16 replies to this topic

[Madam A. D-tor]

Hello every one,

I wonder if there are any technical members in here!
(Maybe employees from Loma or MT or just some people with al lot metal detector knowledge)


What do you think is the best practice in checking the metal detector?

As an auditor I see a lot of different ways to check the metal detector.
First of all there is a different in the frequency. Which is quite understandable, for it depends on the sort of products, the number of different products in a period, the speed of the transportation belt, etc., etc. and no less of all the number of products that the company wants to be redetected if the metal detector fails.
The second differency is in when the check is performed. Some companies do the check before the product run and after setting the parameters. So they first set the program, then check the metal detector, and after that start the productrun. When a new product is coming they again set the program (might be different), perform the check and start the product run. I do not think this is good practice, because you can not tell if the metal detector during the productrun functioned well. It is a veification on the settings and not on the metal detector. Does any one have an other opinion on this? (1)
Other companies just perform the check a few times a day (4, 6 or 8 times ar common) or 1 or 2 times an hour.
The third differency is on how the test is performed. When I worked as a QA- employee in the food, I always make sure that the test bullets are transferred in the metal detector with the food package. On top of it or underneath it, for the resistence of the food and the packaging. In a lot of companies they perform the test by transferring only the test bullets. I even found a company (product = MAP-packed vegetables) where the test is performed with a test bullet in an empty, gas filled and sealed package. Again, I think that these two examples arenot good practices. What do you think? (2)

A few companies told me that there is an article (report) which stated that performing Ferro tests is the same as performing x*2,7 Stainless steel tests. (1,0 Ferro = 2,7 StSt, 1,5 Ferro = 4,05 StSt). Unfortunately, no one can show me this article or tell me where to find it. Does anyone have ever heard of this article and know where to find it? (3)

I know (from my time in a food company) that Mc Donalds provides their suppliers with metal detector guidelines (or requirements). Is someone in here, familair whit this document? Is it usefull? Are there maybe other guidelines available on internet? (4)

Would you please discribe on how your food canpany performs the metal detecort checks and why you think this is best practice? (5)

This item seems to me as a nice discussion thing, where we can share and reflect our experiences. I hope you all join in.

In replying on this topic you can use the numbers after the questions.

[suemal]

Hi
have tried to answer as best as i can!!!
1) we check test pieces alone & metal detectable plaster before production starts, when the product setting changes and at the end of the day, we then check the test pieces with the product also at the start and end of the shift and every 2 hours or after a product change, we only test every 2 hours as we do not have many products going through, once this increases then the frequency of testing would also increase.
2) we place the test pieces on top of the product - we found this was the best position from the metal detector company, we also always test using the heaviest product.
3 & 4) can't help sorry!!!


we have had a few auditors asking us to place the test pieces under the product rather than on top, but once i explain that Loma have told us where to place them, they are happy!!!

hope some of info was helpfull!!!

[Simon]

Any other comments?

[Charles.C]

Dear ADT,

Regarding the MD performance, the installation team always claimed they were the only “absolute” checkers since they measure level / consistency of field strengths and other magical parameters. Good for routine business of course.

Position of test pieces - the classic answer is to take “worst case scenario”. From memory the installer said the lowest position is appropriate but not sure on this one (shd ask the installer I guess, brand variation ??). I originally made dummy packages and tried top, centre and bottom positions but didn’t notice much difference. This method gets involved if you have many sizes / weights etc.
Frequency - as you commented, has HACCP implications if a reject occurs, have seen every variation on this one, it’s nice to do it every hour but it’s also a lot of (paper)work.. Again, auditors get nervous if the frequency has rather long time intervals IMEX.

Yr other queries no idea, never seen a guide with such details on the IT, sorry.

Rgds / Charles.C

[Madam A. D-tor]

Is there no one else who dare to describe their working methods for checking the metal (or foreign body) detector?

[Jarve]

Wherever I have worked I always insist on the metal detection being carrried out pre - production, during production (at suitable intervals), and post production at the end of shift. The test pieces are always inserted into product, replicating worst case scenario - also helps to ensure staff are carrying out checks as the test packs have to be changed as and when product changes.

Of course as all you women know "Size is relevant", always ensure that the test piece sizes are relevant to the product, its pointless testing for 5,6,7mm pieces in small products. Loma have their own guide Loma Guide.

I have personally never seen a McDonalds metal detection guide, but I do have a copy of the BRC guidelines somewhere (will try and find it).

[Ciscokid]

There are some very basic and simple rules to use when testing any metal detector. But first we must identify the weakest links for a metal detector.



1) Metal is hardest to detect in the center (x and y axis) of the metal detector aperture

2) The metal must travel at the same speed as the product

3) The metal test sample MUST be inside the product

4) Testing should happen frequently

5) The reject device must be part of the overall testing program


Ciscokid

[GMO]

I agree Ciscokid. I've always learned that the centre of the aperture is the least sensitive point and the centre of the product is the hardest place to detect. (Note this is different for x-rays as they 'see' two dimensionally.)

IMO it has to be like this (and I'll explain why).

Run three packs through the detector which has been set up for the product. This is to check there is no metal in the pack (of course, you don't know the metal detector is working properly at this stage but it becomes a paradox. You must have a metal free pack for your test pack but you don't know it's working until you test it etc.)

Put each test stick (ferrous, non ferrous, stainless steel) into the product. Make sure the ball bearing end is the end in the middle.

Run each test pack through at normal production distances and check each is detected and rejected.

Start running the line.

Repeat at your desired monitoring frequency and at the end of the run as the very last activity. Note, if the product is changed, there should also be an end of run and start of run check.

Note, new test packs should be made up throughout the run as the product tested should be at the same temperature as the product running.

Other checks to do probably once a day (and as I write this, I've realised I haven't done this and need to set it up on Monday!!!)

During the run, remove alternate good packs and replace with your test packs so it goes:

Good Test Good Test Good Test Good

Check that only the test packs are rejected. This is important to prevent waste and for the operator's reassurance. Otherwise, they may think they're getting false rejects all the time and end up sending out bad product.

Set up some test packs with your stick at the front, middle and back of pack and do a normal challenge test with the products in this position.

This is to make sure that irrespective of where the contaminant is, the pack is still rejected.

Edited by GMO, 11 July 2009 - 07:08 AM.

[Tony-C]

Just like to add the importance of checking rejected packs.

Confirm rejection and inspect product to find reason for rejection.

Any findings should be recorded and investigated.

Regards,

Tony

[Charles.C]

Dear GMO,

Very nice usage document.

IMEX, as you mention, for certain matrices, eg product within protective ice glazes, false positives can be a definite headache.

I hate to be argumentative but for what it's worth, the technician who set up a newly purchased Metal Detector I once used categorically stated that the least sensitive point was not in the middle (I can't frankly remember whether it was claimed to be top or bottom "layer").

However I hv not seen any "official" validation either way so am quite willing to be convinced otherwise ?? I remember an auditor asking me the same question regarding construction of test samples used for checking and he was totally satisfied with my anecdotal reply (probably already encountered / accepted a 33/33/33 distrbution of answers )

Rgds / Charles.C

[Tony-C]

I hate to be argumentative but for what it's worth, the technician who set up a newly purchased Metal Detector I once used categorically stated that the least sensitive point was not in the middle (I can't frankly remember whether it was claimed to be top or bottom "layer").
Rgds / Charles.C

Hi Charles

I have always believed it to be in the centre & All the literature I have read says in the centre - e.g. www.loma.com/docs/Guide_to_Metal_Detection.pdf Like to see something to the contraary if you have it.

Regards,

Tony

[GMO]

Good point re investigation of rejects. We had a hold and some projuct scrapped recently because a reject was investigated, if that hadn't happened, we could have sent it out to the marketplace.

I've never heard anything other than the centre of the pack being the least sensitive point. I suppose you'd have to try it and if you're also doing periodic checks at other points, it's less of an issue IMO.

[Charles.C]

Dear Tony,

Thks yr very useful link. After some considerable searching (couldn’t resist the attractive shade of green ) I couldn’t find any disagreements with yr (et al) statement / reference that the minimum sensitivity is at geometric centre (GC). Just shows one should always demand validations

This one has a summary of the theory - http://www.detectapr...sics/basics.pdf

I hv expanded the discussion a bit more since it seemed interesting and I don’t think has been looked at much before here until this ressurection of an older thread.

Well, , I suspect after studying the Loma link and some others that, despite the simple conclusion, the practical implementation of the GC theory is rather more complicated and variable due to the actual situations encountered.. This will for example be the case (similar to my own) where one machine is used for a range of product packagings of significantly different height.

This link, IMO, expressed the general practical requirements quite concisely and logically –

When testing your metal detector, place the test sample into your product, try to position the test sample at a height inside the product where it would pass through the center of the product tunnel. This is the least sensitive position inside a metal detector tunnel. If it's rejected when it passes through the center then it will certainly be rejected if it passes through the side of the tunnel.

http://www.fastecser...om/tsamples.htm

This seems rather different to the idea in the Loma article (I find their text a bit difficult to understand, maybe me) although the intentions may be the same. The relevant extract is -

When you are testing finished packed products on a conveyor system, place the test piece of metal, where possible, at the extreme end of the pack. If this is impractical - e.g. when you are testing individual small packs or sandwich wedges - place the test piece of metal in the centre of the product. Next, pass both ferrous and non-ferrous test packed (packs?) individually through the search head twice - first with the metal test piece at the leading edge of the pack, and then with the test piece at the trailing edge. In each case, you must observe that the test piece successfully enters the reject bin.

I have doubts that many factories are using such rigorous procedures, especially if they are checking say, every hour. GMO is a very applaudable exception perhaps .

AIB offered another generally quite readable document which contained this -

When conducting performance tests, samples should be positioned where they are least likely to be detected, offering the greatest challenge for the detection system. The importance of the correct positioning of the test piece should not be underestimated. Incorrect positioning may signifcantly impact the performance of the detector and reject device.

https://www.aibonlin...008/10Metal.pdf

Unfortunately the following text then gave no idea as to the appropriate location !!

And one more. Here a method is described which although admitted to be inferior is nonetheless claimed to be a “standard” procedure.

 MD1_2a.jpg   86.59KB   52 downloads

(the underline is mine, seemed to be incorrect or perhaps I've misunderstood ?)
http://www.barnstead...uctPDF_6357.pdf


Rgds / Charles.C

PS - There is certainly a good collection of varied reference terms around for these advertised products. The majority don’t explain what the “sensitivity” means at all, just throw in the “mm”. Some refer to a “best” sensitivity (!?) and just a few state the sensitivity refers to the geometric centre (GC) or more precisely (1 hit out of 10) the ability to respond to a sphere of Xmm diameter existing on its own in the aperture GC.

[GMO]

Thanks Charles, I think I understand the LOMA references but it's different to what I've done in practice in the past.

I think what it's saying is rather than the centre of the product, you should place the stick in the centre of the aperture (wherever that is in your product) so for some products it might be in the middle and some might be near the top. I think that would be impractical to train out to your average operator though (not to suggest that some of them are meatheads or anything...)

Anyway, the second reference talks about putting it in the end of the pack and I can see the confusion; how can it be in the middle and the end? Well it can because if you think three dimensionally, if you're stood at the side of a metal detector with the belt running from your left to your right, on the y and z axis, the stick should be in the middle of the product but on the x axis, it should be at the end. Does that make sense?

Not quite so sure our methods are perfect and after this thread I've realised things could do with improving . I think we still have a way to go.

[Charles.C]

Dear GMO,

Must admit that I got a bit lost after “z axis”. Not much good at lateral thinking, never mind 3D.

I guess if the x-axis is through the aperture centre (GC) and parallel to the belt, and if the sample is a point source, eg a ball- bearing, the latter’s centre can theoretically be placed at any appropriate position in the product - ends, middle whatever, as long as that point passes through the GC. But, presumably, for a fixed belt position, any interference from the material itself may thereby vary. Maybe that’s the reason for the “ends” option. (I assume the “end” is still within the product.).

My guess is that people try putting the test piece at various positions within a representative dummy sample and (hopefully) discover that as long as the point passes near the GC, it doesn’t make much difference for reasonably uniform contents. Or perhaps not ? People’s actual usage procedures are welcomed.

I wonder how many (lucky?) people run a line with the top surface of the product container below the GC ?? (presumably the test piece then goes into the top layer, = minimum "accessible" sensitivity ?)

Rgds / Charles.C

[Eya]

I
Run three packs through the detector which has been set up for the product. This is to check there is no metal in the pack (of course, you don't know the metal detector is working properly at this stage but it becomes a paradox. You must have a metal free pack for your test pack but you don't know it's working until you test it etc.)

I definitely agree on this one speciallly for cartons or boxes used as packaging. I had once experience in my previous company the (LOMA metal detector horizontal type). it keeps rejecting and rejecting product every minute. I check the inside of the product nothing is there, i check all the parameters its be set on the right product. Then when i disect the carton i saw a small aluminum foil in inserted inside. When i ask the supplier they said that they are incorporating some recyled boxes the problem is secondary packaging doesn't have much concern HACCP or GMPs. their argument is we still have primary packaging and carton or boxes is much cheaper if some recycled paper is included.

[gaardendan]

Good point re investigation of rejects. We had a hold and some projuct scrapped recently because a reject was investigated, if that hadn't happened, we could have sent it out to the marketplace.

I've never heard anything other than the centre of the pack being the least sensitive point. I suppose you'd have to try it and if you're also doing periodic checks at other points, it's less of an issue IMO.

I've always understood that it's the center of the aperture that is the challenge on metal detectors and the bottom for x-rays. In other words, in or on top of the unit near the middle of the aperture for MD and in or underneath the unit for X-ray.

Additionally, I have my program set up to test the pieces at the leading edge, the middle, and finally at the trailing edge of the package to eliminate the possibility that the product readings are adding a cumulative reading to the test piece and causing it to be rejected based upon both readings. Does anyone else do this?