing meticulous QA/QC, I can indeed understand and agree with this.



This, and the above quote about assembly being crucial to the turning out of a good product, while very true, are not necessarily directly addressing my question re: the production of individual parts. At no point am I advocating for or against a given manufacturer, or a price point, nor am I attempting to create a false equality where there is none.

But, taken as individual parts, and simpler sub-systems (an assembled upper isn't exactly a demonstration of rocket surgery), parts produced out of known materials to known specs are, in point of fact, the same. The TDP defines the specifications, and those specifications have been proven to function reliably. If they call for 9130 or Carpenter 158 or 6061 or 7075 or 4150 or 4140 or if they call for balsa wood, those specifications, if met, produce a part that works.

Assembly is indeed where a great deal of voodoo comes into things. A BCG, for example, can be assembled or disassembled, with minimum handtools in a matter of minutes. There's a few pins, a couple of springs, some washers, and... well... that's just about it. Those are items that we know how to make, and have been making for years, and can now produce at tolerances far beyond what was available at the outset of the AR-15 pattern. There may be an art to staking, but there's a lot of mojo and voodoo and street magic used to describe hitting a punch with a hammer to displace some material. Maybe I am making light of the process because I've done it a few times, I don't know.



This is, in fact, kind of my point.

So what exactly is your question? You set the preface above but I dont see a question.

Going to OP, yes, all companies produce most of their parts within spec. Anyone can make a part within spec. It's doing it repeatedly and containing all that's not that is important and separates companies. I still don't think I answered your question though, did i?

[JC5188]

The person doing the work controls quality. the inspector can only assure the quality is there



The AR is designed to to eliminate "tolerance stacking". It's designed so that when everything is made within tolerance, the parts will fit. If the parts of an AR don't fit or work together, something is out of tolerance

This is correct. Processes should be in control to the point that QA is unneeded, and a waste of money.

Tolerance stacking is resolved in the engineering phase. Companies that produce their own parts shouldn't experience it, and companies that use "bin parts" from another vendor most certainly will.


Sent from my iPhone using Tapatalk

[MistWolf]

There will be no problem with bin parts if those parts are made to spec. The AR really is a marvel of modern engineering and was far ahead of its time when it was developed. It's the first design to control headspacing to the point all bolts are a drop in fit with all barrel assemblies, a leap forward in rifle manufacturing. It's the first issue rifle to fully embrace modern production methods and it was readily adaptable to manufacturing control methods yet to be developed. That's it's real edge over the M14, AK, FAL and other designs

[nova3930]

It has everything to do with attention to detail. Everyone puts out a bad product once in a while, but not every company has a "where is my order" thread; not every company puts out canted FSBs; not every company sends parts kits out missing parts.

That is ultimately the difference in ANY manufacturing environment. Anybody can produce a 1/100 or 1/1000 item but being detail oriented leads to consistency in producing a quality, functional product. When you have high quality processes that are detail oriented and consistently apply them the end result is that the 1/1000 item is the one that's low quality.

Generally in modern manufacturing 1/1000 is actually bad and you're shooting for the "6 sigma" solution in your statistical process control. That is there are 6 standard deviations between your mean output and the nearest specification limit. In a practical sense that means you produce <4 defects per million opportunities.

Scaling quality up to industrial scale is what separates the men from the boys.

[26 Inf]

Tolerance stacking is resolved in the engineering phase. Companies that produce their own parts shouldn't experience it, and companies that use "bin parts" from another vendor most certainly will.

I would agree that Eugene Stoner and everyone else that had a hand in developing the AR and the final TDP for our battle rifle strove to eliminate tolerance stacking. The weapon was designed to be mass produced with no final hand fitting of parts required.

Your statement - companies that produce their own parts shouldn't experience it, and companies that use "bin parts" from another vendor most certainly will - flies against the basic concept of mass production and the TDP. In an ideal system the parts flow into the assembly line just in time to be installed.

Generally those parts are manufactured by vendors with machines set up to manufacture those parts in high volume. Efficiency in production and cost is achieved by this small number of vendors specializing in the production of a greater number of the part than one individual manufacturer needs. Those shops should be more set up to ensure that tooling remains within spec as well as ensuring the raw material is as specified.

As an example of this I would invite you to run around Wichita, Kansas, and check out all the machine and instrumentation shops that exist along the periphery of the aircraft manufacturer's facilities.

In closing, do any of the AR producers make ALL their own parts?

[JC5188]

I would agree that Eugene Stoner and everyone else that had a hand in developing the AR and the final TDP for our battle rifle strove to eliminate tolerance stacking. The weapon was designed to be mass produced with no final hand fitting of parts required.

Your statement - companies that produce their own parts shouldn't experience it, and companies that use "bin parts" from another vendor most certainly will - flies against the basic concept of mass production and the TDP. In an ideal system the parts flow into the assembly line just in time to be installed.

Generally those parts are manufactured by vendors with machines set up to manufacture those parts in high volume. Efficiency in production and cost is achieved by this small number of vendors specializing in the production of a greater number of the part than one individual manufacturer needs. Those shops should be more set up to ensure that tooling remains within spec as well as ensuring the raw material is as specified.

As an example of this I would invite you to run around Wichita, Kansas, and check out all the machine and instrumentation shops that exist along the periphery of the aircraft manufacturer's facilities.

In closing, do any of the AR producers make ALL their own parts?

I'm well aware of manufacturing processes and JIT. If a part is designed with tolerances that will work with other parts, and those parts are in spec, then tolerance stacking should not occur. It's when you get companies using parts--the manufacture of which they have no control over--from vendors, that you start seeing that.

The company I work for, we make everything but the coil steel the products are made from. Whether that steel is used for a product that is rolled, formed, stamped, welded, or machined...we are in control of its design and manufacture. That includes the engineering and design where the tolerances are spec'd. Therefore, if parts are in spec, they won't stack.

My Division, which I am the manager of, has over 32,000 active part numbers. If we have issues with any of them, they are NCR'd back upstream.

On the occasion we use vendor stuff, it is still designed and spec'd by us.


Sent from my iPhone using Tapatalk

[JC5188]

There will be no problem with bin parts if those parts are made to spec. The AR really is a marvel of modern engineering and was far ahead of its time when it was developed. It's the first design to control headspacing to the point all bolts are a drop in fit with all barrel assemblies, a leap forward in rifle manufacturing. It's the first issue rifle to fully embrace modern production methods and it was readily adaptable to manufacturing control methods yet to be developed. That's it's real edge over the M14, AK, FAL and other designs

Fair enough, parts that are in spec should always work. I'm just not comfortable that all assemblers ensure that.


Sent from my iPhone using Tapatalk