Announcement

Collapse
No announcement yet.

Paraphase inverter question

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

  • #46
    Originally posted by Helmholtz View Post
    O.K., this makes sense as the fundamental frequency is 50Hz.
    The noise pitch is more like 100Hz one, no 50. This induced me in error. I checked first the OT output with a dmm and shows remember 100hz ripple .The power tubes are pretty well balanced.
    Is a very tiny background noise, normal for a guitar aplication can be ignored, ...but should not be there, and definitely was not there with a LTP inverter in same power stage.
    I reworked the chassis few times in different ways till I found it comes from bias supply
    Last edited by catalin gramada; 09-08-2019, 03:19 PM.
    "If it measures good and sounds bad, it is bad. If it measures bad and sounds good, you are measuring the wrong things."

    Comment


    • #47
      The noise pitch is more like 100Hz one, no 50.
      Fundamental frequency of the signal in the scope pictures clearly is 50Hz, but it contains a lot of second harmonic (100Hz). Because of ear sensitivity at low levels and speaker/OT characteristics 100Hz is typically reproduced and perceived louder than 50Hz.

      I checked first the OT output with a dmm and shows remember 100hz ripple.
      How did you determine frequency with the DMM?


      Bias filtering improves with a resistor in series with the diode (of course will require to adjust shunt resistance). Remember a filter cap without enough series resistance is no good LP filter.


      I would have expected more difference between opposite grid signals. If they are identical, I am somewhat at a loss. Leaves power tube gm or OT primary asymmetry - not very likely.


      Do you also hear the hum without NFB?
      Last edited by Helmholtz; 09-08-2019, 03:45 PM.
      - Own Opinions Only -

      Comment


      • #48
        Originally posted by Helmholtz View Post
        Fundamental frequency of the signal in the scope pictures clearly is 50Hz, but it contains a lot of second harmonic (100Hz). Because of ear sensitivity at low levels and speaker/OT characteristics 100Hz is typically reproduced and perceived louder than 50Hz.



        How did you determine frequency with the DMM?


        Bias filtering improves with a resistor in series with the diode (of course will require to adjust shunt resistance). Remember a filter cap without enough series resistance is no good LP filter.


        I would have expected more difference between opposite grid signals. If they are identical, I am somewhat at a loss. Leaves power tube gm or OT primary asymmetry - not very likely.


        Do you also hear the hum without NFB?
        Thanks

        It have a freq meter function.I tested again now.Just like that:

        Click image for larger version

Name:	6..jpg
Views:	1
Size:	91.8 KB
ID:	855513

        It shows the same 50hz fundamental.

        Ok, so have understand improving the filtering of bias supply will be the solution in this circumstances to get rid of noise, please ? I have to add the only other thing I tried to reduce the noise was to raise the amount of nfb. which of course is not something I wish.
        Last edited by catalin gramada; 09-08-2019, 04:02 PM.
        "If it measures good and sounds bad, it is bad. If it measures bad and sounds good, you are measuring the wrong things."

        Comment


        • #49
          And Yes, the hum is much louder without nfb
          "If it measures good and sounds bad, it is bad. If it measures bad and sounds good, you are measuring the wrong things."

          Comment


          • #50
            Originally posted by catalin gramada View Post
            And Yes, the hum is much louder without nfb
            O.K., now I have an idea. Do you now see a significant difference in hum level between opposite grids without NFB? Not sure but maybe the explanation is that the wiring of the second PI tube injects some bias hum to its grid. It will then be amplified by a factor of 50 or so.
            This would require much better bias filtering.
            Last edited by Helmholtz; 09-08-2019, 05:21 PM.
            - Own Opinions Only -

            Comment


            • #51
              This is what I get from power tubes grids. Yellow is first section, Blue the paraphase.

              Click image for larger version

Name:	7..jpg
Views:	1
Size:	342.0 KB
ID:	855516

              At least the measured values are close there is a difference 136-120mV= 16 mV
              Global nfb was not applied
              "If it measures good and sounds bad, it is bad. If it measures bad and sounds good, you are measuring the wrong things."

              Comment


              • #52
                How much ripple do you measure at the speaker output without NFB?
                - Own Opinions Only -

                Comment


                • #53
                  Originally posted by Helmholtz View Post
                  How much ripple do you measure at the speaker output without NFB?
                  It shows 32 mV on scope

                  Click image for larger version

Name:	8..jpg
Views:	1
Size:	426.7 KB
ID:	855523
                  "If it measures good and sounds bad, it is bad. If it measures bad and sounds good, you are measuring the wrong things."

                  Comment


                  • #54
                    Originally posted by Helmholtz View Post
                    How much ripple do you measure at the speaker output without NFB?
                    The real question.
                    And how ripple manifests in speaker to sound.

                    To such demanding "musicians" I have a habit of telling them ... bring the same model of amplifier that sounds different from yours ...

                    EDIT 190908
                    32 mV on scope, is this measured on the speaker?
                    The GA 40 can output max 20W at 8 Ohm.
                    That's 12V~ voltage on the speaker 8 Ohm.
                    Dear colleague, calculate the signal/noise ratio for 12V/0.032V
                    Last edited by vintagekiki; 09-08-2019, 06:04 PM.
                    It's All Over Now

                    Comment


                    • #55
                      The level of ripple in post #42 doesn't seem unreasonable. I'd expect getting on for 1V p-p on the raw bias supply with a diode and 100u capacitor straight off the winding. Try adding a 100u from each bias pot wiper to ground, that should be a lot more effective than adding 100u to the first cap (unless the wipers are right at the top end). Of course it's common mode and should be rejected as H. said.

                      I think a common mode signal applied to the outputs through 220k resistors should produce out of phase signals at the actual outputs. The virtual earth triode has low output impedance compared with 220k so it will overpower a signal applied that way but its input is the output of the other triode and that signal should be inverted and appear at the virtual earth triode's output. Why doesn't that happen? Is there a shed load of phase shift going on because of the capacitor values and low frequency?
                      Last edited by Dave H; 09-08-2019, 06:02 PM.

                      Comment


                      • #56
                        Originally posted by vintagekiki View Post
                        The real question.
                        And how ripple manifests in speaker to sound.

                        To such demanding "musicians" I have a habit of telling them ... bring the same model of amplifier that sounds different from yours ...
                        The noise as it is is not a real issue for a geetar amp, is extremely low, I just can heard it if is completely silent in my room. But as my project is not an amp for sale but for my instructable purpose I wanna push the things as far can be knowing why and how....?
                        "If it measures good and sounds bad, it is bad. If it measures bad and sounds good, you are measuring the wrong things."

                        Comment


                        • #57
                          Originally posted by Dave H View Post
                          The level of ripple in post #42 doesn't seem unreasonable. I'd expect getting on for 1V p-p on the raw bias supply with a diode and 100u capacitor straight off the winding. Try adding a 100u from each bias pot wiper to ground, that should be a lot more effective than adding 100u to the first cap (unless the wipers are right at the top end). Of course it's common mode and should be rejected as H. said.

                          I think a common mode signal applied to the outputs through 220k resistors should produce out of phase signals at the actual outputs. The virtual earth triode has low output impedance compared with 220k so it will overpower a signal applied that way but its input is the output of the other triode and that signal should be inverted and appear at the virtual earth triode's output. Why doesn't that happen? Is there a shed load of phase shift going on because of the capacitor values and low frequency?
                          Thanks Dave. There are a lot of things which complicate my life regards the options in bias circuit. First the voltage is almost in limit to provide a decent ecart. Second ,it is a low current source (see how I did the supply just for the reason to not draw to much current) .Third raising the caps value can be a serious problem for effective charging time, leaving by accident (disrupted supply) the power stage unbiased while caps still charging. Now with all that infos You was kind to provide it is clear what should be done, the next step is how...? Thank You All !

                          Full rectification can be an option...will see.
                          "If it measures good and sounds bad, it is bad. If it measures bad and sounds good, you are measuring the wrong things."

                          Comment


                          • #58
                            Originally posted by catalin gramada View Post
                            This is what I get from power tubes grids. Yellow is first section, Blue the paraphase.

                            [ATTACH=CONFIG]55110[/ATTACH]

                            At least the measured values are close there is a difference 136-120mV= 16 mV
                            Global nfb was not applied
                            There's some phase shift there. Perhaps that's why the ripple isn't rejected more?

                            Comment


                            • #59
                              Originally posted by Dave H View Post
                              There's some phase shift there. Perhaps that's why the ripple isn't rejected more?
                              It was a question at the beginning of the thread. The reason I opened this thread regards how I can configure the floating paraphase adding some caps into circuit. The main reason was to know how I can keep the outputs into right phase over audio frequency range. Maybe shifting poles can help a little bit, but for the moment I.m pretty happy with values I choosed from a sound perspective.
                              "If it measures good and sounds bad, it is bad. If it measures bad and sounds good, you are measuring the wrong things."

                              Comment


                              • #60
                                It shows 32 mV on scope
                                Now this seems to indicate that there should be another reason as power amp gain is typically below 1. Can you please measure grids' and output ripple with both PI coupling caps disconnected.

                                And yes, the phase shift by the 47nF cap is an issue. Does your scope provide differential mode (Channel A - Channel B) operation? If so, what is the differential ripple voltage between the grids without NFB?

                                Regarding bias filtering, please consider that inserting a series resistor before the filter cap in the k Ohm range might be much more effective than doubling capacitance.
                                Last edited by Helmholtz; 09-08-2019, 07:06 PM.
                                - Own Opinions Only -

                                Comment

                                Working...
                                X