5. Product safety plan
CCP – Metal detection
The purpose of the CCP metal detection requires that you understand how your detector works so in this issue we help you learn more about it.
Introduction
Metal detection is an industry wide CCP. In this article we’ll cover everything you need to know regarding metal detection, including how it works, testing, and validation.
How a metal detector works
It’s important to understand how a metal detector works, so that you can validate the CCP and carry out root cause analysis when things go wrong.
Fig.1 The Search Head
The search head
In order to achieve the successful detection of metal in the finished product, the metal detector generates an electric field. When metal passes through it this electric field, it distorts it. The metal detector recognises this distortion and uses this to trigger the rejection system.
This field is generated in and around what we call the ‘search head’. The field of the detector is generated by the interaction between three coils within the search head. The area called the search head is shown in red in Fig. 1, the arrows show the direction of the field.
The signal
When a conductive material (metal contaminant) passes through the search head, it creates a distortion in the field – this is known as the signal.
Fig. 2 The Signal
There are two types of distortion (signal):
- a reactive value – caused by metal of ferrous (Fe) composition,
- a resistive effect – caused by metal of a non-ferrous (NF) composition.
Stainless steel also exhibits a resistive effect but less strongly than a non-ferrous metal – making stainless steel (SS) more difficult to detect.
The signal that is produced from a reactive value and a resistive value can be displayed as a chart (see Fig. 2) to help to explain the effect it has. By drawing it on a chart, we can see the size and position of the signal that is produced by the 3 test pieces (NF: non ferrous, F: ferrous, and SS: stainless steel) and also the signal produced when nothing is passing through the search head.
Fig.3 The Product Signal
The closer to the signal is to R, the less distortion (either +ve or -ve) to the field, the less signal is produced and so, it is more difficult for the detector to ‘see’ the metal. You can see in Fig. 2, that non-ferrous (NF) and ferrous (F) metals are quite some distance from ‘R’, therefore the signal is different to that from ‘R’, and also the arrows is quite long, so the signal is strong. Which means, the detector would recognise the signal as metal.
Stainless steel however, is quite close to ‘R’ and the arrow is only short, and so weak. This is why stainless steel is more difficult to detect. You will find that if any test piece is going to fail to reject when carrying out your tests, it’s normally stainless steel – for this very reason.
The other important thing to understand, is that the product causes a distortion to the field as well. This is because the product has its own conductivity.
Water is a good conductor and is found in most products. But the additional of other ingredients such as sugar, can actually increase the conductivity. This means that when the product itself passes through the search head, it produces a signal. In Fig. 3, we’ve added a product signal to the chart.
False rejects
A false reject is one where there is no metal in the product, but the detector rejects it anyway, as it thinks it contains metal because of the signal it produced.
Fig. 4 The Product Window
We don’t want false rejects, because it creates waste product and also it means that people lose faith in the machine – because they don’t think it’s working. The risk with this, is that if the machine fails its 3 piece test, the operators may decide to continue on anyway, because it’s just the machine ‘playing up’. Or, if the machine actually rejects product because it really has got metal in it, nobody notices, because it’s one of so many rejections. A rejected product, needs to be a really big deal to the operator, so it’s taken seriously – that’s what we need to aim for.
So, to prevent false rejects the metal detector needs to be programmed to ignore the signal created by the product. The machine does this by knowing what part of the signal to ignore. This is known as the metal detector ‘learning’ the product. To explain how the metal detector does this, Fig. 4 Product Window, below shows the product window (the area of signal that the machine ignores).
If the product window is set up incorrectly, as in, not wide enough, it will cause false rejects – because the detector will still ‘see’ the signal from the product. The diagram (See Fig. 5) shows this.
Fig. 5 The Product Window (too narrow)
Fig. 6 The Product Window (too wide)
Or if the product window is too wide, the detector will think any signals generated within it are good product (i.e. do not contain metal). This may mean that stainless steel signal is within the area to be ignored, which will allow product containing stainless steel to pass through the detector without being rejected. The diagram (See Fig. 6) shows this.
So, you can see that getting the ‘ignore’ area of the signal right is really important. We need it to be as narrow as possible, to make sure that only the product is ignored, but wide enough to make sure we don’t reject good product.
This is why, we have to make sure that we have settings of each type of product that we are producing. If we have one setting for all products and the signal that those product produce is wide, then we are at risk of not rejecting product that contains metal.
Some metal detectors will have the functionality to ‘learn’ the product once and then store these settings, under the product name (or product group), allowing the operator to retrieve them on the next run – by picking the required product from the menu.
You don’t need a setting for every single product you make, if they are similar in size and recipe. Go through the products and work out how they can be grouped best and then get the machine to learn each group. If you’re not sure, you can get your metal detector company to come in and help you to work out what the variation in signal is, that each product produces.
The operators of the metal detector then need to make sure that they select the right product, when they are using the detector. There are so many times I’ve gone to a metal detector and checked what product has been selected, and found that it’s not the one that’s being produced at the time.
Some metal detectors may not have this functionality however, or the product may be so variable that storing the settings is not practical. In this situation, the metal detector would need to be set up to ‘learn’ the product prior to each run. And so, this would need to be part of the change over procedures for setting up the metal detector.
Test piece size
The smaller the piece of metal, the weaker the signal and also, the closer to the product window the signal may be. Therefore, it is key to work out what test piece size can be used, to test the machine. There is a balance to be made here, to find the smallest test piece possible, but without triggering false rejects. To do this, trials must be carried out to find out what the smallest test piece size can be used, that is consistently detectable, without causing excessive product rejects. These trials need to make up part of your metal detection CCP validation. In the next article we’ll look at how we should validate your test pieces.
3-piece test
A 3 piece test, where the non-ferrous, ferrous and stainless steel test pieces are tested through the machine is the most basic of tests.
The 3 piece test must be carried out in normal product flow. Normal product flow, means the test packs must go through the detector at, at least the same speed as the normal product would go through it. And, it must also go through at the same spacing too.
Sometimes you will find that when the line is not filled with product (which may be case when the operator is trying to do a test), the packing machine may slow down to allow for the reduction in product throughput. If the operator does the test at this speed, it is not reflective of normal product flow – therefore you have not proven that the machine would work at normal product flow.
This means that you need to consider how you are going to get the test packs onto the line, in normal product flow, if the line is sealed off with guarding.
The memory test
This test is where you have your 3 test packs and in-between them, you have good product, as shown below.
The purpose of this test, is to prove that the metal detector can see metal in the ‘test pack 1’, reject it and then reset itself, so that it can see that there’s no metal in ‘good pack 1’. Then it has to be able to see and reject ‘test pack 2’, then reset itself so it doesn’t reject ‘good pack 2’ and so on.
If the machine cannot reset itself fast enough, it may also reject the good packs. Plus, we are testing the machine, to make sure it does actually reject the test packs and not, the good packs by mistake, allowing the test packs to go through. So, this means that if the machine does not reject all 3 test packs AND, does not allow all 3 good packs to go through, the test has failed.
It is a common mistake to think that the purpose of the test is only to reject the test packs and that it’s ok for the good packs to be rejected too. Rejecting the good packs shows that the machine is not set up correctly and so, you’ll get false rejects.
Also note, how there are 3 good packs – the last pack is there to prove that the machine resets itself after test pack 3, if it’s not there, then you’ve not proven that the reset mechanism worked for all 3 packs. Also, it’s good practice to always have a good pack at the end of the test, as you may find you need this pack to ensure that all the products go through in normal product flow. You may find that, that last pack is needed to ‘nudge’ the test pack through the detector.
Large piece of metal test
One of the most recent tests to be introduced is the large piece of metal test. This test was introduced to replicate a large piece of metal going through the detector, like a spanner for example.
It seems odd that we’d check that the machine can see a large piece of metal, when we’re already checking the machine with smaller pieces of metal – surely, it’s obvious it’s going to reject? Yes, you’re right – but this is where this test is very often misunderstood.
When a large piece of metal goes through the detector, it causes such as large signal, that this can sometimes send the machine haywire! This means, that it will reject the large piece of metal, but once that’s gone through the machine may not work as it should anymore. Which then means, that if a product was to contain metal and go through the machine, it may not be detected and rejected.
So, the key part of this test is that the machine not only sees and rejects the large piece of metal, but actually the most important check – is that it will still see and reject your normal 3 test pieces afterwards.
Passing the 3 test pieces through the machine after this test, is the part that’s commonly forgotten. Make sure your procedure is to do this, because otherwise putting a large piece of metal through the machine on its own could cause the machine to break and you wouldn’t even know it!
Leading, middle and trailing edge test
What we’re going to cover is the leading, middle and trailing edge test and why we do it – because although it’s essentially the same as the 3 piece test (as in, you can use the 3 test pieces), there is a reason why we go to the effort of putting the test pieces in these particular places on the product (as in leading, middle and trailing edges). And, more importantly, why sometimes it’s just relevant, or even possible if your product is square or round!
The why…
The purpose of the leading, middle and trailing edge test is to prove that the reject mechanism is set up and working as it should. To understand this, we need to look at how the metal detector rejects the product. The reject signal is normally triggered in two ways:
- The signal from the metal – causes the metal detector to reject,
- the signal from the metal, plus the signal from the product passing through a sensor which is positioned either before or after the search head – causes the metal detector to reject. (Both things must happen or the rejection won’t happen.)
If the metal detector triggers the rejection sequence using the first option, and long packs of product are being produced, there is a risk that the rejection system may not be set up correctly to ensure that the pack in question is rejected – because the reject arm for example, may trigger too early or too late and therefore either miss the product entirely, or hit the back end or front end, meaning it doesn’t reject fully into the reject bin.
This is because the time and distance from the signal from a metal at the front of a long pack and the time and distance from the signal from a metal at the end of a long pack may differ greatly. This means, that the reject system needs to be set up so the timing works for both cases. This is where a leading and trailing edge test would prove that the timing for metal in the back end and the front end of the product, works – and so, would be rejected.
For metal detection systems that use the second option to trigger the rejection sequence and where packs are smaller than the aperture of the search head, then the leading and trailing edge test does not need to be carried out to prove that the reject mechanism is set up correctly, because the sensor will make sure that the rejection happens at the right time. However, the test should be carried out (perhaps as a fail safe test) to show that the sensor is actually working.
To do the test, or not to do the test – that is the question!
So, this means that the test is not always applicable – if the length of the product is smaller than the length of the search head, as it is moving through the detector. Also, doing a leading and trailing edge test does not add any value, if the product does not have a leading and trailing edge side – in the case of square or round products.
To prove that the test is not applicable, you need to measure the length of the search head. Then, go through all your pack sizes and record the length of each one (in the orientation it would be going through the search head) on your metal detector validation report.
Where the length of the product is less than the length of the search head, the test is not needed.
Metal detection CCP validation
The purpose of validation your metal detection CCP is to ensure that:
- The test pieces are placed in the right place – on or in the product.
- The test pieces are the optimum size.
Validation of the test piece equipment
Where you put the test piece is really important, because we need to make sure that it is in the hardest place for the metal detector to ‘see’ it. To explain, here are the images of the search head again.
Fig. 1 The Search Head
The signal comes from the search head (See Fig. 1) and it’s the signal creates the rejection when metal is detected. So, if we look at the picture again of the search head and draw (green) lines from the corners of the search head, we can see where the weakest part of the detector is.
This spot is where the detector finds it most difficult to see metal (See Fig. 2). Therefore, this is where the test piece of metal should go – so we can make sure we are really challenging the machine.
The signal comes from the search head (See Fig. 1) and it’s the signal creates the rejection when metal is detected. So, if we look at the picture again of the search head and draw (green) lines from the corners of the search head, we can see where the weakest part of the detector is.
This spot is where the detector finds it most difficult to see metal (See Fig. 2). Therefore, this is where the test piece of metal should go – so we can make sure we are really challenging the machine.
Fig. 2 The Search Head (weakest point)
Fig. 3 The Search Head (tall product)
The size and shape of the product on metal detection
This is where the size and shape of the product comes in. If your product is tall, the test piece would need to be in the centre of the product, to make sure it was at the weakest point, as shown in the diagram (See Fig. 3).
Whereas, if your product was small, the hardest place for the detector to be able to see metal would be on the top of the product – so this is where your test piece should be (See Fig. 4). So that’s the first thing you need to validate – where the test piece should be.
Fig. 4 The Search Head (small product)
To do this, measure the detector head height from the conveyor and record it on your validation report. Then, record the product sizes you make and the size of each – to show where the height of the product sits against the detector head. Then work out, from this, if the test piece should be in the centre of the product or on top of the product (depending on which is closes to the weakest point). Then record this for each size of product.
You may find, that if you only have one detector that is used for a wide range of products, which are different sizes, you may need to vary where the test piece goes, depending on what product you are producing.
Test piece size
The next step in the validation is to work out what test piece sizes you can achieve. Many retailers will have set test piece sizes for you to use, but you can use this validation to prove that either:
- You can achieve a smaller size
- You cannot achieve the size they have specified, as it will cause false rejects (learn about false rejects in our previous article
In order to do this effectively, you will need a range of test piece sizes to use and so, it’s generally a good idea to get your metal detection manufacturer to come in and help you – as they will have a test pieces of all the sizes you’ll need, as ideally you don’t want to be buying them! You will need to carry out trials. You can do this on each product you make (if you don’t make many) or you can group your products into similar groups and just pick one product from each group to trial.
Then, with your trial product being produced, learn the detector to the product – to make sure the sensitivity is correct. Then, make up test packs with your current size of test piece and send these through the detector 30 times (I use 30, but if you want to do more or less that’s up to you). If all 30 test packs pass, record this and then move on to the next smaller test piece and repeat.
Repeat this process, until you find you get at least 1 fail (i.e. it is not rejected) out of your 30 test packs. At this point, this test piece has failed, and so, you have proven that the test piece size, one up from this one, is the one that has been validated as successful. This is the test piece size you should use to check the machine on an ongoing basis, for that product or product group. You would then need to repeat this for the other types of test piece.
Record all of your trial results on your validation report. You need repeat this validation when something changes; e.g. the product recipe, size, weight or the equipment producing the product. Ideally, it should also be carried out yearly following the servicing of the machine.
Hello, could someone share the template they use for ccp metal detector internal validation please?
Once again, Thank you, Kassy and team as always for your contributions to the FSQ workstream.
I do have one question and perhaps, it’s been asked before. How often do we have to validate the Metal Detector or X-ray? Our sites are certified with BRCGS and we couldn’t locate anywhere in the standards that it mentioned that requirement. FYI – We have calibration done annually and the vendor also do a performance verification.
Hi Kassy,
the question is about food standard clause 4.10.3.4. Is it acceptable to use some control packs of product with test pieces inside prepared month ago, not those being produced on the line during verification of metal detector?
Thank you!
Hi Mila
No – because the test packs wouldn’t be reflective of what is being packed currently.
Thanks
Kassy
Hi Kassy,
What is the real difference between calibration, verification and validation? My auditor wanted to see different documents for each. If you can help us. Thanks a lot!
Hi Ronak
Calibration of metal detectors are commonly confused with service records. So that would most likely be what your auditor was looking for. Verification records, would be your routine metal detector tests. Validation records would be the report you create (or it can be done by your metal detector provided) which details how you came up with the verification testing routine that you use, e.g. test piece sizes.
I hope that helps.
Kassy
I need to clarify whether the metal detector is a Critical Control Point or is it a monitoring tool for monitoring the Critical Control Point For example, there is a production line on which a critical control point has been placed, which is heating to 100 degrees C, and that point is monitored by a thermometer, so is the thermometer in this case equivalent to a metal detector
Hi Moshen,
Metal detection is a CCP, if your HACCP determines it to be. A metal detection is a process step and the metal detector is a CCP monitoring tool.
Heat treatment to 100C can also be a CCP if your HACCP determines it to be. Cooking is the process step and the probe is a CCP monitoring tool.
Does that help?
Kassy
Hi!
Recently we have had a non conformance in the BRC audit regarding the validation of the metal detector. Despite our MD service carried out a validation by passing different sizes of the 3 types of metal in order to determine the correct size of the test pieces for our operation, the auditor raised that in order to validate that our detector is capable of detecting metal contamination coming in different shape and orientation, we must also validate the detector by passing potential foreign bodies in our factory. The auditor found a piece of wire (about 15 mm long) on the floor and she asked us to put it in the product and pass it through the detector. The outcome was that it was detected in one orientation but not the other. She said that we must include a test of this type in our validation. I would just like to know your opinion on this, since our MD service and we agree that that kind of test can’t be standardized and the industry uses steric balls to eliminate the effect of the shape and orientation.
Thank you
Hi Cristian
Oh dear… I totally agree this isn’t a valid non-conformance. The reason the wire wasn’t detected in all orientations is because of it’s shape and yes, that’s why balls are used. Validation is meant to prove a theory, and putting a piece of wire through the metal detector like this wouldn’t prove anything – it would be good to see what the non-conformance actually says, to understand what she was wanting you to prove.
I would definitely challenge the non-conformance wtih your Certification Body. If you would like any help in doing this, please let me know. And, please let us know the outcome!
Thanks
Kassy
Hi Kassy,
Although I am a provider and service provider of metal detectors, I still think it is very meaningful for users! Even as a service provider, I still recognize the concerns of users.
Because we may have overlooked one of the most important tasks:
If a metal detector is not professionally maintained, its detection capability may not be keep or even attenuated.
Three standard test pieces are used for verification. Even if 100 successful verifications are carried out, the issue cannot be accurately explained, for example:
The signal value of the three metal test pieces may all exceed 1000, but the threshold set is only 200.
However, the detection capability of the MD may be slowly decreasing (almost certainly exists).
If the MD has not been professionally maintained, and when verifying, only observe whether it is rejected without paying attention to the specific signal value.
Then even if the detection ability is reduced to a signal value of only about 400, you can still perform at least 1,000 successful verifications ( I am very sure that I am confident enough on this issue! ).
Therefore, it is not whether the wire itself can be detected, but more importantly, we need to pay attention to the specific signal value, not just how many successful verifications.
I don’t know if I have described this issue accurately, if anyone has further questions, I would be happy to discuss with you.
Many thanks.
Planck
Hi, could you help me telling me what is the limit size of metals allowed in tablets, to know if my metal detector complies with the regulations?
Hi Francisco
I’m not sure where in the world you are, and therefore what legislation you work to – but in the UK there is no acceptable limit. The aim is to not contaminate the product, which means there must not be anything in the product, that shouldn’t be there.
Kassy
hi,
this is very useful. are there any plans to make this available to download? as it would be a very useful tool
Hi, I’d not thought of doing that, but we definitely could. I’ll look at doing this when I’m next writing. Thanks, Kassy
Is metal detection (if present) always a CCP ?
Hi,
No not always. Most customers and auditors would expect end of line metal detection to be a CCP, but if there are additional metal detectors within the process (pre packing/end of line) then these would obviously not be CCPs, as the last point of defence would be the CCP.
Thanks
Kassy
this insight guides are great training tools for younger members of the team. Are there plans to do a “how an x-ray works” ?
Hi, glad you like them! I think so yes, as we’re working on articles related to CCPs. Thanks, Kassy