I want to thank John Stewart for sending me a couple of graphs
for power output vs 2H and 3H vs load for EL84, set up as an SE tube.
However, such simple basic curves for an SE pentode EL84
don't explain what happens when we have two tubes in PP and we want
to know what the increase in thd vs power output is with varying RL
values
with the same op conditions using fixed bias.

In another thread titled "Best way to connect multiple speakers"
the subject discussed drifted to load matching to pentode PP amps and
the effect of load on 3H produced by a pair of pentodes in PP.

Lets see if I have it right. JH alleged its ok to a lower load than the
rated load
because 3H will be reduced.

Common sense would indicate the load lower than that recommended
wouldn't want to be more than 1/2 the rated load for the amp
unless indeed it was set up to be able to take it continuously without
having the plate dissipation exceed the ratings.

Now from the graphs JH has sent me for the SE EL84,
which are for class A SE operation, reducing RL does redice 3H, but
increases 2H.
Reducing load below that where maximum efficiency occurs would also
result in
less power.
From what JH sent me using a pair of EL84 in class A would
produce less 3H in pure class A, but to make sure the power is all class
A
the power has to be reduced as RL is reduced.
But PP operation allows class AB operation, and then the situation
is quite different to the SE situation, and I'd be much obliged if JH
could provide the thd measurements of a PP pair of EL84 set up for
class AB, fixed bias, with less idle dissipation than for pure class A,
to permit a lower load to be used and get more than the possible classA
maximum from a pair,
ie, up to 36 watts if Ea and Eg2 is high enough.

A good starting point for what happens with thd in pentodes such as
EL84/6BQ5
is at
http://www.retrovox.com.au/STC6BQ5.pdf
http://tdsl.duncanamps.com/dcigna/tu...ex/6bq5-5g.gif


I cannot find a set of graphs for the 3H for
a range of RLa-a values, and which shows the
3H increase from 0 watts to full clipping po for each RL.

But everytime I have tested EL84 or other beam or pentode tubes the
conclusion was that I didn't want to use pentodes or beam tubes
connected
as such for my output stages, due to higher levels of thd compared to
UL.

If we do look at the pentode plate curves, the variation in gm
is fairly linear above about 40 mA of idle current.
This would support the idea 3H would not increase much
after becoming high after an early transition from class A to AB.

Let us suppose we had 6k a-a for the PP RLa-a, Ea = 300v, Eg2 = 270v,
Iaq = 30 mA.
The bias point is critical for minimizing 3H during the first few watts,
which is where
we listen most at and without measuring it, I would
say thd would increase if 5 watts was gained from 4k a-a compared to
5 watts into 6k or 5 watts into 15k.
Beyond the first 5 watts, the 3H may not increase to drmatically.

6k a-a seems to be about the lowest load for the Ea/Ia chosen,
and about 15 watts is possible, although PSU sag and OPT losses
usually limit all this wonder to 11 watts, and the PSU B+ sag helps the
thd increase.
For music 6k a-a would be ok but the load lines suggest the
tube dissipation would be exceeded if a continual sine wave was used.
8ka-a will allow sine wave continuously.

As RL goes higher, the 3 H will rapidly increase after the first few
watts of class A.
But its the first few watts that matter so the lower thd with class A is
to be welcomed.
To my mind the best loading would be 12k a-a to a rated load of 6 ohms,
which would suit 8 ohms with its dips in Z.
12 watts is still possible, with 6.5 watts of class A before the
transition
to class AB when there will be a sudden increase in 3H up to 12 watts.

If the EL84 amp is matched to 6k to 8 ohms, and some fool connects 4
ohms,
with dips to 3 ohms, then the power at clipping is only 8 watts,
and a tiny amount of class A is made, and the tubes are likely to
overheat badly with a sine wave. THD is bound to be higher,
watt for watt, but one of you may like to carefully
measure an EL84 pentode amp, but make sure it has fixed bias, and a
stiff power supply.
Cathode bias would make the situation all very much worse, since
sine wave testing of cathode bias PP class AB amps with too low a load
value produces horrendous thd levels since the rise in cathode bias
voltage
mis-biases the output stage by many grid volts.

Patrick Turner, not quite pentoaded out yet.