Ad Widget

Collapse

Announcement

Collapse
No announcement yet.

Better Low Z driver stage design based on our Bootstrapped Gain Stage thread

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

  • #46
    What CF output impedance do you measure without bootstrapping?
    - Own Opinions Only -

    Comment


    • #47
      Originally posted by SoulFetish View Post

      I made that point in my first thread. I know that this is the result you and I expected. We just spent the last page talking about testing this to confirm it.
      But there was some discrepancy about this.
      For instance, in post #3, nick projected that the impedance taken off the cathode would be 370Ω. It was not. In fact it was 3X that amount. I thought it would be helpful to test this, and confirm the result and clear it up.
      I could have been wrong in my understanding about our previous discussion.That's all
      I suspect you have made an error in your experiment. I didn't have a 12AU7 to hand so I tested that with a 12AX7 and

      R1= 1.2k
      R2= 100K
      R3 = 470K
      R4= 0
      Zsrc = ~0
      Vsupply = 280V

      I measured 550 ohms, expected 547 ohms.

      Used 1% measurement resistor, taking scope accuracy as 1% so I'd say 550 +/- 3% i.e. 533.5 to 566.5 ohms

      Technique: Fed 100nF blocking cap 1KHz from signal generator. Blocking cap goes to 1k resistor, adjusted signal to 1V on SG side of resistor (V1) , measured signal at cathode (V2).

      Ro = Rk /(V1/V2-1)

      V1=1
      V2= 0.355
      Rk= 1000
      => Ro=550 ohms

      PS: A very simple sanity test is to calculate 1/gm. For a 12AU7 gm=3 ma/V => Ro = 333 ohms
      Last edited by nickb; 10-11-2020, 07:19 PM.
      Experience is something you get, just after you really needed it.

      Comment


      • #48
        Originally posted by nickb View Post
        A very simple sanity test is to calculate 1/gm.
        True, but it should be noted that gm is not a constant. Rather it depends on plate current and plate (to cathode) voltage. The appropriate value has to be taken from the tube charts.
        For the ECC82 and the circuit in question (7mA/135V) I get roughly gm = 2.2 mA/V corresponding to Ro = 455 Ohm.

        OP used a rather high load current of around 5mArms or 14mA peak-to-peak, which might violate the linearity condition for the measurement.
        Last edited by Helmholtz; 10-12-2020, 12:48 AM.
        - Own Opinions Only -

        Comment


        • #49
          Originally posted by Helmholtz View Post

          True, but it should be noted that gm is not a constant. Rather it depends on plate current and plate (to cathode) voltage. The appropriate value has to be taken from the tube charts.
          For the ECC82 and the circuit in question (7mA/125V) I get roughly gm = 2.3 mA/V corresponding to Ro = 435 Ohm.

          OP used a rather high load current of around 5mArms or 14mA peak-to-peak, which might violate the linearity condition for the measurement.
          Of course its not constant. The point is that under the conditions of this test it's going to be roughly 3mA/V. There is clearly a problem with the experiment. Maybe the 12AU7 is a bad with a gm of 1ma/V or so. One shortcoming I can see if the amplitude of the signal should be small else the non-linearity will become significant. For certain a 10Vrms signal is too large for accuracy but still won't cause the kind of gross error seen.
          Experience is something you get, just after you really needed it.

          Comment


          • #50
            Originally posted by nickb View Post

            Of course its not constant. The point is that under the conditions of this test it's going to be roughly 3mA/V. There is clearly a problem with the experiment. Maybe the 12AU7 is a bad with a gm of 1ma/V or so. One shortcoming I can see if the amplitude of the signal should be small else the non-linearity will become significant. For certain a 10Vrms signal is too large for accuracy but still won't cause the kind of gross error seen.
            First, I ran this experiment several times, with 2 different 5963 tubes I have from a small batch of NOS. Second, I was simulating the operating conditions of the amp by feeding the input to the PI and the plate was driving the cathode follower.
            Perhaps 10V is too large. fine. I was using a 1k/1% resistor measuring 1001.2Ω, and used a Vishay MKP 10µF/500V capacitor in series for the first measurement with a reactance of 16Ω @ 1kHz. For the second test I used a 50µF/200V Electrocube film cap in series with the 1k resistor.

            The power supply for the PI provided +269V with a decoupling filter of 20µF. The power supply nodes for the Driver were +/-127V. Load resistor value of 18k, grid leak of 470k, and grid stopper value of 2.2k.

            If there was clearly an error in my experiment. Where did I make it??

            Click image for larger version

Name:	IMG_6104.jpeg
Views:	228
Size:	3.15 MB
ID:	915579
            If I have a 50% chance of guessing the right answer, I guess wrong 80% of the time.

            Comment


            • #51
              Originally posted by nickb View Post
              The point is that under the conditions of this test it's going to be roughly 3mA/V.
              As stated above I get a somewhat lower gm from the tube charts.


              One shortcoming I can see if the amplitude of the signal should be small else the non-linearity will become significant.
              Just as I said. Load current is too high.

              For certain a 10Vrms signal is too large for accuracy but still won't cause the kind of gross error seen.
              Agree.
              - Own Opinions Only -

              Comment


              • #52
                Wait, first, I want to apologize. I did make one mistake. I was beginning to take this personally for a second, and that will just poison scientific inquiry.
                Nick, you were doing us a solid by checking the work. I appreciate it.
                So, here are a couple of things which I could possibly see which may have been errors.
                First, the possibility that I randomly pulled 2 weak tubes. never tried them before? The second is that I grounded the 1k load to the bottom of the tail resistor, where I connected the grounding terminal of the signal generator?
                If I have a 50% chance of guessing the right answer, I guess wrong 80% of the time.

                Comment


                • #53


                  As the one operating the test and and maintaining the equipment the ball is rather in your court.

                  Checking the condition of the tubes is essential. If you don't have a tester set up a simple DC test by applying a fixed voltage cathode to plate say 250V and a variable negative voltage to the grid let's pick -8 and -9V. Measure the plate current change and that will be the gm. According the the datasheet it should be around 2.2mA/V under these conditions.

                  You can try the approach I used where you inject a small signal into the cathode via a 1K (and cap in series). The nice thing about this approach is the circuit is never changed between measurements so eliminates an uncertainty.

                  Last edited by nickb; 10-12-2020, 05:38 PM.
                  Experience is something you get, just after you really needed it.

                  Comment


                  • #54
                    Originally posted by SoulFetish View Post
                    The second is that I grounded the 1k load to the bottom of the tail resistor, where I connected the grounding terminal of the signal generator?
                    The positive and negative supplies should all be the same from an AC point of view. I'd think twice before connecting the ground on of the signal genny to the end of the tail resistor since that the negative supply ( if I follow you) unless I was certain the SG or supply was floating. Even then I don't see the merit. Connect cap to cathode 1k resistor to ground (0V), sg ground to 0v.

                    One thing just occurred to me. When the 1k is removed and you measure the signal there will be significant DC leakage current through the 10uF that, depending on your measurement device, might be throwing off the open circuit measurement. This is a great example of the advantage of the approach I used as the circuit is the same for both measurements.

                    Experience is something you get, just after you really needed it.

                    Comment


                    • #55
                      Originally posted by nickb View Post

                      The positive and negative supplies should all be the same from an AC point of view. I'd think twice before connecting the ground on of the signal genny to the end of the tail resistor since that the negative supply ( if I follow you) unless I was certain the SG or supply was floating. Even then I don't see the merit. Connect cap to cathode 1k resistor to ground (0V), sg ground to 0v.

                      One thing just occurred to me. When the 1k is removed and you measure the signal there will be significant DC leakage current through the 10uF that, depending on your measurement device, might be throwing off the open circuit measurement. This is a great example of the advantage of the approach I used as the circuit is the same for both measurements.
                      This is the schematic for the exact operational setup. Interestingly, when I tried to measure the loaded voltage 10µF cap (right at the cathode) there was a significant charging pulse, so moving the TP to the other side fixed it.

                      Click image for larger version

Name:	Cathode Bias Driver Zout test.jpg
Views:	173
Size:	141.8 KB
ID:	916376


                      edit: I should have posted that previously to clear up any confusion. plus I uploaded a new one with component reference numbers, because it's annoying not to.
                      Last edited by SoulFetish; 10-23-2020, 04:51 AM.
                      If I have a 50% chance of guessing the right answer, I guess wrong 80% of the time.

                      Comment


                      • #56
                        also, just to confirm; this is your test setup, correct?

                        Click image for larger version

Name:	nickb Cathode Bias Driver Zout test.jpg
Views:	230
Size:	61.7 KB
ID:	915647
                        If I have a 50% chance of guessing the right answer, I guess wrong 80% of the time.

                        Comment


                        • #57
                          V1 must be measured between C1 and R5.
                          - Own Opinions Only -

                          Comment


                          • #58
                            Not that, but this:

                            Click image for larger version

Name:	fetch_2.jpg
Views:	113
Size:	25.8 KB
ID:	915654

                            Have you done the gm test suggested yet?
                            Experience is something you get, just after you really needed it.

                            Comment


                            • #59
                              Originally posted by nickb View Post

                              One thing just occurred to me. When the 1k is removed and you measure the signal there will be significant DC leakage current through the 10uF that, depending on your measurement device, might be throwing off the open circuit measurement. This is a great example of the advantage of the approach I used as the circuit is the same for both measurements.
                              With the method I suggested in post #22 the 10µ cap is not used for open circuit measurement. Of course I'm assuming a decent AC-coupled meter. BTW, the leakage resistance of a good 10µ ecap should be in the 10M range (and he also tried a 50µ film cap).
                              If the meter is not AC coupled, an additional 0.1µ cap in series with the meter should be used.

                              I like the method you proposed. It uses essentially the same principle as an AC-R meter: Injecting an AC current and measuring the voltage drop.

                              If applied properly both methods should give the same results with same circuit and within linear (small signal) range.
                              Last edited by Helmholtz; 10-12-2020, 09:40 PM.
                              - Own Opinions Only -

                              Comment


                              • #60
                                Originally posted by nickb View Post
                                Not that, but this:

                                Click image for larger version

Name:	fetch_2.jpg
Views:	113
Size:	25.8 KB
ID:	915654

                                Have you done the gm test suggested yet?
                                not yet. Im going to redo the test and measure the gm. Im going to do the measurements on my scope for a visual of the P-P voltages and RMS readout. Ill also adjust the signal source to read 1V RMS at the unloaded output.

                                For me the pain in the ass of setting this up is tapping into the heater supply. id like to construct a nice regulated supply at some point, but what about some kind of battery setup? Would a lantern battery be usable in this situation? What about a LiFePo battery ?
                                If I have a 50% chance of guessing the right answer, I guess wrong 80% of the time.

                                Comment

                                Working...
                                X