Magnetic Digital Scale

Above you see a solenoid being used as a digital scale. The magnetic field from the coil in the base levitates the platform above, where a load to be measured is place. This floating platform has a permanent magnet in it, hovering above a hall effect sensor in the base. As the distance between that magnet and the sensor changes, the measurable magnetic field changes as well. The hall effect sensor is linear so the measured value can easily be correlated with a weight. In the video after the break [Vsergeev] demonstrates the device using test weights to show off its 0.5 gram resolution. He thinks that with a few hardware improvements he could easily achieve 0.1g accuracy.


20 thoughts on “Magnetic Digital Scale

  1. Accuracy affected by temperature and friction. If your weight was not exactly on center, it would tilt the platform enough to cause slight friction on one side. Temperature affects magnetism also. A typical load-cell scale uses a load cell in each corner of the platform and they are “summed” together. The weight can be placed anywhere on the platform and the results are constant. One might argue that the metallurgy has temperature affects, but not as unstable as temperature and magnetism.

    But the idea and construction of your magnetic scale is a cool project. Technically, it would be difficult to maintain NIST traceability with the temperature variability alone.

    I wish I had the time to do projects like yours.

  2. Another algorithm for this sort of thing is to use a simple analog feedback loop to keep the platform at a constant height and then use the coil current to indicate total measured weight.

    You could use an optical knife edge beam-break under the center of the platform for instance to define the target height. Then feed the coil current into an ADC to convert into the output value.

    An advantage is that you can replace the concentric cylinders with a simple hinge or bendable spring arm. Since the platform will always be at the same height no matter how much is being measured there are no trigonometry issues to deal with and no mechanical stiction issues.

    Nice project.

  3. I’m not a grammar stickler, but I’d like to think that HaD has not been outsourced, since lately they are struggling with the English language.

    “where a load to be measured is place.”

  4. I work in the scale field and know cells like this exist. They are called MFR cells or Magnetic Force Restoration cells. They are really quick and really accurate. But expensive. I am really impressed he built one himself.

    Most actually do not use a feed back system. They ramp up the power to the coil until the field lifts the weight. And most use a lever arm so the load doesn’t have to be balanced on the cell as well as getting higher capacity or resolution.

    Still a very cool build.

  5. It does have a permanent magnet. It uses the solenoid to vary the force used to push the magnet up and thus push the weight up. Essentially it is finding the amount of force necessary to counteract gravity and thus find the weight.

  6. wow, mlseim beat me to it. While this is a semi-workable scheme, it’s very “semi.” With nearly identical hardware you could get a much better result.

    First, instead of the tube you use an arrangement of flexures to hold the weigh platform parallel as it is allowed to move up and down. This eliminates friction, but does introduce a lot of alignment difficulties; real scale manufacturers have jigs to keep up the tolerances.

    Second, instead of measuring displacement you do as RP suggests and vary the coil current to keep the position constant via a null sensor and feedback loop. This has the advantage of eliminating one of the big drawbacks of flexures, being that they want to torque the platform as it rises and falls. With the position constant the flexures don’t have to flex much.

    This is the way most laboratory balances work, and they can be much more accurate than load cell based scales (which work on the principle of metal deforming a bit as it is stressed, and this deformation slightly changing the resistance of a “strain gage” sensor).

    The accuracy of a scale is described as “n Max” in the trade, the number of reliably discernable “counts” you can repeatably measure from zero to the highest capacity it can weigh.

    Single cell strain gage platform cells with integrated flexures are generally good for nMax of 10,000, so a 100 lb scale can resolve 0.01 lb.

    Multiple cell strain gage based platforms are less accurate for several reasons, and are usually limited to 4,000 or 5,000 counts. This is why a 100,000 lb truck scale weighs in increments of 20 lb.

    But force restoration balances (using the flexures and null sensor and coil current to return weight) are limited only by the precision of their manufacture. If you’ve got the pockets you can buy scales capable of 100,000 or even a million counts. All such electronic scales work on this principle.

  7. This is one of the nicest discussions with such a wealth of feedback and information I have ever seen on HaD. Thanks to Vsergeev for sharing and the rest of the commentary adding useful tidbits.

    Very enjoyable reading

  8. Wow,
    this is a good discussion, now my idea.
    Instead of your coil magnet construction you could take an loudspeaker,then take the analog optical feedback as written above.
    This was a project in an german electronics magazine years ago.

  9. could you use 2 permanent magnets and a pair or so of hall effects as the magnets push together under load the field between/around them would change measuring this change at different points could compensate for uneven loading and would require less power for portability.

  10. Scales made from companies like Acrison use LVDT’s.

    A platform pivots or tilts on torsion bars and
    the “sensor” measure the rod as it moves through
    a coil.

    I just thought of the LVDT scales when Hubert
    mentioned using a speaker. That’s sort of the
    same idea as the LVDT.

    This topic is interesting because there are many
    different types of electronic scales … all used
    depending on the application, amount of weight,
    accuracy, and cost.

  11. Or you could use three Hall sensors, cheap and nasty optical drives use a 4 pin variant which is functionally identical to the expensive and hard to locate 4 pin leaded trapezoid parts often found in vintage 5 1/4″ drives and old record decks.

    iirc the outputs are centred on 2.5V and a simple differential amplifer (one LM324) can then be used to feed the microcontroller’s A-D inputs.

    (ideas swiped from the relevant 4HV forum posts)

  12. Hello, well done on this project.
    I need a high resolution scale (0.1g) with quick dynamic response ( a settling time less than 1 second). I want to use this scale as a reference signal for a control system which controls the dispensing of liquid up to 0.5 ml accuracy. Been looking around, but either the scales settling time is too long or they are too expensive. Any advice? Do you mind sharing the circuit diagram or design details?
    This would be greatly appreciated.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.