Variable Power Using the Windings on a TX

[Walrus]

I was wondering what you guys thought about the power scaling method used by Ampmakers where they take a transformer with a 275 winding, and a 190 winding, and use that to scale the power in the amp by using the 0-275 winding for full power, the 0-190 winding for half-power, and a combination of the 190-275 windings for 1/10th power.

Would that work with an 18-Watt?

Would it be as effective as the cathode switching technique and/or the VVR method?

[HotBluePlates]

I haven't seen a schematic for what you describe, but I looked at the voltages indicated from a math perspective.

If 275vac results in 100% power, then 50% power should result when the voltage is 70.71% lower. That's because power varies with the square of voltage (P= V*I = V²/R), so 1/2 power means you need Voltage * SqRt 2.

275vac -> 190vac does essentially match the reduction of voltage to arrive at 1/2 power.
275vac - 190vac = 85vac does essentially match the voltage reduction for 1/10th power.

The 1/10th power arrangement would require the HV to be derived from the difference of the existing HV windings, and I'd want to see how they do it to be sure it's safe. Other than that, the concept is sound.

Is it as effective? Well, it would work the same as VVR, except that you have 3 dicrete power levels instead of a continuously-variable amount of power. Realistically, Full-, Half- and Tenth-power are probably reasonable points to choose, as tenth-power should be approximately half the ear-volume (hearing is logarithmic).

I don't know nuthin' about the 18watt circuit. 275vac rectified by SS results in about 388vdc, assuming no loss in the rectifier or power supply impedance; I would expect to see 370-380vdc in most cases. That's not necessarily too high for EL84's, but might be higher than you'd prefer. I wouldn't think twice about using that supply voltage with 6V6, 6L6 or bigger tubes.

I would also want to use cathode bias in any voltage-based power switching scheme, because it simplifies proper biasing. The clever voltage-varying schemes which use fixed bias have a method of reducing the absolute value of the bias voltage along with the reduced B+ voltage (say B+ goes from 400vdc to 220vdc, bias goes from - 38vdc to - 18vdc). A cathode bias resistor does this for you automatically.

I would also feel safer switching power levels with the amp on standby (or off). That's not really a requirement with a VVR or Power Scaling circuit, because they basically use a variable voltage regulator to reduce the B+ voltage.

With 275vac, I'd probably look to rate all filter caps at 400-450v. Don't cheat and try to find a way to get away with 350v caps (maybe by using a lossy tube rectifier); the caps will be stressed at turn-on and cap-life will be reduced.