Measurement of Coil Self Capacitance - Terman's method

[Joe Gwinn]

If we had been down this particular road before, why did you not remember how I am measuring the C? I had to tell you twice before you now, apparently, recognize that I am not just computing the C from the resonant frequency as you thought. No, as far as I remember, we have not discussed the details of how I do this measurement.

The road is the kerfuffle about the constancy or non-constancy of inductance in cored coils. There are a number of long and noisy threads on that. We must have made at least 100 postings each in those threads.

I am not sure why you have written this here, but I can give you a proof that measurement of the complex impedance contains the information used in the Terman method. Consider the the complex impedance at a particular frequency. From the negative of the imaginary part one can derive the value of the C that would give the zero phase resonance. One can do this over a range of frequencies and construct the plot made in the Terman method. But why would you do this if you have more information?

The information collected by your instrument is the complex impedance versus frequency taken in a sweep through resonance. While I don't doubt that there is in theory enough data in there to estimate the variation of inductance with frequency, in the inevitable presence of noise, it may prove difficult to do this with adequate precision. This would require analysis and experiment to determine.

Well go right ahead. Certainly the technique that uses less information and a poor numerical technique is the one that needs the support.

Well, this method has been standard in the EE world since the 1930s, and is presented by Terman, one of the giants in the EE world. His book is still considered a classic, and every EE library has a copy.

And Terman's method is simple enough for people without all that equipment and all that mathematics to use. Which may have something to do with the fact that it's still used.


Nor was I the one claiming that one method was better than the other, or attempting to prove it for any definition of "better".

Nobody with any sense does simple extrapolation if there is a better technique available, one using more information and incorporating a model of the process if possible. Take for example the stock market....

Well, there must be a lot of senseless rich people then. By definition, investment implies extrapolation, with real money riding on the quality of that extrapolation. Plus a bit of luck.

[Mike Sulzer]

The road is the kerfuffle about the constancy or non-constancy of inductance in cored coils. There are a number of long and noisy threads on that. We must have made at least 100 postings each in those threads.

That is not how I read your post of 05-15-2014, 09:56 PM. Look at what you are responding to.

The information collected by your instrument is the complex impedance versus frequency taken in a sweep through resonance. While I don't doubt that there is in theory enough data in there to estimate the variation of inductance with frequency, in the inevitable presence of noise, it may prove difficult to do this with adequate precision. This would require analysis and experiment to determine.

There is no sweep, and there is nothing special about the resonant frequency. The SNR is better than it needs to be, as demonstrated by trying shorter codes and still getting the same results. The accuracy of the complex impedance measurements has been verified by measuring various components and comparing with commercial instruments at the frequencies at which they work, and then verifying that the correct frequency variation is measured. You had opportunities to raise such issues in the discussions on this instrument, but you did not. It is extremely low behavior to claim now that there are problems, especially without any evidence and apparently without even a basic understanding of how the instrument works.

Well, this method has been standard in the EE world since the 1930s, and is presented by Terman, one of the giants in the EE world. His book is still considered a classic, and every EE library has a copy.

Did it ever occur to you that it became standard because it was the best you could do in 1930? All EE instruments have evolved since then.

And Terman's method is simple enough for people without all that equipment and all that mathematics to use. Which may have something to do with the fact that it's still used.

For my instrument:

Equipment needed: a recording interface and a computer. How many people in this field have that, but lack any actual EE instruments of any kind?

Mathematics needed: None. It is done for you in the software, just as it is in all modern instruments, full of low level code in FPGAs, up to C++ in embedded Linux processors. I have provided the code for those who want to look. Do not expecxt that on commercial instruments.

Nor was I the one claiming that one method was better than the other, or attempting to prove it for any definition of "better".

Right, you were just casting doubts about my instrument in order to demonstrate that it is not better.

Well, there must be a lot of senseless rich people then. By definition, investment implies extrapolation, with real money riding on the quality of that extrapolation. Plus a bit of luck.

As I said, extrapolation is best used with a model, that is additional information about the process beyond just the time history to be extrapolated. That is how successful investors really work.

[Joe Gwinn]

That is not how I read your post of 05-15-2014, 09:56 PM. Look at what you are responding to.

Well, there is history, and it sounded like we were about to repeat it.

There is no sweep, and there is nothing special about the resonant frequency. The SNR is better than it needs to be, as demonstrated by trying shorter codes and still getting the same results. The accuracy of the complex impedance measurements has been verified by measuring various components and comparing with commercial instruments at the frequencies at which they work, and then verifying that the correct frequency variation is measured. You had opportunities to raise such issues in the discussions on this instrument, but you did not. It is extremely low behavior to claim now that there are problems, especially without any evidence and apparently without even a basic understanding of how the instrument works.

It's true that you actually use random drive (or is it now multi-sine?) and cross-correlation, which is a very good approach. But the resulting data is equivalent to a sweep, albeit collected far more quickly.

Did it ever occur to you that it became standard because it was the best you could do in 1930? All EE instruments have evolved since then.

Sure. We could use present-day standards-lab methods and measure the pickup coils parameters to parts per million to parts per billion. But would there be any point to such accuracy?

For my instrument: Equipment needed: a recording interface and a computer. How many people in this field have that, but lack any actual EE instruments of any kind? Mathematics needed: None. It is done for you in the software, just as it is in all modern instruments, full of low level code in FPGAs, up to C++ in embedded Linux processors. I have provided the code for those who want to look. Do not expect that on commercial instruments.

Have you packaged this up for easy use by musicians, most of whom cannot follow a word of our fights? I think you greatly underestimate the technical background needed to use your instrument and to understand what it is saying. There are a number of people in this forum who do have sufficient background, but there are many more who do not (and yet they somehow manage to make very good pickups).

Right, you were just casting doubts about my instrument in order to demonstrate that it is not better.

In my original posting, I simply provided the pages out of Terman's book that defined the standard method for such measurements. I made no claim that this was the bast method known, or that standards labs would be impressed. You came in, claiming that your method was better. I asked what "better" meant. The point being that the definition of better may include things other than standards-lab levels of accuracy. Simplicity and adequate performance usually wins.

As I said, extrapolation is best used with a model, that is additional information about the process beyond just the time history to be extrapolated. That is how successful investors really work.

Successful investors also have enough self control to not panic, in either direction. My favorite is "The Intelligent Investor" by Benjamin Graham.

The Intelligent Investor CD: The Classic Text on Value Investing: Benjamin Graham, Bill McGowan: 9780060793838: Amazon.com: Books

This is the source of the querulous question "Yes but, where are the Investors' yachts?"

[Mike Sulzer]

It's true that you actually use random drive (or is it now multi-sine?) and cross-correlation, which is a very good approach. But the resulting data is equivalent to a sweep, albeit collected far more quickly.

No, it is a single binary code, longest from the paper I referred to. Not sure why equivalence to a sweep is significant to you. The multi sine has some characteristics I do not like.

Sure. We could use present-day standards-lab methods and measure the pickup coils parameters to parts per million to parts per billion. But would there be any point to such accuracy?

Or we could use a technique that does not require several individual measurements involving multiple components and hand plotting a graph. That is what I have developed.

Have you packaged this up for easy use by musicians, most of whom cannot follow a word of our fights? I think you greatly underestimate the technical background needed to use your instrument and to understand what it is saying. There are a number of people in this forum who do have sufficient background, but there are many more who do not (and yet they somehow manage to make very good pickups).

The instructions I have provided in no way depend on our fights. (I do need to verify that in works on an inexpensive Windows pc with a bottom of the line recording interface, before it is really ready to go.) Musicians know how to use this equipment; that was the point. As for firing up interactive Python, loading a module, and running one program or another, well, this is not so hard to learn. You do need to make a little test setup with a resistor; also not so hard. Understanding the resonant frequency and the coil capacitance is not so hard as well, and the same understanding is needed for Terman's technique, although I think more technician experience is necessary for his. It is not necessary to understand the eddy current plots if you do not want to, but to help, I have provide examples of measurements of pickups with various levels of ec effects.

In my original posting, I simply provided the pages out of Terman's book that defined the standard method for such measurements. I made no claim that this was the bast method known, or that standards labs would be impressed. You came in, claiming that your method was better. I asked what "better" meant. The point being that the definition of better may include things other than standards-lab levels of accuracy. Simplicity and adequate performance usually wins.

"standards labs would be impressed" That is your strawman. I merely claimed higher accuracy and easier use, especially if you have many pickups to measure.

So let's see: you claimed my method derives the C from the resonance and therefore is not accurate, that it might be capable of measuring the apparent inductance as a function of frequency, but that is doubtful just because, and it works by sweeping the frequency. Also that I underestimate how difficult it is to use, and I am apparently claiming "standards lab" accuracy. Obviously you did not take the time to try it out before making these judgements, and apparently did not even bother to remember or look up what you read about it. But other than that, your comments have been most useful in providing an opportunity to explain again how it works. Many thanks.

[Joe Gwinn]

No, it is a single binary code, longest from the paper I referred to. Not sure why equivalence to a sweep is significant to you.

Because it's easier to understand than random drive and cross-correlation.

Or we could use a technique that does not require several individual measurements involving multiple components and hand plotting a graph. That is what I have developed. ... The instructions I have provided in no way depend on our fights. (I do need to verify that it works on an inexpensive Windows pc with a bottom of the line recording interface, before it is really ready to go.) Musicians know how to use this equipment; that was the point. As for firing up interactive Python, loading a module, and running one program or another, well, this is not so hard to learn. You do need to make a little test setup with a resistor; also not so hard. Understanding the resonant frequency and the coil capacitance is not so hard as well, and the same understanding is needed for Terman's technique, although I think more technician experience is necessary for his. It is not necessary to understand the eddy current plots if you do not want to, but to help, I have provide examples of measurements of pickups with various levels of ec effects.

My instinct is that you are grossly underestimating the level of bafflement all that will cause, but there is no harm in trying. But let me put it this way: One thing we often describe is how to measure resonance.

Nor is learning a computer language like Python everybody's cup of tea.

Terman's method requires understanding only the LC resonance equation. This is far simpler than explaining what cross-correlation means, and how to use it with random drive. Or what the difference between random drive and multi-sine is, and why it matters for a pickup.

"standards labs would be impressed" That is your strawman. I merely claimed higher accuracy and easier use, especially if you have many pickups to measure.

The point is that Terman's method is adequate for pickups. I've used it. It's not hard.

So let's see: you claimed my method derives the C from the resonance and therefore is not accurate, that it might be capable of measuring the apparent inductance as a function of frequency, but that is doubtful just because, and it works by sweeping the frequency. Also that I underestimate how difficult it is to use, and I am apparently claiming "standards lab" accuracy. Obviously you did not take the time to try it out before making these judgements, and apparently did not even bother to remember or look up what you read about it. But other than that, your comments have been most useful in providing an opportunity to explain again how it works. Many thanks.

Not exactly. Re-read the original posting, where I posted a scan of the pages of Terman's book where he described his method. You replied with screaming criticisms of a method that has been standard since the 1930s. It was all downhill from there.


Anyway, this thread has gone on long enough. You may have the last word.

[Mike Sulzer]

OK, then I will just drop off this link to a comparison of various methods of measuring the capacitance of an rf coil: Measuring Distributed Capacitance.

Terman's method did not do well, but the reasons for this probably do not apply to its application to pickup coils.

[Mike Sulzer]

Joe G.

When placing a shield around the coil, the full capacitance measured between one coil wire end and the fully isolated shield, is not fully imposed on the resulting coil resonance. Only about 20 percent of the coil to shield capacitance is added to lower the coil resonant frequency.

I would expect that when measuring with a capacitance meter at a frequency well below the coil self resonance, it would not matter which end of the wire you connect the meter to. However, when measuring the self resonance with the shield in place, I think it would make a difference whether the coil start or the coil finish is connected to ground. Is this what you observe?

[Rick Turner]

There's more than one way to shield a coil! Many years ago when I was doing repairs full time, I stopped wrapping copper foil tape directly onto Strat coils and went to painting the insides of the pickup covers to move the shielding farther away from the coils. I never measured the resulting capacitance, but nobody ever complained about a loss of highs, and I had some pretty picky customers.

[Joe Gwinn]

There's more than one way to shield a coil! Many years ago when I was doing repairs full time, I stopped wrapping copper foil tape directly onto Strat coils and went to painting the insides of the pickup covers to move the shielding farther away from the coils. I never measured the resulting capacitance, but nobody ever complained about a loss of highs, and I had some pretty picky customers.

Painting the routed cavity with shielding paint is the best way. The capacitance effect is minimal because the air gap is so large, probably about one inch.

Don't forget to shield the pickup guard as well. Aluminum tape works well.

[Rick Turner]

Got to get them pesky pickup covers, though, in addition to (if a Strat or Tele) 100% of the cavities AND the entire back of the pickguard. You want major ground planes happening. Intercept those buzzy fields before they get anywhere near your pickup coils or any signal carrying wires. I'm a fan of star grounding as much as practical, too, and it seems that it's all the more important in active instruments where ground loops just become little antennae for noise. Capacitive issues (and their effects on treble) aside, my dictum is "too much shielding is barely enough".

The funny thing, of course, is that in the '50's and early '60's, Gibson did a fantastic job of shielding. Then the bean counters took over the design department. And in the early '50's there were far, far fewer sources of radiated noise, and players did not crank the amps to "11", so shielding and even humbucking was far less needed.

[bbsailor]

I would expect that when measuring with a capacitance meter at a frequency well below the coil self resonance, it would not matter which end of the wire you connect the meter to. However, when measuring the self resonance with the shield in place, I think it would make a difference whether the coil start or the coil finish is connected to ground. Is this what you observe?

Mike,

The difference between using either end of the coil was only a few pf but far away from how the total coil-to-shield capacitance should have affected the resonant frequency, using the resonant frequency measurment method shown later in the article using a 1 pf "gimmick" capacitor to isolate the probe loading from causing a lower resonant frequency.

The wire core used was AWG30 with Teflon insulation. The wire OD (including insulation) was .024".

I used a digital frequency display signal generator isolated from the coil by a 1M ohm resistor to accurately measure the resonant frequency observed on an oscilloscope and measured the coil inductance by both a calculated method as well as two LCR meters. When I did these tests I then realized that distributed capacitance is something that is best measured rather than attempting to be calculated. The total capacitance as seen in the mono coil circuit affects the value of the damping resistor (more capacitance requires a lower damping resistor value) and then affects the lowest amount of time when the TX signal can be damped to zero Volts and then become the RX coil thus being able to detect lower time constant targets like gold. Lower coil circuit capacitance damps quicker and then detects the quickly decaying eddy currents in the gold targets. Also, the faster the TX current turns off the better it stimulates smaller targets with lower TCs. The turn-off current TC is governed by the coil inductance divided by the damping resistor value and should be 5 times faster than the TC of the target sought. All of this is of no interest to pickup makers but it expanded my interest in coil time constants.

These experiments led me to believe that guitar pickup time constants (TC) might have something to do with our perception of the initial attack of the initial string transient pulse. A 6,000 ohm pickup that is 2H represents a .000333 second TC or 1/.000333 Hz to reach one TC which is only about 63 percent of the maximum current of that attack transient. Here is where a computer analysis of this situation might reveal some new, ground breaking information. Since there is no data on this issue, I am only trying to better understand why Rick Turner's posted his experience with Litz wire as it offered an audibly detectable coherence in the initial attack. Litz wire does have the ability of preventing the inductance of the coil from changing with frequency as at higher frequencies the current tends to move to the outer sides of the single strand wire and changes the inductance as referenced in the Clifton Labs distributed capacitance discussion. However, Litz wire should only be effective at much higher frequencies, but it might have some subtle effect at audio frequencies near 10KHz per Rick Turner's observatrions.

I am only trying to reach a deeper understand of some of the more subtle, but not documented, effects of guitar pickup design.

I am not trying to jump into your discussion with Joe G.,but I have been following it closely.

Joseph Rogowski

[Mike Sulzer]

Hi Joseph,

I put a shield around a (somewhat underwound) single coil pickup and measured the impedance with one wire connected to the shield, and then the other. The shield and attached wire went to the low side on the jig. There is about a 30 pf difference in the measured C, with corresponding difference in resonant frequency. The lower C value is just one pf more than measured with no shield. The shield wraps nearly all the way around but does not connect to itself to keep eddy currents down.
oneWay.pdf and theOtherWay.pdf

Mike,

The difference between using either end of the coil was only a few pf but far away from how the total coil-to-shield capacitance should have affected the resonant frequency, using the resonant frequency measurment method shown later in the article using a 1 pf "gimmick" capacitor to isolate the probe loading from causing a lower resonant frequency.

The wire core used was AWG30 with Teflon insulation. The wire OD (including insulation) was .024".

I used a digital frequency display signal generator isolated from the coil by a 1M ohm resistor to accurately measure the resonant frequency observed on an oscilloscope and measured the coil inductance by both a calculated method as well as two LCR meters. When I did these tests I then realized that distributed capacitance is something that is best measured rather than attempting to be calculated. The total capacitance as seen in the mono coil circuit affects the value of the damping resistor (more capacitance requires a lower damping resistor value) and then affects the lowest amount of time when the TX signal can be damped to zero Volts and then become the RX coil thus being able to detect lower time constant targets like gold. Lower coil circuit capacitance damps quicker and then detects the quickly decaying eddy currents in the gold targets. Also, the faster the TX current turns off the better it stimulates smaller targets with lower TCs. The turn-off current TC is governed by the coil inductance divided by the damping resistor value and should be 5 times faster than the TC of the target sought. All of this is of no interest to pickup makers but it expanded my interest in coil time constants.

These experiments led me to believe that guitar pickup time constants (TC) might have something to do with our perception of the initial attack of the initial string transient pulse. A 6,000 ohm pickup that is 2H represents a .000333 second TC or 1/.000333 Hz to reach one TC which is only about 63 percent of the maximum current of that attack transient. Here is where a computer analysis of this situation might reveal some new, ground breaking information. Since there is no data on this issue, I am only trying to better understand why Rick Turner's posted his experience with Litz wire as it offered an audibly detectable coherence in the initial attack. Litz wire does have the ability of preventing the inductance of the coil from changing with frequency as at higher frequencies the current tends to move to the outer sides of the single strand wire and changes the inductance as referenced in the Clifton Labs distributed capacitance discussion. However, Litz wire should only be effective at much higher frequencies, but it might have some subtle effect at audio frequencies near 10KHz per Rick Turner's observatrions.

I am only trying to reach a deeper understand of some of the more subtle, but not documented, effects of guitar pickup design.

I am not trying to jump into your discussion with Joe G.,but I have been following it closely.

Joseph Rogowski

[bbsailor]

Hi Joseph,

I put a shield around a (somewhat underwound) single coil pickup and measured the impedance with one wire connected to the shield, and then the other. The shield and attached wire went to the low side on the jig. There is about a 30 pf difference in the measured C, with corresponding difference in resonant frequency. The lower C value is just one pf more than measured with no shield. The shield wraps nearly all the way around but does not connect to itself to keep eddy currents down.
[ATTACH]28944[/ATTACH] and [ATTACH]28945[/ATTACH]

Mike,

Please make two more measurements. (1) the shield capacitance between one pickup coil wire and the free floating shield, and (2) the shield capacitance between the free floating shield and the other pickup coil wire. At frequencies near the 1 Mhz region, the electrostatic interaction of insulation material near the coil affects the resonance more than at near the 10Khz region. Read this document and what they say about coils with high turns, like your pickup coil.
http://www.datatronics.com/pdf/distr...ance_paper.pdf

Your pickup test had several thousand turns of very fine wire, while my pulse induction coil only had 19 turns (about 10.5 inches in diameter about 300 uH) with a Teflon dielectric of 2 thus making other nearby diaelectric interactions create vastly different interactions between the shielded vs no shielded coil results. I used a polyethylene shield-to-coil spacer which has a slightly higher dielectric constant than Teflon. The reason for your different capacitance between using each of the coils wires was due to the following. Your shield was probably laid closely around the pickup coil wire bundle and probably represents a higher measured capacitance as the outer pickup winding is closer to the nearby shield than the inside winding. However, your graphs did not state which connection was inner or outer.

Does your test setup have the capability to impose a pulse on the pickup coil analyze pickup transient time constant relationship to output waveform? Do you have access to 7/44 Litz wire to wind a bobbin coil and compare it to an equivalent single strand wire with the same number of turns doing the same tests as you did above plus the pulse test?

Thanks for your input.

Joseph Rogowski

[Mike Sulzer]

First, I have redone the files from last night. The resistance of the coil was 1.005K high because of a problem in the code (the sensing resistor was left in series). The high frequencies are also a bit different, but that is mostly the result of temperature changes. (No, pickups are not temperature stable, rcoil has changed, too)

The third attachment shows the capacitance from the shield to each wire, measured with my instrument, something it is not really intended for. One way looks good, not too much variation in frequency and not so different from the other measurement. The other way looks through the whole coil and suffers badly from the resonance.

oneway.pdf theOtherWay.pdf sheildCapacitance.pdf

Mike,

Please make two more measurements. (1) the shield capacitance between one pickup coil wire and the free floating shield, and (2) the shield capacitance between the free floating shield and the other pickup coil wire. At frequencies near the 1 Mhz region, the electrostatic interaction of insulation material near the coil affects the resonance more than at near the 10Khz region. Read this document and what they say about coils with high turns, like your pickup coil.
http://www.datatronics.com/pdf/distr...ance_paper.pdf

Your pickup test had several thousand turns of very fine wire, while my pulse induction coil only had 19 turns (about 10.5 inches in diameter about 300 uH) with a Teflon dielectric of 2 thus making other nearby diaelectric interactions create vastly different interactions between the shielded vs no shielded coil results. I used a polyethylene shield-to-coil spacer which has a slightly higher dielectric constant than Teflon. The reason for your different capacitance between using each of the coils wires was due to the following. Your shield was probably laid closely around the pickup coil wire bundle and probably represents a higher measured capacitance as the outer pickup winding is closer to the nearby shield than the inside winding. However, your graphs did not state which connection was inner or outer.

Does your test setup have the capability to impose a pulse on the pickup coil analyze pickup transient time constant relationship to output waveform? Do you have access to 7/44 Litz wire to wind a bobbin coil and compare it to an equivalent single strand wire with the same number of turns doing the same tests as you did above plus the pulse test?

Thanks for your input.

Joseph Rogowski

[bbsailor]

First, I have redone the files from last night. The resistance of the coil was 1.005K high because of a problem in the code (the sensing resistor was left in series). The high frequencies are also a bit different, but that is mostly the result of temperature changes. (No, pickups are not temperature stable, rcoil has changed, too)

The third attachment shows the capacitance from the shield to each wire, measured with my instrument, something it is not really intended for. One way looks good, not too much variation in frequency and not so different from the other measurement. The other way looks through the whole coil and suffers badly from the resonance.

[ATTACH]28953[/ATTACH] [ATTACH]28954[/ATTACH] [ATTACH]28955[/ATTACH]

Mike,

The literature says to minimize capacitance, connect the shield to the coil wire lead with the least voltage potential between them. This would support your trace (the other way) that connected the inner wire to the shield and was looking through the whole coil thickness thus creating a resonant effect on the graph. Please confirm if this is correct according to these measurements.

Joseph Rogowski