|
|
Post by ms on Feb 3, 2019 13:55:21 GMT -5
For $27 you get a pair of these things. Advertised as 10.5K, but really 13.5K, brass cover, brass base plate, ceramic magnet, steel blades in each coil, loose coils waxed into the cover with the base plate shoved on. Just for fun, I measured the impedance in various stages of disassembly. First as is:  Very high eddy current losses, of course. next with the cover and base plate removed:  A lot less loss. Finally with the coils side by side with no steel blades or magnet:  This is a typical air core result. I have ferrite pieces (used for rf suppression) that are about 7/8" by 5/16", made from #43 material. They are a few thousands too thick to fit in the slot in the coils, but sanding them down takes only a few minutes. Six are used to connect the coils, one going under where each string would be if this were made into a sidewinder. (There is no pole piece assembly yet, and so results will change somewhat.) I left enough space between the coils for a ceramic blade as a pole piece. The two coils are sort of permanently wired together; that is, the center conductor of the shielded cable lead goes right inside the tape on one coil, and the pickups are connected together by a black wire going under the tape at each end. Therefore I did not disconnect anything, but simply flipped one col over to made two inductance measurements. The equations for the measurement are 2L + L12 and 2L - L12. The measurements with the Exrtech on 120Hz are 8.77H and 3.5H. There is a variation of about 0.2H between repeated measurements because things are not well secured.) Coil inductance is about 3H, and the mutual is about 2.6H. This might be larger than you expected, but ferrite pieces make a very effective coupler. The impedance with the negative coupling is this:  This is very encouraging; it is very high Q and you can add damping to shape it as you wish. It would give a resonance with cable of maybe 3.5KHz. You could get this up a bit by using some steel damping, which lowers the local L. This might be just what somebody wants, but I want higher like a Fender single coil. It looks like I have to buy bobbins and wind. |
|
|
|
Post by aquin43 on Feb 5, 2019 7:08:07 GMT -5
Hi
Interesting measurements.
Series coil inductance is L1 + L2 -2M, which makes the coupling a more plausible M = 1.3H or K = 0.43
Is there an upper limit for K that still allows the pickup to function? If there isn't, it means that the bandwidth of the pickup could be increased without limit, without loss of output.
Arthur
|
|
|
|
Post by ms on Feb 5, 2019 12:27:17 GMT -5
Hi
Interesting measurements.
Series coil inductance is L1 + L2 -2M, which makes the coupling a more plausible M = 1.3H or K = 0.43
Is there an upper limit for K that still allows the pickup to function? If there isn't, it means that the bandwidth of the pickup could be increased without limit, without loss of output.
Arthur
Right, thanks; you need that factor of two so that with K = 1, and L1 = L2 you have 2L + 2M = 4L, or 2L - 2M = 0 for out of phase. If you want K ==> 1, you have to close the magnetic circuit. Then it is not a pickup up anymore. You can extend the ferrite up the sides to make it less open; I have done this in a tedious 6 coil design and will do so for this as well. This also should reduce the hum along the relevant axis. |
|
|
|
Post by aquin43 on Feb 6, 2019 6:10:22 GMT -5
According to FEMM, even if you close the loop all of the flux returns to the magnet along the two limbs the coils would occupy.
That is an Alnico magnet on top of a Mumetal loop. Mumetal was perhaps not the best choice because it saturates easily but the pattern is essentially the same with mild steel or iron.
I suppose that it is the hum cancelling for fields along the magnet axis that is compromised by increasing the coupling, so the above arrangement would behave like a single coil.
Arthur
Arthur
|
|
|
|
Post by ms on Feb 6, 2019 6:48:47 GMT -5
The permanent magnet magnetizes the strings. (Of course, it also has permeability and conductivity, and so contributes to the pickup magnetic circuit. But that first function, magnetizing the strings, is independent.) The orientation of the field lines of the permanent magnet are not an issue once you have established that they do their job with the string.
Closing the magnetic circuit "below" as you have done, increases the inductance tremendously, but does nothing for increasing the output of the pickup. If you partially close the circuit in the direction up towards the string, you can do some of both. In FEMM, you can simulate the effect of the vibrating string by placing a small piece of permanent magnet material where the string would be and looking at the effect the magnetic circuit has. I forget the details, having not done this for some time. The goal, of course, is to increase the magnetic flux through the coil(s) from the vibrating string without increasing the inductance too much.
|
|
|
|
Post by aquin43 on Feb 6, 2019 8:04:53 GMT -5
I am not in any way advocating such a construction, just pointing out that the flux through the coils from the magnet is, surprisingly to me at least, not diminished by such an extreme geometry. The coil coupling in this case could be above 0.5, diminishing the inductance to below that of a single coil. It seems like getting something for nothing, but I think that the penalty is the loss of humbucking. Now, if humbucking disappears completely in this extreme case, it is probable it is what is being traded off by increasing the coupling generally.
As an aside, I recently came across a very interesting paper that shows how the pickup can be treated as a variable reluctance sensor and the output can be determined from the amount of the pre-existing flux that the string cuts as it moves.
Arthur
|
|
|
|
Post by ms on Feb 6, 2019 11:01:51 GMT -5
I do not think that the hum bucking is affected by the coupling. First, at low frequencies, the inductance is irrelevant. (Same argument as for the output level.) At higher frequencies, as a result of the load on the pickup, current flows and so you have to work out what cancels, or not, with what. I have not, but I bet the answer is that there is no loss of hum bucking, because the same symmetry seems to be there.
As for that paper: They set up a system of flux tubes describing the field of the permanent magnet. Then they consider the change in the reluctance along each tube when the string vibrates. It is non zero when the string passes out of or into flux tube, because the reluctance of the steel string differs from that of air. Even neglecting the incredible complexity of this, is it even right?
Take the strat pickup they discuss, and make an identical coil with no magnets. Now illuminate the string with a magnetic source very far away so that the field lines are very close to straight on the scale of string vibration. Locate the source so that the lines pass directly through the coil. If the string vibrates in the z direction, that is, towards and away from the pickup, No flux tubes of this field are cut. Yet the law of magnetic induction says that there is an output.
Now maybe all this works out if you consider instead the total field, the sum of the applied field and that induced in the string. But I do not think that is what they did, and doing so would make the method even more complicated.
|
|
|
|
Post by aquin43 on Feb 6, 2019 15:11:47 GMT -5
I don't think that the Lemarquand paper can be dismissed out of hand. The authors are well established in the field of magnetics and the paper was published in the IEEE Transactions on Magnetics. It has been referred to in other papers. I had to read it a few times to get the unfamiliar arguments into my head but now it seems to make perfect sense.
The assumption is made that the total flux above the pole is independent of the string position, which is reasonable when you realise that the maximum modulation of the flux through the coil is below one percent. The flux is distributed in the rest state according to the path of least reluctance. The string diverts some of this flux, leaving a smaller amount with the same overall distribution.
It could be interpreted in an analogy between electric and magnetic circuits. The string is one variable resistor in the network that affects all of the voltages in the network.
Arthur
|
|
|
|
Post by ms on Feb 6, 2019 19:21:35 GMT -5
Don't you find it strange that they do not refer to other solutions, for example using the law of magnetic induction? You would think that they would want to know if the results agreed. I do not think that they do. For example, I am sure you are familiar with MacDonald's little paper (Princeton). L&L state that for z vibration (towards and away from the magnet), there is no sensitivity if the string is centered over the magnet because By = 0 (across the face of the magnet). I recall a completely different result in MacDonald, one that makes sense. Bz points through the coil; In fact Bz = B dot da, and the integral over this quantity actually gives the highest sensitivity.
|
|
|
|
Post by ms on Feb 6, 2019 19:55:05 GMT -5
The side winder without the pole structure that I have described above can be tested with the exciter coil by placing it over the space between the two coils. The response is given in the first attachment. There is about 300 pf in addition to the coil capacitance.  If you fill in the space between the coils with ferrite beneath the exciter coil, then the coil excites a field in it, and this adds to the flux through the coils, increasing the output. (That is, a pole piece increases the output of the pickup.) The response with this change is shown in the next attachment.  The output has gone up around 4 or 4.5 db, and so the pole piece is a good addition. The low frequency response is about 1.5 db less than that of the original mini hum bucker (or actually the other member of the pair, which is almost the same as the one I have disassembled). |
|
|
|
Post by antigua on Feb 7, 2019 0:10:16 GMT -5
I'm surprised the inductance didn't increase much with the added ferrite, while the output did.
|
|
|
|
Post by antigua on Feb 7, 2019 2:57:39 GMT -5
As an aside, I recently came across a very interesting paper that shows how the pickup can be treated as a variable reluctance sensor and the output can be determined from the amount of the pre-existing flux that the string cuts as it moves.
Arthur
That model is empirically wrong. They claim that movement perpendicular the pickup's magnetic axis is stronger than movement parallel to the axis, on page 10, referring to figures 6 and 7. That's demonstrably false. Holding a guitar pick against the string above the selected pickup, so as to guide the movement of a plucked string, you can release the string an restrict the string movement to the z or y axis and hear for yourself that the parallel string movement produces strong output "thud! thud! thud!", while the perpendicular movement is almost dead silent, "tink.. tink.. tink..". I'm not strong in the math, so I can't find whatever flaw must exist in it, but it sounds as though they're saying that so long as the string stays within "flux tubes", no tubes are cut, and so nothing happens, hence they believe that perpendicular movement cuts lots of tubes and produces lots of voltage, which obviously doesn't happen in real life. Faraday's Law describes the comparatively simple idea of "voltage = flux change through a surface or loop over time", and that is going to vary based on flux density in one region of space versus another, and not how perpendicular the string movement happens to be with respect the the imaginary flux lines. As the string approaches the coil, the density of flux rises and voltage is produced, but when the string movement is transverse, the flux density across it's path remains about the same, and so very little voltage is produced, despite the fact that it may have cut across many imaginary lines of flux. It also appears that their model places a lot of significance on the permeability of the string, where as the McDonald model makes it out to be insignificant so long as its much greater than one, but does place importance on the radius of the string. The McDonald model produces results that align much more closely with direct observation, so I have a good amount of faith in it. |
|
|
|
Post by ms on Feb 7, 2019 7:20:46 GMT -5
I'm surprised the inductance didn't increase much with the added ferrite, while the output did. In the case of a sidewinder, inductance and output are two very different things, and the ferrite connecting the two coils tends to emphasize this by helping to contain the flux. Thus the "pole piece" does not affect that much since there is not so much flux there. However, for a field coming from the exciter coil the "pole piece" is in the right position to be affected. And then the result of inducing magnetic polarization in it is that the field lines need to go around and return to it, and the ferrite through the coils is in the right position to "help" with that, getting polarized and thus inducing additional voltage in the coil. |
|
|
|
Post by aquin43 on Feb 7, 2019 11:34:10 GMT -5
As an aside, I recently came across a very interesting paper that shows how the pickup can be treated as a variable reluctance sensor and the output can be determined from the amount of the pre-existing flux that the string cuts as it moves.
Arthur
That model is empirically wrong. They claim that movement perpendicular the pickup's magnetic axis is stronger than movement parallel to the axis, on page 10, referring to figures 6 and 7. That's demonstrably false. Holding a guitar pick against the string above the selected pickup, so as to guide the movement of a plucked string, you can release the string an restrict the string movement to the z or y axis and hear for yourself that the parallel string movement produces strong output "thud! thud! thud!", while the perpendicular movement is almost dead silent, "tink.. tink.. tink..". I'm not strong in the math, so I can't find whatever flaw must exist in it, but it sounds as though they're saying that so long as the string stays within "flux tubes", no tubes are cut, and so nothing happens, hence they believe that perpendicular movement cuts lots of tubes and produces lots of voltage, which obviously doesn't happen in real life. Faraday's Law describes the comparatively simple idea of "voltage = flux change through a surface or loop over time", and that is going to vary based on flux density in one region of space versus another, and not how perpendicular the string movement happens to be with respect the the imaginary flux lines. As the string approaches the coil, the density of flux rises and voltage is produced, but when the string movement is transverse, the flux density across it's path remains about the same, and so very little voltage is produced, despite the fact that it may have cut across many imaginary lines of flux. It also appears that their model places a lot of significance on the permeability of the string, where as the McDonald model makes it out to be insignificant so long as its much greater than one, but does place importance on the radius of the string. The McDonald model produces results that align much more closely with direct observation, so I have a good amount of faith in it. s Yes, there is obviously some error. I suspect that the problem stems from considering only the values of the x y z flux components rather than their gradients. Obviously, output can be generated only by changes in the flux cut by the moving "passing reluctance" of the string so it is changes in flux density along the path that would be of significance. So the sensitivity perpendicular to the magnet face should depend on the z variation of the flux density.
The effect of the permeability in the model is a diminishing returns one as in the McDonald model; high permeability helps, higher gives little change but the cross section of the string contributes directly.
There is experimental evidence that even in a strat type that pickup two to three cm of the string are involved above the -3dB level of contribution to the output, depending on the string height above the pickup, so calculating the cut flux wouldn't be that simple anyway.
Arthur
|
|
|
|
Post by ms on Feb 7, 2019 13:21:22 GMT -5
You are not going to fix it easily. It is obvious that only the permanent field in the neighborhood of the string can matter. That is, the details of the shape of the magnet have to drop out. But their claim is that they can get analytic results for square magnets but not cylindrical magnets. Something is very wrong.
Pickups are sensitive over about the width of the pole piece. Theory and experiment agree on this.
|
|
|
|
Post by antigua on Feb 7, 2019 18:36:23 GMT -5
I'm surprised the inductance didn't increase much with the added ferrite, while the output did. In the case of a sidewinder, inductance and output are two very different things, and the ferrite connecting the two coils tends to emphasize this by helping to contain the flux. Thus the "pole piece" does not affect that much since there is not so much flux there. However, for a field coming from the exciter coil the "pole piece" is in the right position to be affected. And then the result of inducing magnetic polarization in it is that the field lines need to go around and return to it, and the ferrite through the coils is in the right position to "help" with that, getting polarized and thus inducing additional voltage in the coil. Is this to say that, the added permeable mass has a two fold effect of improving coupling between the two coils, so that it reduces the inductance it would have otherwise created, all the while reducing reluctance between the coils and the guitar string? |
|
|
|
Post by antigua on Feb 7, 2019 18:45:18 GMT -5
That model is empirically wrong. They claim that movement perpendicular the pickup's magnetic axis is stronger than movement parallel to the axis, on page 10, referring to figures 6 and 7. That's demonstrably false. Holding a guitar pick against the string above the selected pickup, so as to guide the movement of a plucked string, you can release the string an restrict the string movement to the z or y axis and hear for yourself that the parallel string movement produces strong output "thud! thud! thud!", while the perpendicular movement is almost dead silent, "tink.. tink.. tink..". I'm not strong in the math, so I can't find whatever flaw must exist in it, but it sounds as though they're saying that so long as the string stays within "flux tubes", no tubes are cut, and so nothing happens, hence they believe that perpendicular movement cuts lots of tubes and produces lots of voltage, which obviously doesn't happen in real life. Faraday's Law describes the comparatively simple idea of "voltage = flux change through a surface or loop over time", and that is going to vary based on flux density in one region of space versus another, and not how perpendicular the string movement happens to be with respect the the imaginary flux lines. As the string approaches the coil, the density of flux rises and voltage is produced, but when the string movement is transverse, the flux density across it's path remains about the same, and so very little voltage is produced, despite the fact that it may have cut across many imaginary lines of flux. It also appears that their model places a lot of significance on the permeability of the string, where as the McDonald model makes it out to be insignificant so long as its much greater than one, but does place importance on the radius of the string. The McDonald model produces results that align much more closely with direct observation, so I have a good amount of faith in it. s Yes, there is obviously some error. I suspect that the problem stems from considering only the values of the x y z flux components rather than their gradients. Obviously, output can be generated only by changes in the flux cut by the moving "passing reluctance" of the string so it is changes in flux density along the path that would be of significance. So the sensitivity perpendicular to the magnet face should depend on the z variation of the flux density.
The effect of the permeability in the model is a diminishing returns one as in the McDonald model; high permeability helps, higher gives little change but the cross section of the string contributes directly.
There is experimental evidence that even in a strat type that pickup two to three cm of the string are involved above the -3dB level of contribution to the output, depending on the string height above the pickup, so calculating the cut flux wouldn't be that simple anyway.
Arthur
I don't understand the "cutting flux" thing, because when the guitar string moves closer to the magnet, it's not exactly "cutting" the lines of flux, but because the guitar string is of a fixed radius while the flux lines fan out in a gradient, the string envelops more lines of flux as it moves into the region close to the face of the magnet, where the line of flux become more densely packed, and then as the string moves away, it "releases" those lines of flux, and so you get a magnetic delta over time. Maybe that's what you mean by the gradient, I'm not sure, but their conclusion that string movement towards and away from the pole piece is less productive than the side-to-side movement shows that they're making some sort of basic mistake. I just looked at the McDonald equations and I'm proud to report that they make 15% more sense now that the last time I looked at them, but I couldn't determine where they fundamentally diverged. |
|
|
|
Post by aquin43 on Feb 8, 2019 8:43:06 GMT -5
Yes, looking more closely, the "passing reluctance" of the string includes the permeability and the diameter on an equal footing, but experiments confirm McDonald's conclusion that the permeability becomes unimportant when it is reasonably large but the string diameter always matters.
Arthur
|
|
|
|
Post by ms on Feb 8, 2019 9:57:14 GMT -5
Yes, looking more closely, the "passing reluctance" of the string includes the permeability and the diameter on an equal footing, but experiments confirm McDonald's conclusion that the permeability becomes unimportant when it is reasonably large but the string diameter always matters.
Arthur
It is well known that increasing permeability material above moderate is unimportant in magnetic circuits that are open, involving short pieces, rather than closed loops (of permeable material) possibly with small gaps. For example, that is why the inductance of a pickup coil increases just a few times with a core rather than air. It is hard to believe that any professional in the field does not know this. It is not worth spending any more time discussing that paper; it is just a distraction from the topic here. The law of magnetic induction is the proper tool to use for analyzing pickups. |
|
|
|
Post by ms on Feb 24, 2019 7:00:55 GMT -5
Getting good hum cancelation in a sidewinder in the direction along the pole pieces, that is perpendicular to the strings can be achieved by extending the pole pieces away from the string so that there is symmetry above and below the core piece that passes through the coils. This can be seen in some designs where a magnet is used and has this extension below. However, this method has the disadvantage of allowing some of the flux from the vibrating strings to escape out the bottom, following a longer path back to the strings. Another way us to extend the ferrite back up towards the strings at the ends of the cols away from the pole pieces. This also has the advantage of shortening the effective path, increasing inductance and sensitivity for a given number of turns. I did this modification to the prototype with available materials, making an awkward structure that can be measured and also installed in a humbucker cutout. The inductance of each coil increased to 4.8 H. With the two in series in phase, it is 14 H. Wired anti-phase as a sidewinder it is about 5.5 H. This is not what I am after, and so I put the two coils in parallel anti-phase. The inductance is about 1.4 H. This is equivalent to an underwound single col ;pickup. It sounds like an extremely bright single coil, one where you keep the tone control at about 7 using an audio taper 250K pot. The hum is very low.   So this seems like the right way to make a P-90 sized pickup with Fender single coil sound, or rather potentially even brighter sound. I will get parts for the coils and cores that I need. The output level is good; a strummed chord can be as much as about 130 mv, with peak-to-peak a bit over 1 volt. I use 3/16" diameter, 1/32" thick neo magnets on top of the pole pieces. the pickup is set about 3 mm below the strings in the neck position. |
|
