Rout of the amp with FB applied  =                     Ra-a                         
                                                            ZR x ( 1 + [ A" x {µ/TR} x ß ] ) 

Where Ra-a is the anode to anode resistance of the output tubes,
ZR is the output transformer impedance ratio,
A" is the gain of the stages preceding the output tubes,
µ is the amplification factor of the output tubes,
TR is the primary to secondary turn ratio of the OPT = square root of OPT ZR,
ß is the fraction of OPT secondary voltage fed back to be "in series" with the input voltage to V1.

At the working point at Q Ra = 19k, so Ra-a = 38,000 ohms,
OPT impedance ratio = 8k : 6 ohms = 1,333 : 1,
A" = 64,
µ = 190,
TR  turn ratio = sq.rt 1,333 = 36.5,
ß = 0.68

The graph shows the power output for class AB1 6550 beam tetrodes in the above load line analysis example.
This is a fair working point for the 6550 tube, because it offers the possible combination
of over 25 watts of class A and a ceiling of 55 watts. People enjoy tube amps mainly because the first few watts are so blameless. For the best musical performance, the first few watts need to be class A, and without the sudden distortion
increase around the transition from class A to AB known as crossover distortions. From the graph you can see that if we
wanted the amp to produce at least 25 watts of class A, then we would have to chose an anode load of about 8 kOhms, and we would have a class AB ceiling of 35 watts.
Since most folks use no more than 4 peak watts for 95% of what they listen to at home, the idle current could even be reduced to 50 mA, and there would still be enough class A to cover all their needs, without reverting to class AB, which would still provide the same ceiling of 35 watts, regardless of the tube bias current.
With the anode dissipation at 20 watts at idle, the tube life will be longer.

From the curves shown above, one can easily attain 50 watts of class A with a quad of output tubes.

Speakers do not have an even impedance value, and often have high impedance at bass frequencies, and lowest impedance between 200 Hz and 1 kHz, and some have rising impedance above 20 kHz, due to the inductive nature of dynamic speakers.
The load matching for the amp needs to be chosen carefully to ensure the amp works within its class A ability for most of the time, even if the speaker load value has dropped from the maker's nominal 8 ohms down to 4 ohms at say 300 Hz.

If the amp has an 8k : 6 ohm output transformer impedance ratio, then the speaker loads from 4 to 8 ohms will be reflected back to the output tubes as 5.3k to 10.7k and we will always have just enough class A power to keep the music sounding sweet, even with a 3 ohm speaker load. When the anode loading on the tubes goes below 4 kOhms, all the distortions increase very rapidly.
For the above working condition to accommodate most modern speakers I would use an 8k : 6 ohm OPT.

All amplifiers have the same general shape for their graph of maximum output power versus load value, including solid state types. Some makers make a large fuss over the maximum power ability into some small value load but neglect to say in their brochures that the same huge power mentioned is not available at higher load values.
It should be obvious from the graphs here that if an amp can produce 57 watts into 3.3 ohms, then only 18 watts are available into 16 ohms.

Every amp has a voltage swing limitation, and power =            voltage squared            
                                                                                    Load Resistance value in ohms

From the PP graphs, the higher the load value, the greater the % of class A operation. This also gives lower output power, but lower distortion and Rout, so load matching should be carefully considered.
Some tube amps have provision for connecting speakers to different load matching terminals, usually 4 or 8 ohms.
When in doubt, always connect speakers to the lowest impedance terminals.

The choice of impedance matching between 4 ohms or 8 ohms is determined by the square of the turn ratio between primary and secondary windings on the OPT, so if the turn ratio was 27.4 : 1 when using an 8 ohm speaker to the 8 ohm output terminal, then the tube load will be 27.4 x 27.4 x 8 = 6 kOhms.
But if there is a 4 ohm terminal, this will be connected to a tap on the secondary winding,
and there will be effectively less turns on the secondary, giving a turn ratio = 38.7 : 1,
and when a 4 ohm load is connected the tube load will be 38.7 x 38.7 x 4 = 6 kOhms.
The tubes will thus act in the same manner despite the difference in load values connected to the secondary.
 

But operating the tubes in ultralinear, you will get Rout of perhaps 6 ohms, and the further reduction of the Rout to 1 ohm is achieved with some global FB, usually about 16dB.

The use of 12.5% to 20% of the primary in the the cathode windings reduces the Rout to as low as 2 ohms
with a further 8 dB of global NFB needed to reduce Rout to less than 1 ohm. 

The screens are capable of being used as control grids, but as such, they do not have as high an amplification factor, µ, so to alter the electron flow in the tube, a high signal voltage needs to be applied to the screen to have much effect.
There are amps designed to use a fixed grid voltage and have a drive voltage applied via the screens so the output tubes then behave as a low µ triode with a very high Ra similar to the beam tetrode value. Drive voltage needed is in the hundred volts region and must be low impedance because the screen draw dc current at al times which alters non linearly with applied
voltage. Screen drive is rarely ever used even though it is sometimes more linear than conventional beam tetrode operation.
In the UL amp, the screens do receive a high drive voltage from the taps on the OPT, and it applied NFB.
When screens are connected to the anodes, the NFB is the same as occurs in a triode.
 
In beam tetrode mode the screen is kept at fixed voltage to prevent the the anode voltage signal change from affecting the electron flow which occurs in triodes where the grid voltage changes AND the anode voltage changes BOTH have a net mutual effect on the electron current flow, due to electrostatic field effect action. The ultralinear method allows the tube to operate
somewhere between tetrode and triode. With about 43% of the anode signal voltage applied to the screen via the OPT taps on the primary, the tetrode anode resistance will be reduced
from 17 kOhms to about 2-3 kOhms, and the distortions reduced substantially, whilst at the same
time, the maximum output power is reduced by only 15% for the same Ea as the beam tetrode case. With any more than 50% tappings, there are no more benefits to be gained if we wish to keep the output power near the beam tetrode levels;
going over 50% results in reduced power capability towards that of  triode connected tubes when 100% of the anode voltage is fed to the screens.

In the 8585 Integrated amp I have used the distributed load connection, or local cathode feedback windings on the OPT so the control of gain, output impedance and distortions is localised in the grid circuit, as well as the screen circuit.
This method is very effective in reducing the effective anode resistance and distortion, yet maintaining the same power of the beam tetrode. The other major benefit is that the screens can be connected to a fixed voltage of just the right value to suit the output tubes best operating performance setting for lowest distortion. 

I used to think there is no absolutely best single way to arrange the tubes in the PP output stage.
But I have not heard anything better than using cathode feedback well implemented.
( Quad II uses 10% of the anode windings for cathode FB to the KT66, so that the effective Ra-a at the OPT sec is reduced to about 7 ohms and allowing for the winding resistance loss of 17% for the 8 ohm load setting, and with the added global NFB, Ro is reduced to about 1 ohm. )
I now make CFB standard on all my PP amps, unless I make a triode PP stage which I sometimes use.

If 6550 beam tetrodes are set up with a fixed screen voltage with CFB or not, then regardless of gain the Miller input capacitance is low and the output tubes are very easily driven at high frequencies, so little drive current is required in the driver amp. Having said that, I find that the best subjective sound quality is achieved with a driver stage with a low output resistance
and much more current and voltage swing ability than is actually needed than calculations would predict, so in fact the
only time I would use a pair of 12AX7 as a differential drive/input stage would be for a guitar amp when we want the sound to
be rather rosy and "flompy".

I  prefer the CFB option, and when using about 20% of the primary for CFB, the drive voltage to an output grid is raised to about 75 Vrms, or about 7 times the pure beam tetrode drive voltage so then the driver amp needs to have more idle current and low distortion ability.
But I think the end result sounds better with the CFB and mild global NFB.

With all the modes of operation including triode, when even a small amount of global NFB is used, stability will depend on the quality of the output transformer, and the design of the driver stage, and the control of phase shift and gain at the extreme ends of the bandwidth of the amp.
Critical damping is needed, and zobel networks to reduce HF gain and phase shift and
supply a resistive load at HF to ensure stability.

The use of triode connected output tubes is another simple way to apply local feedback in an output stage, but it comes with a penalty of reduced power output capability, and higher Miller input capacitance, and a higher drive voltage of about 35 vrms.
However, the use of a pair of 6550 in triode mode can produce 20 to 30 watts of excellent listening. If 25 watts is more than enough power for you, triodes are the best sounding simple option.
If a slightly higher ceiling is wanted, CFB is the next best, but much harder to implement because
OPTs with CFB windings wound properly with regard to winding geometry don't grow on trees.

The problems of using evermore local NFB in the output stage become counterproductive when the drive voltage required goes over 75Vrms to each output grid, because the distortions in the drive amp begin to become serious, so a balanced approach is required if we want to use a minimum of global feedback, and still get low Rout,
and low distortion, and keep the drive amp simple.

My experience with my 300 watt amps with a dozen 6550 tubes which now have 20% CFB applied in the output stage
indicates that I can still get extraordinarily low thd with mild global NFB even though the drive voltages to the output grids is 75Vrms to each grid. I use a pair of EL84 in triode and these are choke loaded and in a differential amp.
The THD is lower than when trying to 6CG7/6SN7 in a differential amp with resistances to supply the dc.
Schematics and notes about the various techniques are shown in my other amplifier pages.  

Return to Index Page