by paart » Sat May 07, 2016 9:37 am
Being able to independently set a DC bias point for two output tubes doesn’t eliminate the need for matched tubes. While doing so is beneficial in reducing the DC component in the output transformer primary at idle, it doesn’t ensure that the two tubes will conduct identically at 25, 50 or 100 percent power output.
This can be thought of as the AC balance of the output pair. While some designs incorporate an AC balance control for the phase splitter, it ends at that point. If the output tubes do not have matching curves so that the “AC” balance is maintained in that stage, then increased distortion will result.
Due to production tolerances, all tubes deviate from the published plate or transfer characteristic curves in a number of ways. The published curve is an ideal, or average, and individual tubes will deviate from this. Setting the grid bias at idle sets the relationship between grid voltage and plate (and screen grid, if applicable) current at one point only. During operation this relationship will vary from tube to tube, as the grid voltage is varied over a wide range.
The same thought applies to trying to match tubes on a tube tester. A tube tester only measures the transconductance of the tube at one point on the curve. Two tubes that show the same reading on a tester at the specified grid voltage (sometime labeled “bias”) may show considerably different results at a different grid voltage.
Be careful where you buy your “matched tubes”. True tube matching requires special equipment and knowledgeable people!