|
Post by aquin43 on Aug 9, 2019 4:20:31 GMT -5
It is well known that eddy current losses in the coil cause the effective inductance to fall with frequency and the effective series resistance to rise. I thought that a graph showing the degree of variation in a typical humbucker might be of interest. This is an old Dimarzio PAF style with a cover. The losses are inductively coupled to the coil, producing the change in parameters. In practice, the two coils behave differently because of the different core materials. This is the overall effect for the two coils together.
I have plotted the curves out as far as 30kHz to better illustrate their shape. In this case, the cut off frequency, where the the change in both L and R is half of the maximum is 4k7Hz. This also causes a 1.8dB step in the frequency response.
Arthur
|
|
|
Post by ms on Aug 9, 2019 6:33:20 GMT -5
This shows that there would be a significant error when computing the pickup capacitance from the low frequency inductance and the resonance frequency.
|
|
|
Post by perfboardpatcher on Aug 9, 2019 14:24:30 GMT -5
Is it possible to measure (the effect of) eddy current losses with a hall-effect sensor? Wouldn't the flux density at higher frequencies drop due to eddy current losses? (I want to use a hall-effect sensor instead of the pickup coil.)
|
|
|
Post by aquin43 on Aug 9, 2019 16:35:19 GMT -5
Is it possible to measure (the effect of) eddy current losses with a hall-effect sensor? Wouldn't the flux density at higher frequencies drop due to eddy current losses? (I want to use a hall-effect sensor instead of the pickup coil.) No. The coupling of the coil to the field of the magnets is so weak and the coil currents so small that such a measurement would be totally impracticable because of the resolution required.
This measurement is made on the assumption that there is one major source of eddy current loss that is reflected in the resonant impedance of the pickup. The impedance curve is measured with different capacitive loads and the parameters are deduced from fitting L C R and K to the results.
Arthur
|
|
|
Post by ms on Aug 10, 2019 7:31:47 GMT -5
Is it possible to measure (the effect of) eddy current losses with a hall-effect sensor? Wouldn't the flux density at higher frequencies drop due to eddy current losses? (I want to use a hall-effect sensor instead of the pickup coil.) The impedance curve is measured with different capacitive loads and the parameters are deduced from fitting L C R and K to the results.
Arthur
Why? You have a means of measuring the complex impedance as a function of frequency directly. The effective L as a function of f, etc. follow directly from that.
|
|
|
Post by stratotarts on Aug 10, 2019 7:58:21 GMT -5
The impedance curve is measured with different capacitive loads and the parameters are deduced from fitting L C R and K to the results.
Arthur
Why? You have a means of measuring the complex impedance as a function of frequency directly. The effective L as a function of f, etc. follow directly from that. The device is a black box. How can you know what portion of the complex impedance is due to capacitive reactance and what portion due to inductive reactance, at any given frequency? Is there not more than one pair of LC value that will produce the same phase angle, same complex impedance?
|
|
|
Post by ms on Aug 10, 2019 12:34:22 GMT -5
Why? You have a means of measuring the complex impedance as a function of frequency directly. The effective L as a function of f, etc. follow directly from that. The device is a black box. How can you know what portion of the complex impedance is due to capacitive reactance and what portion due to inductive reactance, at any given frequency? Is there not more than one pair of LC value that will produce the same phase angle, same complex impedance? The complex impedance as a function of frequency (Z(f)) is all the information that there is. The measurements with multiple Cs are a low tech way of getting a part of that information in a form that is convenient to use with a simple fit. In general, if you have Z(f), you could assume a C independent of frequency and that Zpc(f) (the impedance of the coil without the C) is a smoothly varying complex function. Then use an inverse technique with regularization to get a very accurate value for C and a smoothed Zpc(f) with most of the measurement noise removed.
|
|
|
Post by aquin43 on Aug 10, 2019 16:49:06 GMT -5
The device is a black box. How can you know what portion of the complex impedance is due to capacitive reactance and what portion due to inductive reactance, at any given frequency? Is there not more than one pair of LC value that will produce the same phase angle, same complex impedance? The complex impedance as a function of frequency (Z(f)) is all the information that there is. The measurements with multiple Cs are a low tech way of getting a part of that information in a form that is convenient to use with a simple fit. In general, if you have Z(f), you could assume a C independent of frequency and that Zpc(f) (the impedance of the coil without the C) is a smoothly varying complex function. Then use an inverse technique with regularization to get a very accurate value for C and a smoothed Zpc(f) with most of the measurement noise removed. Yes, it is just a matter of what is practical. For use with Spice, it is necessary to approximate the inductor impedance using a more or less complex network of simple L and R. Having at least one extra data point well below the unloaded resonance makes this task much easier and can yield a network that gives a good approximation of the inductance variation, the frequency response and the step response.
Arthur
|
|
|
Post by JohnH on Aug 10, 2019 17:58:03 GMT -5
The complex impedance as a function of frequency (Z(f)) is all the information that there is. The measurements with multiple Cs are a low tech way of getting a part of that information in a form that is convenient to use with a simple fit. In general, if you have Z(f), you could assume a C independent of frequency and that Zpc(f) (the impedance of the coil without the C) is a smoothly varying complex function. Then use an inverse technique with regularization to get a very accurate value for C and a smoothed Zpc(f) with most of the measurement noise removed. Yes, it is just a matter of what is practical. For use with Spice, it is necessary to approximate the inductor impedance using a more or less complex network of simple L and R. Having at least one extra data point well below the unloaded resonance makes this task much easier and can yield a network that gives a good approximation of the inductance variation, the frequency response and the step response.
Arthur To see one attempt at that, check out GuitarFreak and the values used in 6-part pickup models. Each of them has has two L's and 3 R's, plus a cap, to get the best correlation that could be found on that basis to match across the range. They relate to Antiguas tests from 2017, so you can compare the values derived with basic L and R and hence C derivation.
|
|
|
Post by perfboardpatcher on Aug 11, 2019 13:12:04 GMT -5
Is it possible to measure (the effect of) eddy current losses with a hall-effect sensor? Wouldn't the flux density at higher frequencies drop due to eddy current losses? (I want to use a hall-effect sensor instead of the pickup coil.) No. The coupling of the coil to the field of the magnets is so weak and the coil currents so small that such a measurement would be totally impracticable because of the resolution required. This measurement is made on the assumption that there is one major source of eddy current loss that is reflected in the resonant impedance of the pickup. The impedance curve is measured with different capacitive loads and the parameters are deduced from fitting L C R and K to the results.
Arthur That's not so good news, then! But I think I'm gonna experiment with a hall-effect sensor I've just purchased. To be honest I don't understand this topic. Might well be my lack of understanding of matter, so don't take this too seriously. You're speaking of "eddy current losses in the coil". You mean core? I consider the pickup coil to be a sensor, not an inductor. Like those cheap round piezos. I had a batch of them and they all measured around 30nF. I think they would make good 30nF caps. I don't know but perhaps a case could be made that irregularities in the frequency response of the piezo as sensor could be framed as cases of effective capacitance.
|
|
|
Post by aquin43 on Aug 11, 2019 15:58:14 GMT -5
The losses are in the cores, magnets and metalwork but are magnetically coupled to the coil so that they appear as part of its impedance and produce a result which behaves as a reduction of inductance and increase of resistance of the coil with frequency. The details depend on the geometry so a small hall effect sensor will not sense what the large coil does. In addition, the hall sensor output will be dominated by the fixed magnetic field which does not register at all with the coil. Assuming that with suitable amplification and filtering the hall output could be used to map the variation in the magnetic signal throughout the space normally occupied by the coil, relating that to the coil impedance would be a major task.
Arthur
|
|