I'm looking for a way to detect a join in a conveyor belt up to a distance of 10 cm away from the the belt. My original idea was to stick magnetised rubber strip on the underside of the conveyor belt and use a Hall effect placed under the conveyor to detect when the strip passes by. When the strip passes by I can determine that this is where the belt has been joined.

The issue with this is that the Hall effect will only detect a field of this magnitude (magnetic field strength similar to a fridge magnet) up to approximately 1 cm. As it is on a conveyor belt, I don't want to use something strong like a rare earth magnet to get a detect. I also can't use a sensor that uses light or relies on reflection as the environments tend to be very dirty.

Is there an alternative way to get a non-touch detect from a sensor up to a distance of 10 cm away? I only need a yes/no binary detect, nothing more.

Edit: As asked for in the comments:

  • Belt material is either rubber or rubber with steel cord for strength Speed is anything up to 5 m/s.
  • Belt thickness can vary up to 30 mm.
  • Sensor needs to be placed to the side of the conveyor or to the underside. It can't be placed above the conveyor.
  • Belt join is made from large steel clips. If a metal detector is placed further down on the belt, I want to be able to warn the metal detector that these clips are approaching in order for it to ignore them. I want to do this with something simpler than an additional metal detector.
  • Dust material would specifically be from somewhere in a quarry or a coal mine. Dust from rocks and rock-like materials
  • \$\begingroup\$ Could you add a picture of what you mean to your question? Perhaps including any physical characteristics of the join, such as alpha-particle transmission compared to the bulk of the material. \$\endgroup\$ Commented Mar 23, 2021 at 20:45
  • \$\begingroup\$ Other details missing from your question. 1. Belt material. 2. Belt thickness. 3. Joint material. 4. Speed. I second the drawing request. It's not clear whether the sensor is detecting the underside of the top of the conveyor (inside the belt) or the underside of the belt return path (outside the belt). \$\endgroup\$
    – Transistor
    Commented Mar 23, 2021 at 20:50
  • \$\begingroup\$ Why is it so important to detect a join in a conveyor belt? The ideal join would be undetectable. Is it a proxy for something else? \$\endgroup\$ Commented Mar 23, 2021 at 20:53
  • 1
    \$\begingroup\$ Do it optically. Super easy actually. You make the joint highly reflective, like by putting a piece of shiny tape on it. Even a line of white paint on a black belt probably be enough. Use a reflective sensor like this: keyence.com/ss/products/sensor/sensorbasics/photoelectric/info You don't need to grow your own, these are standard available off-the-shelf production units where all the engineering has been done for you. All you gotta do is plug 'em in... \$\endgroup\$
    – Kyle B
    Commented Mar 23, 2021 at 21:00
  • \$\begingroup\$ Magnetic detection at 10cm is possible but an order of magnitude more than most sensors are designed for, and almost certain to be finnicky to get reliable triggering at a specific, repeatable point. A reflective optical sensor with a target of retroreflective tape (from any hardware store) would work, but more details are needed to give a definitive answer. \$\endgroup\$
    – vir
    Commented Mar 23, 2021 at 21:01

4 Answers 4


Here's an inductive sensor with a claimed sensing distance up to 20cm:

https://www.seeedstudio.com/Grove-2-Channel-Inductive-Sensor-LDC1612.html It uses the TI LDC1612 chip.

MORE inductive finds:

(totally SICK, dude.)

  • \$\begingroup\$ I'd say this is right type of sensor, but would recommend more industrial inductive proximity sensor. \$\endgroup\$
    – Ernesto
    Commented Mar 24, 2021 at 2:44
  • \$\begingroup\$ These all seem to have detection distances in the range of a couple of mm.. \$\endgroup\$
    – ChrisD91
    Commented Mar 24, 2021 at 12:25

If you can warn the magnetic detector that the metal clips in the joiner are coming, then then you should just be able to outright tell the magnetic detector to ignore its own detection of the staples without secondary sensors of any sort. Especially if the magnetic detector gives an output more detailed than a simple ON/OFF output and especially if the conveyor belt speed is known and if it is the kind that has no slip. I don't see why you need a second sensor of any kind at all.

  • \$\begingroup\$ This isn't possible as tramp metal could provide a very similar signature to the metal belt clips. The clips however are in a fixed location on the belt and using a sensor to determine that it is indeed the clip at this point, the detector can ignore it. A secondary sensor for this is 100% needed. \$\endgroup\$
    – ChrisD91
    Commented Mar 24, 2021 at 7:16
  • \$\begingroup\$ @ChrisD91 Hmmm I disagree if the belt has motion feedback. The problem with all your other methods is that they are either inductive and would react the same way to the tramp metals, or are magnetic and there's nothing guarantee that other metal won't stick the magnet and muck up the words or that the metal coming isn't magnetized itself. Can this conveyor belt slip? Or use brass or austentic stainless steel clips on the belt? \$\endgroup\$
    – DKNguyen
    Commented Mar 24, 2021 at 7:18
  • \$\begingroup\$ Unfortunately this is for any conveyor that uses this steel clip method to provide a join. So this conveyor could slip, but would use mild steel for clips. Brass wouldn't be strong enough and stainless would be an unnecessary expense. As it is for any conveyor we also wouldn't know the belt speed without something like the addition of a tachometer. \$\endgroup\$
    – ChrisD91
    Commented Mar 24, 2021 at 7:36
  • \$\begingroup\$ The current solution is to use a miniaturised version of the main metal detector and have it placed near the side/edge of the belt or to vulcanise a metal piece into the belt where the clips are positioned (if they are in the middle). This seems overly complicated as it required double the circuitry. I was hoping for a much simpler method using a third party module of some kind. \$\endgroup\$
    – ChrisD91
    Commented Mar 24, 2021 at 7:38
  • \$\begingroup\$ @ChrisD91 Is the additional cost of a stainless joiner really an unnecessary expense in the face of adding on a whole bunch of electronics or a second metal detector? \$\endgroup\$
    – DKNguyen
    Commented Mar 24, 2021 at 8:03

Inductive sensors would trigger on all conductive materials to some degree, so they are not very selective. Hall effect detection is very sensitive and selective as only few materials are permanently magnetic. If the field from the magnet drops off too fast at distance, you can use some iron object to funnel the magnetic field into the sensor and increase it by orders of magnitude.

This sometimes goes by the name flux concentrator. E.g. A conical shaped piece of permeable material right near the sensor.

Ideally what you want to create is a single iron loop core, which has two slits: one for the sensor, and one for the magnet (i.e. the belt).


Could a passive RFID tag work? (Same thing as used in theft prevention) Placed on the belt's underside, should last for quite a while? Tags are cheap but unsure on the electronics cost on those.

  • \$\begingroup\$ How would this work? Could you explain more, in detail and more specifically how RFID in particular could solve the problem in the question? \$\endgroup\$
    – ChrisD91
    Commented Apr 5, 2021 at 23:02
  • \$\begingroup\$ They use "Pit tags" for fish, RFID tags for goods that could be shoplifted, and so on; Not my area of expertise but aware of them. A tuned RF tank in the tag resonates with the detector and sends a signal, en.wikipedia.org/wiki/Radio-frequency_identification for a bit more. Detection ranges are on the order of a meter easily, and you could place several tags into or on the belt (For fish they use "pit tags" which are waterproof, less range though I would think.) \$\endgroup\$
    – M W
    Commented Apr 7, 2021 at 4:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.