December 2002

The first half of my column this month is devoted to letters I've received over the last twelve months.

Some time ago I mentioned the ECH83 - a 12V h.t. triode-heptode intended for use in car radios. It was widely believed to be nothing more than a specially selected ECH81. Chris Colebrook has since measured an ECH83 on a valve characteristics meter and found its characteristics to be identical to those of an ECH81. But if they're the same valve, why are the maximum ratings of the ECH83 so much lower than those of the ECH81. Were Mullard just being cautious, or did the different ratings add weight to the deception. Perhaps one day, Mullard will own up.

The Infamous Backward Volume Control

In the circuit to the right, you'll see that the volume control has been wired backwards. The anode is usually connected to the top of the volume control and the slider feeds the control grid. But this arrangement is the reverse. The practice appeared in the late 1950s and continued into the early part of the 1960s. Even some transistor circuits had the volume control wired like this. In the column I give some reasons why this connection should not be used. Here are two graphs to illustrate the points I make. (Open them in another window or save them to disk and print them out.)

Graph 1

Graph 2

Graph 1

I haven't yet found a mathematical formula that gives the percentage (of total) resistance of a logarithmic-law potentiometer at any given angle of rotation. (If anyone has such a formula, can you please let me know.) The curve marked 'Log' was copied from a graph I found on the 'Web and gives the voltage at the slider of a logarithmic-taper potentiometer when it's wired in the usual way. The x-axis is the rotation of the volume control in degrees. The y-axis is the output voltage in percent. The curves marked '10k source' and '100k source' show the voltage at the top of a backwards-wired 1M ohm volume control when the source impedance (that drives the slider) is 10k ohm and 100k ohm respectively. The top of the volume control is open circuit. (This is valid at low and medium frequencies when the top of the volume control feeds the grid of a valve.)

You'll see that when the source impedance is low, the output from the volume control varies almost anti-logarithmically. This is exactly the opposite of what is needed. Even with a moderate source impedance the output tends to become linear, still far from ideal. The only way the output could approach the usual logarithmic curve is for the source impedance to be numerically similar to the resistance of the volume control itself. In this case, 1M ohm.

Graph 2

Just in case I was fooling myself, I produced two new curves based on the output from the top of a backwards-wired 1M ohm volume control (as in the previous graph) when the source was an ECC81 and an ECC83 amplifier stage. For the ECC81 I've taken ra=20k ohm and an anode load resistor (Rl) of 100k ohm. For the ECC83 I've taken ra=80k ohm and an anode load resistor (Rl) of 220k ohm. The curves are scaled so that 90% output is reached at 285 degrees rotation. The curve marked 'Log' is the same as in Graph 1. I think I've done the calculations correctly; if anyone disagrees with my figures, let me know.

You'll see that the curves lie between the 10k and 100k curves of Graph 1, just as you'd expect. Both are still far from the ideal logarithmic response. Maybe a pentode with a 1M ohm anode load would be better, but why bother. Just wire the volume control the way it should be wired. I still wonder why this backwards connection ever caught on.

Return to the contents page