A Tip When Designing Power Supplies

[steve]

As a tip, try to eliminate chokes in the B+ power supply filtering stages. The reason is a choke causes
distortion in multiple ways. Any alteration to the musical signal is by definition distortion as the output
is different than the input.

Not only does the musical signal see a reactive part, but sees three reactive parts, the decoupling capacitor,
the previous inductor/choke, and the previous filter capacitor.

That decoupling capacitor you see in the schematic is not just a capacitor. That decoupling capacitor reacts
with the choke and previous filter capacitor to create a mess.

Besides the inductive reactance, which varies with frequency, the inductor also has winding resistance
which does Not vary vs frequency.

Below is a photograph taken, showing a thevenin equivalent circuit, and what happens to a signal of varying frequencies across a

typical 1 Henry choke,
200 ohm winding resistance and
200 ohm load resistor.

X horizontal line represents increasing frequency, Y vertical line represents increasing signal voltage amplitude.

1. A perfect choke with zero dc winding resistance would show an upward angled straight line with no deviation.

2. If we only consider the winding resistance, Zero inductance, one would see only a horizontal straight lline.
This would create constant isolation between filter capacitors. Maybe not perfect isolation, but
constant vs frequency.

3. With both inductance and 200 ohm winding resistance, the line is curved vs frequency. That will
cause the musical signal mess which cannot be compensated for.

Of course some circuits need a choke to minimize hum, but those designs/circuits are inherently flawed.

It is best to eliminate chokes completely if possible. If one cannot, try to eliminate the
choke nearest to the gain stage and use a non-inductive power resistor of choke winding resistance
instead.

cheers

steve

[Folsom]

Resistance and/or chokes mostly work for tubes though... A little voltage point is not a big deal if you have 400vdc, but if you have 26vdc then it can be a potentially large amount of output wattage lost.

Anything with an inductor is complicated and tends to run the risk of sounding life sucking... There are ways and not ways to use them.

[steve]

Yep, chokes mess up the sound.

For general public, a choke connected to the decoupling capacitor in the power supply (in this case with tubes) causes a change in RL vs frequency in a non-linear way (similar to the decoupling capacitor changing RL). Of course the decoupling capacitor also changes RL. So two reactive components that causes RL to change value.

The equation for gain for a common cathode triode stage is: -mu x RL/rp + RL

Another equation for the gain is: -gm x rpRL/rp + RL

mu = amplification factor, often called just u.
gm = transconductance
rp  = plate resistance of the tube
RL  = plate load resistance (from plate to B+ voltage)

As RL changes vs frequency, the gain of the stage changes vs frequency.

When changes of 1 part in a million in frequency response (FR) is perceived,
that is a change in the -120db range.

Books for reference.

"Semiconductor and Tube Electronics An Introduction" by James G Brazee (One of my college textbooks.)
RCA "Radiotron Designers Handbook" edited by Langford and Smith, by over 2 dozen engineers, 1960.

cheers

steve