I understand how error proofing stations work, but as part of a layered audit I need to verify my error proofing stations. I know how to verify with simple stations ie. make a bad part, insert into the error proofing station and verify that it fails.

The question is how do I verify a station makes a part live? I have a station sets a voltage on a part, and another one that welds to a certain value. I can't force the machine to make a bad part. How do I verify that these stations work? Are these stations that I really need to error proof?

Thanks for the help!

Hello, and welcome to the Cove!

As far as forcing error, I'm not sure what your part is, and whether Destructive Testing is an option.

However, I would submit to you that you can vary voltage and weld parameters. I'm not a welder, but have been around long enough to know there's a reason why you pay for a good welder. There are a lot of variables associated with welding. I would think those could be introduced.

Thanks for the reply and the welcome!

I am looking how to verify an error proofing test in general where the test is part of a prcess. We have one automated process where a potentiometer is turned to a specific voltage. The error proofing is that the machine will fail a part if the specific voltage is not hit. I can't force it to hit the incorrect voltage (and therby fail the part), so I don't know how to verify that the station's error proofing is working.

I may not have to error proof these stations - I may be diving into this too deep...

Not enough info.

Thanks for the reply and the welcome!

I am looking how to verify an error proofing test in general where the test is part of a prcess. We have one automated process where a potentiometer is turned to a specific voltage. The error proofing is that the machine will fail a part if the specific voltage is not hit. I can't force it to hit the incorrect voltage (and therby fail the part), so I don't know how to verify that the station's error proofing is working.



If you have a potentiometer to dial in the voltage, why can't you vary the voltage? I'm just confused, I guess. If you know that incorrect voltage can fail the part, then adjust it and fail the part. By varying the voltage, you can determine what the acceptable range is.


Every process has error; one just needs to figure out how to identify it, measure it, and make decisions if it is worth controlling.

How about this.

If that part of the process is 'off' will the line still cycle and process the part to the next position. Error proofing may not be variations of voltage etc., it could just mean the (weld) cycle is skipped and not noticed.

How about this.

If that part of the process is 'off' will the line still cycle and process the part to the next position. Error proofing may not be variations of voltage etc., it could just mean the (weld) cycle is skipped and not noticed.

...simple is better...I agree....

How is verifying a bad part was made "error proofing?" You have not prevented any error. Error proofing prevents errors from happening. Detecting errors already made, in my opinoin, in not error proofing. That might be escape proofing but for sure not error proofing.

In welding, a common error proofing methodology is to create assembly jigs that only allow parts to be welded together in one orientation.

If the electrical system on the welder stop the welder from making a bad part, that may be error proofing.

Regards,

Dirk

It may just be semantics, but I think what you are referring to is mistake proofing which is the preferred approach where applicable. Error proofing just doesn’t let the error (already made) move on.

I should have kept my question more generic… Suppose you have a process where you push a pin in a hole to a depth of 1 inch. I was considering whatever mechanism controls the press depth (lvdt, encoder etc…) to be the error proofing device. My question was how do we verify that this is working. I think the obvious answer is to under press a part (.9 inches) and make sure it catches it, but what if you can’t intentionally make a bad part? Maybe this is not a proper definition or example of error proofing.

Dirk - escape proofing as you called it is traditionally included in the generic class of "error proof devices" or "mistake proofing" The wording isn't exaclty correct, but it is still calssified that way in most of the literature.

on the other hand the OP's situation isn't so much an error detection device where one could introduce a bad part and see it is detected; nor is it a typical error proofing device where one could try to make an error and see if you coudl create the defect (placing a part in the wrong orientation); it sounds like more of a 'tool control'. (if he really can't vary the voltage and see if the equipment 'shuts down' or 'refuses' to perform the function) he could verify by tracking the performance over time...it's more of a validation but sometimes that's all you can do.

Suppose you have a process where you push a pin in a hole to a depth of 1 inch. I was considering whatever mechanism controls the press depth (lvdt, encoder etc…) to be the error proofing device.

IMO, error proofing is performing A DMAIC or other type analyses to ascertain the applicable variables and controlling them. Above, there are multiple facets that would need to be identified to minimize/eliminate error. The mechanism to measure distance, depth, etc. would be one tool. The sharpness of the pin, temperature, density variations of the material, etc. would all come into play.

Does that sound close?

This thread has taken an interesting 'twist'.

One of the challenges that automotive suppliers have is that we don't get to discuss terminology, 'correctness' of application etc. or other, deeper, philosophies. The customer wants error proofing to be done as part of LPA's and that means 'testing' the system with a reject part (or similar) to see if the equipment etc is capable of letting it go on, unchecked.

This thread has taken an interesting 'twist'.

One of the challenges that automotive suppliers have is that we don't get to discuss terminology, 'correctness' of application etc. or other, deeper, philosophies. The customer wants error proofing to be done as part of LPA's and that means 'testing' the system with a reject part (or similar) to see if the equipment etc is capable of letting it go on, unchecked.


I agree. Let's not overthink this. Any device that protects the customers, or reduces the possibility of making bad parts, is a good thing. Go with it.

Thank you Andy, very much. I have been rambling a bit in this thread, as I have been unclear what the intentions/ requirements are of the OP. NOTE: That is not meant negatively towards anyone, just means I have a lack of understanding on this particular application.

Andy (or others), please if you don't mind, elaborate on this process/application in the automotive industry.

Too, a bit about LPA would benefit me also

I am not over thinking anything. The differnece between error prevention (poka yoke, mistake proofing, error proofing) and escape controls (error detection) is not an advacned concept. It is a very important concept.

In all of the literature I have read, the main purpose of mistake proofing is to ensure no errors are made. If no errors are made, then there is no need to detect errors and stop them from escaping.

Advance Product Quality Planning put the emphasis on the process design to ensure quality output. Same thing for mistake proofing - the main prupose is to ensure quality output so the organization should not have to dedicate excess resources to inspection and escapes. Spend time and effort to prevent erros as opposed to spending time detecting them.

The Layerd Process Audit question should concentrate more on verifying that controls are in place to prevent errors and less emphasis on inspections to detect errors already made.

Regards,

Dirk

I'll try to help....I've got some similar fixtures.
Let's say that your automatic fixture twists a potentiometer clockwise until it gets a signal from a multimeter that the voltage is in specification. Can you unplug the connection between the two? Can you put an insulator between your part and the measuring fixture so it never gets a signal? So you've got a pot twister and a meter, somewhere there must be a power supply to turn the part on. Can you turn off the power supply to your parts so they never turn on? What happens? Be creative.

We had an "error-proof" system that worked something like this...until we figured out that the power supply had failed and was running at 1/2 the set input and every pot was set wrong. We had to add a check of of the input voltage to the existing output check in our test software.

No problem, Brad:

Simply put, a Layered Process Audit is a check of a workstation (or similar) in a manufacturing process (assembly, welding, etc) by various 'layers' of management from line supervision to the top manager of the facility, to watch the operator 'doing' the process. These are done at a frequency which ranges from per shift (supervisor) to weekly/monthly. A simple audit checklist ensures the same topics are covered each time plus any actionable items found, which are, in effect, escalated to top management each time they do an LPA.

An error proofing audit is a type of LPA but done by specific individuals (process technicians etc.) Typically it consists of running a known reject part down the process to 'trigger' any error proofing systems such as position detectors, vision systems, etc. Other systems rely on proximity detectors which can be 'fooled' and tested to see if the line still functions and will 'accept' defective parts.

The underlying reason for these 'audits' is to re-inforce operators 'following work instructions'............because some people think that's why we get poor product quality!

There are other threads here about LPA's (I don't know how to link them:o )

I am not over thinking anything. The differnece between error prevention (poka yoke, mistake proofing, error proofing) and escape controls (error detection) is not an advacned concept. It is a very important concept.

In all of the literature I have read, the main purpose of mistake proofing is to ensure no errors are made. If no errors are made, then there is no need to detect errors and stop them from escaping.

Advance Product Quality Planning put the emphasis on the process design to ensure quality output. Same thing for mistake proofing - the main prupose is to ensure quality output so the organization should not have to dedicate excess resources to inspection and escapes. Spend time and effort to prevent erros as opposed to spending time detecting them.

The Layerd Process Audit question should concentrate more on verifying that controls are in place to prevent errors and less emphasis on inspections to detect errors already made.

Regards,

Dirk


What I meant by overthinking was that the terms are used somewhat interchangeably, to the point where the differences are lost in context.

However, I trust we all agree there are different levels of execution, regardless of which term is applied.

The best approach is a system that prevents the error from even occuring. Sometimes that is not possible, or feasible. Many times, the best we can do is put automated inline sensors and inspection equipment to 100% check the output, without adding labor. This does not prevent the making of bad parts, but prevents them from passing.

Regardless of which term we assign to that, it still protects the customer, but is not as ideal as preventing it in the first place.

Either way, the customer receives benefit. The Layered Audits are supposed to verify these systems work.