Self designed tube amp problem

[Helmholtz]

.The maximum output undistorted voltage was 4 or 4.5 volts rms on 3.9 ohms ( a 15 watt resistor) and this is a little more than 4 watts rms I don't think this is normal.
The transformer was hand wired and when I measured it was 9.6kOhms anode to anode to 4 ohms at secondary.

What is the EL84 cathode idle current? Should be around 9mA.
What is B+/HT at full output?

The transformer was hand wired and when I measured it was 9.6kOhms anode to anode to 4 ohms at secondary.

How did you verify/measure the 9.6k primary impedance? What is the OT primary DCR?

[The Dude]

Reposting this tweaked for easier viewing.

[eschertron]

Pentode mode or triode mode when the 5W was measured?

[Manowar1985]

The HT is about 293 volts but I did not measure it with full output. I will do that.
I have variable Grid bias. Ranging from 0 volt down to-14 volts.I had set the idle current to 20 mA .When the bias is colder the nonlinearty is worse and when it is hotter it is improved a little bit.I also don't see that grid current limiting distortion when I experimentally set the bias to 0 volts.
The nonlinearty happens to the same output level for both modes.
Also I apply input in the two inputs of the output stage. Like a differential input

The output transformer is measured as follows

I apply the mains voltage to the primary (anode to anode) and I measured the voltage at the secondary. I then calculated the ratio of the 2 windings and calculated the the ratio squared. I finally multiplied it by 4 ohms to determine the a to a impedance

[Manowar1985]

I measured again the HT and saw the following things

At maximum negative bias (that is -14.5 volts) I get about 7.8 mA of anode current from each valve.
I put some music on really loud and the HT did not seem to change much probably half a volt. It was about 300 volts at -14.5V bias.Only when I increase the bias up to 0 volts thd HT goes down to 280 volts but it doesn't seem to change when thd music is really loud.

I also did the same thing the the 3.9 ohms resistor with a sine wave signal and I saw that I had an HT of 296 at zero signal, 291 at full clean signal and 289 at the point that it begins to distort.

[Helmholtz]

Also I apply input in the two inputs of the output stage. Like a differential input

Here might be your problem.

A PP amp is no differential amplifier. Each side needs its own signal between grid and ground.
So if you apply a floating differential signal between the 2 inputs, ideally no output should result.

I assume that one side of your signal had ground reference, so you were actually driving only one of your tubes.

I apply the mains voltage to the primary (anode to anode) and I measured the voltage at the secondary. I then calculated the ratio of the 2 windings and calculated the the ratio squared. I finally multiplied it by 4 ohms to determine the a to a impedance

Good.

[pdf64]

...Only when I increase the bias up to 0 volts ...

What makes you think that's a reasonable thing to do to power valves? They will probably now be damaged.

[Manowar1985]

I tried applying input voltage between the input of the 1st valve and ground and the 2nd valve and ground. The output for each case was about 2.7 volts rms clean.

When I put the bias - 14.5 volts the were working again just like before and same as different pair of new el84 s .So the pair is not damaged.Any suggestions.?
Should I rewind the opt?

[Helmholtz]

I tried applying input voltage between the input of the 1st valve and ground and the 2nd valve and ground. The output for each case was about 2.7 volts rms clean.

Not sure what you mean with first valve.

You definitely need to apply the signal before the phase splitter. A pp amp is not complete without a PI.

[Manowar1985]

I have 2 el84s for the output stage. The first and the second or the upper and the lower valve.

I need to make it as isolated as possible because I will have to check at the same time if the phase Splitter distorts too
I see them as separated common cathode amplifiers with "almost" same output and separated inputs like they are in parallel except for the inputs that are separated so I test each one as a normal transformer coupled common cathode amplifier.

Let's say that the output stage works as expected.I then add the phase Splitter and if that works as expected I finally add the μ follower.

Anyway the phase Splitter that I plan to use in this design is a long tailed pair with a constant current source as the tail resistor

[Helmholtz]

I have 2 el84s for the output stage. The first and the second or the upper and the lower valve.

I need to make it as isolated as possible because I will have to check at the same time if the phase Splitter distorts too
I see them as separated common cathode amplifiers with "almost" same output and separated inputs like they are in parallel except for the inputs that are separated so I test each one as a normal transformer coupled common cathode amplifier.

Let's say that the output stage works as expected.I then add the phase Splitter and if that works as expected I finally add the μ follower.

Anyway the phase Splitter that I plan to use in this design is a long tailed pair with a constant current source as the tail resistor

As said before, you need to feed both power tubes simultaneously with equal but opposite phase signals - that's why you must use a phase splitter.
You cannot test the PP stage without it. The PI is an integral part of the PP power stage.

If you drive only one power tube tube, you get a poorly biased single ended amp, putting out very low power before clipping.

[Manowar1985]

Ok.Is this correct?

[eschertron]

Ok.Is this correct?

I see in the schematic that you've added 10 Ohm resistors to the cathode connection. When did you do this? Earlier drawings did not show them. So, I wonder if these are not 10 Ohms, but closer to 1k? 10R will drop less than 1/10v if at max negative bias you have 7.8ma current. Insignificant compared to the negative 14-point-something bias voltage you measured. And -14v may be a little cold, but the EL84s should still produce a large output. Added to the statement that you can set the bias to zero volts and not have the tubes run away with over-dissipation supports the notion that there's a much larger cathode resistance than what's drawn. Just curious, and throwing that out for consideration.

[Manowar1985]

I also forgot to tell you that I got the 4 volts rms on the output when I had all the amplifier stages together. These were

μ follower
Phase Splitter (the one above)
and output stage

Then after I saw that I could not get more than 4 or 4.5 volts at the output I decided to check each stage separately starting from the output backwards to the input

[Manowar1985]

Eschertron forget about the previous drawings this is a different one.I added the 10 ohm resistor to keep both currents of the output tubes equal without changing the operating conditions significantly ghats why they are only 10 ohms.Do you think that they cause distortion problem (still talking about cathode resistors)