# Recommendation for metal finding method

I will start by explaining the problem to make my question clear.

During the production of diapers , the machine need continuous supplying material. So Butt Splicer is used to attached the end of expended roller with the start of the next roller using double sided tape. So in the production their will be a defected diaper ( a diaper which have a material with double sided tape on it. There is some methods to reject the defected diaper during the process . But to ensure that this diaper is rejected , a Metal Foil Tape is used with the double side tap, and a metal detector used after the packaging to reject the bag with the defected metal.

And here is the question,The client want to find these diaper s with metal while they are in the diaper bag without opening the bags and checking every diaper with the metal detector. A normal metal detector which will generate a magnetic field will detect if the metal is on the bag because of the shape of the magnetic field. So I need a sensing method with linear detection ( to scan diaper by diaper , So the scanning depth most by more that 20cm and the thickness at least 3mm.

The project summery:

A device to identify the position of a defected diaper in the bag.

A motor which is controlled by a microprocessor will hold the sensor (which I am asking about) to scan the diapers bag from the top side

(This picture is only for showing how the plastic poly-bag will be align from the top side) and once a metal is founded, a marker will mark the position of the diaper in the bag , so the operator could find the defected diaper and identify to the technician which material is not rejected in the process.

• I feel like you might be overengineering the part where the person that has to take the box out of the belt, and open it, would have to look through nine diapers. The "look for metal" in closed box part makes sense to me, I'm not so sure about the "identify which diaper exactly". – Marcus Müller Apr 23 '17 at 15:30
• Probably an obvious question, but why can't the diapers be rejected after forming, but before they end up in the box? – Peter M Apr 23 '17 at 15:40

Pretty much all industrial/food metal detectors I've come across will easily detect the presence of metal foil tape in one diaper in a box of several diapers. It will probably register as a massive signal. I've designed them so I know how sensitive they are. For instance, a metal detector with an aperture of 14" x 7" can easily detect a 1 mm diameter piece of iron running through the centre of the aperture. On one application I remember, you could regularly detect the presence of iron naturally contained in a cabbage.

Finding a thin sheet of conductive tape will be a breeze.

They will also be able to stop the converyor (that you should feed them through) at a place indicative of the diaper that is containing the rogue material. Standard metal detectors like this have an output that can trip a belt motor or activate a reject arm. Pretty standard stuff.

I just don't think you've done your homework enough. Try looking up Apex metal detector (Thermo Scientific): -

The most basic search heads work with three coils like this: -

As the metal contaminate passes through the coils, there is a large "positive" signal followed by a sharp zero signal (metal in centre) followed by a large anti-phase (negative) signal. The positive and negative signals largely coincide with the positions of the outer receive coils and, because the geometry of those coils is known, it's a small matter to compute the (fairly) precise position of a rogue piece of metal and stop the conveyor at a fairly precise inspection point.

There are a plethora of Radar movement detectors available such as this which should suit your need. These detectors use a metal horn to diverge the beam so several options exist to make the results usable:

1. Use a flat plate collimator lens to concentrate the beam
2. use a digital sensor to measure the box position and length relative to a narrow beam return
3. Or use a camera to get the box position relative to the beam return
4. Or use multiple movement detectors with flat plate deflector behind the target box. This will reduce the RCS and with for example 3 detectors you could calculate the X,Y,Z position in the box. (though I seriously doubt you'd get closer than Diaper +/-1 accuracy)
5. As in 4 above (3 movement detectors) and collimator lens so you have narrow beams. If you know the speed of the box then you can use a break edge detector and speed to calculate position in box.
• I like the radar approach. Also, thinking about it, I frankly fail to see why I'd need to know exactly which diaper in the box is damaged without opening it. It's not like taking the box out of the production flow, opening it, would be much longer when you'd have to look through all nine diapers in the box.In fact,I somehow doubt any of our systems pay for themselves in finite time buy saving work. Realistically, put all boxes with broken diapers aside, and sort them when you've got spare time. Done. – Marcus Müller Apr 23 '17 at 15:26
• Thanks @MarcusMüller. I agree with you that finding the exact diaper seems a step too far. From my perspective I'd would simply want to find the box with a bad diaper in it, and take that whole box out of production. Then the problem reduces to perhaps one or two collimated movement detectors and a punch arm to direct the bad box off the line ....no pesky operators to pay for. – Jack Creasey Apr 23 '17 at 15:33
• @MarcusMüller, the importance of finding the crappy nappy in the box of happy nappies depends upon the manufacturer's volumes, so disregarding is second-guessing what the OP's seen and we haven't. This is a serious volume product and they could be churning out enough to have a bad box, say, every two minutes on bad days. You're dead right that the investment has to be weighed against manual inspection time but the OP may well have done that. Good ideas on technology :-) – TonyM Apr 23 '17 at 15:42
• You, @TonyM, put a smile on my face. And I'm currently doing microwave stuff, so that's not that easy :) – Marcus Müller Apr 23 '17 at 15:47
• Solidarity, fellow worker @MarcusMüller :-) I'm in the middle of reviewing safety VHDL for a medical application, so I could do with a laugh an'all. (or d'you mean you're reheating what's left of last night's madras? :-D ) – TonyM Apr 23 '17 at 15:56

What's wrong with the standard method of detecting metal, a conveyor through a portal with a central exciting coil, and two antiphase detection coils one fore and one aft.

As the metal item approaches the portal on a conveyor belt, it changes the coupling with the fore coil first, creating an output. As it passes the central plane, both coils are affected equally, and the output is zero. As the item leaves the portal, it couples more strongly with the aft coil, creating the opposite of the initial output.

The trick will be to calibrate that central null point very carefully. By knowing the timing of the null output compared to the position of the box as it passes through, the controller will know where in the box the metal tape is located.

• upvote, for obvious answer! – Marcus Müller Apr 23 '17 at 15:28
• I'd suggest that this would not allow identification of which diaper in a box is bad, its a great gross detector, but with low accuracy. – Jack Creasey Apr 23 '17 at 16:10
• I'd suggest it would. Have you ever worked on portal metal detectors? How many patents do you have in the field? – Neil_UK Apr 23 '17 at 18:56
• Hee hee I have 4 patents. Do I win? (US 7,061,236 B2, US 2003/0164766 A1, US 2004/0155651 A1 and US 2008/0055080 A1) plus the equivalent EU ones. I totally agree with Neil. A portal MD will breeze it. +1 to cancel the uninformed musings of those who don't know. – Andy aka Apr 24 '17 at 7:27
• @Andyaka Yes, you win, I have only one in metal detection. Small engineering world isn't it? It's so old that the main thrust of my contribution was implementing matched filters on the newly available Z80, dragging the analogue team screaming into the white heat of the digital revolution. – Neil_UK Apr 24 '17 at 7:36

Your depth/width ratio makes the classical magnetic approach to finding metal unusable.

So, the intuitive way for me here would be a simple x-ray system: Cathode ray tube above, an array of more or less sensitive detectors (might simply be diodes in reverse bias) below.

However, x-rays are dangerous, so let's keep that as plan B.

You can do the same via detection of radio wave transmissions. My hypothesis here is that diapers and cardboard boxes don't have significanty different $\epsilon_r$ or $\mu_r$ than air, and thus, practically won't deflect an incident wavefront.

Let's look from above onto the system I imagine:

   radio waves
|  |  |
|  |  |
v  v  v
+-----------+
|  box of   |
|  diapers  |
+-----------+
|  |  |
|  |  |
v  v  v


Now, if you have a piece of metal in that box, the following will happen:

   radio waves
^    |  |  |
\   |  |  |
\  v  v  v
+-----------+
<-|  box of   |
|  diapers  |---->
+-----------+
|     | \
|     |  \
v     v   >


I.e. reflections in random directions, and less transmission straight through the box.

You could either measure the transmitted power and look for changes, or try to catch power going into other directions.

By choosing a transmit antenna (will probably be a horn antenna) and a suitable frequency (probably some 60+ GHz ISM band), you can get sufficient resolution.

A much cheaper, but less accurate method would be to take the whole box, put it in to a metal chamber (needs to be "kind of RF tight", ie. not really a locking door thing) of known dimensions, excite the chamber with a frequency that should nicely resonate in its lowest mode within that chamber, (ie. chamber probably something like half a wavelength) and see how well that works.

Another idea: excite the box of diapers with a strong RF radiation (prototype: microwave oven) that you modulate at an audible frequency. That should make the metal patch vibrate audibly, and you can detect that with microphones.

Hope that helps you get your client happy :) (why are you making money with this and I write answers? Strange.)

• I think radio wave will have the same disadvantage as the magnetic field. As both of them will cover more than one dipper as they will spread non linearly, and the strength of the signal cannot by use to find the metal because the meal may be in different position of the diaper. – John Alawi Apr 23 '17 at 14:52
• @JohnAlawi " they will spread non linearly" <--- you're wrong here. That's why I said "suitable antenna"!! You can form the beam however you like, if you have an antenna that's large enough / a wavelength that's small enough; use a 70 GHz oscillator, a sufficiently large horn antenna, and you can get beam widths < 3°, which should totally be sufficient for your purpose. – Marcus Müller Apr 23 '17 at 15:12
• depends on your wavelength. There's plenty of ways to calculate the beam width of different antennas. A 20x40 cm flared horn antenna at 70 GHz would have a beam width of about 1°. So very little, easily enough to focus on a single diaper. – Marcus Müller Apr 23 '17 at 15:33
• I'm sorry to have to say this but the classical method of using magnetic fields will work very well in this application. Detecting a piece of metallized tape in a three coil inductive balance search head running at (circa) 300 kHz will easily pin-point the rogue diaper. – Andy aka Apr 23 '17 at 18:38
• @Andyaka you're very probably right; see the discussion under Neil's answer, for example; upvoted yours :) – Marcus Müller Apr 23 '17 at 19:06

take a look at many of the metal detector designs, especially those ones built on a mcu.

my suggestion would be to drive the mcu with an oscillator whose frequency is dependent on the presence of metal. all the code does is to increment a counter.

Pick a mcu with a watchdog timer who typically has an independent oscillator. Activate the watchdog to reset the mcu.

the first thing the code does is to compare the value of the counter (which shouldn't be reset at start) vs. a known value (calibrated to the non-presence of metal). And light up a led if the deviation is too big.

After that, keep incrementing the counter -> that increment will be stopped by the watchdog reset.

The diameter of any low cost cheap metal detector that operates 50-100kHz or so must be greater than the depth of the box metal foil plus gap for ease of detection. The consumer or commercial units can go to 10x diameter of loop.

||-----------------||

/


A differential coil method is best for detecting foil passing under gap between coils. and can be made for < \$10.