Measure presence of (strong) magnet @ 1 meter.

[mvdswaluw]

Hello all,

I would like to be able to detect the presence of a strong magnet at 1 - 1,5 meters.
My idea was to hook an arduino to a HMC5883L-board, sample the magnetic field in 3D with a slow moving average per axis and compare that against a fast moving average with some trigger levels. But the HMC5883L board could probably not be sensitive enough to have a proper detection at the wanted distance.

Is there a way to improve detection? Perhaps by gluing a ferrite rod upright onto the chip to concentrate the magnetic fields into the chip?
Or do you have any other ideas.

Goal is to detect shoplifters with detaggers in fitting rooms, thus the device should not be larger than say a cigarette pack to be able to mount it out of sight in each room.

Thanks in advance.

[T3sl4co1l]

It's possible to calculate or estimate the largest reasonable magnetic field such a device might produce, but the smallest (to be able to ensure detection) is impossible.  All it takes is a few pieces of iron to concentrate the field into a smaller space, and your sensible external field goes way down.  For cost, size and weight reasons, I expect these devices already use such design (pole pieces), so that the expected change in field due to their presence (~nT?) will not be distinguishable from steel belt clips, or daily variation in the Earth's magnetic field.

A metal detector might be a better screening method, but it will generate quite a lot of false positives.

Tim

[ElectroIrradiator]

It will probably be hard to do what you are asking. Far from a magnet, the magnetic field decreases in proportion to the cube of the distance between the magnet and the detector. Additionally, this relationship is only valid outside the near field of the magnet. So close to the magnet the field is stronger than one would expect from this relationship.

Imagine you can build a widget, which can detect a small yet powerful permanent magnet at, say, 10 cm separation. If you ignore the complication of the near field being stronger, then you'd still need to increase the sensitivity by a factor of 1000 to detect the same magnet at a separation of 1 m.

[mvdswaluw]

... need to increase the sensitivity by a factor of 1000 to detect the same magnet at a separation of 1 m.

Hence why I would go with oversampling and a moving average. To get optimal sensitivity. Any deviation from the average should/could trigger an alarm.
The extra thing I need is a way to focus the view of the sensor in the direction of interest.

I can already see with an app on my mobile phone that I can detect magnetic variation if a weak magnet get within say 10cm, so 50cm should be possible with a good sensor setup and some tweaks.

[tggzzz]

... need to increase the sensitivity by a factor of 1000 to detect the same magnet at a separation of 1 m.

Hence why I would go with oversampling and a moving average. To get optimal sensitivity. Any deviation from the average should/could trigger an alarm.
The extra thing I need is a way to focus the view of the sensor in the direction of interest.

I can already see with an app on my mobile phone that I can detect magnetic variation if a weak magnet get within say 10cm, so 50cm should be possible with a good sensor setup and some tweaks.

Do you need to unambiguously distinguish a strong magnet at 1m from a weak magnet at 1cm or 10cm?

[mikeselectricstuff]

Inverse square law, Earth's magnetic field plus other local magnetic influences mean this probably isn't realistically doable.
You might be able to detect motion of a magnet as opposed to presence, but likely to give plenty of false alarms. e.g. some handbags have magnetic clasps.

[Jeroen3]

Before continuing you should set some goals in Tesla.
Then you can search for applicable sensors.

If you know the minimum strenght required to detag, you can use that as reference.
For example, a small rare-earth magnet can be 1T. But not at 1 meter.

[mvdswaluw]

Do you need to unambiguously distinguish a strong magnet at 1m from a weak magnet at 1cm or 10cm?

No, just want to know if someone is likely to have a magnet in the fitting room so we can ring a bell for people to go and "offer assistance".

[mvdswaluw]

Before continuing you should set some goals in Tesla.
Then you can search for applicable sensors.

If you know the minimum strenght required to detag, you can use that as reference.
For example, a small rare-earth magnet can be 1T. But not at 1 meter.

Can do that with the detag fixture in the counter, but I already have some HMC5883L-board available in the hobby drawer.
And it's just for my personal gratification to catch some shoplifters in the act. My wife works in the shop and gets more and more frustrated when she finds yet again three security-labels on one item. Knowning that she and/or her colleagues have again lost two other items.

I already installed two wireless door alarms on two drawers where they found clothing hangers with regular interval. Didn't take more than a week and they have never seen that (elderly) couple again. 

[mvdswaluw]

... false alarms...

Better to have false alarms than no alarms. And a false alarm also makes the girls aware that they can go and offer assistance and try to sell extras

[tggzzz]

... false alarms...

Better to have false alarms than no alarms.

Probably not, as has been demonstrated many many times before in investigations as to why The Bad Event occurred. Too many false alarms => ignored alarms and alarms switched off.

But that is application dependent and dependent on the individuals responding to the alarms.

[Siwastaja]

If you could install three or four devices per stall, on each wall and possibly door, you could halve the distance requirement to 50 cm, therefore making it 8x easier to detect. This could make design so much easier and lower cost that having multiple units wouldn't be a problem. You can design it to the size of a coin cell!

In addition to the point about false positives made by tggzzz, remember that people also will be pissed off from false alarms, and for a good reason. And it is almost guaranteed that most employees won't handle these cases as they should because they are often already stressed from shoplifting, and the principle of presumption of innocence is not stressed enough. So, really try to avoid false positives. I'd say, if more than 5% of alarms are false, forget about it.

How do you plan on handling the situation of someone having very weak magnet, heck, even some magnetized metal, who happens to put it near the detector? If you need to detect a strong magnet @ 50 cm, you'll detect almost anything @ 5 cm. You could add some proximity sensor that would lower the detection level, but it will get very complicated soon.

[miguelvp]

Loosing sales from pissed off customers (if a false positive situation is not handled right) might be more detrimental than shoplifters. And you might not even catch them because they might remove the tags outside the dressing rooms to begin with.

[mvdswaluw]

... will be pissed off from false alarms...

Goal is to have a stealth alarm near the counter, for heads up purpuse. Perhaps if a trail is succesfull with an indicator number per fitting room.
Will try something in a few weeks. Need to gather parts an take them with me to the camp ground. Forgot that this week...

[free_electron]

what if i bring in a horse shoe magnet with a shunt on the poles. virtually no leakage. good luck finding that.

or u could use a magnet that traps the field in an air gap ... i can make very strong fields that way and there is no leakage. same principle as used in the voice coil magnet of a harddisk. The field is trapped.

You can't sense it even an inch away ...

The solution for your problem : an assistant that counts how many pieces you take in and how many you bring back out ... you get a numbered tag indicating how many pieces you took in.

That's how they did it in the 80's and 90's ...   But i guess employing someone is a bridge too far...

[Jeroen3]

Most shops calculate the loss of shoplifting in their prices anyway. It's a known risk with a shop.
If you don't want customers to steal stuff you should have a webshop. But also webshops have a whole lot of other potential risks.

On topic:
You could install a larger number of sensors around the shop to create located magnet tracking.
Great for a university project, unaffordable for your average clothing store.

[T3sl4co1l]

RFID tags would be much more sensible.  They can still be shielded from sensors, but in the usual case at least, will have better range and sensitivity than the proposal.

[richard.cs]

The design of detagging magnets lend themselves to a tightly concentrated field anyway,  they will be extremely hard to detect from that range.  I'm sure you could see them but like the others I think you will have too many false positives, quite possibly more than actual shoplifters.

Given detagging is only one of several ways of getting items out pay the alarms I'm not convinced. Can you tell us why counting items in and out doesn't work in your situation?

[mvdswaluw]

The design of detagging magnets lend themselves to a tightly concentrated field anyway,  they will be extremely hard to detect from that range.  I'm sure you could see them but like the others I think you will have too many false positives, quite possibly more than actual shoplifters.

Given detagging is only one of several ways of getting items out pay the alarms I'm not convinced. Can you tell us why counting items in and out doesn't work in your situation?

Counting doesn't work, not enough people in the shop. Budget thing. My wife's contract was not renewed, and she was replaced by a younger (cheaper) employee. Just a few months later the company filed for bankruptcy in the Benelux. Just heard today they are getting back in business, so I will pick up the project (only) if they hire my wife again ;-)

[CatalinaWOW]

Maybe a different approach to the problem.  The tags themselves are designed to be detected.  If their position can be localized you might find that the motion signatures are very different for tags moving with normal fittings than for those being removed and re-applied to another garment.  There might be other clues.  The frequency of resonator might wiggle due to the influence of the removal magnet.  Localization isn't trivial either, but seems easier than detecting the magnet.

[f5r5e5d]

leaving "strong magnets" lying about isn't risk free - pinching/crushing from pairs snapping together to wiping your (formerly) potential customer's credit card mag strips...

[BrianHG]

Maybe a different approach to the problem.  The tags themselves are designed to be detected.  If their position can be localized you might find that the motion signatures are very different for tags moving with normal fittings than for those being removed and re-applied to another garment.  There might be other clues.  The frequency of resonator might wiggle due to the influence of the removal magnet.  Localization isn't trivial either, but seems easier than detecting the magnet.

This is actually a better idea than you think.  I would create tags, with a lithium battery with an RF pic 455mhz transmitter which fires off a 455mhz ID encoded alarm if it is removed from the garment and continues to do so until it is reset at the counter.  At the counter or security office, there is a RF receiver listening for the ID number of that tag & has a photo of the garment related to that ID item.  This tag should also have the old fashioned alarm for the entrance/exit of the store included as well.

[CatalinaWOW]

Being able to change the tags is a game changer.  There are lots of options.