|
Post by antigua on May 6, 2020 14:11:58 GMT -5
Someone on Strat Talk was asking about the effect of pole piece beveling on the magnetic field. I've been messing with FEMM trying to get more comfortable with it, and made a quick a dirty demo of without/without beveling. As one would expect, the difference is negligible. There's also been some recent conversation about what sort of magnet the guitar string becomes, and the conclusion was that it's like two long magnets, perpendicular to the axis of the coil and pole pieces, and if you look carefully, this FEMM model appears to reflect that, with the flux lines running very parallel to the length of the string. Even directly above the pole piece, it appears the lines of flux immediately turn left or right.
|
|
|
Post by ms on May 6, 2020 15:20:40 GMT -5
This appears to agree with what Aquin43 said. Is this the "recangular" mode of FEMM? That is, where the pole ;piece an d string actually go forever into and out of the page?
|
|
|
Post by antigua on May 6, 2020 15:36:48 GMT -5
This appears to agree with what Aquin43 said. Is this the "recangular" mode of FEMM? That is, where the pole ;piece an d string actually go forever into and out of the page? It's the default settings. I barely know how to use the software yet. The UI is so crude that simple tasks take a lot of time to carry out. I managed to make some rectangles, assign them materials and then run the simulation. Do you think the results would be substantially different if it was modeled in 3D? Can't it at least be said that this model is true for the immediate cross section of the pole piece and string? Regarding the string magnetism testing, I have a test in mind, I want to position near identical AlNiCo poles above and beside the pickup's AlNiCo pole piece, and see if there is a configuration which induces a magnetic null in the guitar string, and make note of what geometries it takes for that to happen.
|
|
|
Post by ms on May 6, 2020 15:46:33 GMT -5
This appears to agree with what Aquin43 said. Is this the "recangular" mode of FEMM? That is, where the pole ;piece an d string actually go forever into and out of the page? It's the default settings. I barely know how to use the software yet. The UI is so crude that simple tasks take a lot of time to carry out. I managed to make some rectangles, assign them materials and then run the simulation. Do you think the results would be substantially different if it was modeled in 3D? Can't it at least be said that this model is true for the immediate cross section of the pole piece and string? Regarding the string magnetism testing, I have a test in mind, I want to position near identical AlNiCo poles above and beside the pickup's AlNiCo pole piece, and see if there is a configuration which induces a magnetic null in the guitar string, and make note of what geometries it takes for that to happen. Remember, it is a two dimensional program, and the default settings are what I described in my previous post. In effect, the "string" is a plane that is acting as a magnetic shield. Nonetheless, something similar would happen with an actual 3D representation and program, but the details are unclear.
|
|
|
Post by antigua on May 6, 2020 15:58:38 GMT -5
It's the default settings. I barely know how to use the software yet. The UI is so crude that simple tasks take a lot of time to carry out. I managed to make some rectangles, assign them materials and then run the simulation. Do you think the results would be substantially different if it was modeled in 3D? Can't it at least be said that this model is true for the immediate cross section of the pole piece and string? Regarding the string magnetism testing, I have a test in mind, I want to position near identical AlNiCo poles above and beside the pickup's AlNiCo pole piece, and see if there is a configuration which induces a magnetic null in the guitar string, and make note of what geometries it takes for that to happen. Remember, it is a two dimensional program, and the default settings are what I described in my previous post. In effect, the "string" is a plane that is acting as a magnetic shield. Nonetheless, something similar would happen with an actual 3D representation and program, but the details are unclear. The string is a "magnetic shield", though, right? But maybe only with respect to whatever is directly opposite the string. If the steel were a thing membrane, like a stretched out steel ribbon, I think everything would be the same, with one less axis of movement. Since the static field and the changing field are different concerns, where only the latter results in voltage being produced, and all of the movement originates at the guitar string, would it correct and/or informative as a model, to have the guitar strings be as two long magnets over a permeable pole piece?
|
|
|
Post by antigua on May 6, 2020 16:27:06 GMT -5
I tried it for fun; AlNiCo "strings" over a steel pole piece. The main difference appears to be that the lines of flux, in the area where the coil would be, are more perpendicular to the pole piece, not as parallel with it.
|
|
|
Post by aquin43 on May 7, 2020 10:29:50 GMT -5
An observation: in 3D space, the flux from a pole falls off according to the inverse square of distance. In 2D space, it falls off only with the inverse of the distance. This must surely affect the flux pattern in the modelling close to a pole.
Arthur
|
|
|
Post by antigua on May 7, 2020 11:55:04 GMT -5
An observation: in 3D space, the flux from a pole falls off according to the inverse square of distance. In 2D space, it falls off only with the inverse of the distance. This must shurely affect the flux pattern in the modelling close to a pole.
Arthur
I'm not quite understanding; isn't 2D just a cross section of what might a 3D object?
|
|
|
Post by aquin43 on May 7, 2020 12:50:48 GMT -5
An observation: in 3D space, the flux from a pole falls off according to the inverse square of distance. In 2D space, it falls off only with the inverse of the distance. This must shurely affect the flux pattern in the modelling close to a pole.
Arthur
I'm not quite understanding; isn't 2D just a cross section of what might a 3D object? Unfortunately it is not. The lack of the third dimension changes the way the flux diverges since it only has two dimensions to diverge into. The third dimension is not just ignored as in making a cross section - it simply does not exist.
Arthur
|
|
|
Post by antigua on May 7, 2020 15:40:48 GMT -5
I'm not quite understanding; isn't 2D just a cross section of what might a 3D object? Unfortunately it is not. The lack of the third dimension changes the way the flux diverges since it only has two dimensions to diverge into. The third dimension is not just ignored as in making a cross section - it simply does not exist.
Arthur
I'll have to research the matter further, but some questions come to mind; why then is any 2D modeling of value of real objects are 3D? What sort of difference to do expect to see if this were a genuine 3D cross section? You noted that the flux runs lengthwise along the string and it seems that at the very least the 2D model validated your statement. Thanks for providing these insights as well as the sophisticated SPICE model.
|
|
|
Post by aquin43 on May 7, 2020 16:05:07 GMT -5
Unfortunately it is not. The lack of the third dimension changes the way the flux diverges since it only has two dimensions to diverge into. The third dimension is not just ignored as in making a cross section - it simply does not exist.
Arthur
I'll have to research the matter further, but some questions come to mind; why then is any 2D modeling of value of real objects are 3D? What sort of difference to do expect to see if this were a genuine 3D cross section? You noted that the flux runs lengthwise along the string and it seems that at the very least the 2D model validated your statement. Thanks for providing these insights as well as the sophisticated SPICE model. The two dimensional behaviour is valid in many cases in ferromagnetic structures where there is little variation of flux in the third dimension. In these cases it does behave like a cross section of the 3D object. The flux distribution above a rail type pickup can be described by a 2D cross section, for example. Any slice perpendicular to the rail and reasonably far from the ends will have as much flux entering as leaving from adjacent slices in the 3rd dimension, so the behaviour will be two dimensional.
The 2D string and pole model is like a cross section of a ribbon over a rail. The flux distribution is not the same in detail as a string over a pole, but the result is similar in form to what is observed in a real 3D system, so it provides an interesting insight into the general shape of the field.
Arthur
|
|
|
Post by antigua on May 7, 2020 16:44:17 GMT -5
I'll have to research the matter further, but some questions come to mind; why then is any 2D modeling of value of real objects are 3D? What sort of difference to do expect to see if this were a genuine 3D cross section? You noted that the flux runs lengthwise along the string and it seems that at the very least the 2D model validated your statement. Thanks for providing these insights as well as the sophisticated SPICE model. The two dimensional behaviour is valid in many cases in ferromagnetic structures where there is little variation of flux in the third dimension. In these cases it does behave like a cross section of the 3D object. The flux distribution above a rail type pickup can be described by a 2D cross section, for example. Any slice perpendicular to the rail and reasonably far from the ends will have as much flux entering as leaving from adjacent slices in the 3rd dimension, so the behaviour will be two dimensional.
The 2D string and pole model is like a cross section of a ribbon over a rail. The flux distribution is not the same in detail as a string over a pole, but the result is similar in form to what is observed in a real 3D system, so it provides an interesting insight into the general shape of the field.
Arthur
This is about what I would expect. The main reason I'm interested in this is because of conversation that revolves around "aperture" or "sensing width", and so I suppose, I would want to know if the 2D model creates any false impressions with respect to magnetic aperture.
|
|
|
Post by aquin43 on May 8, 2020 3:10:12 GMT -5
The two dimensional behaviour is valid in many cases in ferromagnetic structures where there is little variation of flux in the third dimension. In these cases it does behave like a cross section of the 3D object. The flux distribution above a rail type pickup can be described by a 2D cross section, for example. Any slice perpendicular to the rail and reasonably far from the ends will have as much flux entering as leaving from adjacent slices in the 3rd dimension, so the behaviour will be two dimensional.
The 2D string and pole model is like a cross section of a ribbon over a rail. The flux distribution is not the same in detail as a string over a pole, but the result is similar in form to what is observed in a real 3D system, so it provides an interesting insight into the general shape of the field.
Arthur
This is about what I would expect. The main reason I'm interested in this is because of conversation that revolves around "aperture" or "sensing width", and so I suppose, I would want to know if the 2D model creates any false impressions with respect to magnetic aperture. Given that the rate of change of flux with distance is one dimension slower in the 2D system ( 1/r for poles and 1/r^2 for dipoles) one would expect that the apparent aperture would be larger than in a real system.
I tried modelling the string in FEMM using an array of small magnets which produced the same internal flux density as in the real string, based on the simulations and the measurements which have been carried out by POTEG.
This allows a 2D representation of the flux in the coil, but I couldn't work out how one calculates the aperture from that.
POTEG measurements set an upper limit to the sensing aperture of a pole and show it to be very narrow, hardly larger than the pole itself and quite negligible in terms of the frequency response of the pickup. The aperture of the pickup seems to become important when the pickup contains more than one sensing pole, such as a humbucker or one of the single coils on a flanged backplate.
By the way, If you want to do much in this line with FEMM it is worth looking at OCTAVE FEMM, which is an interface to FEMM from the Octave programming language. It is a lot to learn but in the end it can make things so much easier.
Arthur
Arthur
|
|
|
Post by ms on May 8, 2020 7:04:47 GMT -5
An observation: in 3D space, the flux from a pole falls off according to the inverse square of distance. In 2D space, it falls off only with the inverse of the distance. This must shurely affect the flux pattern in the modelling close to a pole.
Arthur
I'm not quite understanding; isn't 2D just a cross section of what might a 3D object? I think if helps to consider how the two dimensional drawing such as yours above corresponds to a 3D case in both geometric modes of FEMM. In the mode with cylindrical symmetry, your drawing consists of a cylindrical pole piece with a large flat disk over it. So you could run the same drawing in that mode and you would get a somewhat different answer.
|
|
|
Post by ms on May 8, 2020 7:32:06 GMT -5
This is about what I would expect. The main reason I'm interested in this is because of conversation that revolves around "aperture" or "sensing width", and so I suppose, I would want to know if the 2D model creates any false impressions with respect to magnetic aperture. Modeling the aperture is hard with a 2D program. I am thinking that a multi step process with various approximations is necessary. For example, start with an Alnico pole piece. It has cylindrical symmetry and so can be done in the cylindrical mode. The result is a field with axial and radial components. From the symmetry in Aquin43's measurement, we know that the axial component does not do anything since it makes no difference when the magnet is placed below or to the side of the string, with respect to the pickup. So now think of putting a string above this pole piece and plot out the radial component of the field along the string. Now make a FEMM model (cylindrical symmetry) of the string and somehow construct a field (using coils and magnets or whatever) that has this field along the axis of the string. Solve the problem and get a field that is due to that and the magnetization induced in the string. Then use this information to find the magnetization along the string. At each point along the string, the local magnetization induces a field in the coil. (A small bit of magnetization makes a dipole field and so you can figure out how much each bit contributes relative to the bit in the center.) If we think of the magnetization as lots of atomic dipoles (rather than atomic size current loops) a simple derivative of the field at a point along the string determines the dipole intensity. Since the field changes very quickly where the pole piece would be if it were in the model, we get a large contribution there, which falls off quickly with distance along the string.
|
|
|
Post by aquin43 on May 8, 2020 8:43:27 GMT -5
This is about what I would expect. The main reason I'm interested in this is because of conversation that revolves around "aperture" or "sensing width", and so I suppose, I would want to know if the 2D model creates any false impressions with respect to magnetic aperture. Modeling the aperture is hard with a 2D program. I am thinking that a multi step process with various approximations is necessary. For example, start with an Alnico pole piece. It has cylindrical symmetry and so can be done in the cylindrical mode. The result is a field with axial and radial components. From the symmetry in Aquin43's measurement, we know that the axial component does not do anything since it makes no difference when the magnet is placed below or to the side of the string, with respect to the pickup. So now think of putting a string above this pole piece and plot out the radial component of the field along the string. Now make a FEMM model (cylindrical symmetry) of the string and somehow construct a field (using coils and magnets or whatever) that has this field along the axis of the string. Solve the problem and get a field that is due to that and the magnetization induced in the string. Then use this information to find the magnetization along the string. At each point along the string, the local magnetization induces a field in the coil. (A small bit of magnetization makes a dipole field and so you can figure out how much each bit contributes relative to the bit in the center.) If we think of the magnetization as lots of atomic dipoles (rather than atomic size current loops) a simple derivative of the field at a point along the string determines the dipole intensity. Since the field changes very quickly where the pole piece would be if it were in the model, we get a large contribution there, which falls off quickly with distance along the string. I already have the axisymmetric model of the string, constructed from a double array of short magnets whose magnetisations are chosen by inspection to match as near as possible shape of the POTEG measurements. This allows the flux at any 3D point in air to be calculated. It should be possible to estimate the aperture by setting all except one of the magnets in turn to zero magnetisation and plotting the change in the field at a representative point. I chose to make the string permeability reasonably high in order to smooth out the step transitions between the magnets. I am not sure whether the result would be correct with an individual magnet and a permeability above one in the rest. Perhaps it would, since the system is linear.
Extending this to include a pole piece would seem to be impossible in a 2D system.
Arthur
|
|
|
Post by antigua on May 9, 2020 2:12:39 GMT -5
This is only half on topic, but close enough. I tried experimenting with the placing an another almost identical AlNico 5 pole piece over the string and pole piece of a pickup in situ, and I found that it would only cause a strong cancellation of the fluxes if the other pole piece was magnetically aligned with the pole in the guitar, (both north up or south up), and this fits with the theory, because even though the flux in the dead center of the guitar string is of a common axis as the magnets, the return paths that spread wayward from both pole pieces created a null of equal(ish) positive and negative flux in the same points of space, also where the guitar string happens to be. But what is confusing to me is that I had to have the pole piece very close to the string, a lot closer than are the pickup's pole pieces. To get a strong cancellation, I had to have the external pole piece about 1mm away from the string, where as the pole piece in the pickup is about 4mm away from the string. I would have expected the cancellation to be strong even if the other pole piece was also 4mm away from the string. Also, the cancellation is never complete, it sounds like it's about 20% of it's full output at the maximum of cancellation. Orienting the external pole piece to be upside down, so that its like-on-like polarity, the output increased somewhat, maybe inversely to what had cancelled with the pole piece right side up. Note in this picture the pole pieces are facing the same direction polarity-wise. The external pole piece's bevel makes it look like it's opposing orientation, but magnetically it isn't. Also I had forgotten a test I did a long time ago to get a feel for aperture with, and I just did it again, which is holding a pickup external of the guitar in hand, hooked up to a little amp, and seeing how loudly it reads the string when held parallel to the guitar strings instead of perpendicular. I was surprised (because I had forgotten since the last time) that the pickup received a lot of sound from the moving guitar string, even when the string is not aligned over the pole pieces. In fact, I'd say it was about 50% loudness when the string was parallel to the edge of the Strat pickup in hand. It just goes to show that the aperture is a strong gradient, and that trying to mentally model it as a discrete width probably leads to erroneous results. In reality, it's some % of voltage reduction with respect to distance from center, and it's obviously not a linear gradient, either.
|
|
|
Post by ms on May 9, 2020 5:52:46 GMT -5
Maybe the other pole pieces affect the observation. Also Alnico is a permeable material, and different types have different permeabilities.
|
|
|
Post by antigua on May 9, 2020 14:42:53 GMT -5
Maybe the other pole pieces affect the observation. Also Alnico is a permeable material, and different types have different permeabilities. That's a good point about the other pole pieces bringing in fields that aren't being negated, I'll have to think of a way to get around that without without taking any pickups apart. I noticed the effect was strongest with the high E, which has one neighbor, a bit less with the B, G and D, and a lot harder to induce in the A and low E, which tended to do the "warbling" thing with a magnet being really close to them. All the pole pieces in question are AlNiCo 5.
|
|
|
Post by pablogilberto on Jul 22, 2020 21:20:02 GMT -5
Hi antiguaExcellent work on this. Do you have a post on how to do these simulations? I'm interested in trying this as well. Thanks a lot!
|
|
|
Post by antigua on Jul 24, 2020 2:30:26 GMT -5
Hi antigua Excellent work on this. Do you have a post on how to do these simulations? I'm interested in trying this as well. Thanks a lot! The FEMM simulations? I just watched a video and read some documents to do what little I did. It's extremely hard to use and I'm not sure how informative the modelling is for pickup modelling, but I think it shows the relative unimportance of whether or not the magnet is beveled. People seem to want to attribute a lot of the difference to that which they can see with their eyes, and discount things which are hidden or completely invisible. I think if the pickups were entirely hidden beneath the pick guard, there would be a lot less interest in them.
|
|