I heart a middle size company (Tier 1) which have several assembly processes said it's PPM is 2.0. I want to know is it truth.

As I know one or two assembly process probably get very low PPM, but whether the PPM of a finished parts(have 5 assembly processes)can reach 2PPM? Becuase if only 2PPM, how is it's six sigma level?


thanks

Michael

I heart a middle size company (Tier 1) which have several assembly processes said it's PPM is 2.0. I want to know is it truth.
Right this minute? Who knows? It's possible, but it's also probable that A) the number will change, and B) it most likely bears no real-world relationship to what would happen after actually making 1,000,000 things.

Right this minute? Who knows? It's possible, but ..Agreed, it is not impossible, merely improbable. An exceedingly simple assembly process would help too.

/Claes

Another consideration is the way the PPM is determined. If, for example it is the number of actual rejects per million parts as JSW05 says, that's one thing, but consider this:

If the number is based on Cpk values determined from a standard "fall-out chart", 2.0 PPM represents a Cpk of 1.6 assuming the process is centered, stable and never shifts. If, however, the 1.6 Cpk is viewed assuming a normal 1 1/2 sigma shift over a given time period the expected fall-out will jump to 460,000 PPM (illustrating the "B" scenario in JSW05's post). This is why a six sigma program is so critical to a process. With a 6 sigma process, even counting the normally expected swing of +/- 1 1/2 sigma, there would only be an expected fall-out of 3.4 PPM.

Dave

I heart a middle size company (Tier 1) which have several assembly processes said it's PPM is 2.0. I want to know is it truth.

As I know one or two assembly process probably get very low PPM, but whether the PPM of a finished parts(have 5 assembly processes)can reach 2PPM? Becuase if only 2PPM, how is it's six sigma level?

Michael

One can imagine zero defects even in a complex assembly process <BIG IF> a serious product and process design effort was done and serious poka yokes were put in place. And no, I have never seen it done.

We came close in a simple operation for one characteristic with an effective poka yoke. Zero defects for that charactersitic for 18 months followed by 100% defective. Turns out the poka yoke was re assembled wrong after maintenance. So we poka yoked the poka yoke and went back to zero defects. However this was one characteristic on a simple operation.

Surely someone, somewhere has done well on more complex processes?

I sure hope NASA has complex assemblies under control for the sake of the next team to fly the shuttle.

We came close in a simple operation for one characteristic with an effective poka yoke. Zero defects for that charactersitic for 18 months followed by 100% defective. Turns out the poka yoke was re assembled wrong after maintenance. So we poka yoked the poka yoke and went back to zero defects. However this was one characteristic on a simple operation.Whups :mg: .Well, there we go. Nothing is ever 100% foolproof, but nigh on everything can be improved... Good tale, Caster. Something to remember.

/Claes

98.5% of all our produced parts are hand built. We have enjoyed the status of being able to claim a PPM of 18 or less for the past 3 years from our OEM customers.

While the processes are not perfect and we do make mistakes, being a hand built operation we have the luxury of correcting prior to final assembly and shipment to the customer. One true measure of our quality system at this time is our level of produced parts vs. labor hours. We know that one team member can make .92 parts per hour. If on any given day we have a team member that only made 7 parts or less we know there was an issue that needs investigation.

Just some thoughts.....

While the processes are not perfect and we do make mistakes, being a hand built operation we have the luxury of correcting prior to final assembly and shipment to the customer.

This highlights another of my objections to the whole PPM concept. If I'm the customer and I ask for PPM data (and feel free to shoot me if I do), I want to know about the output of the process, not outgoing quality level. I know that if a supplier has to make 10 units in order to yield 9 good ones, I'm going to wind up paying for the tenth one anyway, one way or another.

This highlights another of my objections to the whole PPM concept. If I'm the customer and I ask for PPM data (and feel free to shoot me if I do), I want to know about the output of the process, not outgoing quality level. I know that if a supplier has to make 10 units in order to yield 9 good ones, I'm going to wind up paying for the tenth one anyway, one way or another.

Let me assure you, you are preaching to the choir. I agree, but our system is what it is. Apprarently in the old days of my company, they were producing in upwards of 11 parts to get 8 good ones, then racking their brains to figure out why they were using so much material, but still claiming a PPM of 0 (they all were repaired prior to autoclave and shipment). One of the things we have done is to accurately define all our scrap. In our industry there is always a little (12 - 18 sq. in) scrap per piece. When we started noticing 100 - 120 sq. in. we knew there was a problem.

One thing our TS registrar mentioned to us, we are preventing rather than detecting.

IEGeek wrote: 98.5% of all our produced parts are hand built. We have enjoyed the status of being able to claim a PPM of 18 or less for the past 3 years from our OEM customers.



We don't use PPM. We use DPMO (Defects per Million Opportunities) & DPU (defects per unit). We count the opps on an assy then track defects recorded whether they be visual inspection defects or test defects - they both count in total DPMO/DPU. Since we have certified assembly personnel, our inspection DPMO is quite low. Most of the inspection defects are recorded based on audits or when an assembler finds a problem with a piece part or prior sub-assy they are trying to put into the next higher assy. Most of the impact to my program DPMO are issues found in testing.

I have a fairly complex product with a surprisingly low opportunity count - we only count those opps in the assemblies we assemble (number of processes applied, number of torque requirements, number of parts on the parts list at that level, etcetera) - the opps in lower assemblies by our internal sub-contractors (i.e., circuit card assemblies) are counted at a different level.

If my programs ever reach a DPMO of 18, I'd say I've worked myself out of a job. What a goal....

Whether 2 PPM or 200 PPM is good or bad is highly dependent on the business you're in. And it's pretty much the same idea and intention whether we call it PPM or DPMO or DPU or PTDNWPMTD (Parts that do not work per million that do).

Another problem comes from how defects are defined! In reliability engineering some disagreement always arises as to what exactly a "failure" is in the MTBF (Mean Time Between Failures) statistic. The same problems exist in having a clear definition of "defect", i.e. if you can quickly pass a file, cloth, or your hand over a marred surface, is it no longer considered a defect, or if the shop people know that an engineering tolerance is unrealistic and go safely outside of it, is it considered a defect?

Then there's always the politics involved when the focus is on numerical targets (see Deming's points 10 and 11) rather than understanding and improving systems (in the long term).

Somebody stop me :bonk:

Maybe it is true.our finished product always are 0PPM.It is really simple and having good control methods.

Too many skeptics.

I had 13 product lines (electronic switches for the automotive industry) which produced about 480,000 finished good/month. 3 of my lines were consistent in producing at a PPM level of zero on a reqular basis.

Total average PPM for 13 lines was 4106 across a two shift operation.

Is it possible... yes it is. Is it sustainable... yes it is.

How do I know this is true. My customer return rate for these three lines over 12 months was zero.

Bragging...... yes I am, and pleased to do so.

Welcome to the Cove!:D I'm sure your experience will prove valuable here. Now, a few questions:

3 of my lines were consistent in producing at a PPM level of zero on a reqular basis.
All production processes are capable of producing at a PPM rate of zero if you only include the periods of time when they are producing at that rate, no?

Total average PPM for 13 lines was 4106 across a two shift operation.
Is 4106 total or average? 4106 PPM doesn't seem a very lofty goal; was there a typo, perhaps?

How do I know this is true. My customer return rate for these three lines over 12 months was zero.
That's admirable performance! I don't understand though how you're using the PPM number. It sounds as if there was a great deal of sorting going on if your PPM rate was 4106 and your customer return rate was zero. How were you able to find all of the defectives?

How do I know this is true. My customer return rate for these three lines over 12 months was zero.

Bragging...... yes I am, and pleased to do so.These are great numbers. But don't forget that Automotive OEMs are continuously doing recalls affecting millions of vehicles, due to faulty components that were never caught during inspection, either at the supplier nor at the OEM.

Not sure if this posted. Will try again....

Thank you for the welcome.

We define a defect as any product that does not meet fit, function, or appearance, of the customer expectations. No missing parts, class A apperance level, no miss assembly, etc.etc.
This is based on our assembly inprocess defect collection data. Any defect found in random Q.C. audits resulted in an automatic 100% sort.
Of the three lines that repeatedly produce at a rate of zero defects, one line produced 36,356 pieces, one line produced 19,031 pieces, and the other line produced 11,920 pieces. (Evening shift produces zero defects on 5 of 11 lines that also produce on day shift, with the lowest volume of 1,438 to the highest volume of 34,386).

For this 30 day time period we produced 418,627 finished goods with 1,719 inprocess defects. (1,719 X 1M)/418,627) = PPM 4106.28

Our assembly method is build -check, build-check, build-check. Our last function is electronic testing. Any fall-out between the first part assembled and the final inspection is recorded as a defect. If we drop an piece/part and we do not know if it fell on the floor, the table or in the assembly, we reject as a defect.
No repairs are authorized on the line, and no repair tools are ever on the line.

My highest fall out line for this monthly time period was 68 defects over 2642 finished goods resulting in a PPM of 25,738.

This is all high pace hand assembly, pull system. No build up is ever on the line. (well, almost never. I think it has something to do with the fact that if I see build up on the line, I take it off the line and throw in the trash, and have the Team Leader count it as scrap.)

With 5 team leaders I would typically throw 2 of 3 pizza parties each month. (many 3 per months) The criteria is zero customer returns for the month, both shifts (team work) for every line in that teams cell. (lines range from 5 to 10 product lines).

And if your really curious, 62 of the 68 defects were machine defects, not man defects. And yes we did fix the machine. (took us two months to find the root cause... tough little bugger).

RE: fall out in the after market. Yes, pain in butt. We spent 3 to 4 month chasing down the root cause of an electronic failure after 2 1/2 -3 yrs of operation across different parts of the U.S. We were finely able to define that the root cause was "armor-all" was being sprayed on the switch surface, then leaking down between the contact and contact point causing discontinunity. This was verified by our customer and a letter was sent out to the auto detailers. (I dont know if as a result of this we began to see armor-all come out in toweletts, but I'm glad they did).

Hello again ASutherland....
Sound like good numbers to me :D . Another question just occurred, hope you'll indulge me....
It sounds like you are 100% testing, maybe this is a legislative requirement in your industry - if so perhaps you'd like to answer this hypothetically. If it weren't a legal requirement, at what point would you consider reducing your test rate? What pattern of testing would you use?
I'm interested in this - in a previous job my boss used often to say "well of course ideally we'd only test 5%" at the stage we were at though this used to scare the bejeezers out of me!

Although I live in an ideal state. I find I must work in the real world. Of course, 100% testing is only 85% effective. So, I can realistically see that 100% testing is ineffective.
So how to combat this? We verify our processes are being followed. If we verify we are following our process, we know that the products we make are 100% good without testing.

Would I send an electronic component to the customer without being tested if it wasn't an industry requirement?

If I verified the process was stable, If I audited on a regular basis and found that we were following our processes. If I see that statistically my defects were consistently zero. Yes, I would feel very confident in sending this finished good to the customer without testing, knowing that it is a good reliable product.

Of course, 100% testing is only 85% effective. So, I can realistically see that 100% testing is ineffective.
The efficacy of 100% inspection is dependent on so many different variables that there is no generalization about it that can be considered reliable in every case.

So how to combat this? We verify our processes are being followed. If we verify we are following our process, we know that the products we make are 100% good without testing.
Amen. Several years ago Honda had TV commercials for their lawnmowers which featured an end-of-the-line inspector named Martha. Martha was responsible for starting lawnmowers before they were packaged and shipped and, as the commercial copy explained it, if Martha pulled on the cord and the mower didn't start the first time, the mower didn't get shipped. I remember the commercial having the opposite of the intended effect on me, because I thought I'd rather buy a lawnmower made by a company whose processes are under control to the extent that Martha isn't necessary.