Mitigating eddy current losses in pickup covers, and other metal parts

[Antigua]

An electrical engineer and guitar hobbyist named Ken Willmott has done some experimentation to figure out what geometries cause more or less eddy current damping in guitar pickup covers. It's very informative, and he suggests some specific modifications that can be made to existing covers in order to greatly reduce eddy currents without making a radical change in their appearance.

Here is a PDF he prepared http://kenwillmott.com/blog/wp-conte...r_Geometry.pdf

The tl;dr is that, because eddy currents require continuous metal wherein a current can move in a circular pattern, if you cut the metal and break continuity wherever current could flow in a circle, on a plane that is parallel to the coil, you greatly reduce eddy current damping.

Ken Willmott has a blog here kenwillmott.com | My projects, ideas and stories

I've seen it discussed here that a laminated cover or core would reduce eddy losses, but it appears that the audio spectrum is of low enough frequencies that it's not necessary to have a cover or core material finely laminated. Simply cutting a pole piece radially, for example, might eliminate the majority of eddy current losses that it would otherwise cause, while still performing the function of increasing inductance and bringing flux to the guitar strings.

[Jason Rodgers]

So, if the cuts to the top of a brass cover make it essentially audibly transparent, would similar cuts have the same effect on a brass baseplate (which are maligned for killing high frequencies)?

[Antigua]

In theory, yes, in practice, no. The majority of eddy current losses are caused by the flux change that is represented by the moving guitar string. There is a flux change brought about by current moving through the coil, but it's small compared to that of the moving guitar string, so the cover, screws and slugs have a significant impact, but the base plate is too far removed to impact the Q factor or the overall high end audibly.

For example, the greatest losses I've seen are in Filter'trons, and it's been proven that nearly all the losses disappear when the fillister screws are removed. It goes from having no Q factor to speak of, to having a +7 to +10dB rise at the resonance with the screws removed. The metal base isn't very consequential.

I hadn't heard this said about the base plate, but if that's a popular belief, I can do a quick test with a PAF clone, with and without a base plate to see exactly what happens.

[Jason Rodgers]

I hadn't heard this said about the base plate, but if that's a popular belief, I can do a quick test with a PAF clone, with and without a base plate to see exactly what happens.

I tried to find a specific thread to quote, but yes, that seems to be the conventional wisdom around here. I've not done A-B tests myself to observe this, but as I have begun to make my own baseplates, it's a topic that interests me.

[Mike Sulzer]

But which eddy currents affect the frequency response?

First, consider the eddy currents resulting from current flow in the coil. These eddy currents load the coil, and they damp the coil-cap resonance. They have the most effect where the pickup impedance is high and the frequency is high. It takes less energy loss to affect the response where the impedance is high.

Now consider eddy currents induced directly by the motion of the string. These currents rise at the usual 6db per octave, and thus we expect to see gradual differences with frequency with the cover/no cover cases, etc.

Now look at illustration 30 in the pdf file that you linked to. It shows the full range response for a tele neck pickup for the various cases (cover, no cover, slotted cover). The only significant differences are in the resonant region. There are no gradual 6db/octave differences. This is experimental evidence that all significant effects from the eddy currents in the cover are from the damping of the resonance resulting from currents flowing in the coil.

This implies that a base plate also has a significant effect since it is no further from the coil than the cover.

In theory, yes, in practice, no. The majority of eddy current losses are caused by the flux change that is represented by the moving guitar string. There is a flux change brought about by current moving through the coil, but it's small compared to that of the moving guitar string, so the cover, screws and slugs have a significant impact, but the base plate is too far removed to impact the Q factor or the overall high end audibly.

For example, the greatest losses I've seen are in Filter'trons, and it's been proven that nearly all the losses disappear when the fillister screws are removed. It goes from having no Q factor to speak of, to having a +7 to +10dB ride at the resonance with the screws removed. The metal base isn't very consequential.

I hadn't heard this said about the base plate, but if that's a popular belief, I can do a quick test with a PAF clone, with and without a base plate to see exactly what happens.

[Joe Gwinn]

An electrical engineer and guitar hobbyist named Ken Willmott has done some experimentation to figure out what geometries cause more or less eddy current damping in guitar pickup covers. It's very informative, and he suggests some specific modifications that can be made to existing covers in order to greatly reduce eddy currents without making a radical change in their appearance.

Here is a PDF he prepared http://kenwillmott.com/blog/wp-conte...r_Geometry.pdf

The [idea] is that, because eddy currents require continuous metal wherein a current can move in a circular pattern, if you cut the metal and break continuity wherever current could flow in a circle, on a plane that is parallel to the coil, you greatly reduce eddy current damping.

Ken Willmott has a blog here kenwillmott.com | My projects, ideas and stories

Ken W's analysis is correct so far as I can see.

What's missing is that one can make covers out of metals that don't support significant eddy currents at audio frequencies. The classic example is nickel silver. Non-magnetic stainless steel also works. The issue is to comper materials by their skin depth at say 1,000 Hertz. Both magnetic permeability and electrical resistivity of the bulk metal matter.

There were a number of threads on this, so some searching is in order. I recall publishing some tables.

I've seen it discussed here that a laminated cover or core would reduce eddy losses, but it appears that the audio spectrum is of low enough frequencies that it's not necessary to have a cover or core material finely laminated. Simply cutting a pole piece radially, for example, might eliminate the majority of eddy current losses that it would otherwise cause, while still performing the function of increasing inductance and bringing flux to the guitar strings.

Laminating the cover will not work - orientation is wrong. Each lamination will have eddy currents circulating, so the effect of laminating is nil. If the magnetic lines are perpendicular to the plane of the laminations, there will be full eddy currents. If perpendicular, no eddy currents.

The cover (and baseplate) slits will work, as shown by Ken W.

Laminating a blade core does work, although there is an acoustic optimum, to be determined using golden ears. Too few too thick and the sound is too dull. Too many too thin and the sound is too sharp. Three thin soft steel laminations seems to be the answer, based on what I've seen in ads.

One can visualize the path the eddy currents will take by regarding their path as a blurred image of the pickup coil in all nearby pieces of metal. If you can only see the coil reflection edge-on, the eddy currents are reduced. I did a thread on this some time ago, using heat maps of induction-heated steel objects as the example.

The baseplate can definitely have an effect, if it's close enough. If it's made of stainless stell, there will be little effect.

[Antigua]

Nickel silver causes eddy losses, they just aren't as steep as brass. As a matter of fact, just last night I was comparing a nickel silver cover from Tonerider with a brass Seymour Duncan cover, and this is how it broke down:



And here's a pic of me doing this



Ken's modifications take the losses down to a fraction of a decibel, with either metal.


As far as laminations making for a harsh sound, the Q can be damped with parallel resistance, so my thinking is start with a high Q and then damped in to taste. From a pickup maker's perspective, it seems to be heretical to hardwire a resistor or a cap to a pickup, so I can see why they might want to leave the losses as is.

[Antigua]

But which eddy currents affect the frequency response?

First, consider the eddy currents resulting from current flow in the coil. These eddy currents load the coil, and they damp the coil-cap resonance. They have the most effect where the pickup impedance is high and the frequency is high. It takes less energy loss to affect the response where the impedance is high.

Now consider eddy currents induced directly by the motion of the string. These currents rise at the usual 6db per octave, and thus we expect to see gradual differences with frequency with the cover/no cover cases, etc.

Now look at illustration 30 in the pdf file that you linked to. It shows the full range response for a tele neck pickup for the various cases (cover, no cover, slotted cover). The only significant differences are in the resonant region. There are no gradual 6db/octave differences. This is experimental evidence that all significant effects from the eddy currents in the cover are from the damping of the resonance resulting from currents flowing in the coil.

This implies that a base plate also has a significant effect since it is no further from the coil than the cover.

I'm willing to do a with and without base plate test, if that's helpful in coming to a conclusion. The closest thing I had done before was to compare a brass base plate of a Chinese humbucker to a DiMarzio 36th Anniversary's steel or nickel silver base plate, whatever it is they used. The difference in that case was minimal, less than 1dB.

[Antigua]

I just did a with/without base plate test, here is the result:



It looks like there is a potential for about half a decibel improvement in the resonance if you were to make the base plate completely transparent to eddy currents.

They say that one decibel is the smallest difference in amplitude that a person can discern, so if a difference is seen that doesn't exceed 1dB, there's cause to conclude that it's inaudible by itself, though several of these small factors could combine together to total more than one decibel of difference, so if you start out with a cheap, thick copper cover, as opposed to a better nickel silver base plate, and then find a way to totally eliminate eddy current damping, that combination could equal one decibel. But even then, it's hovering on the threshold of human hearing.

Here's the test setup:

[Mike Sulzer]

Ken W's analysis is correct so far as I can see.

Yes, his "slot solutions" are right and very useful, but the physical cause of the losses is not "signal attenuation", but rather damping of the resonance in the pickup electoral circuit. I have discussed the effect in illustration 30, but you also can see this in others, such as 26. Note that the unmodified brass cover wipes out the resonance nearly completely. The full slotted cover has a bit more than 1db at resonance, but the effect is almost entirely gone outside the neighborhood of the resonance, more than can be explained by a 6db/octave effect.

[Mike Sulzer]

There are some things still to understand here. According to this table: https://www.nde-ed.org/GeneralResour...ivity_Misc.pdf, brass is more than 5 times as conductive as nickel silver. There are variations amongst specific alloys, so who knows for sure, but based just on conductivity, I would expect the difference between NS and brass on the plot to be much more. So I am not convinced we understand what is really happening.

Nickel silver causes eddy losses, they just aren't as steep as brass. As a matter of fact, just last night I was comparing a nickel silver cover from Tonerider with a brass Seymour Duncan cover, and this is how it broke down:

[Mike Sulzer]

Thanks. I do not understand these results, but I wonder if the distance from the exciting coil is so important, then is its geometry also important? That is, how different would the results be with a tiny exciting coil which would make a field more like that from a string? Also I think it would be interesting to try this with pickup coils with fewer turns which would put the resonance well up above the audio range. Then it might be easier to see what the effects of eddy currents as a function of frequency are. That is, try to make the circuit effects simpler.

I just did a with/without base plate test, here is the result:

[Mike Sulzer]

Let's look at a further case of interest in regard to the loss by damping the pickup resonance. This is a case when we would expect that there could be no resonant damping, and there is not. Look at Illustration 7. Take the case with the loop below the cover, and consider it the reference. (It is not a perfect measure of "no effect", but it is close.) Now compare this to the case with the loop at the exciter coil. Notice that at about 10 KHz, this is about 2db below the reference. At 5KHz it is still about two db below the reference. Thus in this case the pickup, to a reasonable degree of accuracy, is just passing on what is happening to the coil. This is very different to what happens when the loop or cover is at the pickup; in those cases the loss is selectively higher at the resonance.

By the way, it does not look as though the loops and covers are exactly the same in what they do. Possibly the skin effect is a small factor in what the loops do, but not the covers, which are a thinner flat conductor. (However, it is very useful to have both!)

[Antigua]

There are some things still to understand here. According to this table: https://www.nde-ed.org/GeneralResour...ivity_Misc.pdf, brass is more than 5 times as conductive as nickel silver. There are variations amongst specific alloys, so who knows for sure, but based just on conductivity, I would expect the difference between NS and brass on the plot to be much more. So I am not convinced we understand what is really happening.

You're probably right, it's probably not brass. I think I saw a copper undercoat and was confused. It might even be differences in the thickness that caused the difference.

[Antigua]

Thanks. I do not understand these results, but I wonder if the distance from the exciting coil is so important, then is its geometry also important? That is, how different would the results be with a tiny exciting coil which would make a field more like that from a string? Also I think it would be interesting to try this with pickup coils with fewer turns which would put the resonance well up above the audio range. Then it might be easier to see what the effects of eddy currents as a function of frequency are. That is, try to make the circuit effects simpler.

That would take a little more work, so I'd have to test this at a later time, but I have done other tests to see if proximity between the exciter coil and pickup makes a difference, and I had found that it didn't.

For this test, the exciter coil had 1 Vpp driving it, and I've found that increasing the voltage up to 5 Vpp doesn't generally change the response curves. Perhaps the driver coil can create a stronger flux field that is able to reach and interact with the further ends of the pickup, but the flux is weak enough at that point that those effects are very small, and disappear beneath the more immediate and powerful effects caused by the more proximate cover, poles and screws. It seems that proportionately, between weaker strings or a stronger exciter coil, scale rather linearly. In fact, that's easy to demonstrate, so I will do so. If the exciter coil was causing an added distortion, if anything the 0.4dB difference would likely be even smaller.

A little array of tiny coils that resemble guitar string would be nice to have, I'll have to think about how to construct that.