Announcement

Collapse
No announcement yet.

Paraphase inverter question

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Originally posted by Dave H View Post
    220k AC load
    Left triode 2.62V, Right triode 2.53V

    110k AC load
    Left triode 2.256V, Right triode 2.173V

    Edit:
    220k Left triode, 110k Right triode
    Left triode 2.62V, Right triode 2.248V

    Schematic

    [ATTACH=CONFIG]55157[/ATTACH]
    I am unsure how to interprete the data and just realized that the 47nF output caps present an impedance of ca. 68K @50Hz. So it makes a significant difference if output voltages are measured at the plates or across the load resistors.
    As we want the impedances as seen by the power tube grids, it would be best (and easiest for calculation) to know the voltages across the 220k/110k load resistors. This would allow to easily calculate currents by V/R and effective impedance Z= dV/dI.
    This way the 47nF caps are treated as integral parts of the PI.

    To find the ouput impedances directly at the plates, I would need the voltages taken at the plates + load currents from the sim, as calculating the load currents with a complex load would get somewhat awkward.
    Last edited by Helmholtz; 09-10-2019, 02:27 PM.
    - Own Opinions Only -

    Comment


    • It is something which intrigues in Rod Elliot floating paraphase description , as: "The second form is shown shown below (version 2), and it works by using the second valve as a unity gain inverter, which improves matters somewhat. Different value plate resistors may still generally needed." And have to ask, why? as time second stage act as unity gain? If he refers to feedback resistor it have sense, otherwise I don't get it. What is the catch ?

      https://sound-au.com/valves/analysis.html#s5
      "If it measures good and sounds bad, it is bad. If it measures bad and sounds good, you are measuring the wrong things."

      Comment


      • Originally posted by catalin gramada View Post
        It is something which intrigues in Rod Elliot floating paraphase description , as: "The second form is shown shown below (version 2), and it works by using the second valve as a unity gain inverter, which improves matters somewhat. Different value plate resistors may still generally needed." And have to ask, why? as time second stage act as unity gain? If he refers to feedback resistor it have sense, otherwise I don't get it. What is the catch ?

        https://sound-au.com/valves/analysis.html#s5
        I think he just did not consider other balancing options (like increasing the gain of the second stage).
        - Own Opinions Only -

        Comment


        • Originally posted by Helmholtz View Post
          I am unsure how to interprete the data and just realized that the 47nF output caps present an impedance of ca. 68K @50Hz. So it makes a significant difference if output voltages are measured at the plates or across the load resistors.
          As we want the impedances as seen by the power tube grids, it would be best (and easiest for calculation) to know the voltages across the 220k/110k load resistors. This would allow to easily calculate currents by V/R and effective impedance Z= dV/dI.
          This way the 47nF caps are treated as integral parts of the PI.

          To find the ouput impedances directly at the plates, I would need the voltages taken at the plates + load currents from the sim, as calculating the load currents with a complex load would get somewhat awkward.
          The voltages I posted were across the 220k/110k resistors but I forgot to increase the cap values (it was late), When I calculated the 5D5 output impedance I used huge caps. Doing the same with Catalin's circuit gives output impedances of 15k for the left triode and 1k for the right.

          Comment


          • This is not fast moving game chess game, so conclude within 10 minutes.

            I hold that I am qualified enough (telecommunications engineer for 40 working life) to follow this thread, so for the pure curiosity I would ask a dear colleagues, on which amplifier are make measurements, with which measuring instruments (signal source, oscilloscope) make measurements, and photographed measurements, perform calculations, all within a half hour.

            All of the above measurements listed in these 140 posts are only valid if performed on the same amplifier and with the same measuring instruments.
            I got myself thinking, what the writer of this thread actually wanted from the thread.
            A bunch of graphs from an oscilloscope, measured results that no one needs, and no conclusion is in sight. The whole thread is more and more like to collective spam.

            It would be good to go back to the first post of the this thread

            https://music-electronics-forum.com/showthread.php?t=49645

            and what J M Fahey said
            ... Ignore what said before ... reset the counter to 0 ... start with a white sheet of paper ...
            It's All Over Now

            Comment


            • The voltages I posted were across the 220k/110k resistors but I forgot to increase the cap values (it was late)
              That's fine as it gave me a chance to calculate the actual impedances that load the grids (i.e. with 47nF coupling caps and 220k resistors at 50Hz). I get 42k for the left and 32k for the right tube. Caveat: This only holds for the 50Hz fundamental. For the higher harmonics the impedance of the 47nF is lower and impedance asymmetry and differential grid ripple increase.

              I also theoretically calculated the impedances directly at the plates (with 220k resistors and "infinite" coupling caps), assuming plate currents of 1.5mA and rp=35k from the 12AT7 datasheet. This gives me 21k for the left and 1.5k for the right tube. The calculated loaded tube gains are 27.3/27.5 .


              BTW, I think that the split-load PI should have the same issue. Its output impedances can differ even more, e.g. Zplate: 56k, Zcathode: 1.2k.
              - Own Opinions Only -

              Comment

              Working...
              X