Most of the AM'ers endevour to improve their audio, at least in the area of distortion. One area which is seldom examined is whether the class C final reproduces an output signal in exact accordance with the modulating waveform.
 I examined this carefully a while back, with my then 6146B powered DX100 final. (It's since been changed to a pair of 813's, another subject!!) The trapezoid scope test is invaluable for this test. The modulation is applied to the horizontal, and an RF sample is applied to the vertical. Thus, in a perfectly modulated signal, a triangular trapezoid pattern will be produced, with perfectly straight sides.
 In the case of the 6146's, it turned out to have a noticable slope which meant the output amplitude was NOT directly proportional to the modulating waveform and hence, distortion. Through experiments, I determined the modulated screen voltage, derived from a dropping resistor from the plate supply ( modulated B+), had too much influence on the output signal. In other words, when the modulating waveform was on the negative swing, additional screen voltage was needed, and conversely, on the positive swing less voltage was needed.
 Using the trapezoid scope pattern for verification, I changed the screen circuit to a network where part of the screen voltage is from the fixed plate supply (unmodulated B+), and the other part from the normal screen dropping resistor (modulated B+). Thus, achieving the goal of more screen voltage on the negative swing and less for the positive. This resulted in a perfect waveform except for a small "bump" that was still present near cutoff or the bottom of the modulation swing. I determined that increasing the grid drive from 2.5 mA per tube to 3 mA per tube removed the "bump" which left an exquisitely perfect modulated waveform!

Top of Next Column

 

 Using the trapezoid scope pattern for verification, I changed the screen circuit to a network where part of the screen voltage is from the fixed plate supply (unmodulated B+), and the other part from the normal screen dropping resistor (modulated B+). Thus, achieving the goal of more screen voltage on the negative swing and less for the positive. This resulted in a perfect waveform except for a small "bump" that was still present near cutoff or the bottom of the modulation swing. I determined that increasing the grid drive from 2.5 mA per tube to 3 mA per tube removed the "bump" which left an exquisitely perfect modulated waveform!
 With proper voltages now on the final, another overlooked source of distortion is the RC set up by the screen dropping resistor (R1) and the screen bypass capacitor (C1) which can cause a phase difference between the plate and screen modulation signals on the higher audio frequencies. A technique I used to eliminate this was to place a proportionally valued capacitor (C2) across the plate to screen dropping resistor (R1). Distortion from the phase shift can be readily seen as a loop effect on the sides of the trapezoid pattern.
 In short, for the 6146, I recommend the following . . .

* 3 mA per tube grid drive.

* Derive 60 percent of the screen voltage from the modulated B+ (using a dropping resistor.) and 40 percent from the unmodulated B+ (using a second dropping resistor, R2 on schematic).

73's and Clean signals!!

Dean


from The AM Press/Exchange, May 1989

 

Back to the Mods Page


23 January 1998