Measuring pickups Capacitance?

[Antigua]

I agree that the uneven layering of turns is the more probable explanation of secondary resonances, but there is no solid proof yet - only some probability. I have found new strat PUs that show both an unusual low Q as well as a very strong antiresonance-resonance pair. And I am convinced that such secondary resonances could be modelled from the equivalent circuit of partial shorts shown by Zollner if distributed capacitances are included. Fact is that winding shorts do happen but are only likely to be detected if output is noticeably reduced and/or the sound is really dull. So I suspect some dark figure.

Hence my wish for a bifilar wound PU. There were requests from other posters also.

You say winding shorts "do happen" but that uneven layering resulting in secondary resonances lacks solid proof. Both of these things lack solid proof, IMO.

A bifilar coil wouldn't necessarily prove a relationship between uneven winding and secondary resonances, though. I would be interested in making a pickup that deliberately emphasizes a secondary resonance, in order to better understand existing pickups, but I don't think a bifilar coil truly approximates that, since existing pickups are not bifilar. I personally don't find winding pickup to be an enjoyable process, and it hurts my eyes to guide the wire onto the spinning bobbin, so it's not something I do readily.

I think a better test pickup might be one where the coil is intentionally wound extremely lopsided at first, and then balanced out as the wind is completed, so that the majority of the early winds would be on one side of the coil former, and the latter winds on the other. This would be like a stacked coil, except but without a significant division between the two halves.

I have also found Strat pickups to have unusually high or low Q factors, but that can also be attributed to defective formulations of AlNiCo, which was supported by the fact that the pole pieces had lower residual flux values than was to be expected.

[Joe Gwinn]

Not necessarily. A single shorted turn won't show at all. Even 100 hard-shorted outer turns change Rs@1kHz by less than 10%. A more sensitive method is to compare the Q values of the resonance in the impedance plots. But verified reference PUs are rarely available.

I'll grant that Q is more sensitive, but it's harder to measure than the excess of AC over DC resistance. For pickup makers in production, they can easily measure the AC and DC resistance of every unit of a given model they make, and then units with too much excess resistance will stick out. And if it doesn't stand out, it's likely to be inaudible as well.

You may want to read this article by Prof. Manfred Zollner:
https://gitec-forum.de/wp/wp-content...he-winding.pdf

There are several variables that influence the results: number of turns shorted, position of the shorted section within the coil influencing coupling - and contact resistance of the "short". The article only covers perfect coupling.

One may argue that only shorts that noticeably effect the loaded transfer response will be audible. But I would be alarmed by other evidence as well, because this indicates that something is wrong with the wire quality or the winding technique.

There are a number of old threads on this very subject. The core problem is that if one winds 4000 turns of thin wire on a form with winding pressure of say 20 grams, the total hoop stress is 4000*20= 80 kilograms, in a very small space, so the unit pressure at the bottom is quite high. This is enough to force perfectly good magnet wire onto the core and/or other strands with sufficient pressure to pierce the insulation layer on the wire - shorts to the magnets are a common problem in strat pickups. It turned out that no kind of varnish film was stiff enough to prevent this (the wires would creep through the varnish film), but taping the core with paper of kapton or the like would work. These are soft enough to give a little, and yet won't keep on creeping. Likewise, if bobbins were not perfectly smooth and deburred near the core, there could be shorts. (Tape also helps with rusty magnets.)

[Mike Sulzer]

The outer coil shows about +2dB over the inner coil. Both are 4,000 turns, but the outer coil involves more area and wire length, of course. Ultimately the deciding factor for the sum voltage comes down to which coil catches a greater overall flux change. I'm not sure of the exact geometry or integrations that are at play, but it intuitively stands to reason that bigger net catches more fish.

But as you make this net bigger, some of the fish that swim in, turn around and swim back out. That is, field lines from the vibrating string must return to it, and so as you make the coil very big, flux begins to cancel out. But apparently you have not reached that point yet with your inner and outer coils. I am a bit surprised that the outer has 2db more output. Since downward going flux is not completely confined to the core, I do expect loops right at the core to contribute less than ones somewhat larger, but I would have thought not so much. Maybe some FEMM modeling is required, a cylindrical pole piece with a tiny cylinder above it to act as the string. Or perhaps MacDonald's integration could be used with different limits to look at this.

[Helmholtz]

I'll grant that Q is more sensitive, but it's harder to measure than the excess of AC over DC resistance. For pickup makers in production, they can easily measure the AC and DC resistance of every unit of a given model they make, and then units with too much excess resistance will stick out. And if it doesn't stand out, it's likely to be inaudible as well.

Completely agreed. Measuring loss resistance or the Q of L at 1kHz is by far the most convenient way to detect serious defects.

There are a number of old threads on this very subject. The core problem is that if one winds 4000 turns of thin wire on a form with winding pressure of say 20 grams, the total hoop stress is 4000*20= 80 kilograms, in a very small space, so the unit pressure at the bottom is quite high. This is enough to force perfectly good magnet wire onto the core and/or other strands with sufficient pressure to pierce the insulation layer on the wire - shorts to the magnets are a common problem in strat pickups. It turned out that no kind of varnish film was stiff enough to prevent this (the wires would creep through the varnish film), but taping the core with paper of kapton or the like would work. These are soft enough to give a little, and yet won't keep on creeping. Likewise, if bobbins were not perfectly smooth and deburred near the core, there could be shorts. (Tape also helps with rusty magnets.)

Thanks, this makes a lot of sense and essentially confirms what I had in mind. I have seen several Fender strat PUs from the late 60s and early 70s showing this kind of problem. All of them had (period correct) PE insulation. Formvar is a tougher material according to MWS.

Another critical area could be the start wire crossing the bottom of the coil.

Such creepage induced shorts may need some time (and thermal cycles) to develop and show. Thus pre-delivery inspection might not detect them.

[Helmholtz]

But as you make this net bigger, some of the fish that swim in, turn around and swim back out. That is, field lines from the vibrating string must return to it, and so as you make the coil very big, flux begins to cancel out. But apparently you have not reached that point yet with your inner and outer coils. I am a bit surprised that the outer has 2db more output. Since downward going flux is not completely confined to the core, I do expect loops right at the core to contribute less than ones somewhat larger, but I would have thought not so much. Maybe some FEMM modeling is required, a cylindrical pole piece with a tiny cylinder above it to act as the string. Or perhaps MacDonald's integration could be used with different limits to look at this.

This is exactly what Zollner explains in his book. Based on simulations of the ac component of B, he shows that the induced µV/turn value of a Jazzmaster PU first increases with the radius (coil half-width) up to ca. 7.5mm, but decreases above.

[Joe Gwinn]

Thanks, this makes a lot of sense and essentially confirms what I had in mind. I have seen several Fender strat PUs from the late 60s and early 70s showing this kind of problem. All of them had (period correct) PE insulation. Formvar is a tougher material according to MWS.

PE isn't all that bad, and while Formvar is tougher, it still can be penetrated.

Another critical area could be the start wire crossing the bottom of the coil.

Such creepage induced shorts may need some time (and thermal cycles) to develop and show. Thus pre-delivery inspection might not detect them.

Yes. As for the start wire, one remedy is to tape over it before winding the full coil. Or a groove or alternate path for the start. And so on.

As for creepage, that has to be solved by design, and metric of success is the absence of warranty repair requests.

We have had reports of bad wire, with brittle enamel that invariably cracked under the stresses of winding (where this wire is stretched). If the enamel doesn't respond to stretching more or less the same as copper, the enamel pops off the copper.

[Antigua]

But as you make this net bigger, some of the fish that swim in, turn around and swim back out. That is, field lines from the vibrating string must return to it, and so as you make the coil very big, flux begins to cancel out. But apparently you have not reached that point yet with your inner and outer coils. I am a bit surprised that the outer has 2db more output. Since downward going flux is not completely confined to the core, I do expect loops right at the core to contribute less than ones somewhat larger, but I would have thought not so much. Maybe some FEMM modeling is required, a cylindrical pole piece with a tiny cylinder above it to act as the string. Or perhaps MacDonald's integration could be used with different limits to look at this.

I didn't mean to imply that if the pickup was infinite in size that you'd get a maximum potential output.

BTW, my testing coil is shaped like the end of a Popsicle stick, which I usually have oriented as though it were a guitar string, but I also tried turning it sideways, and the overall output increased by 1dBV, but the difference between inner and outer coils was still 2dBV:



I suspect that if the coil had steel pole pieces, the difference between the inner and outer coils would be smaller than they are having AlNiCo pole pieces.

[Antigua]

I was doing some experiments with the exciter coil over a Stratocaster pickup, so see how the output level drops off as offset distance is added between the exciter and the pickup, and one thing I see is that when the exciter is about two millimeters off the edge of the pickup, there is a dead spot where the voltage output drops, and then it increases again at 3mm, and a little more at 4mm before dropping off again. What is happening is that in this space it's transitioning from the primary path to the return path, and that can see in the phase of the size wave.

Exciter over pickup:


Exciter 2mm off the edge


Exciter 6mm off the edge


The blue line is the voltage from the pickup. The red line is the voltage from the function generator. The frequency happens to be 5kHz, which is below resonance.

You can see that once the exciter is 6mm away, the voltage is a opposite phase from what it is when the exciter is over the coil. This means the pickup coil is generating voltage from the return path of the exciter coil. The dead spot at 2mm must represent the distance at which the primary and return paths cause a perfect cancellation.

The insight I get from this is that there must be some ideal ratio of pole piece to coil width, because the width of the pole piece will determine the width of the magnetic field, and the width of the coil will determine what ratio of primary to return path flux manifests as a voltage. If the coil is too small, the primary flux path will exceed the size of the coil itself, wasting flux. If the coil is wider, it will integrate all of the primary flux, and some non-productive lines of flux which point sideways, but if the coil is a multiple of the pole piece width, say a P-90 or a Jazzamster pickup, it will capture primary flux, unproductive sideways flux, and return path.

Another thing I can see from the oscilloscope is that the return path of flux, though weaker than the primary flux, does not drop off as quickly with distance, for example, here's the reading with the exciter 20mm away:



Even at a larger distance, the return path is still generating a consistent voltage. Assuming the magnetic field off of a guitar string has a similar overall geometry, then as the coil is made wider and wider, the more and more return path flux you end up capturing, causing an increase in phase cancellation.

I'm not sure what the ideal pole piece width to coil width ratio is, but it would also have to account for the distance between the coil and the string, since if the string is further away, that would influence the area of magnetization. My guess should be that the typical geometry of a Strat coil, where the coil's outer radius is about three times the inner radius, or the radius of the pole pieces, is probably close to idea.

[Helmholtz]

I was doing some experiments with the exciter coil over a Stratocaster pickup, so see how the output level drops off as offset distance is added between the exciter and the pickup, and one thing I see is that when the exciter is about two millimeters off the edge of the pickup, there is a dead spot where the voltage output drops, and then it increases again at 3mm, and a little more at 4mm before dropping off again. What is happening is that in this space it's transitioning from the primary path to the return path, and that can see in the phase of the size wave.

In this experiment you are using the PU as sensor for the flux distribution of the exciter coil. Thus your results will depend on the shape of the exciter. The real ac field between string and PU has a spatial distribution different from a solenoid (albeit squeezed) and therefore most probably penetrates the PU coil in a somewhat different way.

I generally doubt that such measurement can give precise information about the PU's relative sensitivity (H. Lemme does not recommend his method for sensitivity measurements).

But does anybody really care about a +/- 2dB difference in sensitivity as long as the sound does not change? Anyway, the more effective way to increase sensitivity with the same number of turns is a flatter, wider coil - or simply a pickup structure that allows to position the coil (not necessarily the magnets) closer to the strings.

[Antigua]

In this experiment you are using the PU as sensor for the flux distribution of the exciter coil. Thus your results will depend on the shape of the exciter. The real ac field between string and PU has a spatial distribution different from a solenoid (albeit squeezed) and therefore most probably penetrates the PU coil in a somewhat different way.

I generally doubt that such measurement can give precise information about the PU's relative sensitivity (H. Lemme does not recommend his method for sensitivity measurements).

Precision was not the goal. The goal is merely to demonstrate that, at a certain width, the coil ceases to be productive, and becomes counter productive. Even though this might be obvious to some, seeing it in a real bode plot proves it, for people who have to see to believe. Since all magnetic fields have a return path, it need only be a small magnetic field, as a string is small, in order to demonstrate the effect.

But does anybody really care about a +/- 2dB difference in sensitivity as long as the sound does not change?

Why make a pickup inefficient if you don't have to? The goal is simply to know all there is to know about pickup design, because among guitar players, there's just a lot of whimsy and guessing going on. Whether someone chooses to apply the information in their own pickup designs is for them to decide.

[Helmholtz]

The goal is merely to demonstrate that, at a certain width, the coil ceases to be productive, and becomes counter productive.

I don't think that your results can confirm this statement. Even if the local ac flux changes polarity at same distance from the PU's center, this doesn't mean that the outer turns do not contribute positively to output voltage. What matters is the integrated net flux through the loop area of each turn. And this stays positive (i.e constructive) even for the outer turns of a coil having a width well above 1.5''. This means that the outer turns of a fat coil don't diminish output, they just contribute somewhat less than the inner turns.

[Antigua]

This means that the outer turns of a fat coil don't diminish output, they just contribute somewhat less than the inner turns.

Something doesn't sit right with me; suppose you were to take this to an extreme, you have a pickup with a coil that is one foot in diameter, and it magically fits in the guitar. So you have about six inches of air gap, and the pole pieces are standard size. Will this pickup produce a good output, even though the turns of coil wire are six inches away from the pole pieces?

[Mike Sulzer]

Something doesn't sit right with me; suppose you were to take this to an extreme, you have a pickup with a coil that is one foot in diameter, and it magically fits in the guitar. So you have about six inches of air gap, and the pole pieces are standard size. Will this pickup produce a good output, even though the turns of coil wire are six inches away from the pole pieces?

He is not referring to a very wide coil, just a "fat" pickup. All flux cancels out if the loop is big enough, since the lines that leave the string must come back to it.

[Helmholtz]

All flux cancels out if the loop is big enough, since the lines that leave the string must come back to it.

Yes, and for this reason very wide loops (turns) produce near zero induced voltage. But as the voltages of all turns add up, the total EMF will not decrease by additional outer turns.

[Mike Sulzer]

Yes, and for this reason very wide loops (turns) produce near zero induced voltage. But as the voltages of all turns add up, the total EMF will not decrease by additional outer turns.

Yes, you are right, as long as you have the narrow turns, you get something from them, but the wide turns make a low Q low pass filter with a low frequency cutoff, and so as a practical matter, you do not get what you want.