A significant precursor to the major changes was the introduction of weight loader technology, where loose weights were replaced by captured weights that were raised from and lowered onto a weigh pan extension.  The loaders were driven by manually operated dials that read out in decimal values.  This did not involve any real change in weighing technology but eliminated loose weights and much of the drudge involved in equal arm weighing.  It also got users in the habit of being able to read the final weight result directly as a decimal value, rather than adding up the values of multiple loose weights.  It made weighing easier and faster.


Along came the small electronic microprocessor, a far cry from today’s versions, but enough to straighten out the curvature in the force vs. current characteristic of a magnetic force motor.  These motors themselves had only recently been greatly improved as stronger and less temperature sensitive permanent magnets became available.  OK, we have a replacement for the weights, but what do we do for a suspension that suits our force motor.


Once again Roberval to the rescue.  The Roberval suspension gets rid of what are called the end loads and side loads and makes the top loading balance a reality.  End loads and side loads are caused when the unknown weight is placed off center on the weigh pan causing torque forces that must be compensated for.  The Roberval suspension calls for friction free pivots in order to produce a structure that is a parallelogram when viewed from the side and a rectangle when viewed from either end.  It can operate as a full equal arm balance, or it can operate as half an equal arm balance.  The Torsion Balance Company produced Roberval mechanisms using taut bands in a torsion mode to act as the pivots.  It sold them as equal arm balances for the laboratory and a half equal arm in school scales.  In the new force motor designs the pivots were made from flexures which were necked down and squeezed thin at the flexure point.  These remain the physical weak point of most lab and analytical scales even today.  


All the weights are gone and the computer has turned the force motor current into a digital value, scaled it properly based upon a known calibration weight, offset it to provide the zero and tare features, linearized the readings to compensate for the sag in the force versus current curve, temperature compensated for the changes caused by magnetic motor temperature changes, has a mode which can remove long term drift, drives a digital display that displays the weight in any one of many units, does a whole bunch of different applications (percent weighing, parts counting, totalizing, etc.), and more.   
The logical question is what comes next.  Laptops run the show.  Mechanisms get smaller and lighter.  Software comes on discs.  As far as basic technology changes, I’m drawing a blank.