Announcement

Collapse
No announcement yet.

Measuring pickups Capacitance?

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

  • #91
    But this kind of artefact would be seen without a cover as well, and so it is easy to determine if it is a serious problem.
    Right, but I am not sure if everbody cares to check. The depression might also hide below the resonance and cause a too low resonance peak (Q) and a high frequency drop-off steeper than -12dB/octave in the integrated response. I would recommend to use a low loss control PU with very high resonance frequency (low/medium impedance type) for this kind of check.
    But it is easier to verify that the exciter coil current does not change over the whole frequency range.
    - Own Opinions Only -

    Comment


    • #92
      Originally posted by Helmholtz View Post
      I may have misinterpreted you, as it is not clear to me what you mean by "driven". Both measurements, impedance as well transfer response require a drive (signal).

      -The impedance measurement (direct current feed) is sensitive to eddy current effects in cores but much less so to eddy current losses in conductive covers.

      -The field drive method for transfer response much better (more realistically) reveals the effect of lossy covers at medium and high frequencies by producing a drecrease (loss) in output voltage. After integration, this typically shows as a depression below resonance.
      But this same kind of response with depression can be produced as an artefact, if the exciter current starts decreasing in this frequency range, caused by the increasing impedance/reactance of the exciter coil over frequency. In this case you need to increase the series resistor value and/or reduce exciter coil inductance.
      Right, this is what is at issue. The Filter'tron plot in post #80 shows that depression in dramatic form, but we know it's caused by eddy currents, because other plots of Fender single coils are very flat before they reach resonances, for example:

      Click image for larger version

Name:	uLK8U8g.png
Views:	1
Size:	11.1 KB
ID:	849419


      Also, the plot in this post http://music-electronics-forum.com/t46007-4/#post493813 has a plot that gets to 16kHz without any dips. Also, that plot shows the effects of dramatic eddy current losses when the secondary coil was closed and continuous.


      Here's something I don't understand: why can't I get the field drive coil to generate a -6dB/oct slope by putting capacitance in parallel with the field drive coil? Shouldn't the voltage drop with frequency? I tried it and it seems to have no such effect.

      Comment


      • #93
        Here's something I don't understand: why can't I get the field drive coil to generate a -6dB/oct slope by putting capacitance in parallel with the field drive coil? Shouldn't the voltage drop with frequency? I tried it and it seems to have no such effect.
        The natural high frequency slope of the field coil drive response (aka bandpass transfer reponse) is -6dB/octave, i.e. capacitive behaviour The integrator transforms this to the low-pass response and adds another -6dB/octave. The result is -12dB/octave.
        What is the point of putting a capacitor in parallel with the exciter coil? What is the bandwidth of the integrator?
        - Own Opinions Only -

        Comment


        • #94
          Originally posted by Helmholtz View Post
          What are all these needle artifacts? I don't see them in my impedance measurements. Also the high and low frequency slopes appear not to be correct. Did you measure as I proposed without the field coil?
          We call them "spurs" in English. They look like external interference to me, probably from clocked digital logic with very short rise and fall times on the edges. Plotting amplitude versus linear (not log) frequency will reveal any harmonic ladders, which may be a clue as to source. Be suspicious of nearby test equipment.


          The zigzag anomality is the result of an additional series and parallel resonance with higher resonant frequencies. I could show in simulations that such behaviour can be the result of partially shorted windings. Another explanation could be a very sloppy wind, where the winding is not carefully layered but outer turns are used to fill lower spaces thereby causing an uneven distribution of the distributed capacitance. I have not found a way yet to prove this idea wrong or right, as I am not winding.
          One way to dig into this mathematically may be found in the System Identification literature, which is large, and mostly directed at things like modelling chemical plants the better to control them. But we can use their methods in miniature, to deduce likely equivalent circuits. Note that eddy currents will likely need to be modeled directly, as no lumped-component circuit does it justice.

          Comment


          • #95
            The current in the field or exciter coil creates an ac magnetic field that induces a voltage, not a current, in series with the pickup coil.
            This is nonsense. Induced voltage always develops across the inductor and not in series. This follows from Maxwell's equations as well as Faraday's law of induction.
            - Own Opinions Only -

            Comment


            • #96
              Here's something I don't understand: why can't I get the field drive coil to generate a -6dB/oct slope by putting capacitance in parallel with the field drive coil? Shouldn't the voltage drop with frequency? I tried it and it seems to have no such effect.
              Well, this is easy, if you understand resonant circuits. The exciter coil and the parallel C form a parallel resonant circuit. In the vicinity of its resonant frequency, coil current and coil voltage rise, while the outer current through the series resistor decreases caused by the increased impedance. The normal voltage decreasing effect of the added C only shows above this resonance.

              The most probable reason for a high frequency drop-off flatter than -6dB/octace before and -12dB/octave after integration is noise floor of the Velleman. S/N ratio is improved by higher drive current. I use a power amplifier connected to the generator output to drive the coil.

              You may also want to check the integrator for errors at low signal levels caused by offset.

              BTW, the 6dB/octave slopes below and above resonance prove that the current through the inductance at low frequencies and the current through the capacitance at high frequencies must be constant (independent of frequency). Those who understand filters will know what I mean.
              Last edited by Helmholtz; 04-29-2018, 06:22 PM.
              - Own Opinions Only -

              Comment


              • #97
                Originally posted by Helmholtz View Post
                This is nonsense. Induced voltage always develops across the inductor and not in series. This follows from Maxwell's equations as well as Faraday's law of induction.
                No. Take the turns one at a time. Integrate the E field around each turn. (The E field has a non-zero curl.) This gives the voltage around each turn. The turns are all in series; so add up the individual voltages to give a single voltage in series with the coil.

                If a voltage source appeared in parallel with the coil, it could drive the coil inductance as well as any external load. This is obviously impossible; it would be able to provide an infinite amount of energy.

                Comment


                • #98
                  No. Take the turns one at a time. Integrate the E field around each turn. (The E field has a non-zero curl.) This gives the voltage around each turn. The turns are all in series
                  Correct. In consequence the sum of the voltages appears at/between the terminals of the real coil.

                  so add up the individual voltages to give a single voltage in series with the coil.
                  The conception of an ideal inductor in series with a voltage source is one valid model. But this voltage source needs to be controlled to give f-proportional voltage, as does the real inductor. (Simulations using a constant voltage source will not give the real PUs frequency response, which shows an output voltage proportional to string velocity (frequency*amplitude) below resonance.)

                  I prefer Zollner's method of producing the induced voltage with a constant current source wired in parallel with the ideal inductor.

                  If done correctly both methods/models give the same simulation results and thus can be considered equivalent.
                  Last edited by Helmholtz; 04-29-2018, 03:17 PM.
                  - Own Opinions Only -

                  Comment


                  • #99
                    Originally posted by Helmholtz View Post
                    The natural high frequency slope of the field coil drive response (aka bandpass transfer reponse) is -6dB/octave, i.e. capacitive behaviour The integrator transforms this to the low-pass response and adds another -6dB/octave. The result is -12dB/octave.
                    What is the point of putting a capacitor in parallel with the exciter coil? What is the bandwidth of the integrator?
                    The point of this is that I'm trying to see if the integrator can be removed from the testing process, by duplicating its effects by passive means, because it there's a way to get a similar plot without specialized equipment, it would make pickup testing more accessible to hobbyists like myself.

                    Just so you know, the integrator is between the pickup and the oscilloscope, as this is how Ken Willmott designed the device. I understand that Helmuth Lemme put the integrator in between the function generator and field coil.

                    I'm having partial success using this circuit on the field coil side. You can see from the simulation that I across L1 "field coil" slopes off past 1kHz, so that's good, but it's still far from being great.

                    Click image for larger version

Name:	JGVKdiz.png
Views:	1
Size:	99.2 KB
ID:	849422

                    Here are actual plots using an SSL-1, using the schematic above on the field coil side. It looks more like an integrated plot as the value of C is increased, but at lower frequencies it's not flat, and the noise increases as it's made flatter with higher values of C.

                    Click image for larger version

Name:	zqxoZzu.png
Views:	1
Size:	12.8 KB
ID:	849423

                    I notice that if R2 is a higher value, the slope drops off at a lower frequency, but doing that causes the field coil's strength to become too weak.
                    Last edited by Antigua; 04-29-2018, 06:56 PM.

                    Comment


                    • Originally posted by Helmholtz View Post
                      Correct. In consequence the sum of the voltages appears at/between the terminals of the real coil.



                      The conception of an ideal inductor in series with a voltage source is one valid model. But this voltage source needs to be controlled to give f-proportional voltage, as does the real inductor. (Simulations using a constant voltage source will not give the real PUs frequency response, which shows an output voltage proportional to string velocity (frequency*amplitude) below resonance.)

                      I prefer Zollner's method of producing the induced voltage with a constant current source wired in parallel with the ideal inductor.

                      If done correctly both methods/models give the same simulation results and thus can be considered equivalent.
                      The sum of the voltages does not appear at the terminals of the real coil, although it can approach it at very low frequencies. The sum of the voltages is modified by the various impedances I mentioned before. In particular at the resonance, the output voltage across the coil is greater than the series voltage at that frequency.

                      Of course you can always use a current source, properly located. But this is just a distraction; neither of us was talking about this equivalent current source.

                      Of course the series voltage increases with frequency; that is what the physics tells us to do.

                      Comment


                      • Originally posted by Antigua View Post
                        The point of this is that I'm trying to see if the integrator can be removed from the testing process, by duplicating its effects by passive means, because it there's a way to do get a similar plot without without specialized equipment, it would make pickup testing more accessible to hobbyists like myself.

                        Just so you know, the integrator is between the pickup and the oscilloscope, as this is how Ken Willmott designed the device. I understand that Helmuth Lemme put the integrator in between the function generator and field coil.

                        I'm having partial success using this circuit on the field coil side. You can see from the simulation that I across L1 "field coil" slopes off past 1kHz, so that's good, but it's still far from being great.

                        [ATTACH=CONFIG]48657[/ATTACH]

                        Here are actual plots using an SSL-1, using the schematic above on the field coil side. It looks more like an integrated plot as the value of C is increased, but at lower frequencies it's not flat, and the noise increases as it's made flatter with higher values of C.

                        [ATTACH=CONFIG]48658[/ATTACH]

                        I notice that if R2 is a higher value, the slope drops off at a lower frequency, but doing that causes the field coil's strength to become too weak.
                        Did you read my post #96? It explains, why the C across the exciter coil is not a good idea. This method can never replace a real integrator.

                        The integrator can be placed between PU and scope as well as between generator and exciter coil. The latter arrangement provides a stronger signal for the integrator, which is generally good. But I am not sure, if the integrator is able to directly drive the low impedance exciter coil load. In any case I strongly recommend to drive the coil via a linear power amplifier. I use one channel of a 60W stereo amplifier.

                        For integration I use a passive integrator after the power amplifier, consisting of a huge air core inductor of around 20mH and a DCR of 2Ohms. This is wired in series with the exciter coil and takes care of the 1/f drive current.
                        - Own Opinions Only -

                        Comment


                        • Originally posted by Helmholtz View Post
                          Did you read my post #96? It explains, why the C across the exciter coil is not a good idea. This method can never replace a real integrator.

                          The integrator can be placed between PU and scope as well as between generator and exciter coil. The latter arrangement provides a stronger signal for the integrator, which is generally good. But I am not sure, if the integrator is able to directly drive the low impedance exciter coil load. In any case I strongly recommend to drive the coil via a linear power amplifier. I use one channel of a 60W stereo amplifier.

                          For integration I use a passive integrator after the power amplifier, consisting of a huge air core inductor of around 20mH and a DCR of 2Ohms. This is wired in series with the exciter coil and takes care of the 1/f drive current.
                          I did read your post, the -6dB slope comes after the resonance between the cap and the field coil, but I used very high C values, 1uF, so the resonant peak was very low, so that in and of itself didn't appear to be a problem. It appears to me, based on LTSpice modelling, that if R2 is a higher value of resistance, the roll off occurs at a lower frequency, but a more powerful amplification would be required to drive the field coil. The goal is to come up a more simple test procedure, and a power amp would add complexity, but at least it's a much (much much) more common device than an integrator circuit.

                          It sounds like your series inductor integration past a power amp is an easy setup to create, I'd be interested in seeing plots made with the integration, and pictures of the coil itself. I do keep a power amp on hand for testing purposes, I could give it a try.

                          Ken Willmott was of the opinion that putting the integrator between the pickup and oscilloscope was less prone to noise, but I don't have a strong opinion one way or the other.

                          Comment


                          • Here is a test showing the result of two different methods of bode plotting the transfer function of a pickup. The first method has the pickup hooked up directly to the function generator and oscilloscope, and treats the pickup as though it were just a low pass filter, a series inductance. The second method treats the pickup like a very poorly coupled transformer, where the inducing coil is a primary connected to the function generator, and the pickup itself is treated as secondary, hookup up to the oscilloscope.

                            This is what the test setup looks like when being directly driven by a function generator, and having the voltage difference measure across a 1meg resistor in series:

                            Click image for larger version

Name:	oDWmH8x.jpg
Views:	1
Size:	948.7 KB
ID:	849448


                            And this is what the external field coil test setup looks like. There is a tiny excitation coil wrapped around a wood stick:

                            Click image for larger version

Name:	djEzIkZ.jpg
Views:	1
Size:	189.5 KB
ID:	849447


                            For this test, I used an Epiphone 57CH PAF style humbucker with a brass cover, so that the "worst case scenario" of a brass cover that causes a high amount of eddy currents.

                            Here are four plots lines comparing difference measurements:

                            Click image for larger version

Name:	Pm2dPog.png
Views:	1
Size:	20.4 KB
ID:	849446

                            All four plots are made with the integrator in front of the first channel of the Velleman PSCU2000 to get -6dB/oct.

                            The two taller peaks (green and red) are with the brass cover removed, while the two lower peaks (blue and black) have the brass cover in place.

                            Of the two taller peaks with the brass cover removed, the taller green peak was made using the external field coil, while the shorter red peak was produced by driving the pickup directly with a function generator. For some reason, these two peaks show fairly different amplitudes, a difference of 3.5dB, with the external coil test method yielding a higher Q.

                            Then, of the two shorter peaks, for which the cover was in place, the higher blue line is the pickup being driven directly with a function generator, where as the lower black line was made using the external coil.

                            There are a few things going on. Interestingly the two testing methods show the greatest difference with the brass cover off, instead of on, which is the opposite of what I would have expected. The Q factors of the "cover off" plots are about 3.5dB apart at the peak. Even though there are differences in need of an explanation, they two test methods are more similar than I thought they would be, possibly suggesting that trying to mimic the geometry of a guitar string with an external field coil is not strictly necessary to create useful data plots, eddy currents and all.

                            Comment


                            • The sum of the voltages does not appear at the terminals of the real coil, although it can approach it at very low frequencies. The sum of the voltages is modified by the various impedances I mentioned before. In particular at the resonance, the output voltage across the coil is greater than the series voltage at that frequency.
                              This statement is correct. Mike Sulzer is/was right and I apologize. I was on the wrong track.

                              The total induced voltage/EMF would show at the inductor's terminals only in an open loop situation without any current. But any real world inductor like a PU is terminated at least by its own distributed capacitance. Just as you said.
                              Last edited by Helmholtz; 04-30-2018, 03:23 PM.
                              - Own Opinions Only -

                              Comment


                              • Is anybody going to wind a bifilar (two wires in parallel) pickup coil and measure its impedance curve in series configuration?
                                - Own Opinions Only -

                                Comment

                                Working...
                                X