Bob Lunn
08/26/97 11:18 AM


> A scale that could measure down to fractions of
> a gram would be somewhat useful - counting things
> - measuring correct ratio of sulphur / charcoal /
> potassium chlorate etc (no, I don't make this in
> 5k lots!) <g>

     Laboratory scales and commercial counting scales
     are most certainly useful things.

     You might care to price these items.  This will
     give you a first order estimate of the difficulty
     of building one.

     In considering mechanisms to use for load or mass
     determination please keep in mind:

          span drift
          zero drift
          hysteresis
          creep

     ...just for starters.

     Jacques requires a mechanism to measure a load
     from 0 to 5,000 grammes with a resolution of 5
     grammes, with a dead load not exceeding 500
     grammes.

     We should still clarify the temperature range over
     which the scale is to operate, the repeatability
     of readings, long term drift of readings, need for
     zero and tare mechanisms, and requirements (if any)
     for standards certification.

     These days, the conversion of an electronic signal
     to a digital value is easily solved for reasonable
     cost.  The greatest problems usually arise in the
     physical load sensing mechanism.

     There's no point in having a 16-bit A/D converter
     if an elastomeric mounting imposes 5% hysteresis,
     or a knife edge continually binds.

     Anyway, the most common sensing techniques for
     electronic load measurement are:

          metal foil and silicon strain gauges

            (configured to respond to shear, bending,
             torsion, compression, or tension)

          piezoelectric bending beams

          capacitive measurement of displacement

          optical measurement of displacement

     Strain gauges are most commonly used for loads in
     the range of 50kg to 25,000kg.  The most common
     implementation would be a shear beam, with a nominal
     accuracy of 2000 to 3000 grads.

     Physical constraints on the load sensing element to
     which the strain gauge is bonded makes the use of
     strain gauges for very small, and very large, loads
     more difficult.

     Piezoelectric systems have probably had most success
     with very large loads requiring modest accuracy, such
     as gross tonnage measurement of trucks.

     The difficulty with piezoelectric elements is that
     they only respond to changes in load.  A static load
     produces no output.  They can, however, be used to
     build very robust (and thus safe) weighing systems.
     Hence their success with high capacity weighing.

     Capacitive and optical measurement of displacement
     are the most common mechanisms for small loads.  For
     a 5kg scale these are certainly the approaches I
     would be considering.  I would not use a strain
     gauge based load cell for this purpose.

     Scale design and manufacture is a _very_ competitive
     industry, and has been so for many decades.  There
     are a large number of patents covering scale design
     and manufacture.  An excellent education in these
     issues can be drawn from reading patent documents.

     Given that many of these patents have expired, the
     techniques they describe are available for use.

___Bob