by dcgillespie » Mon Jun 28, 2010 12:59 am
I have not searched for a link to the article I referred to, but it is entitled "New Amplifier with KT88's" by David Hafler, and appeared in the January, 1958 edition of Radio-Electronics magazine on pages 58 & 59. He provides a graph in the article showing how the chosen value (11.2 ohms) reduces IM distortion from .5% to .2% @ 27.5 watts output, and from .8% to .5% at 55 watts output -- over that of simply grounding the cathodes, or using values up to double that of which was chosen.
As to your other question and where negative voltages can come from; regarding output stages and specifically audio work, if they are negative, they must be supplied. This of course is all as referenced to ground. Under quiescent conditions, as far as the tube is concerned, all it needs is for it's control grid to be more negative than it's cathode. Now whether this is created by referencing the grid to ground and then allowing the cathode to become more positive by way of a voltage drop across a cathode resistor, or by grounding the cathode and then supplying the grid with a negative voltage, it makes no difference. It could even be a combination of the two. In either case, the grid is made negative with respect to the cathode so as to control current flow through the tube.
When a cathode resistor is used, it is called cathode bias. When a negative supply is used, it is called fixed bias. Both approaches have drawbacks and benefits which go far beyond mere biasing of a tube under quiescent conditions. Understanding which is best used and when starts with the overall design goals for a given amplifier. High power and low distortion favor fixed bias, while economy and low maintenance favor cathode bias.
As for the ST-70, it is a fixed bias amplifier. Yes, there is a small amount of bias created by way of the Biaset resistor, but the vast majority is provided by the negative bias supply. And yes, you can adjust the negative bias voltage to the tubes, but it is still considered "fixed" as once set, that is the bias the tube operates with. With cathode bias, it is typically not adjustable in classic designs, and is not considered as fixed since the bias voltage will depend on the current flow through the tube at any given moment. That of course varies with signal conditions, and so is not considered as fixed in the same manner that use of an adjustable negative bias voltage is. I hope this helps.
Dave