5 replies to this topic

[Mimi2021]

Good afternoon

I was wondering if there is a minimum number of passes that we need to do with the wands, when we are calibrating our metal detector we pass each wand 3 times for each kind of metal, is that really necessary? or can i just pass each wand once? Is FDA has a minimum requirement?

[Jim E.]

We require three passes of each type of test piece, front middle and back of finished bags.  We also, complete a check with two wands in separate bags butted against each other. Tests are conducted every 60 minutes to reduce potential hold time in event of a failure. 

[SQFconsultant]

I am not aware of a minimum, however I have almost always seen operators do their tests running each one 3 times.

 

Except for Bubba at the shrimp plant and he only did it once.

Edited by SQFconsultant, 10 December 2024 - 07:02 PM.

[jfrey123]

3 runs of each wand is what everyone seems to default to.  It's not FDA established in writing, but you are required by them and your GFSI to ensure your verification of a CCP is effective.  Running it and getting a 3 out of 3 hit for each wand pretty well establishes it is functioning within your parameters, whereas running each one just once doesn't prove it will see the metal reliably.  Most of us are using the 3x3 runs to also spot check various positions on the belt or in the box to ensure there are no "gaps" in the detection aperture.  

[G M]

Good afternoon

I was wondering if there is a minimum number of passes that we need to do with the wands, when we are calibrating our metal detector we pass each wand 3 times for each kind of metal, is that really necessary? or can i just pass each wand once? Is FDA has a minimum requirement?

 

The minimum you can get away with is probably once at the beginning of the run and once at the end.  I wouldn't recommend doing the minimum though - if you get a failure you're looking at retesting/reworking an entire day of production.

 

Most third party standards like to see tests done in the 1-4 hour range.  And when it comes to confidence in verification results, repetition greatly increases it.  With that in mind, something along the lines of three passes with each challenge material once per hour is probably 'best practice'.

[Planck]

1,Each test piece is verified 3 times, as well as at certain intervals, and the essence of this work is to verify the reliability of the MD.

Simply put, you must expect MD to detect 100% of your test piece.

For example, if you repeat the test 100, 1,000, 10,000 or more times, can you be sure that MD can do it?

So, this is a simple and acceptable method of verification.

 

2, This of course includes management reasons as well, as it is only necessary to isolate for an hour rather than a day.

( It's not actually a simple equivalence, but the details are complicated to explain. )

 

3,Each package needs to be tested front, middle and back for a technical reason and it is also affected by the rejection method at the same time.

1), The immediate reason for this test is that when the reject device needs to accurately reject a package, then it is necessary to add photoelectricity to the entrance of the MD.

2), The MD must also detect only one package and not consecutively.

3), So the MD needs to know the length of the package or calculate the length himself.

4), So, the MD may be set to an incorrect length, especially if the actual length exceeds the set length.

5), Excesses may not be detected by the MD and may also lead to inaccurate rejection.

6), So, MD would need to verify the front, middle and back.

 

4, A verification failure should not occur in MD.

In general, if you can calculate the reliability of MD, then it should be at least less than 1 in 10,000

When it's greater than 1 in 1,000, we can already feel the instability of the MD more clearly.

When it's less than 1 in 100,000, MD has been fairly reliable.

 

5, A situation that is easy to overlook is this:

If a plant has very little risk of metal foreign bodies, or an extremely low likelihood of them occurring, it is all the more important to ensure that the MD is almost absolutely reliable.

Let's make an extreme assumption: if a plant has 100 occurrences of metal foreign bodies per day, then he would only have to detect 99 of them to be very successful.

If a plant is likely to have this happen only once a month or more, how would you handle the situation?

So, this leads to more specialised companies paying more attention to MD.

This situation requires a more specialised verification of the MD, which usually involves analysing and calculating the reliability of the MD.

 

Looking forward to seeing more and more specialised questions about MD, we are happy to provide as detailed an explanation as possible.