This article is from the Antique Radios And Phonographs FAQ, by Hank van Cleef vancleef@netcom with numerous contributions by others.
My radio plays, but the audio is distorted. Announcers sound like
mush-mouths, and music sounds as though gravel is rattling in the
instruments. I checked it with another speaker, and it sounds just as
bad.
The most common causes of distortion are in the power amplifier
circuit.
(Note that in the following I am assuming class A or AB1 operation,
where tubes do not draw grid current. If the grids of your power
amplifier tubes are driven by power tubes, such as a 6N7 or 6V6's, most
of the following applies to operation at low output).
1. Check that the coupling cap (or caps, in the case of
push-pull) are not leaking DC from the preceding stage and pulling the
output tube grid high. Most circuits use a 180K to 500K grid leak to
ground and a .05 or .1 microfarad coupling cap. At low-moderate audio
output, there should be no measureable DC voltage across the grid leak
resistor. Check the grid leaks themselves for proper value and good
connections (typically to ground). Wax paper coupling caps here are
notorious for giving problems, and are candidates for replacement even
if they appear to be good. The tube itself may be developing excess gas
current in the grid circuit. Disconnect the coupling cap, turn the set
on, and make sure there is no voltage developed across the grid leak. If
there is, replace the tube. Note that most tube testers won't disclose
this problem. With larger tubes (6V6, 6L6), replacement tubes made
after the mid-70's often had poor gas current characteristics, and some
designs were built with higher-value grid leak resistors than specified
by the manufacturers on the assumption that replacements would "never be
that bad." Most beam tubes specify a maximum impedance in the grid
circuit of 500K for cathode bias, 100K for fixed bias operation.
2. Check the value of the cathode resistor. Be careful here,
because a resistor that has overheated may not only have changed value,
but have charred the color bands so that they look like a very different
value resistor. If the circuit uses a cathode bypass capacitor (usually
an electrolytic, 20 mfd. 25 volt typical), check that it isn't leaking
current, and check that it has capacitance.
3. Check grid bias with the set running. Proper bias for
various tubes can be estimated from tabular data in tube manuals, and
ranges from around -7.5 volts for a small high-gain beam pentode like a
50L6 to around -60 volts for a large low-gain triode like a 2A3.
4. On a push-pull output stage, check that both sides are
operating. An easy check is to jumper across the grid leak resistor
with a clip lead, and see if things change. If jumpering one input
kills the audio, the other side is inoperative. Prime things to suspect
if one side is dead are the power tube on that side, open transformer
plate winding (no B+ on that tube), open coupling cap, or problems in
the voltage amplifier ahead of the output stage.
5. If you haven't found the problem yet, check the quality of
the audio coming out of the preamplifier stages.
6. DC imbalance can cause problems in push-pull circuits. Most
old radios don't have any place to measure this. You can wire 100 ohm
resistors into the plate circuits, in series with the output
transformer, and measure the quiescent DC voltage across them. For most
old radios, a 20% imbalance is tolerable. Keep in mind that the
voltage developed across a cathode resistor is total cathode current,
both screen and plate, and that a common cathode resistor in a push-pull
circuit is looking at the effects of two tubes simultaneously.
 
Continue to: