The impact of the resonant peak upon the tone of a pickup

[antigua]

A question that has been bugging me for a long while is.. if a P-90 has a resonant peak of only 2.1kHz, how is it not an extremely dark pickup? It is dark, but not necessarily muddy. It's still clear enough that it's been used and accepted for it's clean tones since 1946.

I think I've found the answer to this question by taking a clean Stratocaster audio sample. The Strat contains Lollar Blonde pickups, and was ran through a Boss ME-80 multi effect process. It appears the key is the fact that the low pass filter is effectively only 6dB per octave; rather gradual. It is not heard as a hard cutoff.

If the pickup were used as a dedicated LPF, it would have a -12dB / octave drop, but the induction of voltage by way of a moving magnetized guitar string imposes a +6dB/oct roll off that combines with the -12dB/oct to become -6dB/oct. Eddy currents increased this rate of roll off, proportionate to the amount of eddy currents that are produced by the pickup.

I've made some sound samples to demonstrate the point. The sound samples start with the neck pickup, and work through the five positions of a Strat, ending with the bridge pickup.

It seems to be the case that a -6dB to -12dB/oct roll off is gradual enough that a lot of harmonic content beyond that particular peak frequency remains audible, and is important to defining the "tone" of the pickup. In listening to a -48dB/oct cut of the same frequency, you can get a sense for that fact very easily. By comparing -6dB with -12dB, you can also intuit the impact of eddy currents inducing a third order low pass.




First, here is how the audio amplitude and frequency spectrum looks. I included hard cuts at 6kHz, 7kHz and 8kHz, sort of for no reason, as you can see from the spectral display that there is no content below 5kHz, and only a brief peak at 6kHz in the transient of the neck/middle position. For the -12dB/oct and -6dB/oct samples I omitted sound samples of those higher cut offs. 






The take away I get from these sound samples is if I listen to the sound clip with a 48dB/oct roll off, it sounds quite unlike a P-90, but if I listen to the 12dB/oct roll off, it sounds quite like a P-90, even with Strat pickups as the original sound source. 

The 6dB/oct roll off, as opposed to the 12dB/oct roll off, sounds a bit too bright for a P-90, IMO. I suspect the eddy current losses caused by the steel screw core cause it to have a roll off that falls somewhere in between.

I also think the -6dB/oct roll off at 3kHz gives the pickup a "meaty" tone that corresponds with pickups that dip into this range, such as Telecaster pickups, or Stratocaster Texas Specials. The -6dB/oct at 4kHz roll off retains a light of chime and presence by contrast. 

Another mystery seems to have cropped up though: the 48dB/oct roll off sounds suspiciously similar to what I get when I put a 3nF to 15nF cap in parallel with a pickup, producing a particularly nasal tone. I've also wondered why putting a 3nF cap (or whatever drags it's resonant peak down to that of a P-90) doesn't make a Strat pickup sound like a P-90. The -12dB attenuated audio clip actually does sound like a P-90 to my ears, but the result I get with a 3nF cap sounds a lot more like the the -48dB/oct audio clip. It suggests that the drop off is steeper with the parallel cap, but as far as I know, the cap should not make the drop off any steeper, as it's still a second order low pass, not a third order. 






Sound samples:




48dB / Octave Roll Off




First I'll start with an extreme 48dB/oct roll off, just so you can hear what a "hard" cut off sounds like at a given frequency. 

48dB roll-off at 1kHz
soundcloud.com/andrewflanders/lollar-blonde-1000khz-lpf-48db?in=andrewflanders/sets/lollar-blonde-various-hpf-profiles

48dB roll-off at 2kHz
soundcloud.com/andrewflanders/lollar-blonde-2000khz-lpf-48db?in=andrewflanders/sets/lollar-blonde-various-hpf-profiles

48dB roll-off at 3kHz
soundcloud.com/andrewflanders/lollar-blonde-3000khz-lpf-48db?in=andrewflanders/sets/lollar-blonde-various-hpf-profiles

48dB roll-off at 4kHz
soundcloud.com/andrewflanders/lollar-blonde-4000khz-lpf-48db?in=andrewflanders/sets/lollar-blonde-various-hpf-profiles

48dB roll-off at 5kHz
soundcloud.com/andrewflanders/lollar-blonde-5000khz-lpf-48db?in=andrewflanders/sets/lollar-blonde-various-hpf-profiles

48dB roll-off at 6kHz
soundcloud.com/andrewflanders/lollar-blonde-6000khz-lpf-48db?in=andrewflanders/sets/lollar-blonde-various-hpf-profiles

48dB roll-off at 7kHz
soundcloud.com/andrewflanders/lollar-blonde-7000khz-lpf-48db?in=andrewflanders/sets/lollar-blonde-various-hpf-profiles

48dB roll-off at 8kHz
soundcloud.com/andrewflanders/lollar-blonde-8000khz-lpf-48db?in=andrewflanders/sets/lollar-blonde-various-hpf-profiles





12dB / Octave Roll Off



Next is 12dB/oct samples:

12dB roll-off at 1kHz
soundcloud.com/andrewflanders/lollar-blonde-1000khz-lpf-12db?in=andrewflanders/sets/lollar-blonde-various-hpf-profiles

12dB roll-off at 2kHz
soundcloud.com/andrewflanders/lollar-blonde-2000khz-lpf-12db?in=andrewflanders/sets/lollar-blonde-various-hpf-profiles

12dB roll-off at 3kHz
soundcloud.com/andrewflanders/lollar-blonde-3000khz-lpf-12db?in=andrewflanders/sets/lollar-blonde-various-hpf-profiles

12dB roll-off at 4kHz
soundcloud.com/andrewflanders/lollar-blonde-4000khz-lpf-12db?in=andrewflanders/sets/lollar-blonde-various-hpf-profiles

12dB roll-off at 5kHz
soundcloud.com/andrewflanders/lollar-blonde-5000khz-lpf-12db?in=andrewflanders/sets/lollar-blonde-various-hpf-profiles






6dB / Octave Roll Off




Finally, 6dB/oct roll-off:

6dB roll-off at 1kHz
soundcloud.com/andrewflanders/lollar-blonde-1000khz-lpf-6db?in=andrewflanders/sets/lollar-blonde-various-hpf-profiles

6dB roll-off at 2kHz
soundcloud.com/andrewflanders/lollar-blonde-2000khz-lpf-6db?in=andrewflanders/sets/lollar-blonde-various-hpf-profiles

6dB roll-off at 3kHz
soundcloud.com/andrewflanders/lollar-blonde-3000khz-lpf-6db?in=andrewflanders/sets/lollar-blonde-various-hpf-profiles

6dB roll-off at 4kHz
soundcloud.com/andrewflanders/lollar-blonde-4000khz-lpf-6db?in=andrewflanders/sets/lollar-blonde-various-hpf-profiles

6dB roll-off at 5kHz
soundcloud.com/andrewflanders/lollar-blonde-5000khz-lpf-6db?in=andrewflanders/sets/lollar-blonde-various-hpf-profiles

[antigua]

I was doing some rough testing tonight with an actual Strat which I have wired with a variable cap dial. What I'm looking to find out is a broad answer to the question of why it is that simply putting a cap in parallel with a pickup doesn't give you a "hot" pickup.

The two obvious facts are that a "hot" pickup has a higher voltage output and a lower resonant peak, but it appears a third important fact is that a naturally hot pickup has a more subtle roll off than any pickup with a cap in parallel with it.

This plot below shows that the Strat experiment, full open, with various cap values, and one black plot line shows a 3n cap with the tone knob at "4", putting some resistance between the 3n cap and the pickup. It's a little ugly looking, but the important take away is that the full open pickup has a -14dB/oct cut, but when the 10nF cap is put in parallel, the cut becomes 19dB/oct. It can also be seen that series resistance with the cap has the opposite effect of softening the slope, to 8dB/oct in the test below.



(I forgot to put a minus sign in front of the dB/oct values, just imagine it's there)


Keep in mind that for these tests I'm not running through the Ken Willmott integrator, which is overlaying a -6dB/oct filter over everything, where as the first post of this thread is talking about a raw, unfiltered signal, so -6dB/oct in that post is equivalent to -12dB/oct in this post.

Now, here's a P-90 in a Les Paul with 250k load resistance. Now this hot pickup has a resonant peak similar to the Strat pickup with the 3 nanafarad capacitance, but where as that cap causes the pickup to have a steep -19dB/oct roll off, this hot P-90 has only a -12dB roll off, thereby retaining a lot of treble that the Strat pickup drains away:


(I forgot to put a minus sign in front of the dB/oct values, just imagine it's there)


This leads directly to one useful conclusion: a cap doesn't make a pickup hot, because the cap creates something like a sloppy 3rd order low pass filter that more aggressively attenuates treble than a "hot" pickup of similar resonance. 

And there's a second useful conclusion. In the first plot I hard set the cap to 3nF and tweaked the tone knob to get as close to a P-90 as possible, but there's a new problem; the series resistance of the tone pot has the opposite effect, showing an attenuation of only -8dB/octave, very close to a first order low pass filter. I'm not even sure why that happens. So now the problem is we have a Q factor that looks good, but a rate of attenuation that lets too much treble pass, causing the Strat pickup to retain a brighter character than the P-90, with the 3nF cap and the tone knob pulled back.



The combined fact that the slope of attenuation differs greatly by adding parallel caps and resistance, and the prominence these slopes have on the tone, as demonstrated by the audio samples in the first post, deals a blow to the concept of passive pickup modelling, unless these filter profiles can be addresses and somehow passively tailored further. 

Another take away is that this cripples modelling to some extent, because if you apply the Lemme model and add caps in parallel, you pretty much maintain a 12dB/oct slope no matter what, because the model is not correctly placing any of the pickup's LCR values where they truly are in relation to the parallel cap, it's only a rough approximation.

I was thinking of feeding audio samples into LTSpice and getting samples back out with various attenuation profiles, but these findings suggest that it might be a waste of time, because if the dB/oct slopes are wrong, the sound samples will not be closely representative of reality. 

The Fishman Fluence is a modeled pickup, in that it takes a flat response as masks a resonant LPF over the top of it. I'd be curious to know what dB/oct attenuation they arranged for after the resonant peak. By all accounts, the pickup sounds pretty good, so I imagine they got the slope pretty close to -12dB/oct, as it appears that there are readily audible consequences to deviating from that slope.

[gitpiddler]

You really give me tool envy. My bastard strat has a visibly low resonant peak when I thump on it. The Tele has almost no thump but sounds badass an octave lower. I would like to swap pickups between them, but I enjoy having one that will slice your earlobes off if asked, and the other to make things move around as if the room were an air hockey table.

When the pickup peaks higher than the majority of notes you can play, is that not relatively flat? The bigger question is the impact of the resonant peak(s)of the instrument. Leo Fender aimed for getting the lows and highs from the instrument, and the mids from the pickups.

That's why I'm a leg man. The rest can be modified. Me caveman, you rocket scientist. Excellent work BTW.

[ms]

Mar 27, 2017 23:09:39 GMT -5 antigua said:

I was doing some rough testing tonight with an actual Strat which I have wired with a variable cap dial. What I'm looking to find out is a broad answer to the question of why it is that simply putting a cap in parallel with a pickup doesn't give you a "hot" pickup.

I think what you need to switch in is a simple network, maybe a cap with a series and parallel resistor.  For each C that use have on your switch, you select the two resistor values that give you the best match to the tone you want, with, say, the tone and volume controls in their maximum positions.

[gitpiddler]

I recall the resonant peak of the human ear is around 3 kHz at lower volumes, hence the old loudness button on stereos to compensate. Lower volumes? What? I want to know the frequency that makes her clothes fall off! The resonant frequency of milk maybe? Maybe Santana, when he said he wanted to hear a baby crying in there somewhere, had it right.
I kept trying to express a thought earlier and the board kept feeding me a "new" word - Jubblies! That was a bodacious start to my day-a reference to one of my hometown heroes David Keith's line in "An Officer and a Gentleman". We are also the home of the original Mountain Dew. The factory was across the street from the bus station and in the path of I-40.

Gitwhittler to standby. Y'all have a goodun

[antigua]

Mar 28, 2017 6:33:00 GMT -5 ms said:

Mar 27, 2017 23:09:39 GMT -5 antigua said:

I was doing some rough testing tonight with an actual Strat which I have wired with a variable cap dial. What I'm looking to find out is a broad answer to the question of why it is that simply putting a cap in parallel with a pickup doesn't give you a "hot" pickup.

I think what you need to switch in is a simple network, maybe a cap with a series and parallel resistor.  For each C that use have on your switch, you select the two resistor values that give you the best match to the tone you want, with, say, the tone and volume controls in their maximum positions.

The purpose of the selector is just to variably select tone caps, not to make a hot pickup, but nevertheless it shows that more than just a cap is required.

I'm not sure if a combination of resistor and cap can get it to where it needs to be to adequately curve match. It might be the case that once you get the correct Q, you have the wrong slope, and vice versa. It would take some trial and error. If it could  be correctly modeled, it would be very easy to demo, but the modeling won't help with the slope apparently, so it would require some tedious trial and error with an analyzer, and an assortment of caps and resistors. 

[ms]

Mar 28, 2017 9:33:54 GMT -5 antigua said:

Mar 28, 2017 6:33:00 GMT -5 ms said:

I think what you need to switch in is a simple network, maybe a cap with a series and parallel resistor.  For each C that use have on your switch, you select the two resistor values that give you the best match to the tone you want, with, say, the tone and volume controls in their maximum positions.

The purpose of the selector is just to variably select tone caps, not to make a hot pickup, but nevertheless it shows that more than just a cap is required.

I'm not sure if a combination of resistor and cap can get it to where it needs to be to adequately curve match. It might be the case that once you get the correct Q, you have the wrong slope, and vice versa. It would take some trial and error. If it could  be correctly modeled, it would be very easy to demo, but the modeling won't help with the slope apparently, so it would require some tedious trial and error with an analyzer, and an assortment of caps and resistors. 

The series resistor affects the Q and the HF slope.  The parallel resistor affects the Q, but not the HF slope in the high frequency limit.  You can start outside the guitar with pots (instead of fixed resistors) for a given value of C.  It will take some back and forth.  I have mostly just done parallel resistors; it seemed good enough to me, but it did seem that there was more that could be done.

[antigua]

Mar 28, 2017 9:57:59 GMT -5 ms said:

Mar 28, 2017 9:33:54 GMT -5 antigua said:

The purpose of the selector is just to variably select tone caps, not to make a hot pickup, but nevertheless it shows that more than just a cap is required.

I'm not sure if a combination of resistor and cap can get it to where it needs to be to adequately curve match. It might be the case that once you get the correct Q, you have the wrong slope, and vice versa. It would take some trial and error. If it could  be correctly modeled, it would be very easy to demo, but the modeling won't help with the slope apparently, so it would require some tedious trial and error with an analyzer, and an assortment of caps and resistors. 

The series resistor affects the Q and the HF slope.  The parallel resistor affects the Q, but not the HF slope in the high frequency limit.  You can start outside the guitar with pots (instead of fixed resistors) for a given value of C.  It will take some back and forth.  I have mostly just done parallel resistors; it seemed good enough to me, but it did seem that there was more that could be done.

That's a good idea, I'll give that a try. Even if it works out though, it sort of changes the game; we can't just say "instead of buying a new pickup, just put a cap across it", because it would turn out that you need a special recipe of capacitance and resistance to get to a particular goal. That's not entirely a bad thing, but it's much less simple and and straight forward, and would spawn a whole new pastime of figuring out pickup mod recipes.

[Charlie Honkmeister]

Mar 28, 2017 10:53:11 GMT -5 antigua said:

Mar 28, 2017 9:57:59 GMT -5 ms said:

The series resistor affects the Q and the HF slope.  The parallel resistor affects the Q, but not the HF slope in the high frequency limit.  You can start outside the guitar with pots (instead of fixed resistors) for a given value of C.  It will take some back and forth.  I have mostly just done parallel resistors; it seemed good enough to me, but it did seem that there was more that could be done.

That's a good idea, I'll give that a try. Even if it works out though, it sort of changes the game; we can't just say "instead of buying a new pickup, just put a cap across it", because it would turn out that you need a special recipe of capacitance and resistance to get to a particular goal. That's not entirely a bad thing, but it's much less simple and and straight forward, and would spawn a whole new pastime of figuring out pickup mod recipes.

Antigua, with a EDA design program (LTSpice, Circuitlab, etc.)  it's easy as pie to tweak a value and do a simulation run to look at the response.

Since JohnH has confirmed that pickup models in general do work well to a reasonable level of accuracy, given that the design in question has "reasonably" low eddy losses,  it's perfectly legitimate for experimentation to use a modeled pickup with parameters which are derived from real measurements of the pickup.  

I've been experimenting in this way for approximately two years,  on the way to developing the electronically variable capacitor/buffered approach I have been talking about on posts. 

Specifically, what I've come up with for the input network is:


where the pickup DCR is lumped into the simulation model of the inductor L1.   So,  ignoring C2,  which is the cap for the variable capacitance feature,  besides inductance,  we are only looking at pickup DCR,  stray capacitance (C_pickup),  voicing capacitance (C1),  series resistance (R1)  and shunt/parallel resistance (R2)  to provide voicing for the pickup,  which covers resonant frequency and Q value before rolloff.   

I'm ignoring amp input resistance and capacitance in this model because I'm feeding a JFET buffer with 10 Megohms input impedance, and very low capacitance, with it.  

If this looks like JohnH's  4 and 6 component models,  that's because we ended up in the same place.  But we're interested in modeling the whole instrument, cable and all,  alll the way to the amp input, to relate the component changes anywhere in the chain to how the instrument will sound.  

Just an extremely simple model like this will give lots of experimentation,  and you already have some of the answers to the question,  "What do I make it sound like?"   since you have been doing some great work on testing and publishing pickup electrical parameters.  

From building, modding, and testing,  I came up with some general observations:

1.  Resonant peaks on the order of 5 to 8 dB provide familiar tonality and are closest to emulating the RC loading of a passive instrument connected to a cable with significant capacitance.  This is true for the entire resonant frequency range of interest,  approximately 1.4 KHz through about 4.8 KHz.   You might need Q in the upper end of the 5-8 dB range  in certain situations like, for example,  emulating a particulary twangy Tele pickup.  Parallel/shunt resistance (R2) is the most important controlling value for setting this Q value.  

2.  Series resistance (R1 in the pic)  is needed to lower the Q into about the same 5-8 dB range, at low (less than 2  Khz or so) frequencies where the shunt capacitance is high. 

I'm sure that Mike's (ms's) observations are valid as well, and with a circuit simulator you can get very rapid feedback as to how a component change affects the response.
Also,  don't forget that with LTSpice you can run a simulated signal through the simulated circuit, and get output data files.  I've heard this is pretty slow, though.

I haven't particularly looked at the real-world slope of the rolloff above resonance but hope to do so when I set up a test bench similar to yours  (Circuitgear CGM-101,  driver coil,  Stratotarts integrator board, etc.,  all of which I already have on hand. ) 

What you have here is very interesting and I'd like to do some experimentation like yours to see how the rolloff slope affects tonality.

-Charlie

 

[antigua]

Mar 28, 2017 14:44:57 GMT -5 Charlie Honkmeister said:

Mar 28, 2017 10:53:11 GMT -5 antigua said:

That's a good idea, I'll give that a try. Even if it works out though, it sort of changes the game; we can't just say "instead of buying a new pickup, just put a cap across it", because it would turn out that you need a special recipe of capacitance and resistance to get to a particular goal. That's not entirely a bad thing, but it's much less simple and and straight forward, and would spawn a whole new pastime of figuring out pickup mod recipes.

Antigua, with a EDA design program (LTSpice, Circuitlab, etc.)  it's easy as pie to tweak a value and do a simulation run to look at the response.

Since JohnH has confirmed that pickup models in general do work well to a reasonable level of accuracy, given that the design in question has "reasonably" low eddy losses,  it's perfectly legitimate for experimentation to use a modeled pickup with parameters which are derived from real measurements of the pickup.  

I've been experimenting in this way for approximately two years,  on the way to developing the electronically variable capacitor/buffered approach I have been talking about on posts. 

Specifically, what I've come up with for the input network is:


where the pickup DCR is lumped into the simulation model of the inductor L1.   So,  ignoring C2,  which is the cap for the variable capacitance feature,  besides inductance,  we are only looking at pickup DCR,  stray capacitance (C_pickup),  voicing capacitance (C1),  series resistance (R1)  and shunt/parallel resistance (R2)  to provide voicing for the pickup,  which covers resonant frequency and Q value before rolloff.   

If this looks like JohnH's  4 and 6 component models,  that's because we ended up in the same place.

Just an extremely simple model like this will give lots of experimentation,  and you already have some of the answers to the question,  "What do I make it sound like?"   since you have been doing some great work on testing and publishing pickup electrical parameters.  

From building, modding, and testing,  I came up with some general observations:

1.  Resonant peaks on the order of 5 to 8 dB provide familiar tonality and are closest to emulating the RC loading of a passive instrument connected to a cable with significant capacitance.  This is true for the entire resonant frequency range of interest,  approximately 1.4 KHz through about 4.8 KHz.   You might need Q in the upper end of the 5-8 dB range  in certain situations like, for example,  emulating a particulary twangy Tele pickup.  Parallel/shunt resistance (R2) is the most important controlling value for setting this Q value.  

2.  Series resistance (R1 in the pic)  is needed to lower the Q into about the same 5-8 dB range, at low (less than 2  Khz or so) frequencies where the shunt capacitance is high. 

I haven't particularly looked at the real-world slope of the rolloff above resonance but hope to do so when I set up a test bench similar to yours  (Circuitgear CGM-101,  driver coil,  Stratotarts integrator board, etc.,  all of which I already have on hand. ) 

What you have here is very interesting and I'd like to do some experimentation like yours to see how the rolloff slope affects tonality.

-Charlie

 

I've done a lot of LTSpice modelling, and in fact I wrote a how-to guide guitarnuts2.proboards.com/thread/7842/modeling-electric-guitar-ltspice

Pretty much the entire issue at hand as to do with attenuation slopes past resonance. Getting the correct resonant peak and Q is not really that difficult, it's what happens beyond that point that is more tricky, the rate of attenuation by frequency. To be honest, I never gave it much thought, but the sound samples in the first post of this thread demonstrate how crucial the slope is to modeling the particular voice of a pickup.

As I mentioned above, with LTSpice and the models we've all been using, the capacitance is assumed to be lumped, but in reality it's interlaced with the resistance and the inductance of the coil, so an external capacitor doesn't cleanly combine with the coil to form a clean second order -12dB/oct LPF.  The steepness in the first "Strat" plot with the various caps shows a 19dB/oct slope, suggesting that the external cap combines with the pickup to form a third order LPF - that's a very big problem.

MS has a good suggestion about trying to curve match an actual pickup with resistance in series and parallel to the cap, so I ordered an assortment of trim pots and I'll give it a try.

I know you're doing active circuitry, so that's kind of a different thing. I'm still intent on passive solutions. Trying to make a Strat pickup sound like a P-90 is probably the most far-out nuts attempt as passive tweaking, it might be much more reasonably to simply make one Strat pickup sound like another, but if such a limitation exists, it's good to find out sooner than later.

[Charlie Honkmeister]

Antigua,

Oops,  forgot that it was you who did the excellent LTSpice tutorial ,  which I have turned multiple people on to.  Mea culpa.  I really appreciate your work on that.   

I don't think it's nuts at all to want  to get a P-90 sound in a Strat pickup, especially bridge position;  several makers have tried to go this way to fatten up the midrange honk and make the PU better suited for crunch and distortion.  

The ideal passive solution would be to switch components in and out to get the maximum versatility.  But you're going to run into inconsistency in output levels with all passive when you switch voicing,  and you still have the cable capacitance to deal with.  That's what finally drove me to active buffered.   

If you can tolerate lower output level in general,  if you go for a lower inductance, lower DCR pickup,  the job is easier because the cable capacitance will affect the pickup circuit less and allow you more tweaking range on setting resonant frequency and cutoff slope.

   

[Charlie Honkmeister]

The other approach that might work out for a single coil is possibly a tapped pickup where the full coil is a lower resonant frequency (P90)  and the fractional coil is more traditional Strat. 

That's done on both coils in a humbucker,  by the Joe Barden Two Tone.   It does get a pretty nice single coil sound in "switched/tapped"  mode, without much of a perceptible volume drop.  

[antigua]

Mar 28, 2017 15:37:40 GMT -5 Charlie Honkmeister said:

Antigua,

Oops,  forgot that it was you who did the excellent LTSpice tutorial ,  which I have turned multiple people on to.  Mea culpa.  I really appreciate your work on that.   

I don't think it's nuts at all to want  to get a P-90 sound in a Strat pickup, especially bridge position;  several makers have tried to go this way to fatten up the midrange honk and make the PU better suited for crunch and distortion.  

The ideal passive solution would be to switch components in and out to get the maximum versatility.  But you're going to run into inconsistency in output levels with all passive when you switch voicing,  and you still have the cable capacitance to deal with.  That's what finally drove me to active buffered.   

If you can tolerate lower output level in general,  if you go for a lower inductance, lower DCR pickup,  the job is easier because the cable capacitance will affect the pickup circuit less and allow you more tweaking range on setting resonant frequency and cutoff slope.

   

I think a passive solution might improve output issues. Currently if you put a P-90 in a Strat bridge, it would be a lot louder than the middle and neck. If you look at this plot, the amplitudes are all nearly identical, with nothing more than a parallel cap. I'm not sure if added resistance in parallel and series with the cap will change this, or to what degree. 



People are used to a P-90 "pushing" their amp, though, and that wouldn't happen.

If it can really be done I'd foresee cases where one "hot" pickup can be cast as another "hot" pickup, such as sending a Texas Special from 3.5kHz down to 2kHz in a way that is convincingly "P-90". Currently, a 3nF cap will get it there, but with a very steep drop off.  

[antigua]

Mar 28, 2017 15:53:33 GMT -5 Charlie Honkmeister said:

The other approach that might work out for a single coil is possibly a tapped pickup where the full coil is a lower resonant frequency (P90)  and the fractional coil is more traditional Strat. 

That's done on both coils in a humbucker,  by the Joe Barden Two Tone.   It does get a pretty nice single coil sound in "switched/tapped"  mode, without much of a perceptible volume drop.  

Speaking of, I discovered (insofar as I've never seen it mentioned elsewhere) that tapping causes huge parasitic capacitance. The unused portion of coil couples with the used half. With the SD SSL-4 I measured something around 600pF additional capacitance caused by the unused coil. In turn, SD makes the tap point really low, so that you have a very low inductance combining with a very high capacitance, for an overall low output. Add to that the sharper 3rd order slope of an external capacitance, and it's not wonder the tapped tone is so unpopular. The solution there is to disconnect the tapped coil at both ends. This requires 4 conductor cable, while most tapped pickups have only three conductors. It would also require a DPDT instead of an SPST switch, which is not a real big deal. If this was done, you could have a tapped single coil with two (or more) convincing voices. If I owned and operated a pickup company, I would have done this yesterday. 

[Charlie Honkmeister]

Mar 28, 2017 16:59:04 GMT -5 antigua said:

Mar 28, 2017 15:53:33 GMT -5 Charlie Honkmeister said:

The other approach that might work out for a single coil is possibly a tapped pickup where the full coil is a lower resonant frequency (P90)  and the fractional coil is more traditional Strat. 

That's done on both coils in a humbucker,  by the Joe Barden Two Tone.   It does get a pretty nice single coil sound in "switched/tapped"  mode, without much of a perceptible volume drop.  

Speaking of, I discovered (insofar as I've never seen it mentioned elsewhere) that tapping causes huge parasitic capacitance. The unused portion of coil couples with the used half. With the SD SSL-4 I measured something around 600pF additional capacitance caused by the unused coil. In turn, SD makes the tap point really low, so that you have a very low inductance combining with a very high capacitance, for an overall low output. Add to that the sharper 3rd order slope of an external capacitance, and it's not wonder the tapped tone is so unpopular. The solution there is to disconnect the tapped coil at both ends. This requires 4 conductor cable, while most tapped pickups have only three conductors. It would also require a DPDT instead of an SPST switch, which is not a real big deal. If this was done, you could have a tapped single coil with two (or more) convincing voices. If I owned and operated a pickup company, I would have done this yesterday. 

I remembered that you (I believe) had posted on the unused part of a tapped coil before.  It makes sense that that that would be the case for the unused windings. 

The Joe Barden basically just shorts the two tap points, one on each humbucker coil, together.  I have no idea why this wouldn't create response drop or high frequency loss,  but it does work sonically.   I have a HB Two/Tone bridge that I can send you if you like at some point, to test. 

FWIW,  Gibson has a few patents on a bifilar pickup coil.  Basically,  wind the whole thing with 2 wires laid together instead of one.  Besides creating a very exact 50% tap point if you wanted it,   you could disconnect the second winding and because the second winding is not active,   the effective spacing between and around the remaining winding  that is working is increased,  and that would reduce interwinding capacitance somewhat.  Also, by passive loading of the second winding, either connected or not connected to the first winding,  it might be possible to tailor the PU response in a more flexible way.

Gibson Bifilar Coil Patent       

[antigua]

Mar 28, 2017 22:11:18 GMT -5 Charlie Honkmeister said:

Mar 28, 2017 16:59:04 GMT -5 antigua said:

Speaking of, I discovered (insofar as I've never seen it mentioned elsewhere) that tapping causes huge parasitic capacitance. The unused portion of coil couples with the used half. With the SD SSL-4 I measured something around 600pF additional capacitance caused by the unused coil. In turn, SD makes the tap point really low, so that you have a very low inductance combining with a very high capacitance, for an overall low output. Add to that the sharper 3rd order slope of an external capacitance, and it's not wonder the tapped tone is so unpopular. The solution there is to disconnect the tapped coil at both ends. This requires 4 conductor cable, while most tapped pickups have only three conductors. It would also require a DPDT instead of an SPST switch, which is not a real big deal. If this was done, you could have a tapped single coil with two (or more) convincing voices. If I owned and operated a pickup company, I would have done this yesterday. 

I remembered that you (I believe) had posted on the unused part of a tapped coil before.  It makes sense that that that would be the case for the unused windings. 

The Joe Barden basically just shorts the two tap points, one on each humbucker coil, together.  I have no idea why this wouldn't create response drop or high frequency loss,  but it does work sonically.   I have a HB Two/Tone bridge that I can send you if you like at some point, to test. 

FWIW,  Gibson has a few patents on a bifilar pickup coil.  Basically,  wind the whole thing with 2 wires laid together instead of one.  Besides creating a very exact 50% tap point if you wanted it,   you could disconnect the second winding and because the second winding is not active,   the effective spacing between and around the remaining winding  that is working is increased,  and that would reduce interwinding capacitance somewhat.  Also, by passive loading of the second winding, either connected or not connected to the first winding,  it might be possible to tailor the PU response in a more flexible way.

Gibson Bifilar Coil Patent       

I just don't know enough about the Bardens to speculate, because the capacitance in a humbucker is greatly reduced by the fact that the capacitve clumps are in series with each other, which might mitigate the unused coil capacitance quite a lot. Though, even with a humbucker, if you truly want the unused coil to "be gone", you have to disconnect both ends. This is true even if you have a bifilar tapped coil.

[antigua]

Mar 28, 2017 9:57:59 GMT -5 ms said:

Mar 28, 2017 9:33:54 GMT -5 antigua said:

The purpose of the selector is just to variably select tone caps, not to make a hot pickup, but nevertheless it shows that more than just a cap is required.

I'm not sure if a combination of resistor and cap can get it to where it needs to be to adequately curve match. It might be the case that once you get the correct Q, you have the wrong slope, and vice versa. It would take some trial and error. If it could  be correctly modeled, it would be very easy to demo, but the modeling won't help with the slope apparently, so it would require some tedious trial and error with an analyzer, and an assortment of caps and resistors. 

The series resistor affects the Q and the HF slope.  The parallel resistor affects the Q, but not the HF slope in the high frequency limit.  You can start outside the guitar with pots (instead of fixed resistors) for a given value of C.  It will take some back and forth.  I have mostly just done parallel resistors; it seemed good enough to me, but it did seem that there was more that could be done.

I went ahead and tried this tonight with caps and trim pots, and had good results. The goal was to curve match a Strat pickup with 1.9H inductance to a P-90 with closer to 7.5H inductance, and what it took was 100k ohms parallel resistance and 330pF parallel capacitance. The gray line is the Strat pickup, and the black line is the P-90. Both have very close resonant peaks, and a slope that hits right around -12dB/oct.



The P-90 was in a Les Paul, and  the Strat pickup in a Strat, so the 330pF cap and the 100k ohm resistance are completely circumstantial values, and different ones would have to be used for different pickups and different existing loads, but it's a proof of concept anyway.

To show the difference the parallel resistance makes, here is the same graph, but I added a plot with the 330pF cap and no parallel resistor, the pink line, and you can see that the it both has a high Q and a steeper slope, resulting in a "cocked wah" sort of tone. 




One odd thing is that when I plugged into an actual amp with this cap and resistor connected (externally with alligator clips), the values both seemed too low. I have a Les Paul with P-90's on hand so I could go back and forth, between that and the modded Strat, and it seemed that I had to use a 2nF cap and up the parallel resistance to about 150k ohms to get it to sound like the P-90 / Les Paul. The bad news is I'm not sure why such higher values were needed, but the good news is that with adjusted values, I though it sounded close enough to a P-90, warm jazzy tone, albeit much quieter. The graph above shows a 5dB difference in amplitude, but I think it was even more than that in situ. 

I have a few guitars modded with variable cap tone controls, and based on these findings, I think I need to install a push pull that can throw in a 100k to 150k ohms parallel resistor in order to exploit the low value caps as a means of faking "hot" pickup tone. These trim pots will be great for fine tuning that value.



I also tried series resistance with the cap, which is essentially the same thing as a tone control at some value other than zero, and this is the closest I could get it, shown in the pink line below:



Getting the series resistance to this point was very difficult; compared to parallel resistance, the series resistance needed very exact values just to get this close. When it comes to curve matching, series resistance with capacitance might be capable of doing it, but using parallel resistance with capacitance seems like a much more sane way to go about doing it.

[Charlie Honkmeister]

Great persistence and nice work, Antigua.   You didn't mention when you did listening tests with an amp,  the cable you used and maybe an idea of its capacitance.  

If you used a typical cable length,  the cable (even if it's good quality)  and amp input Z may have affected the lower Z output Strat pickup less than the higher output Z P-90,  when actually plugged in.   (Just a WAG for why you had to add additional RC loading to the Strat pickup to match tonality.) 

[antigua]

Mar 30, 2017 0:21:05 GMT -5 Charlie Honkmeister said:

Great persistence and nice work, Antigua.   You didn't mention when you did listening tests with an amp,  the cable you used and maybe an idea of its capacitance.  

If you used a typical cable length,  the cable (even if it's good quality)  and amp input Z may have affected the lower Z output Strat pickup less than the higher output Z P-90,  when actually plugged in.   (Just a WAG for why you had to add additional RC loading to the Strat pickup to match tonality.) 

It probably does owe to the guitar cable or amp input. The fact that every rig is different would make it very difficult to associate any set of R and C values with a particular outcome. 

[ms]

Mar 29, 2017 22:52:14 GMT -5 antigua said:

Mar 28, 2017 9:57:59 GMT -5 ms said:

Getting the series resistance to this point was very difficult; compared to parallel resistance, the series resistance needed very exact values just to get this close. When it comes to curve matching, series resistance with capacitance might be capable of doing it, but using parallel resistance with capacitance seems like a much more sane way to go about doing it.

I was thinking you might need both the parallel and series resistor (which would be small if you use both) at once to do the job with two sounds so different, but you seem to have done it.  Now, a possible next step.  The tone circuit is not really what you say.  The .022 cap is nearly a short circuit at frequencies near the resonance, and so from 10 down a significant way, the tone control is just a Q control very much like the resistor you have been adjusting.  But just when it gets interesting, that capacitor gets in the way.  So, unless you are really in love with that ultra bassy sound you get with the tone on zero, replace the cap with a resistor, say 20K or whatever you like (just to stop a dead short at zero), and then you have a tone control that IMO is much more useful for generating a wide range of useful tones.  It complements your capacitor switching without having to put Rs across each C. You adjust to what you want at the moment.  Of course, you run out of holes in the guitar body if you want to do everything, so you have to compromise somehow.  But volume, cap switch, and Q adjust tone control is one way to go.  But it is a matter of personal preference.

[Charlie Honkmeister]

Mar 30, 2017 1:38:31 GMT -5 antigua said:

Mar 30, 2017 0:21:05 GMT -5 Charlie Honkmeister said:

Great persistence and nice work, Antigua.   You didn't mention when you did listening tests with an amp,  the cable you used and maybe an idea of its capacitance.  

If you used a typical cable length,  the cable (even if it's good quality)  and amp input Z may have affected the lower Z output Strat pickup less than the higher output Z P-90,  when actually plugged in.   (Just a WAG for why you had to add additional RC loading to the Strat pickup to match tonality.) 

It probably does owe to the guitar cable or amp input. The fact that every rig is different would make it very difficult to associate any set of R and C values with a particular outcome. 

Antigua,  you know that you're just talking yourself, bit by little bit, into an active buffered approach, right?          No worries,  I like passive, too.  

[antigua]

Mar 30, 2017 8:26:08 GMT -5 Charlie Honkmeister said:

Mar 30, 2017 1:38:31 GMT -5 antigua said:

It probably does owe to the guitar cable or amp input. The fact that every rig is different would make it very difficult to associate any set of R and C values with a particular outcome. 

Antigua,  you know that you're just talking yourself, bit by little bit, into an active buffered approach, right?          No worries,  I like passive, too.  

This is a case where my interests and the current limits of my knowledge really compliment each other; I don't understand the implications of active circuitry very well, and I hate batteries.

I just prototyped an external mount cap based filter and kill switch, which I posted a bit about in another forum. I haven't mentioned it here because it's not pickup specific, although I did I can see this possibly leading to external mount pickup modding. The crucial aspect of it is though, is that your guitar, which never required batteries, still won't. The idea that you can take a guitar out of a case in fifty years and it will still work, almost like new, is part of the beauty of it all. Some guys don't care, and they happily use EMG and onboard active electronics, and they're not wrong, we just have different ideals. The thing of it is, if you have a battery, you go down the rabbit whole of eventual digital manipulation, and forget modelling pickups, at that point you can make the guitar sound like a piano, or whatever. There's no longer a purity of concept. 

[frankfalbo]

What does having a battery onboard have to do with digital manipulation? Or purity of concept? Literally the very next thing your guitar signal contacts on the other side of the cable is powered.

[antigua]

Mar 30, 2017 17:27:02 GMT -5 frankfalbo said:

What does having a battery onboard have to do with digital manipulation? Or purity of concept? Literally the very next thing your guitar signal contacts on the other side of the cable is powered.

The idea that you can take a guitar out of a case in fifty years and it will still work, almost like new.

[stratotarts]

I think the active/passive choice has a lot to do with the cost/benefit trade off. The active choice has other limiting factors besides the cost of the circuitry. There is the hassle of the mods to the guitar to accommodate it and also the battery. If it is a stock feature of the guitar, it adds cost. It also adds a maintenance overhead for the player, as they have to maintain the freshness of the battery and replace it when it fades. Then there is the aesthetic - it conflicts with the common preference for things traditional and "vintage". All this for real advantages that most players don't really understand and might not really need. It would be more attractive, at least to players with some technical interests, if it were possible to add a few more useful design features to the basic concept. In fact, I want to do this. The "rabbit hole" is where the features get out of control and start escalating the cost and maintenance overhead again. Such systems have already been marketed and have some remarkable abilities, but I think deviate from the basic guitar concept to a degree that it almost becomes a different instrument. I want to respect the intrinsic nature of the instrument while using active circuitry to improve the player's control options, at the same time as providing the already well known basic advantages such as indifference to cord length. The important thing is to hit a sweet spot where numerous and useful advantages can be had with the maximum simplicity and lowest cost.

Oh, and most importantly, as with all my projects so far, I want to share my work freely and openly. Free as in both "free beer" and "free speech".

[ms]

Mar 30, 2017 18:26:18 GMT -5 stratotarts said:

I think the active/passive choice has a lot to do with the cost/benefit trade off. The active choice has other limiting factors besides the cost of the circuitry. There is the hassle of the mods to the guitar to accommodate it and also the battery. If it is a stock feature of the guitar, it adds cost. It also adds a maintenance overhead for the player, as they have to maintain the freshness of the battery and replace it when it fades. Then there is the aesthetic - it conflicts with the common preference for things traditional and "vintage". All this for real advantages that most players don't really understand and might not really need. It would be more attractive, at least to players with some technical interests, if it were possible to add a few more useful design features to the basic concept. In fact, I want to do this. The "rabbit hole" is where the features get out of control and start escalating the cost and maintenance overhead again. Such systems have already been marketed and have some remarkable abilities, but I think deviate from the basic guitar concept to a degree that it almost becomes a different instrument. I want to respect the intrinsic nature of the instrument while using active circuitry to improve the player's control options, at the same time as providing the already well known basic advantages such as indifference to cord length. The important thing is to hit a sweet spot where numerous and useful advantages can be had with the maximum simplicity and lowest cost.

Oh, and most importantly, as with all my projects so far, I want to share my work freely and openly. Free as in both "free beer" and "free speech".

And it does not take long if you live in a hot damp tropical climate for the battery to corrode, leak, and destroy most of what is in the same compartment with it. 

[stratotarts]

The most standard external power is a phantom powered XLR - would that be so bad as a guitar cable? You could plug right into any mixing board. Or a mic preamp like this:
link

[pablogilberto]

This is well written antigua
Kudos!

I really like your approach.

This answers the question why pickups of different types (strat, tele, P90, HB, etc) will sound differently even if we wind them to have the same Resonant Frequency and Q.

We really need to consider the effects of Eddy Currents due to metal parts PLUS we need to take a look at what happens with the slope past the Resonant Freq.


I just have additional questions:

1. If we consider a strat SC pickup, can we make a generalization that 2 pickups with the same Resonant Freq will really sound alike because they have no Eddy Current Issues and the Roll-Off past the Resonant Freq will be the same?

Are we expecting the same roll-off because they are the same (geometrically) or do we still need to examine the roll-off for they still might differ significantly?


2. Where can we attribute the sound a particular pickup type, say Telecaster Pickups? I mean the Funky Trebly sound that we get from it compared to the clean and glassy tone of Vintage (cool) strat pickups?

3. I also noticed that adding a parallel capacitance seems to slightly boost the resonant peak (dB) while lowering the resonant frequency.
I did some simulation in SPICE to check what happens when we increase a guitar cable capacitance from 300pF up to 1000pF.
The higher capacitance sounded so muddy because of the lowered resonant freq. But I'm also interested about the sonal effect of the resonant peak being slightly boosted as the capacitance is increased. Does it impact the tone? Any thoughts on this?

4. Regarding your experiment of adding a series/parallel resistor and parallel cap... Have you concluded something about it? Is it worthwhile to try shaping a freq response using this method to recreate another pickups freq response so we can get a similar tone?

5. "but the induction of voltage by way of a moving magnetized guitar string imposes a +6dB/oct roll off" - I want to understand this better. Can you please point me to resource which discuss this further? Or maybe you can just enlighten me on this?
I think this is connected to why we use the Integrator Circuit in generating the freq response curve, correct?

6. How do we define and measure the, Q (Quality Factor) in our use case here?

7. "with LTSpice and the models we've all been using, the capacitance is assumed to be lumped, but in reality it's interlaced with the resistance and the inductance of the coil, so an external capacitor doesn't cleanly combine with the coil to form a clean second order -12dB/oct LPF. "
- Can you please explain this further?

Thanks a lot!

[antigua]

Jun 22, 2020 9:31:38 GMT -5 pablogilberto said:

This is well written antigua
Kudos!

I really like your approach.

This answers the question why pickups of different types (strat, tele, P90, HB, etc) will sound differently even if we wind them to have the same Resonant Frequency and Q.

We really need to consider the effects of Eddy Currents due to metal parts PLUS we need to take a look at what happens with the slope past the Resonant Freq.


I just have additional questions:

1. If we consider a strat SC pickup, can we make a generalization that 2 pickups with the same Resonant Freq will really sound alike because they have no Eddy Current Issues and the Roll-Off past the Resonant Freq will be the same?

Are we expecting the same roll-off because they are the same (geometrically) or do we still need to examine the roll-off for they still might differ significantly?


2. Where can we attribute the sound a particular pickup type, say Telecaster Pickups? I mean the Funky Trebly sound that we get from it compared to the clean and glassy tone of Vintage (cool) strat pickups?

3. I also noticed that adding a parallel capacitance seems to slightly boost the resonant peak (dB) while lowering the resonant frequency.
I did some simulation in SPICE to check what happens when we increase a guitar cable capacitance from 300pF up to 1000pF.
The higher capacitance sounded so muddy because of the lowered resonant freq. But I'm also interested about the sonal effect of the resonant peak being slightly boosted as the capacitance is increased. Does it impact the tone? Any thoughts on this?

4. Regarding your experiment of adding a series/parallel resistor and parallel cap... Have you concluded something about it? Is it worthwhile to try shaping a freq response using this method to recreate another pickups freq response so we can get a similar tone?

5. "but the induction of voltage by way of a moving magnetized guitar string imposes a +6dB/oct roll off" - I want to understand this better. Can you please point me to resource which discuss this further? Or maybe you can just enlighten me on this?
I think this is connected to why we use the Integrator Circuit in generating the freq response curve, correct?

6. How do we define and measure the, Q (Quality Factor) in our use case here?

7. "with LTSpice and the models we've all been using, the capacitance is assumed to be lumped, but in reality it's interlaced with the resistance and the inductance of the coil, so an external capacitor doesn't cleanly combine with the coil to form a clean second order -12dB/oct LPF. "
- Can you please explain this further?

Thanks a lot!

1) yes, if there are no significant eddy currents, or differences in terms of eddy currents, the rate of roll off will be the same. If the DC resistance was different, that would affect it to, but such a big difference is uncommon in practice.

2) Telecaster bridge pickups are usually higher inductance, 3 henries or greater, Strats usually closer to 2.5 henries or lower.  Tele neck pickups often have metal covers, and both have flat pole pieces where as Strats are usually staggered.

3) boosting the resonant amplitude definitely makes a pickup sound more muddy. In order to model a higher inductance pickup with low inductance and a capacitor, you have to also add parallel resistance to bring the resonant amplitude back down. It happens because the ratio of inductance to capacitance come closer to a 1:1 ratio (but is still far from it) as you add capacitance. 

4) I've got low inductance single coils to sound like a Seymour Duncan Little '59, which was a goal I had. It requires a precise amount of capacitance, and while you can do it via measurement, it's also fairly easy to do by ear. If you put the cap and resistor combo on a push pull pot, you can get that fatter tone on demand. 

5) the 6dB roll off is a consequence of reactance, and the "AC resistance" being a frequency dependent effect. As frequency goes up, the reactance become less and the voltage drops. Check this page out www.electronics-tutorials.ws/filter/filter_2.html

6) the most common way to define Q factor is bandwidth between 0 and -3dB at the resonant peak, but I don't usually bother to calculate that out, I just note the amplitude at resonance above the flat portion of the signal. The integrated bode plots make that easy to do. The -3dB value is more time consuming to figure out.

7) I'm not as sure about that, I would have to refresh my memory and go over it all again to see if I still agree with what I said a few years ago. I think the eddy currents of the P-90 might be the bigger reason for a difference in slope and not the series resistance of the coil.

[aquin43]

One major difference between the P90 and the typical Strat pickup is that eddy currents make the P90 inductance and resistance vary considerably with frequency in the region around 3kHz but the Strat pickup not so much.

An interesting side effect of P90-fying a Strat pickup with a parallel network is that the frequency response varies much less with the volume setting than it does with a real P90:

The brown lines are the P90 and the red the typical Strat pickup with the conversion network.