Pickup Testing

[R.G.]

This issue came up in my tester proposal threads.

What do you test on a pickup to characterize it and why?

There have been some obvious things proposed in this forum's history:
(a) number of turns
(b) Wire resistance
(c) Inductance
(d) Self capacitance

as well as some things I didn't see:
(e) Self resonant frequency
(f) Frequency response (and how it's done)
(g) Impulse response

My logic in this is that
(1) there is something about a pickup's response that affects it's "tone" for lack of a better word
(2) this something, or somethings, is measurable if you know what and how to measure
(3) if we can't measure things that correlate with a pickup's sound, then we're not doing the correct measurements.

So - what do we measure? And how?

I'd be happy to propose test rigs when we know what to measure, and I'm sure other people would too.

As an example, (g) Impulse response. You can pretty much draw out a speaker's steady state response by whipping an impulse of current into it and then recording the response to the impulse. A little digital filtering math turns the impulse response into a steady-state sinewave response graph.

One could, for instance, use a single turn driver coil with X current running through it, and then interrupt the current in the driver coil. This will decay very, very quickly and approximate an impulse of MMF into the pickup coil. This gives us the impulse response to measure and do math games on.

[soundmasterg]

The things I can think of that would be useful in measuring would be (in no particular order):

1. DC resistance
2. AC resistance
3. Inductance
4. self capacitance
5. self resonant frequency
6. Turns count
7. Frequency response
8. Magnetic strength
9. Magnetism pattern
10. Q

Number 1 above can be measured with a standard DMM.
Numbers 2, 3, and 10 can be measured with the Extech 380193 meter.
Number 4 can be measured with the Extech also. I don't think it is accurate in that test, but I could be wrong.

I'm not sure what the Impulse response is that you brought up, but it brought to mind another possible measurement. I would guess that based on their various construction aspects and magnetic strengths, each pickup will have a varying response to a disruption of the magnetic field as far as the speed and strength of reaction is concerned. So if you disturb the magnetic field with a string, a P90 may react quicker and produce a stronger response than something like a Strat pickup perhaps? I would think there would be a way to measure the reaction strength and reaction speed, and it may be a useful comparison to make between pickup types and give a further understanding of the differences in magnet type or strength and how that coorelates to the sound of the pickup in question? Or maybe I'm thinking too hard again?

Greg

[R.G.]

2. AC resistance

AC resistance and DC resistance are identical if the decrease in conduction area due to skin effect is ignored. Skin effect does make for a rise in resistance with frequency, but for audio frequencies, it's generally something that can be ignored; it can especially be ignored in fine gauge wires, as the "skin depth" is so large that it goes all the way through the wire. I'm thinking I don't understand what you mean by "AC resistance". If this is just what the Extech reads when you set it to read AC resistance, it's probably an artifiact of the meter trying to read the resistance in a big inductor with distributed capacitance.

3. Inductance
4. self capacitance
5. self resonant frequency

These are the obvious things to look at. A pickup is a distributed capacitance component, with the internal self capacitances separated by resistances. Has anyone looked at whether the measured inductance and measured capacitance produce the measured self resonant frequency?

7. Frequency response

Is typically done with a driver coil?

8. Magnetic strength

Yep.

9. Magnetism pattern

Yep. Hall effect sampling?

10. Q

Yep.

Number 1 above can be measured with a standard DMM.
Numbers 2, 3, and 10 can be measured with the Extech 380193 meter.
Number 4 can be measured with the Extech also. I don't think it is accurate in that test, but I could be wrong.

Yeah, that's the distributed capacitance thing I was mentioning.

I'm not sure what the Impulse response is that you brought up,

If you take a hammer and whack a bell, what happens then is the impulse response of the bell.

If you do the math, it turns out that you can backwards-compute the entire frequency response of the bell from a digitized recording of the impulse response. Some speaker testers have started using impulse response instead of swept tones for measuring the response of speakers. I'd have to do some digging to find out whether it can be done accurately enough at high frequencies to be applicable to pickups or not. I mentioned it because we could do a magnetic "whack" with a one-turn coil pretty easily and then record the voltage out of the pickup. That will be mostly the self resonance, but the other stuff tells the rest of the frequency response.

but it brought to mind another possible measurement. I would guess that based on their various construction aspects and magnetic strengths, each pickup will have a varying response to a disruption of the magnetic field as far as the speed and strength of reaction is concerned. So if you disturb the magnetic field with a string, a P90 may react quicker and produce a stronger response than something like a Strat pickup perhaps? I would think there would be a way to measure the reaction strength and reaction speed, and it may be a useful comparison to make between pickup types and give a further understanding of the differences in magnet type or strength and how that coorelates to the sound of the pickup in question? Or maybe I'm thinking too hard again?
Greg

Not really - you just described part of why an impulse response works. The ideal impulse to use in an impulse response is a one-unit high disturbance of zero time duration. The shorter and closer a real impulse is to a zero-time impulse the more accurate the result.

When oil companies set off explosives in the ground and listen, they are doing an impulse response test on the planet.

If we hit a one-turn driver coil with a 1A pulse just long enough for the current to come to 1A and then turn off the current, it slaps the magnetic field of the pickup with one Ampere-turn of MMF. By recording what happens from then on, you get the speed of response of the magnet/coil structure.

You might want the driver coil oriented in an oval like a string, not over the whole pickup. Don't know.

I have a lot of training in math, physics, and electronics, but very little in pickups, so don't mistake my blathering on about pickups as me thinking I know this stuff.

But a pickup still obeys the laws of physics, however complexly, so I think it's both measurable and learn-able. If we can figure out what to measure that correlates to how they sound, we can then go measure really, really good ones and figure out how to make more of them. Or make even better ones.

[Joe Gwinn]

AC resistance and DC resistance are identical if the decrease in conduction area due to skin effect is ignored. Skin effect does make for a rise in resistance with frequency, but for audio frequencies, it's generally something that can be ignored; it can especially be ignored in fine gauge wires, as the "skin depth" is so large that it goes all the way through the wire. I'm thinking I don't understand what you mean by "AC resistance". If this is just what the Extech reads when you set it to read AC resistance, it's probably an artifiact of the meter trying to read the resistance in a big inductor with distributed capacitance.

No, AC resistance is in fact resistance (being in phase with the drive). The Extech was calibrated against a Maxwell-Wein Impedance Bridge, and gives the same answer, to within experimental error. This was published on the old AMPAGE Forum. See http://home.comcast.net/~joegwinn/ for the details of the bridge.

While it's true that there is negligible eddy current effect in the coil wire, it does not follow that eddy current effects are negligible in a pickup. There is still a lot of metal nearby, some of which is magnetic.

These are the obvious things to look at. A pickup is a distributed capacitance component, with the internal self capacitances separated by resistances. Has anyone looked at whether the measured inductance and measured capacitance produce the measured self resonant frequency?

Yes. They do match, but with a twist: because most pickups have significant amounts of metal nearby, the inductance depends on the frequency.

[R.G.]

No, AC resistance is in fact resistance (being in phase with the drive). The Extech was calibrated against a Maxwell-Wein Impedance Bridge, and gives the same answer, to within experimental error. This was published on the old AMPAGE Forum. See http://home.comcast.net/~joegwinn/ for the details of the bridge.

OK. Did that. Please bear with me, I'm trying to learn.
(a)Does the measured AC resistance match the DC measurement?
(b)What happens to the measured AC resistance as you vary the measurement frequency?
You're right that the real part of the vector impedance will be different from the DC resistance. That should also vary with frequency as well.

While it's true that there is negligible eddy current effect in the coil wire, it does not follow that eddy current effects are negligible in a pickup. There is still a lot of metal nearby, some of which is magnetic.

So you're also measuring eddy current loss as well as skin effect?

Yes. They do match, but with a twist: because most pickups have significant amounts of metal nearby, the inductance depends on the frequency.

That makes sense.

Along the line of testing, what are your suggested tests?

[soundmasterg]

OK. Did that. Please bear with me, I'm trying to learn.
(a)Does the measured AC resistance match the DC measurement?
(b)What happens to the measured AC resistance as you vary the measurement frequency?
You're right that the real part of the vector impedance will be different from the DC resistance. That should also vary with frequency as well.

Here's a couple examples of some Extech measurements, and a DCR measurement from a true RMS B & K meter. The Extech measures at 120Hz and at 1000Hz for the AC resistance, inductance, and Q.


Pickup 1 = Wolfetone custom neck pickup for a Strat. Wound on his old drill press winder, hand guided and tensioned, scatter wound using polysol 42 gauge wire.


DCR = 5.76k

@120Hz

ACR = 5.97k
Inductance = 2.236 Henries
Q= .2891

@1000Hz

ACR= 7.39k
Inductance= 2.212 Henries
Q= 1.882

Pickup # 2 = Rickenbacker HB2 humbucker using samarium cobalt magnets, epoxy potted, using 44 gauge wire, machine wound.


DCR = 14.45k

@120Hz

ACR = 14.55k
Inductance = 7.053 Henries
Q= .3724

@1000Hz

ACR= 19.07k
Inductance= 6.869 Henries
Q= 2.193

Pickup # 3 = 70's Harmony single coil pickups rewound by TV Jones in mid 90's. Pickups are arranged like a P90 in a way, but use more metal, and ceramic magnets on sides, very odd design and not sure how it works. Hums just as much or more as P90 design. Used 44 gauge wire, don't recall which type.


DCR = 6.91k

@120Hz

ACR = 7.079k
Inductance = 3.193 Henries
Q= .3467

@1000Hz

ACR= 12.53k
Inductance= 2.494 Henries
Q= 1.223

Possum was telling me a bunch of stuff about how the AC resistance reading is used, and if I recall right, it can be used to spot defective winds on a given coil, but the number is only useful within it's own pickup family. Every pickup design will have a different number range that is useful for that measurement. You can see that at 120Hz the DCR and ACR are close, but the spread widens as the frequency goes up. I don't know if that continues as the Extech can only measure at those two frequencies. It would be really cool to find something that could measure accurately higher up as 1000Hz is still pretty low on a guitar, and 120Hz is almost at the bottom range of what a guitar can reproduce. I don't know of anything that can do higher frequency measurements though.

Greg

[Joe Gwinn]

OK. Did that. Please bear with me, I'm trying to learn.
(a)Does the measured AC resistance match the DC measurement?

Yes, but only at very low frequencies. The total AC resistance equals the DC resistance plus the eddy-current load (expressed as a resistance).

(b)What happens to the measured AC resistance as you vary the measurement frequency?
You're right that the real part of the vector impedance will be different from the DC resistance. That should also vary with frequency as well.

The eddy-current load increases with the square root of frequency.

So you're also measuring eddy current loss as well as skin effect?

Yes, but the two effects are intertwined. Skin effect causes the current to crowd near the surface, raising the resistance to flow seen by that current.

If the current is in a wire, travelling close to the surface, one will see the AC resistance of the wire increasing. However, with #42 wire, this is not the cause of the increase in AC resistance seen in pickups. The wire is far too thin for this effect to be important at audio frequencies.

If the current is an induced eddy in a hunk of metal, skin effect will keep the eddy close to the surface of the hunk, and somewhat change the details, but a finite current is being pushed through a finite resistance, which costs energy, and by the law of conservation of energy somebody has to pay. There is only one source of energy available, and that's the coil generating (when driven by the LCR meter) the time-varying field that's causing the eddy currents. So, the coil sees an added AC resistance. And it is a resistance (not a reactance) because real energy is being dissipated in the hunk. In high-powered equipment, such as induction stovetops, the hunk of metal will get very hot.

To take extremes, if the hunk of metal were superconducting, with zero resistivity, the eddy currents would be confined to the surface, but there would be no energy loss and thus no increase in AC resistance. By the same token, if the "metal" were plastic, with infinite resistivity, no currents would flow, and there would be no eddy current loading.

Along the line of testing, what are your suggested tests?

When I first started thinking about pickups, I was trying to measure everything to 1% or better, and got nowhere. Things just aren't that precise in pickups, the inductance actually varies significantly with frequency and with string tuning (which varies the DC magnetic bias and thus moves one around on the permeability curves of the materials), and the Q of pickup inductors is in any event quite low. What I ended up with is Rdc, Rac(1KHz), L(1KHz), resonant frequency, and self-capacitance.

The sequence is thus:

1. Measure Rdc and Rac(1KHz).

2. Measure L(1KHz).

3. Measure resonant frequency, both alone (zero added capacitance), and with three or so different mica or film (not ceramic) capacitors in parallel.

This information will allow one to deduce the self-capacitance, which cannot be measured directly. When you do the computations, especially if you graph the datapoints (capacitances), things won't quite line up. This is due to the inductance changing with frequency (which is due to the eddy currents). One could take more points and also estimate the change in inductance near the resonances, but I never bothered.

[SteikBacon]

I posted this before. http://online.physics.uiuc.edu/cours...asurements.pdf

What do people here think of this way of testing?

[R.G.]

@Joe -
OK, that makes sense. The energy in eddy current losses comes from somewhere, and it's got to be energy of the moving string.

When I first started thinking about pickups, I was trying to measure everything to 1% or better, and got nowhere. Things just aren't that precise in pickups, the inductance actually varies significantly with frequency and with string tuning (which varies the DC magnetic bias and thus moves one around on the permeability curves of the materials), and the Q of pickup inductors is in any event quite low. What I ended up with is Rdc, Rac(1KHz), L(1KHz), resonant frequency, and self-capacitance.

The sequence is thus:
1. Measure Rdc and Rac(1KHz).
2. Measure L(1KHz).
3. Measure resonant frequency, both alone (zero added capacitance), and with three or so different mica or film (not ceramic) capacitors in parallel.

This information will allow one to deduce the self-capacitance, which cannot be measured directly. When you do the computations, especially if you graph the datapoints (capacitances), things won't quite line up. This is due to the inductance changing with frequency (which is due to the eddy currents). One could take more points and also estimate the change in inductance near the resonances, but I never bothered.

That makes sense as well, and presumably one could make use of the measured stuff to semi-model a frequency response.

How does one then relate the derived model of the pickup back to freuquency response from the pickup with a real string? What I was after with the moving-string model was to mechanically operate the string in a repeatable way, acquire the response info in a repeatable way, then use the method for comparing one pickup to another based on actual string movement.

There seems to be a gap in the middle of pickup winding. On one hand we have number of turns, wire thickness, spacing, coil dimensions and magnetic properties. Those seem to relate to the resulting sound only by human anecdotal experience. On the other hand, we have electrical measurements: inductance, resonance, DC and AC resistance, capacitance, etc., which also seem to relate to sound only anecdotally.

Neither path seems to tell us repeatably "there's a hump in the midrange and a little dip at 4kHz which will make this sound less present that this one where there is no dip, but less output level for the same string motion." It seems like this kind of data would be very useful because one could then relate both the mechanico-winding data and the electrical measurements to what signal you get out.

Or am I thinking too hard again?

[David Schwab]

(a)Does the measured AC resistance match the DC measurement?

AC resistance increases with frequency. Another way to look at it is AC resistance is the impedance. Because of impedance, high frequencies need more energy to get out of the coil, which is one reason ceramic magnets sound brighter.

So as Joe pointed out, only at very low frequencies.

This is a very interesting thread!

[R.G.]

AC resistance increases with frequency. Another way to look at it is AC resistance is the impedance. Because of impedance, high frequencies need more energy to get out of the coil, which is one reason ceramic magnets sound brighter.

So as Joe pointed out, only at very low frequencies.

This is a very interesting thread!

Yeah, it increases with frequency because the power dissipated in eddy currents increases with frequency as do all magnetic materials driven by a changing field.

The impedance is more that that - it's the sum total of the real part (resistance) plus the complex impedance generated by the reactances, the (nonlinear) inductance and the distributed capacitances.

I still have to do some more thinking about what happens if you buffer a pickup coil so essentially no current flows. There are still losses, from the string dragging the M-field around, but since no current can flow in the pickup winding, I'm not sure what effect that has on the signal out of the coil. It may only cause the secondary effect of damping string motion but not affect the coil voltage in a primary fashion.

In either case, what I'm after is that bridge I was talking about - how to repeatably measure things that are direct indicators of a pickup's performance so that direct cross-correlation between pickups can be made, not just inferences based on humans picking out things and constructing a map.

[Joe Gwinn]

How does one then relate the derived model of the pickup back to frequency response from the pickup with a real string? What I was after with the moving-string model was to mechanically operate the string in a repeatable way, acquire the response info in a repeatable way, then use the method for comparing one pickup to another based on actual string movement.

Frequency response can be measured many ways, but only fairly large changes in response curve seem to cause a change in sound that all observers agree to.

There seems to be a gap in the middle of pickup winding. On one hand we have number of turns, wire thickness, spacing, coil dimensions and magnetic properties. Those seem to relate to the resulting sound only by human anecdotal experience. On the other hand, we have electrical measurements: inductance, resonance, DC and AC resistance, capacitance, etc., which also seem to relate to sound only anecdotally.

Neither path seems to tell us repeatably "there's a hump in the midrange and a little dip at 4kHz which will make this sound less present that this one where there is no dip, but less output level for the same string motion." It seems like this kind of data would be very useful because one could then relate both the mechanico-winding data and the electrical measurements to what signal you get out.

Or am I thinking too hard again?

At the 50,000-foot level, there are three things going on.

First, eddy current damping reduces high frequencies, but the effect is very broad.

Second, resonance of the pickup inductance with the self-capacitance plus the cable capacitance will cause a peak in the spectrum. Where this peak lands depends on inductance and total capacitance. If the resonance is slightly above the guitar range, the effect is to enhance highs, leading to a brighter sound. With more capacitance, the resonance moves down, and I suppose a lot of cable could enhance mid tones.

Third, the higher the total AC resistance, the lower the Q of the resonance, the broader and shorter the resonant peak.

[R.G.]

Thanks, Joe. I'm good with that as a 50,000 foot description.

What I'd like to do is get down to 10,000 feet. Any pointers there? Even if on the map we have to put notations in some places like "Here there be nonlinearities."

[soundmasterg]

This post is like walking through the halls at a college listening to two professors talk in detail about something you have an intense interest in, but don't have the technical chops to quite follow the conversation. I'm surprised some others haven't contributed yet to this one.....

Greg

[R.G.]

I picked up a new "string mover" tester trick. I saw a report where someone glued a paper tube to a speaker cone and used that to move a length of wire.

They were looking for how to separate horizontal versus vertical sensitivities in pickups, but it seems like a good way to move a string a fixed distance and see what signal you get out of the pickup.