• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Push-pull 807 amplifier without global NFB

Ok I am no expert in making transformers. Are the primary and secondary separate or interdigitated. If they are separate you will get a lot of phase lag which will make any form of feedback very difficult. So the LT spice model is not accurate I see with K=1 - its more of an ideal transformer. For LT spice to work you would have to measure the primary cap and leakage inductance.
 
baudouin0,

I believe by interdigitated primary and secondary windings, you mean Bifilar, Trifilar, Quadrifiler, etc.
Those all work best with 1:1, 1:1:1, and 1:1:1:1 ratios, etc. Trifilar is as close to perfect coupling as there is.
Trifilar and Quadrifiler do not couple as well, but are still pretty good.

Most other windings have higher leakage reactance from primary to secondary; versus 1:1, etc. ratios.

And output transformers are usually high primary turns to secondary turns ratio.
Example, 6,600 Ohm plate to plate, and 8 Ohm secondary. Total primary turns to secondary turns ratio = 28.7 : 1
Lets change it to 28.0 : 1
For this example, only 1 turn of secondary can be interleaved with 2 of 28 turns of primary. Not very close coupled with the other 26 turns.

Just my opinions
 
Last edited:
Interleaved windings where you sandwich secondary's between primaries. So you would have a layer of primary, insulation, layer of secondary, insulation and the process would repeat a few times. This gives the lowest leakage inductance and good primary secondary insulation.
 
Baudouin0,

Now I see what you meant. You are correct.

Sorry, that was just my bad American language interpretation of the term. Most Americans would not make that mistake.
Interleaved Layers (well defined for me); not interleaved windings (less well defined for me).

And the only transformers I ever wound were on a round core (an Amidon core, if I remember correctly from many decades ago).
They used Trifilar twisted wires, to convert balanced DSL signals to un-balanced spectrum analyzer inputs; and to convert from un-balanced signal generators to balanced DSL circuits.
 
Last edited:
Yep that is an interesting read. I guess CFB is just another form of negative feedback, and stability needs to be considered. Maybe the turns ratio is too small and/or the leakage inductance is to big with separate windings. 10-17KHz is quite low for an additional 180deg phase shift. It does require more voltage drive to the the grids too as the cathode tends to track the grid. Sowter talk about 14 separate winding sections for very wide bandwidth and two chamber bobbin for full geometric balance. They may also contain a screen between the primary and secondary to reduce the capacitance between the windings. The only transformers I've wound were for SMPS.
 
Last edited:
Patrick Turner also has a interesting write up on the EAR 509 - such a shame there is not anyone hosting this site in an official capacity anymore. (I took a snapshot a couple fo years a go).

The EAR509 is using cathode feedback, and has some caps from one output tube anodes to other output tube cathode ...
The two windings do not need to be bifilar wound, because the leakage inductance
between the two windings can be eliminated by use of shunt C across ends of anode and cathode windings with the
same phase and amplitude of Vac. EAR used 22uF elcaps, but I found 2uF plastic caps were entirely adequate.

I have a Michaelson&Austin TVA-10, also designed by TP with a similar output topography. It rocks!
 

Attachments

  • Screenshot 2022-03-18 081419.png
    Screenshot 2022-03-18 081419.png
    51.5 KB · Views: 187
I built a proto with cathode feedback and had a peak of oscillation at 17kHz, so your Problem sounds familiar. I ended up changing the amp, but I have the suspicion of the problem. I use a dual chamber OPT, and the way it is wired I was not using the primary and secondary from the same chamber on a tube. https://www.diyaudio.com/community/threads/oscillation-with-cathode-feedback.381803/
My transformer also has two chambers. In each chamber the winding sections are P1-S1-P2-S2-P3. All windings in a section belong to the same tube. P1-P2-P3 are connected in series, S1-S2 are connected parallel. Would it be better to interleve P1-S1'-P2-S2'-P3 where S1' and S2' are from the other chamber? Or some other mix?
 
Let me share my current project, a push-pull amplifier with 807 output tubes. Since this is a rebuilt of my old construction, I wanted to reuse some parts, especially the chassis, power transformer and output transformer, at least the iron. I set some design rules for myself:

B+ should stay below 400V​
There will be no global negative feedback​
Local feedback in each stage​
The output tubes will be driven from cathode followers​
Perfectly symmetrical driving of the output tubes​
Possibility to connect in ultralinear, triode or pentode operation​
AB1 class of operation​
Around 10W or more output power​
Below 1% THD at full output if possible​
I know the 807 shines at 600...800V anode voltage, but I wanted to build as simple as possible. Also I don't feel comfortable with that voltage level.
The simulation is promising, see attached circuit diagram. I will add some measuring data later.
Question to ask, why would you not use the S8 tap. With S4 grounded, it would then be symmetrical, correct?
 
I don't know. You have thought about this. I think I would be sweeping the transformer with the correct primary and secondary load measuring gain and phase. First thing it to analyze the problem.
I did it without CFB. The gain from g1 to secondary is -12dB. There is a resonance at 89.5kHz and huge phase shift up to 360°. Is there any way to suppress this resonance?