Shop OBEX P1 Docs P2 Docs Learn Events
Make your own load cell (force sensor) from stuff you already have. — Parallax Forums

Make your own load cell (force sensor) from stuff you already have.

Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
edited 2011-07-31 14:19 in General Discussion
Here's a project you can complete in a half hour or less with materials you probably already have at home or can obtain locally. It's a "load cell" or force sensor whose resistance is inversely proportional to the compressive force applied to it. It uses a "pellet" of black anti-static foam. This kind of foam is loaded with carbon particles. When the foam is squeezed, the carbon particles become compressed together, conducting electricity more readily, providing a variable resistance that can readily be measured.

Required Materials (Illustration below.)

1" x 2" piece of copper-clad circuit board.
Small piece of black anti-static foam.
1½" x 1½" piece of springy foam, the same thickness as the anti-static foam.
Some insulated wire.
Double-stick tape (e.g. carpet tape).
Solder.

Required Tools

1/4" hole punch.
Scissors.
Hacksaw to cut circuit board.
Soldering iron.
Fine-grit sandpaper or Scotch-Brite® pad.
X-Acto® or other razor knife.

Construction Steps

1. Polish the copper-clad circuit board with the sandpaper or Scotch-Brite® pad until it's shiny.
2. Cut the circuit board into 1" x 1" squares. Then nip one corner off of each one. (Illustration below.)
3. Solder wire ends to identical corners adjacent to the cut corners of both circuit boards. (Illustration below.)

attachment.php?attachmentid=56763
Materials and Circuit Board Preparation.

4. Punch a hole in the black foam, retaining the little round piece that gets punched out.
5. Cover both sides of the springy foam with double-stick tape, leaving the paper liners on the non-foam sides.
6. Draw a 1" square on one of the liners.
7. Punch a hole in the foam/tape sandwich about 1/3 of the way into the marked square. (Illustration below.)

attachment.php?attachmentid=56764
Holes Punched in Both Foam Pieces

8. Trim the foam to size along the lines you drew.
9. Peel the paper off one side of the foam sandwich and adhere it to one of the circuit board pieces. The off-center hole should be positioned opposite the wire/notched end of the board.
10. Insert the black foam pellet into the hole. Make sure it's pushed in all the way.
11. Remove the other paper liner. (Illustration below.)
12. Adhere the other circuit board to the newly exposed adhesive. Squeeze the two circuit boards together tightly to get a good bond with the tape.
13. With the razor knife, trim away the excess foam exposed by the notched circuit board corners. This completes construction. (See illustration below.)

attachment.php?attachmentid=56765
"Open-faced Sandwich" with Pellet Installed; Finished Load Cell

With an ohmmeter connected to the wires, you can test your load cell with various finger pressures to see how the resistance changes. I used a postage scale in conjunction with an ohmmeter and a drill press to measure the resistance at various measured forces and plotted the results, shown below. (See explanation below for why there are two curves.)

attachment.php?attachmentid=56774
Graph of Load Cell Resistance vs. Force

You can interface your load cell to a BASIC Stamp, using the RCTIME setup to measure its resistance. If your sole interest is measuring the force against a fixed threshold, a resistive divider, with the load cell as one leg, will be adequate. Just pick a resistor value that places the divided voltage at the logic threshold of your input when the applied force is at its desired threshold.

Load cells (the expensive kind) are often used in electronic scales. So a natural question would be, "How suitable is this load cell for weighing stuff." The answer is, "Not at all." The reason is two-fold: 1) Both foams have memory — especially the antistatic foam. This means that it won't bounce back quickly — or even completely — like a spring does and will eventually fatigue altogether. 2) The conductive properties of the anti-static foam will change over time as the foam fatigues and the carbon particles shift. Both these properties will conspire to compromise the repeatability needed for precision weighing.

The memory property mentioned in 1) above is also called hysteresis, the tendency fo a system to stay in the state it's in, even when external conditions change. That's why there are two curves on the above graph. One curve shows the resistance as the applied force is increasing; the other, as the force is decreasing. As you can see, there's a wide disparity between them, rendering any absolute measurement questionable, at best.

Nonetheless, for sensing things like relative finger pressure, bumps, footfalls, robotic grip pressure, etc., a simple sensor like this one may well suffice. In any event, they're simple and cheap to build and fun to play with.

-Phil

Update: 'Took new measurements using a drill press to stabilize the applied force, this time in both directions to demonstrate the foam's memory property.

Post Edited (Phil Pilgrim (PhiPi)) : 11/16/2008 7:36:08 AM GMT

Comments

  • SRLMSRLM Posts: 5,045
    edited 2008-11-16 03:25
    Nice start to the sensors thread!
  • Oldbitcollector (Jeff)Oldbitcollector (Jeff) Posts: 8,091
    edited 2008-11-16 03:28
    Awesome! Nice instructional!

    OBC

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    New to the Propeller?

    Getting started with a Propeller Protoboard?
    Check out: Introduction to the Proboard & Propeller Cookbook 1.4
    Updates to the Cookbook are now posted to: Propeller.warrantyvoid.us
    Got an SD card connected? - PropDOS
  • ZombieBotZombieBot Posts: 2
    edited 2008-11-16 17:56
    Very cool. I'll have to try this for a gripper sensor. It's a moneysaver (which for a broke college kid like me is a very good thing).
  • Chris SavageChris Savage Parallax Engineering Posts: 14,406
    edited 2008-11-17 17:35
    Hello,

    This sensor question is being moved to the Sandbox forum. Sorry for any inconvenience this has cause. The sensors forum is being removed. Please see announcements for further details. Take care.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Chris Savage
    Parallax Engineering
  • DosManDanDosManDan Posts: 179
    edited 2008-11-17 19:43
    What a great idea! I have to try this out.

    Dan
  • bambinobambino Posts: 789
    edited 2008-11-19 15:05
    This really gave me a, "Duh (smacks forehead) Why didn't I think about that?" , Moment!
    Nice one one Phil, Thanks!
  • mehran4868mehran4868 Posts: 1
    edited 2011-07-30 08:17
    Many thanks to Phil.
  • localrogerlocalroger Posts: 3,451
    edited 2011-07-30 11:12
    Well this must be some kind of a plot. I come here on the weekend to relax and forget about work and ... did I mention that I work for a scale company?

    Semiconductive foam sensors like this are good for things like bump detectors on robots. A good source for the foam is an old cheap keyboard; most modern cheap keyboards use this stuff to determine that you are pressing a key. But for actually measuring force, it's very meh. It's about as accurate as eyeballing something and making a guess as to how much it weighs.

    The silver standard for actually measuring force is the strain gage, a resistive element that is epoxied to a piece of metal that flexes in response to force. 15 years or so ago I know there were inexpensive self-adhesive strain gages you could buy to do your own DIY force measurement, but they never worked very well and you still needed to form up a Wheatstone bridge and use an instrumentation op-amp to read them. A proper load cell with internal flexures to cancel out side loads and such will set you back USD$200 at minimum. There's a lot of engineering involved.

    Incidentally, the load sensors used in the cheap electronic bath and diet scales that can be bought retail are Smile. Play with the conductive foam before wasting money on those things.

    The gold standard for force measurement is what is called "force restoration." In this scheme there is a lever system which applies the measured force to an arm that moves an element through a null detector, and a magnet which resists the applied force. The amount of current that must be put into the magnet to center the null detector reveals the measured force. A strain gage scale can reliably measure to one part in 10,000 over its capacity, but a force restoration balance can measure to one part in 1,000,000 or better -- only limited by the machining of the lever and flexure system. All electronic lab balances work by force restoration.

    At the very low end, you can also measure force by having two sheets of metal spring-loaded to get closer to each other when force is applied, and measuring the capacitance between them. That's how a lot of those electronic diet scales work, but there are all kinds of errors that creep in (air has a dielectric potential which changes depending on the humidity, and are the springs temperature compensated?) but they don't have the hysteresis of conductive foam.
  • K2K2 Posts: 691
    edited 2011-07-30 11:33
    Hmmm... Force Restoration. I always wondered how the US Mint could weight a barrel of pennies to within a single penny. Thanks for the explanation. BTW, I once used a load cell and an embedded controller to make a dynamometer for a large mass transit company. Curiously enough, they still use it. I'm glad someone else designed the load cell (Omega) for the very reasons you mentioned.
  • LeonLeon Posts: 7,620
    edited 2011-07-30 12:04
    Many years ago a company with which I had some involvement used a load cell in a bank note counting machine. It was very impressive - with a big stack of simulated notes removing a single note reduced the count by one.
  • Mark_TMark_T Posts: 1,981
    edited 2011-07-30 15:54
    localroger wrote: »
    The gold standard for force measurement is what is called "force restoration." In this scheme there is a lever system which applies the measured force to an arm that moves an element through a null detector, and a magnet which resists the applied force. The amount of current that must be put into the magnet to center the null detector reveals the measured force. A strain gage scale can reliably measure to one part in 10,000 over its capacity, but a force restoration balance can measure to one part in 1,000,000 or better -- only limited by the machining of the lever and flexure system. All electronic lab balances work by force restoration.

    Apparantly the new gold standard is the watt balance, where the current balancing step is augmented by a calibration of the coil via the motor/generator equation giving perhaps upto 1 part in 10^8, see http://en.wikipedia.org/wiki/Watt_balance

    This might be quite a daunting project for the home constructor ;)
  • localrogerlocalroger Posts: 3,451
    edited 2011-07-30 21:22
    Mark_T, I think I would consider that tech something like the platinum standard. :=)
  • Dave HeinDave Hein Posts: 6,347
    edited 2011-07-31 07:30
    localroger wrote: »
    Incidentally, the load sensors used in the cheap electronic bath and diet scales that can be bought retail are Smile. Play with the conductive foam before wasting money on those things.
    Hmm, I used a bathroom scale load cell to measure rocket thrust, and it seemed to work fairly well. I would be interested to know the deficiencies of the bathroom scale. I assume it has an accuracy of a couple of percent. Is the accuracy worse than that, or is there something else about it that makes it "Smile"?
  • RavenkallenRavenkallen Posts: 1,057
    edited 2011-07-31 09:51
    I work at a supermarket and we have this scale that can literally weigh the money from the register and it knows exactly how much is in the stack. It is pretty cool:)
  • localrogerlocalroger Posts: 3,451
    edited 2011-07-31 14:19
    @Dave -- A couple of percent is about right, only worse if you actually want to use the scale platform. They tend to have huge corner and milk errors that can't be adjusted out. Bear in mind that 0.2% is considered maintenance tolerance for a real scale. Measuring rocket thrust is a special case because you can strip away the crappy platform and lever system, which helps a bit.

    @Ravenkallen that's a special case of what is called a counting scale. You put a sample on it, such as 10 dimes, and it figures out from the sample how much a dime weighs and when you dump the rest of the dimes on it, it figures out how many there are based on the calculated piece weight. Money scales sometimes have the sample weights pre-programmed and can show the results in monetary units.
Sign In or Register to comment.