Hi:

I'm the one that mentioned "practical sense". I was referring to two issues
here. Issue #1: In your posts circa 1999 you speculated that the the
common cathode resistor scheme might produce more power under certain
conditions in an AB output stage (such as the AA151 or Dyna ST/SCA 35 are)
which could be an advertising advantage at the time these amps were
marketed. What I found by actual measurement was that there was no
appreciable difference in max power output between two tubes on a cathode
resistor vs all four tubes on a cathode resistor with either both channels
driven or one channel driven. This was on Dyna ST35. Signal was a sine
wave at various frequencies. Admittedly, music is not a continuous signal
so the results might be different with transients or tone bursts. At any
rate, I was unable to verify the theory that the common cathode resistor
would allow more power in any of the various cathode bias schemes.

Issue #2: You also state that best output tube balance is achieved with
individual cathode resistors, next best with four to a resistor and worst
with two to a resistor. By balance I assume you're referring to the
relative equality of bias current at idle. My understanding and experience
regarding the various biasing arrangements is that the ranking you give them
is really more by order of the degree of influence one output tube has on
the others tied to the common cathode resistor. Obviously, individual
cathode resistors provide the greatest degree of isolation. Due to
averaging, four tubes to a resistor is next with two to a resistor being
worst. So if one inserts four randomly chosen tubes into the amp, tubes
with the individual cathode resistors will bias up to some idle current
value dictated by individual tube characteristics at the chosen operating
point. However, each tube will probably idle at a different current. In
the case of two randomly chosen tubes to a cathode resistor, the tube
tending to draw more current at the operating point will tend to turn off
it's lower current mate thus exacerbating any imbalance between the two.
The case of four randomly chosen tubes to a cathode resistor is intermediate
between the two cases in that the influence of any one tube on the others is
reduced by averaging. However, aren't most PP amps designed on the
assumption that current imbalance betw output tubes is minimized? While
most OPTs will tolerate some degree of imbalance, on the whole, idle current
balance is desirable.

If one agrees that idle current balance is important then one will tend to
used matched pairs in the output stage of a PP amp. This is where real
world conditions may reduce the potential benefits of the four into one
cathode resistor. Assume two matched pairs of tubes where matching betw
tubes in a pair is close but matching betw the two pairs is not: a real
world situation in my experience. If this set of tubes is inserted into a
PP stereo amp with individual cathode resistors everything is fine. If
these tubes are inserted into the same amp but with individual cathode
resistors per pair of tubes, everything is still fine even though one pair
draws more current than the other. Finally, if these tubes are inserted
into the amp with all four tubes on a single cathode resistor, the hot pair
will tend to turn off the cooler pair similar to the effect of two
unbalanced tubes on one cathode resistor. The difference here is that
presumably, tube to tube balance will be maintained on each PP pair although
not at the expected values. I'm reasonably sure this is why Dyna
recommended a matched quad of 6BQ5s for the ST/SCA35 although they stated
that two matched pairs would work. By your statement "I am not obsessive
about tube balance,some people even tolerate single ended amplifiers, the
ultimate in unbalance." you obviously give output stage balance a lower
priority than is traditional. As you're aware most higher end "golden age"
PP designs had provisions for output stage bias balancing (inc those using
cathode bias) as do most modern PP designs (notable exception is McIntosh).
Also, although a single ended output stage may be the ultimate in imbalance,
the circuit and/or magnetics are designed to handle the imbalance: most PP
output stages assume a fairly good balance for best performance.

The reason I used the term "practical sense" in the first place is that I
more or less did what you suggested, I collected about 80 new mfg, NOS and
"good used" 6BQ5s by various mfgrs and characterized bias at the ST35 and
AA151 operating points. The idea here is that a mix like this is now what
is available the the 21century tube user. All tested good or better on a gm
type tube tester. What I found was that bias varied over a very wide range,
even when gm was fairly closely matched. The results explained why one tube
in a "matched" quad" I purchased from a reputable source would go red plate
in my ST35. Obviously the quad had been matched for gm only. The hot tube
would run at about 20 mA more than the others in the quad at operating
conditions. I suppose it's debatable that a sample of 80 is statistically
significant. OTOH, the quantity of tubes produced today is miniscule
compare to the good ol days so maybe 80 is significant esp if considering
the population of 6BQ5s available for my use.

Summarizing, I fully understand what you were saying in you earlier posts
about the theoretical advantages and disadvantages of the various cathode
biasing schemes. No disrespect was intended when I stated that your theories
were unverified. However, based on my experience, the real world variation
in tubes combined with my understanding that PP output stage balance is
desirable means that the theoretical advantages of four into one cathode
biasing will probably not be realized: Power output capability is not
increased and the requirement for bias matching remains. Therefore, I still
contend that Heath's and Dyna's use of the single cathode resistor was
driven primarily by cost considerations i.e save the cost of a cap and
resistor and indirectly burden the end user with the additional costs of the
matched quad nec. for optimum operation. Any functional advantages were
incidental and secondary.

As an aside, one poster commented that it was doubtful that a company like
Heath or Dyna would attempt to save a few cents on the addl parts nec for
individual bias resistor if that was a better way of doing things. My day
job is with one of the few remaining manufacturers of "physical objects" in
the US. I can assure you that a few cents per unit over a production run of
thousands of units will get attention.

As always, if my logic is warped I welcome enlightenment as to the error of
my ways.

--
Steve


John Byrns wrote in message ...

Hi All,

Since I saw my name mentioned I thought I would throw my two cents into
this thread and also comment on the other AA-151 thread.

snip