In the 1980s, pneumatic-clamped ultrasonic welders evolved from controlling only trigger force, clamp force, and exposure time (and amplitude by way of boosters) to a variety of strategies. The first of these was welding by energy. From a machine design standpoint, this is relatively easy to do, so when microprocessors first started making their way into ultrasonic machines, many machines were marketed with this feature as more advanced machines. Energy is the amount of work expended doing something, while power is the rate at which energy is expended. So what those units do is sample watts at regular intervals, from, in early systems 20 milliseconds, down to the current indsutry best of 0.5 milliseconds (shorter sample time is generally better). A running total is kept of these wattage readings, and when the total of all readings divided by the sample rate equals a certain predetermined amount of energy, ultrasound stops and the machine goes on to the hold part of the cycle. If these machines were put in weld by energy mode, they also could test the actual weld time against limits to see if a typical cycle was run. The converse was also available, welding by time and testing the actual energy against preset limits. The theory here is that if you know how much energy you put into a weld you know how much plastic you have actually melted. In practice, for most applications the best control mode remained weld by time, but energy data was collected, used to identify suspect parts, and the science of the process was advanced.