Most tube testers test at only a single point. Even if you used real world voltages, single point matching leaves a lot to be desired.
When performing a curve trace, the Computerized Cardmatic software performs 8 to 15 separate tests. For each test the plate voltage is held constant, the plate current is measured at various grid voltages, which is then entered into the computer program. This data is used to plot a line on a graph.
The gradient of this line shows the tube's mutual conductance (the ratio of the change in plate current to the change in grid voltage). This line also shows the tube's plate current at any given grid voltage. The closer the plotted lines are for any given tubes, the closer these tubes are to being matched, for both mutual conductance and plate current.
These tests are extremely sensitive, and show even minute differences in the characteristics of the tubes being tested.
Ideally, these tests would be performed at the same plate voltage that the tube will see in circuit. The results would then be so accurate that you could actually use these curve charts for not only matching, but even designing circuits. However, designing circuits is not what we are trying to do with these charts, we are simply trying to match tubes.
What these curves do show, and show well, are the tube's trends. The word trend is important here. Extensive testing has shown that tubes which plot similar curves at 150 plate volts will be closely matched even at higher plate voltages. The curves produced at 150 volts cannot tell you the tube's exact parameters at any other voltage, but they are so good at predicting the tube's trends that it is very rare for tubes which plot similar curves at 150 volts to not be matched at higher voltages. This method of matching tubes, even at voltages well below real world voltages, really does work.
Phil Frakes, designer/programmer of the computerized version of the
Cardmatic Tube Tester