Actual output amplitudes of various pickups

[antigua]

For my pickup database's RLC plotting feature, I needed some realistic offsets to express the loudness difference between different pickup classes. Pickups have RLC characteristics that are relatively easy to acquire, but the actual amount of output voltage is dependent upon the amount of magnemotive force that traverses the pickup's coil(s).

Here's some dB levels I got from five guitars, three Strats, one with A3 and one wih A5 pole pickups, one with SSL-5's that can tap in all three positions, one guitar with Filter'trons and one with PAF clones. 

For each guitar, the pole pieces were set to be exactly 4.5mm away from the strings (which is rather low). On the Strats with staggered poles, I went by the outer most pole pieces. The guitars were plugged straight into a Boss ME-80 (1 meg input impedance), then into a laptop via the USB connection. I recorded the sample with Adobe Audition and then acquired the average RMS amplitude from exactly the first two seconds of the strum. There are no extra gain stages or EQ, at least none that are user facing. I briskly strummed an open E chord, getting the string about an inch away from the neck pickup of each guitar, one inch in the direction of the bridge, so on a Strat, that's right in between the neck and middle pickup. All of the pickups are in different guitars, but all have the same strings, Elixir .010-.046 coated electric guitar strings.


The MIM A5 Strat and SSL-5 tapped pickups bridge positions had the lowest output, so I made that pickup the baseline, it shows zero, and the rest are some dB above that pickup. That value was -42.27, so +42.27 dB has been added to all the values to make them easier to compare. 

As an aside, I made a sort of discovery in the process; in the past I would pluck a single string, usually the D just to keep things clean, but I got very mixed results between humbuckers and single coils, and it occurred to me that the comb filtering of a wider PAF will selectively make particular harmonics weaker, so even though a PAF might be stronger than a single coil overall, comb filtering might make the "open D" weaker than it would be otherwise. The wider a pickup is, the more comb filtering occurs. Therefore, to compare a single coil to a humbucker, you have to either find a note that doesn't fall into the comb filter, or much easier, strum all six strings so that you get an assortment of harmonics that don't fall into the filtering.

The phenomena of comb filtering is covered by J Donald Tillman www.till.com/articles/PickupResponse/



position / dB / inductance L & DC resistance R

Strat MIM AlNiCo AlNiCo 5 in Fender Strat
B        0    (L: 2.2 R: 5.8k)
BM       3.56
M        8.57 (L: 2.2 R: 5.8k)
MN       8.54
N        8.44 (L: 2.2 R: 5.8k)

Strat 60th Anniv  AlNiCo 3 in Fender Strat 
B        0.07 (L: 2.5 R: 6.0k)
B        2.15
M        4.52 (L: 2.5 R: 6.0k)
MN       8.61
N        8.52 (L: 2.5 R: 6.0k)

Strat SSL-5 (AlNiCo 5) high output pickup in a Strat
N        2.59 (L: ~6 R: 12.9k )
BM       6.93
M        11.54 (L: ~6 R: 12.9k )
MN       12.69
N        12.46 (L: ~6 R: 12.9k )
Strat SSL-5 (AlNiCo 5) tapped pickup in a Strat
N        0.00 (L: ~2 R: 6.6k )
BM       2.73
M        5.98 (L: ~2 R: 6.6k )
MN       7.49
N        6.24 (L: ~2 R: 6.6k )

Filter'tron, Gretch in Fender Cabronita
B        5.38  (L: 2.3 R: 5.0K)
BN       7.83
N        9.8   (L: 1.7 R: 4.2K)

PAF clone Tonerider AlNiCo IV in Epiphone Les Paul
B        7.08   (L: 5.5 R: 8.2k)
BN       12
N        12.45  (L: 4.6 R: 7.1k)

Strat Lollar Blondes AlNiCo 2, flat poles
N        2.85 (L: 2.7 R: 6.0)
BM       5.22
M        6.21 (L: 2.5 R: 5.8)
MN       7.30 
N        9.50 (L: 2.2 R: 5.6k)





There is a large amount of variability in the data as far as what it represents, since I can't be sure I strummed with the same intensity for each sample, so the best I can do is look for broad trends. I would have liked to sampled more pickups, and using more methods, but this is tedious work and time is limited. Hopefully I can get more data later. I will also have a set of P-90's soon, and I'm looking forward to adding them to the mix.

A major complication is that these pickups all have different RLC values, so it can't be clearly see what difference owes to geometry, and those which owe to differences in impedance.  In general, Strat pickups rarely come close to PAF RLC values. I have an SSL-5 test subject, but it's inductance actually shoots past the PAFs. The Filter'trons have inudctances that are closer to the Strat pickups and serve as a decent point of comparison. 

First thing of note is that the Strat AlNiCo 3 and AlNiCo 5 pickups appear to be equally loud. The AlNiCo 5 produces twice as much flux density as AlNiCo 3, but it doesn't matter too much, apparently. The explanation for this is probably that while the absolute flux might be a lot higher, but the flux difference within the string's travel isn't all that great. When you have a stronger magnet, it's not just the maximum flux that is higher, but also the minimum flux. The AlNiCo 3 is wound a little hotter than the AlNiCo 5 pickups (and the AlNiCo 3 increases the inductance by itself), and that might cause a compensation that brings the AlNiCo 3's output closer to the AlNiCo 5 pickup, but they're still very close regardless. 

In general, the two lower output Strat pickups have a dB spread 0dB to 8dB, bridge to neck, while the PAF clone was 7dB to 12dB, and the Filter'tron was 5dB to 10dB. Note that the while the PAF clone has an inductance of 5.5H and 4.6H, the Filter'tron only has 2.3 AND 1.7H, even less than the average Strat pickups. From that it can be inferred that the voltage output is hugely increased by the side-by-side humbucker layout. Based on this data, you might assume that the humbucking layout adds 5 to 7dB at the bridge position, and 2 to 4dB at the neck. This greater dB difference seen in the bridge makes sense, since the humbucker bridge pickup extends away from the bridge, exposing it to greater string displacement. 

The SSL-5 has the highest inductance of all the pickups tested, around 6 henries, and so it's interesting to see how that high inductance Strat pickup fits with the observation of the previous paragraph, and in full 6 henry mode, the bridge to neck spread is 3dB to 12dB. So in the bridge position, it's only 3dB louder than the "average Strat pickup" and still 3 to 5dB quieter than the two humbuckers in the bridge position. But the SSL-5 in the neck is another story, at 12dB it's just as loud as the PAF clone neck with 4.6H inductance, and louder than the Filter'tron neck with 1.7H inductance. This goes to show that a high wind count can on a Strat pickup will allow it to achieve the voltage output of a typical PAF in the neck position, but perhaps not in the bridge. 

Another data point is the SSL-5 full versus tapped. Tapped the SSL-5 is a little weaker than the "average Strat pickups", but it also likely has a lower inductance, as that's only 6.6k DC resistance of 43 AWG wire. The tapped inductance is likely right around 2 henries, based on measurements of the very similar tapped SSL-4. The tapped SSL-5 shows output levels that are only slightly lower than the "average Strat pickups". An interesting fact about tapped coils is that they have a crazy high capacitance do to the presence of the unused portion of coil, so despite the low inductance, their have a resonant peak similar to a higher inductance Strat pickup, and that likely contributes poor opinions about tapped pickup tone. 

I also have middle pickup and in between dB levels in the table, but I've chosen not to analyse those for now. 

There's way more to be said about this, but my first post conclusion is that it's safe to say a humbucker gives you a boost in the ballpark of 3dB in the bridge, and the 6dB in the neck, all other things being equal. There's also clearly a scale difference between the bridge and neck. The extreme spread of the SSL-5 goes to show that doubling the nearly tripping the inductance of a Strat pickup gets you an additional 6dB in the neck, but only an additional 2 to 3dB boost in the bridge. 

[JohnH]

Excellent, and a very credible spread of values. The way the B, BM M MN and N values vary supports and is supported by the mathematics. Here are some GF plots using a CS69 (similar to your base case), showing how strumming of all strings between M and N pickups is expected to vary in amplitude across the selector 5 positions:



The approx. 8db spread is as expected. These are enveloped plots of the overall magnitudes vs frequency, allowing for harmonics and comb effects as well as electrical.

[antigua]

That's an impressive overlap between measured and calculated values. That's good news, it means my test setup worked out, as far as keeping things consistent.

This test covers a lot of ground in terms of pickup geometry and placement, RLC values, etc. it's all of these things distilled to a single dB value, but the most important thing here is trying to derive a source current. 

In the spice modelling we just set the AC source voltage to whatever, a value of one volt perhaps, because everything can be analyzed without being concerned about the actual source current, but now when we want to know, even as a rough guess, how much louder a 4 henry neck PAF will be than a 2.5 henry Strat pickup, we have to know what the relative values of "AC current" are for these two pickups. V =  I * Z, we know Z very well now, but I is somewhat of a mystery. 

If you can think of any math that can help with that, let me know, but I fear that it involves some pretty complex math and might even require FEM modelling to see how much flux is displaced within different coil geometries. Maybe we can instead collect data points and try to "curve match" it somehow. For example, suppose the PAF bridge gives you +5dBV over a Strat pickup, what is the difference in current?

[JohnH]

I think we are learning a lot, and have a fair handle on the maths around electrical response, and also string vibration. But I suspect that a mathematical model of the electromagnetic characteristics of the pickup, to the extent that we could actually calculate absolute amplitudes, is too much to attempt with any confidence (at least for me). I think the empirical testing, such as you did here, is the best way to fill in this gap, and it forms a basis for any future theorizing about the absolute physics.

[antigua]

Here's some model based guestimating with LTSpice. I'll compare the neck and bridge of the Fender Strat 60th Anniversary to the Gretsch Filter'tron, because I have the most reliable specs for these two pickups. 

With the AC amplitude set to 1 volt paired with the 60th Anniversary, the simulation shows -120dB at 120Hz with L 2.5H and R 6.0k ohms. What I'm finding is that LTSpice reports -120dB no matter what, unless the values of R, L or C, unless the values are such that the impedance curve dips down to 120Hz, otherwise if the peak and Q are high, the slope is flat at 120Hz and it's just -120dB 



The Filter'tron measures 2.3H and R is 5.0k ohms, and leaving the AC source at 1 volt, the dB at 120Hz is still -120dB since it's flat at 120Hz. An actual Filter'tron has some pretty dramatic eddy current losses though, so looking at 120Hz would exclude losses at higher frequencies. 

In the real test results, the Filter'tron bridge shows +5.3dB overall, so what does the LTSpice AC source voltage have to be to increase the output at 120Hz to -114.7dB? The difference between L and R with these two pickups apparently doesn't matter at 120Hz, but I put those values in anyway, and here's the answer, 1.82 volts brings the dB to -114.757dB at 120Hz: 



If you have to increase the AC source to 1.82 volts to "curve match", does this mean the current produces by the Filter'tron bridge is just shy of double that of the Strat 60th Anniversary? That seems intuitive, since you have two coils instead of the one. 

For the neck pickups, we have 8.4dB for the Strat, and 9.8dB for the Filter'tron. To match these offsets, it takes an AC source of 2.63 volts to get to -111.6dB for the Strat neck, and 3.1 volts to get to -110.2dB for the Filter'tron neck.

[JohnH]

At low frequency, the L and c parts have negligible effect, though there is a very slight reduction due to the two resistors in the model. In terms of output voltage , this is a small reduction of R2/(R1+R2) = 1000/1005 = 0.995. In terms of db, this is a reduction of 20log10(0.995) = -0.04db, ie not much.

To get a db increase of +5.3 by changing input, input voltage needs to increase by a factor 10^(5.3/20) = 1.84, which is about what you found.

[antigua]

Do you think a necessary voltage of 1.82 or 1.84 in order to match the single coil with a Filter'tron in the lower frequency ranges suggests that a Filter'tron produces about twice as much current as a Strat pickup, or would that be a misreading of the math?

[JohnH]

I dont think we need to worry about current specifically at all. It is all implied by taking care of voltage and impedance. Yes more voltage leads to more current, by the same factor, for a given impedance, but i dont see it as a seperate independent variable.

[ms]

The comb filtering affects the very high harmonics only which contribute very little to the total power.  The comb filtering is audible on the D string, but to a lesser extent that the other two wound strings, but even on them it does not affect the power much..

[antigua]

Feb 13, 2017 8:33:29 GMT -5 ms said:

The comb filtering affects the very high harmonics only which contribute very little to the total power.  The comb filtering is audible on the D string, but to a lesser extent that the other two wound strings, but even on them it does not affect the power much..

That's true, although for some reason I didn't get such prominent dB spreads with the D string alone. Maybe it's something else particular to the D string, but in any case I think a full chord has several advantages of plucking particular strings, but it's much harder to mechanically plunk a whole chord with good consistency. Since the dB spread matched predictions so closely, I'm guessing the hand strumming was sufficient, though. 

[antigua]

Feb 12, 2017 18:22:22 GMT -5 JohnH said:

I dont think we need to worry about current specifically at all. It is all implied by taking care of voltage and impedance. Yes more voltage leads to more current, by the same factor, for a given impedance, but i dont see it as a seperate independent variable.

What I'm trying to narrow down is a constant (fuzzy as it might be) that generically states the power difference between a Fender single coil and a PAF or Filter'tron, because a PAF and single coil can have identical everything in terms of RLC values, even impedence, but nevertheless the PAF will have higher voltage because of that coefficient that comes about from the fact that the PAF is a lot wider and seeing more flux change.  

So voltage equals V = I * Z for AC, but as you say the L and C are not very reactive at lower frequencies, so would we refer back to V = I * R?  In any event, we know the voltage is what is measures out to be, and "I" changes from one pickup form factor to the next, but as I proposed in the paragraph above, a slim coil and a wide coil might have identical Z, or R, therefore the crucial difference in predicting actual voltage output will be "I", the amount of current that is accumulated as a consequence of the geometric size of the pickup.

[JohnH]

It might be of interest to see a predicted set of outputs from just one string, in this case a D plucked open.



These curves don't distinguish between the characteristics of different strings other than as a frequency shift (which happens before the electrical model). Each is a 'perfect string'. The curves show the comb filtering per Tillman, and phase cancellations/interactions on the B+M and M+N settings.

In making the enveloped plots shown earlier, it is simply the maximums of any string output at a particular frequency, ie, I assumed that when you strum all strings, the maximums are not greater, but the curve is more filled in by all th epeaks. I think that is a better approximation than adding amplitudes from different strings, which will be all slightly out of sync, but it is an assumption.

Very interesting how singles and PAFs with the same RLC properties have different outputs. I had assumed that such a case would be because humbuckers are normally set higher than singles, but in your tests you equalised this parameter. And also, isn't it true that the max flux on a HB is usually less than for a single? So I guess it must be the wider field on the HB getting more string length involved in creating flux change. And/or, is the more permeable steel core on a PAF better at bringing the flux changes due to the strings down deeper into the body of the coil as compared to alnico slugs?

[antigua]

Feb 13, 2017 14:12:54 GMT -5 JohnH said:

Very interesting how singles and PAFs with the same RLC properties have different outputs. I had assumed that such a case would be because humbuckers are normally set higher than singles, but in your tests you equalised this parameter. And also, isn't it true that the max flux on a HB is usually less than for a single? So I guess it must be the wider field on the HB getting more string length involved in creating flux change. And/or, is the more permeable steel core on a PAF better at bringing the flux changes due to the strings down deeper into the body of the coil as compared to alnico slugs?

I have some flat poled Strat pickiups and Lace Sensors that I should probably record also. A problem with measuring from pole tops is that the coil distance is still variable, and that's probably the more important geometry. I also have "rail" single coil sized humbuckers that I should test out as well. 

Regarding flux density, it seems to make only a small difference, because you can see that my A3 and A5 Strat pickups have such close values. A5 measures 1050G at the top, A3 is around 600G, so a little over half. The A3 Strat pickup has a slightly higher impedance, but only by a small amount, so it's pretty much a wash.

The fact that the Filter'tron has a lower impedance than either Strat pickup, but still produces 5dB more at the bridge and 2dB more at the neck reinforces the notion that having two coils close to the strings gives you more current and power than a single coil affords. The Filter'tron and the PAF are not too far apart, despite the PAF having perhaps twice the impedance.

The SSL-5 is sort of counterpoint on the single coil side; with it's 6H inductance and 13k DC resistance, it produces output that is comparable to the PAF and Filter'tron, but only in the neck position, in the bridge it is weaker. 

[antigua]

I've just added in measurements for flat poled Lollar Blondes. Since the distance from the string is based on the height of the pole piece, this means the coil is about a couple millimeters closer to the strings for the same pole piece distance. It looks like there is about a ~2dB boost as a result of the closer coil proximity. The flux density of the AlNiCo 2 is comparable to the AlNiCo 3 of the Fender 60th Anniversary set, so it makes a good point of contrast. 

What I'm seeing is that different briskness of strumming with a pick easily varies the output by +/- 3dB, and most of the "average" Strat pickups are this close together, so it will be hard to tell is one is louder or quieter than another, but then you plug in a Les Paul a Gretsch, you get a 5dB bump due to the humbucking layout, and that's a difference that surpasses what varied strumming accomplishes, and that matches with my real world experience with swapping around guitars. The exception is where the inductance of the Strat pickup is MUCH higher, the SSL-5 with it's 6 henry inductance does get it's dB level into humbucker output range. 

[Charlie Honkmeister]

Thanks for some great persistent work here, Antigua.

Do you want to pursue any greater consistency on the "strum" or string excitation, or do you think that most folks will need anything more accurate than what you have done?  

If you wanted to chase any more consistency,  would you think that you would need a mechanical plucker (which would be not that hard to do with group help, with steppers or radio control type servos and some sort of 3D printed armature)?  I could really see whipping something up with an Arduino and a stepper. 

Or would you be thinking you would go back to refining the driver coil measurement where you might standardize your driver coil  (geometry, winds, etc.),  drive voltage, power, and input signal (pink noise maybe, or ms's sequences)   and standardize the position and height in both single coil and humbucker measuring modes?

Just curious where you might want to go next or whether you even want to at this point.

-Charlie   

[JohnH]

A idea I was wondering about, which I think Ive heard of elsewhere, would be to mount a length of piano wire eccentrically in the chuck of a drill, and run the drill at a consistent height and speed over the pickups. Im not sure if it would be fast enough and would need to be at least 6000rpm to cteate 100 hz or more.

[reTrEaD]

Sounds scary.

You can probably find drills that get to about half that rotational speed. Above that, you'd need to start looking at die grinders.

An offset mass means vibration. Bad for the bearings and could make holding it stationary extremely difficult. The longer the piano wire is, the greater the tendency to bend. Seems like it could turn into a whip even at fairly low rotational speed.

[antigua]

Feb 14, 2017 23:32:23 GMT -5 Charlie Honkmeister said:

Thanks for some great persistent work here, Antigua.

Do you want to pursue any greater consistency on the "strum" or string excitation, or do you think that most folks will need anything more accurate than what you have done?  

If you wanted to chase any more consistency,  would you think that you would need a mechanical plucker (which would be not that hard to do with group help, with steppers or radio control type servos and some sort of 3D printed armature)?  I could really see whipping something up with an Arduino and a stepper. 

Or would you be thinking you would go back to refining the driver coil measurement where you might standardize your driver coil  (geometry, winds, etc.),  drive voltage, power, and input signal (pink noise maybe, or ms's sequences)   and standardize the position and height in both single coil and humbucker measuring modes?

Just curious where you might want to go next or whether you even want to at this point.

-Charlie   

When I did all the measurements in one day things seemed sort of consistent, but then I did the Lollar Blonde a couple days later, and I think I might have used a different pickup because the dB was higher, so I plugged the MIM Strat back in, tested with that, then found that the MIM was about ~2dB louder in all positions this time around, so I subtracted 2dB from the Blonde values to "normalize" it with respect to the measurements from the day before. I feel I got the basic information I need, which is, about how much louder is a humbucker than a single coil, and how much louder is a 6 henry single coil than a 2-3 henry single coil. I'd have loved to see how much louder an 8 henry PAF type is than a 4.5, but I don't have a guitar with such hot pickups in it at the moment. So I think hand strumming really only works if you do it in a single session so you can be aware of exactly how you're strumming. Even then, it's kind of a mess. I tried to do a Strat with Lace Sensors, but for reasons I'm unsure of, the variability was too high to make any kind of sensible observations about the outcome. 

I'd take you up on ways to do something consistent, but I'm not planning to revisit this too much for the time being. This kind of testing that involves recording sound clips and analyzing the resulting mess of audio is very tedious. I had recorded one guitar and put it away, starting measuring dB levels, only to realize I had to go back and do it all over again because I forgot to lower the pickups to the standard height first.

I think you'd have a hard to getting consistency, even from a servo. The problem is that 1dB of difference is meaningful when comparing pickups, but it's extremely hard to get a consistent mechanical pluck that stays within that range. The string(s) get's pulled back, that amount of tension is important, and then the way the string breaks away from the plectrum makes a difference in the amplitude. At least with hard strumming you're subjecting the string to a near-physical maximum, but with mechanical methods that's generally not the case.

There are still some things that might be worth trying; one idea could be to put capacitance across the guitar so that only low frequencies are measured, which might alleviate variations that result from the more random harmonic amplitudes, and that might prove more consistent, but the dB results might also be misleading. 

A good solution would preferably be very simple to set up and carry about, because it's a tedious process aside, making this task a little less tedious would go a long way. I had hopes that breaking fine wire over the string would be consistent, but the variation still exceeded several decibels from go to go, and threading the wire around the string, and lining it up in a particular spot, hitting record, and the pulling on the wire was a tiring routine. I had a little pile of hair-like wire after a few minutes. The main problem is that the force with which you pull the wire determines how long it holds out before it breaks, being consistent with that pull is just like trying to be consistent with a strum. 

It's not really possible to evenly apply the driver coil to humbuckers and single coils because for humbuckers the driver has to be placed sideways and in between, which decreases the flux density for both coils, where as a single coil get's much closer coupling. I supposed two identical driver coils could be placed over a humbucker, but that's a lot of trouble for an imperfect test scenario, since it's still not actually a guitar string.

I have an eBow, I might give that a try next time. Maybe applying a vibration to the bridge would work similarly, like one of those Sonicare tooth brushes. Maybe after my nerves have calmed in a couple weeks I'll give those things a try. 

[antigua]

Feb 15, 2017 0:10:33 GMT -5 JohnH said:

A idea I was wondering about, which I think Ive heard of elsewhere, would be to mount a length of piano wire eccentrically in the chuck of a drill, and run the drill at a consistent height and speed over the pickups. Im not sure if it would be fast enough and would need to be at least 6000rpm to cteate 100 hz or more.

Along these lines, I saw someone tape a length of guitar string to a speaker and then place it over a pickup. Not bad, but the string is not in situ. I wouldn't have thought it mattered until I saw such variation between bridge, middle and neck positions. The way a pickup relates to the strings seems to differ in the bridge versus the neck, it's not a linear thing. For example, a humbucker bridge will be louder than a single coil bridge because the humbucker reaches out towards the neck, but this "handicap" doesn't exist for the neck position, where both types of pickups are more or less on equal ground (save for the neck anti node). Complicating matters even more, if the pickup has a coil that is close to the top of the pickup (Lace Sensor), this bridge/neck disparity appears to be magnified. 

For my purposes, all I really need are ball park figures, because if two pickups are within, say 3dB, you can close that gap by just adjusting the heights of the pickups a little. What's important are conditions that represent huge differences, such as three more henrys of inductance, or two side by side coils instead of one. 

[ms]

Feb 15, 2017 2:22:32 GMT -5 antigua said:

Feb 15, 2017 0:10:33 GMT -5 JohnH said:

A idea I was wondering about, which I think Ive heard of elsewhere, would be to mount a length of piano wire eccentrically in the chuck of a drill, and run the drill at a consistent height and speed over the pickups. Im not sure if it would be fast enough and would need to be at least 6000rpm to cteate 100 hz or more.

Along these lines, I saw someone tape a length of guitar string to a speaker and then place it over a pickup. Not bad, but the string is not in situ. I wouldn't have thought it mattered until I saw such variation between bridge, middle and neck positions. The way a pickup relates to the strings seems to differ in the bridge versus the neck, it's not a linear thing. For example, a humbucker bridge will be louder than a single coil bridge because the humbucker reaches out towards the neck, but this "handicap" doesn't exist for the neck position, where both types of pickups are more or less on equal ground (save for the neck anti node). Complicating matters even more, if the pickup has a coil that is close to the top of the pickup (Lace Sensor), this bridge/neck disparity appears to be magnified. 

For my purposes, all I really need are ball park figures, because if two pickups are within, say 3dB, you can close that gap by just adjusting the heights of the pickups a little. What's important are conditions that represent huge differences, such as three more henrys of inductance, or two side by side coils instead of one. 

For relative levels within 3 db, I think you could just use careful measurements made with your small exciter coil.  And having made all these measurements, I think you have the data to show that, or to find out what needs to be done in order to make it work.

[antigua]

Feb 15, 2017 6:13:25 GMT -5 ms said:

Feb 15, 2017 2:22:32 GMT -5 antigua said:

Along these lines, I saw someone tape a length of guitar string to a speaker and then place it over a pickup. Not bad, but the string is not in situ. I wouldn't have thought it mattered until I saw such variation between bridge, middle and neck positions. The way a pickup relates to the strings seems to differ in the bridge versus the neck, it's not a linear thing. For example, a humbucker bridge will be louder than a single coil bridge because the humbucker reaches out towards the neck, but this "handicap" doesn't exist for the neck position, where both types of pickups are more or less on equal ground (save for the neck anti node). Complicating matters even more, if the pickup has a coil that is close to the top of the pickup (Lace Sensor), this bridge/neck disparity appears to be magnified. 

For my purposes, all I really need are ball park figures, because if two pickups are within, say 3dB, you can close that gap by just adjusting the heights of the pickups a little. What's important are conditions that represent huge differences, such as three more henrys of inductance, or two side by side coils instead of one. 

For relative levels within 3 db, I think you could just use careful measurements made with your small exciter coil.  And having made all these measurements, I think you have the data to show that, or to find out what needs to be done in order to make it work.

That's a good point; for comparing gives types of pickups, like low output HB versus high output, the exciter method should be good enough.

And it just occurred to me that I can probably compare a single coil and HB by just orienting the exciter over the single coil as though it were one half of a humbucker, at 90 degrees with only half the exciter over the single coil. That should impose a similar flux field. It will be interesting to see how close those measures come to the hard strumming approach. 

[ms]

Feb 15, 2017 11:12:36 GMT -5 antigua said:

Feb 15, 2017 6:13:25 GMT -5 ms said:

For relative levels within 3 db, I think you could just use careful measurements made with your small exciter coil.  And having made all these measurements, I think you have the data to show that, or to find out what needs to be done in order to make it work.

That's a good point; for comparing gives types of pickups, like low output HB versus high output, the exciter method should be good enough.

And it just occurred to me that I can probably compare a single coil and HB by just orienting the exciter over the single coil as though it were one half of a humbucker, at 90 degrees with only half the exciter over the single coil. That should impose a similar flux field. It will be interesting to see how close those measures come to the hard strumming approach. 

I think the right way to excite a hum bucker is with two small coils, over corresponding pole pieces in each of the pickup coils. (Exciter oils must be out of phase, of course.)  The string is magnetized most strongly and in the right direction (field pointing through the coil) right over the pole piece, and the coil should do the same thing.

[antigua]

Feb 15, 2017 15:16:39 GMT -5 ms said:

Feb 15, 2017 11:12:36 GMT -5 antigua said:

That's a good point; for comparing gives types of pickups, like low output HB versus high output, the exciter method should be good enough.

And it just occurred to me that I can probably compare a single coil and HB by just orienting the exciter over the single coil as though it were one half of a humbucker, at 90 degrees with only half the exciter over the single coil. That should impose a similar flux field. It will be interesting to see how close those measures come to the hard strumming approach. 

I think the right way to excite a hum bucker is with two small coils, over corresponding pole pieces in each of the pickup coils. (Exciter oils must be out of phase, of course.)  The string is magnetized most strongly and in the right direction (field pointing through the coil) right over the pole piece, and the coil should do the same thing.

It would be fairly tricky to make another identical exciter coil. I think the sideways mode will work out ok though. 

[antigua]

One caveat to the exciter coil as a means of comparing output levels is that it will ignore differences in magnetic strength of the pickup's pole pieces, but it's looking to me like that's not really much of a determinant anyway, since the Strat A3 and A5 show the same numbers. Another indicator of this truth is that is that it's never been sufficient to make a pickup "hot" simply by putting stronger magnets in it alone, they invariably feature a higher inductance, quite a lot higher in the 70's/80's hot rod pickups such as the SSL-5 or Super Distortion. 

[Charlie Honkmeister]

Feb 15, 2017 6:13:25 GMT -5 ms said:

Feb 15, 2017 2:22:32 GMT -5 antigua said:

Along these lines, I saw someone tape a length of guitar string to a speaker and then place it over a pickup. Not bad, but the string is not in situ. I wouldn't have thought it mattered until I saw such variation between bridge, middle and neck positions. The way a pickup relates to the strings seems to differ in the bridge versus the neck, it's not a linear thing. For example, a humbucker bridge will be louder than a single coil bridge because the humbucker reaches out towards the neck, but this "handicap" doesn't exist for the neck position, where both types of pickups are more or less on equal ground (save for the neck anti node). Complicating matters even more, if the pickup has a coil that is close to the top of the pickup (Lace Sensor), this bridge/neck disparity appears to be magnified. 

For my purposes, all I really need are ball park figures, because if two pickups are within, say 3dB, you can close that gap by just adjusting the heights of the pickups a little. What's important are conditions that represent huge differences, such as three more henrys of inductance, or two side by side coils instead of one. 

For relative levels within 3 db, I think you could just use careful measurements made with your small exciter coil.  And having made all these measurements, I think you have the data to show that, or to find out what needs to be done in order to make it work.

After thinking about this a little bit, I would side with ms on this,  in that the output of a pickup is a "secondary"  measurement and is not as important as the other things you want to measure.  

If you had the task of  reverse engineering a pickup without destroying it,  IMHO you would have the core information you need just by running your (or ms's, or both) suite of tests,  without needing any output measurement referenced to some standard.   

The only pieces of data that would be immediately useful numbers if you were trying the reverse engineering thing, would be the wind count on the coils, and the AWG of the wire.   But since we are playing in a relatively small sandbox  (almost always #41 to #44 AWG, known overall geometry, etc.)  there's ways to either directly infer the wind count (inductance, DCR proportional to number of winds for a given AWG, using Salvarsan's coil calculator), or nondestructively inspect the coil wire,  to get the AWG.   So the electrical measurements other than absolute output are really the key and you guys have gotten that covered really well.  

If we were doing a "fundamental physics" investigation, we would be chasing off after magnetic core materials, temperature effects, and all that stuff.  But you guys (Antigua, ms, JohnH) have absolutely nailed the right set of methods and theoretical/practical balance to be the most useful,  as things stand now.     

[antigua]

Feb 16, 2017 10:30:27 GMT -5 Charlie Honkmeister said:

Feb 15, 2017 6:13:25 GMT -5 ms said:

For relative levels within 3 db, I think you could just use careful measurements made with your small exciter coil.  And having made all these measurements, I think you have the data to show that, or to find out what needs to be done in order to make it work.

After thinking about this a little bit, I would side with ms on this,  in that the output of a pickup is a "secondary"  measurement and is not as important as the other things you want to measure.  

If you had the task of  reverse engineering a pickup without destroying it,  IMHO you would have the core information you need just by running your (or ms's, or both) suite of tests,  without needing any output measurement referenced to some standard.   

The only pieces of data that would be immediately useful numbers if you were trying the reverse engineering thing, would be the wind count on the coils, and the AWG of the wire.   But since we are playing in a relatively small sandbox  (almost always #41 to #44 AWG, known overall geometry, etc.)  there's ways to either directly infer the wind count (inductance, DCR proportional to number of winds for a given AWG, using Salvarsan's coil calculator), or nondestructively inspect the coil wire,  to get the AWG.   So the electrical measurements other than absolute output are really the key and you guys have gotten that covered really well.  

If we were doing a "fundamental physics" investigation, we would be chasing off after magnetic core materials, temperature effects, and all that stuff.  But you guys (Antigua, ms, JohnH) have absolutely nailed the right set of methods and theoretical/practical balance to be the most useful,  as things stand now.     

Part of induction relates to the area of the loop and the amount of flux change, so the fact that these dimensions and quantities change from one pickup to pickup is a metric that should be quantified somehow. According to this en.wikipedia.org/wiki/Electromotive_force#Notation_and_units_of_measurement it would be a measure of voltage, but I'm not sure how we can practically measure that, since the voltage that comes away from the pickup is also dependent on the impedance, but I suppose it's possible if you somehow factor out the impedance. I have several pickup with identical dimensions which only vary in terms of impedance, so even from that some inferences can be eyeballed. Intuitively I thought flux density was more analogous to current, but I must be mistaken because all the talk is of voltage when it comes to induction.

The practical reason for it is, say you have a "balanced" set of pickups, and you want a bright pickup, but not one that will be overpowered. This issue comes up a lot especially with guitars that have a mixture of single coils and humbuckers. The issue can be remedied a few ways, it depends on what sort of compromises a guitarist is willing to make. 

[antigua]

Here's an initial measurement with the exciter coil; I compared an SSL-1 and a Fender Fideli'tron because the have very similar L and R values. 

The exciter coil is placed in a uniform way over both types. The popsicle stick exciter coil is perhaps 150 turns of 41 AWG, and is being driven with 0.5 Vpp. For a humbucker it goes like this:



So for the single coils I have it oriented the same way so that the amount of EMF through will be apples to apples. I took the cover off to try to get a similar exciter to coil proximity. 



And this is the loaded / unloaded result. The higher amplitude lines are the Fideli'tron, the low is the SSL-1. 



I'm not sure which frequency is ideal for contrasting loudness, but you can see it's pretty flat for the most part. I set the marker at 501Hz, and the dB difference there is 6.9dB, or we can just round off and say 7dB at 500Hz.  The absolute amplitude of the Fideli'tron is 1.9dBV, and the SSL-1 is -4.9dBV. 

As far as the noise floor, it's a little hard to tell, I range the plotter with the exciter coil far away from the pickup just to see how low it goes, but the plot is very choppy:



It looks like the bottom is around -10dBV.

So we have the SSL-1 5.4dB above the floor, and the Filter'tron 12dB above the noise floor, which is over double, but close to double, so it seems to me that for having two coils below the string, you get roughly double the output. 

It also looks like the loaded and unloaded amplitudes are close enough, so I have another six pickups to test, but for each I will only measure loaded amplitudes. 

[JohnH]

It looks like that small exciter coil has a horizontal axis. Should it not be vertical to align with the most sensitive direction of the pickup coils? The relative sensitivity of the pickup to other axes may not be consistent with their main axes. I'd suspect the single pickup is being relatively under reported in this case compared to the Hb.

[ms]

Feb 16, 2017 14:20:37 GMT -5 JohnH said:

It looks like that small exciter coil has a horizontal axis. Should it not be vertical to align with the most sensitive direction of the pickup coils? The relative sensitivity of the pickup to other axes may not be consistent with their main axes. I'd suspect the single pickup is being relatively under reported in this case compared to the Hb.

Yes, the pickup coil is sensitive to the field component pointing through the coil.  The magnet magnetizes the the string in this direction right over the center of the pole piece.  As you go along the string away from the center, the direction of the magnetization changes, and so the contribution to the signal drops.


Results should be more consistent using a vertical coil also since a small tilt should have less effect. 

[antigua]

Feb 16, 2017 14:20:37 GMT -5 JohnH said:

It looks like that small exciter coil has a horizontal axis. Should it not be vertical to align with the most sensitive direction of the pickup coils? The relative sensitivity of the pickup to other axes may not be consistent with their main axes. I'd suspect the single pickup is being relatively under reported in this case compared to the Hb.

I got the idea for measuring humbuckers this way from Helmuth Lemme's materials:




This is not much different from having two separate exciter coils over the two coils, because you have north return path in one coil and south return path in the other, satisfying the phase of the coils. The flux density pattern is not identical since there is a central source of flux instead of two sources, but this should be effective enough in terms of placing equal amounts of EMF through both coils of a humbucker, and the one coil of the single coil. 

For the single coils I usually do place the driver directly over the pole piece to mimic a string, but in order to make it apple to apples with a humbucker, I have to treat the single coil as though it were half a humbucker, because placing the driver perfectly over the single coil will give it a boost of direct EMF that the humbucker will not receive.

The 7dB spread between the Fideli'tron and the SSL-1 is pretty close to thge difference I saw with pickups in the neck position, so even though this method of excitation and measurement is a lot different, at least the difference in output are in the same ball park. The major drawback of the exciter method is that the loudness boost you might get from a ceramic magnet instead of AlNiCo is not represented. I think I will have to go back to the strumming strings method to get values for magnet types. My belief is that the traditional AlNiCos are all very close, but that AlNiCo 8 or ceramic magnets will boost the output. Based on my experience with magnet swapping I think it could easily be a 3dB boost. When you slide a ceramic magnet into a PAF, the boost in output is fairly audible.