guitar pots

[uvacom]

Not so easy. You will need a leakage inductance in series with a resistor that varies with the square root of frequency across the pickup. But the effects are large enough to matter.

A new model the impedance of a humbucker, compared to measurements

It's totally possible to create component models in pspice with frequency dependency - I was just reading about it the other day here.


Also, I found the problem with my simulation - it's not with the model, but with how I was formatting the results. I was plotting the voltage using the LOG (which PSPICE actually interprets as the natural log, usually expressed as ln) function instead of LOG10 (on my TI-89 LOG10 is just LOG, so that's how I'm used to typing it).

Anyway, that caused the scaling of the Y axis to be off, the difference between the peak at resonance is just a little less than 3dB. Still significant enough to take into account, considering that both our ears and guitar speakers are quite sensitive in this range.

[Steve Conner]

Well, it's the difference between getting into my 3dB Club, or not.

Why exactly do amp builders need to know this? The standard 1M/100pF input of a guitar amp is negligible compared to the guitar's volume pot and the cable capacitance, so you have a neat "separation of concerns" at the input jack. The guitar pots and bright caps are the luthier's problem, and the cable capacitance is the owner's problem. (Yes, guitars sound noticeably different with different lengths and brands of cable, for sound scientific reasons.)

Although some old amps with multiple input jacks have 136k input impedance when only one input is used. That would negate the benefit of a 500k volume pot. I call these Type S after my old Selmer.

Yet other old amps had 136k if you plugged into one jack, and 1M if you plugged into the other one. Sometimes they were labelled Lo and Hi. I'd call this Type F because it's how Fenders were wired. I found that this was a nice tonal option to have: the low impedance input would subdue hot pickups a bit by loading them down, whereas the Hi-Z one would let them yowl out.

If you have a Type S, you can just stick an unwired 1/4" jack plug in the other hole for a gain boost.

[Mike Sulzer]

That looks like a really good way to do it.

There still remains the problem of finding out the values of the leakage inducdance and the frequency variable resistance. I think the modeling approach attempting to match the measurements gives reasonably accurate values, but there is not universal agreement.

If you take the values I found, even though they are for a different pickup, and put them in a pspice model, yoo should see the dip in the response below the resonance.

resonant peak: Yes, 3db at the peak is quite audible. The resonance location and its height are very important to the tone of the pickup, probably the most important factor. it is surprising that more people do not work directly with it, rather than relating a list of physical properties (shape of coil, type of core, etc.) to the tone that they hear.

It's totally possible to create component models in pspice with frequency dependency - I was just reading about it the other day here.


Also, I found the problem with my simulation - it's not with the model, but with how I was formatting the results. I was plotting the voltage using the LOG (which PSPICE actually interprets as the natural log, usually expressed as ln) function instead of LOG10 (on my TI-89 LOG10 is just LOG, so that's how I'm used to typing it).

Anyway, that caused the scaling of the Y axis to be off, the difference between the peak at resonance is just a little less than 3dB. Still significant enough to take into account, considering that both our ears and guitar speakers are quite sensitive in this range.

[Steve Conner]

resonant peak: Yes, 3db at the peak is quite audible. The resonance location and its height are very important to the tone of the pickup, probably the most important factor. it is surprising that more people do not work directly with it, rather than relating a list of physical properties (shape of coil, type of core, etc.) to the tone that they hear.

I argued years ago that pickup makers should forget all of this magic physical property voodoo and assume the pickup to be modelled by a 2-pole lowpass filter. I got a pasting for it.

It opens up all sorts of possibilities like a "Linkwitz Transform" pedal for turning one pickup into another.

[uvacom]

That looks like a really good way to do it.

There still remains the problem of finding out the values of the leakage inducdance and the frequency variable resistance. I think the modeling approach attempting to match the measurements gives reasonably accurate values, but there is not universal agreement.

If you take the values I found, even though they are for a different pickup, and put them in a pspice model, yoo should see the dip in the response below the resonance.

resonant peak: Yes, 3db at the peak is quite audible. The resonance location and its height are very important to the tone of the pickup, probably the most important factor. it is surprising that more people do not work directly with it, rather than relating a list of physical properties (shape of coil, type of core, etc.) to the tone that they hear.

Yeah, I was just playing around here with different values for the Rvar and the inductance and it does appear to produce those results, but the values I was choosing were purely a shot in the dark.

I don't want to diminish what pickup winders do because I couldn't do it myself, and I'm fascinated by the somewhat arcane approach which is taken, but it has always surprised me a little that the different variables (as you said, coil shape, wire gauge, etc.) aren't better parameterized in terms of the response.

[Chuck H]

Why exactly do amp builders need to know this? The standard 1M/100pF input of a guitar amp is negligible compared to the guitar's volume pot and the cable capacitance, so you have a neat "separation of concerns" at the input jack. The guitar pots and bright caps are the luthier's problem, and the cable capacitance is the owner's problem. (Yes, guitars sound noticeably different with different lengths and brands of cable, for sound scientific reasons.)

The same reason a chef may want to know where the beef came from, or a mechanic may want to know the hardness of a chassis bolt. It's not the chefs job to make beef, but it is his job to prepare it. It's not the mechanics job to engineer the car, but it is his job to do an adequate repair that won't fail again.

There are a couple of other input circuits on guitar amps. But yes, for the most part it's the standard 1M load in behind a 68k resistor to the input stage. However, some amps seem to handle different types of guitars with more grace. Same input circuit, different first amplifier circuit.

For me, as an amateur who found himself in a position to do some design work, I'm overwhelmed often by what I don't know. One problem I've had on occasion has been limited access to different guitars for amp testing purposes. So the more understanding I have of the resonant characteristics of guitar circuitry the better. At least that's how I see it.

Chuck

[Mike Sulzer]

Yes, the resonant low pass filter is very important for determining the sound. There are other things, of course. I was somewhat surprised that the eddy currents in the core are responsible for a dip in the response below the peak, as uvacom and I were discussing. Unless you explicitly account for the resonant peak, it is very difficult to determine the physical causes of other effects on the sound. For example, changing the core material could affect both the inductance and the eddy current losses. You need to be able to separate the two effects to understand what is happening.

I argued years ago that pickup makers should forget all of this magic physical property voodoo and assume the pickup to be modelled by a 2-pole lowpass filter. I got a pasting for it.

It opens up all sorts of possibilities like a "Linkwitz Transform" pedal for turning one pickup into another.

[Glass Snuff]

Do you see any glaring discrepancies? I use terry downs's model as a starting point, but I've just started making the model more elaborate by modeling the pickup as a network of distributed elements (instead of just lumped elements), and eventually I'd also like to model effects such as eddy currents, etc.

I plugged your values into my Spice program and got similar results to yours. My model was closer to a Duncan Ant humbucker than a Tele pickup.

A few years ago I did a lot of work with LSpice, and found Terry's model didn't always conform to what I was hearing. As I recall, I did away with the transmission line for the cable. Sadly, I had a hard drive crash, and lost all my tweaks. It would be good to come up with more accurate models, especially for the pickup.

[uvacom]

I could bring a guitar cable into the lab and use one of the network analyzers to determine the actual characteristic impedance and group delay. But at audio frequencies I tend to think that modeling it as a lumped capacitance would be sufficient. I've just been using the transmission line model for "completeness".

[Steve Conner]

The electrical length of a guitar cord is much less than a quarter wavelength of the highest frequency of interest, and the circuit impedances are much higher than whatever the characteristic impedance of guitar cord is, so it's quite all right to use the lumped capacitance model. The transmission line model is essentially just a waste of CPU cycles (like 99% of other software these days )

[Mike Sulzer]

The electrical length of a guitar cord is much less than a quarter wavelength of the highest frequency of interest, and the circuit impedances are much higher than whatever the characteristic impedance of guitar cord is, so it's quite all right to use the lumped capacitance model. The transmission line model is essentially just a waste of CPU cycles (like 99% of other software these days )

If you do specify a transmission line, you need to specify its characteristics correctly, or you will not get the correct value of the capacitance in the computation. It is much easier just to use some typical value like 500 pf.

[BatRastard]

Interesting stuff. What should the capacitance read on a good cable? I ask because I recently purchased two cables on "special" and get tons of noise
in both when plugged in. They read 1.6nf while my other cables are in the .60-.80nf range.

[billyz]

Noise from cables is usually due to poor construction. Particularly the shielding. High quality braided shield has worked the best for me. It could also be the dielectric insulator is causing noise especially when moved. Good quality cable is pretty noise free regardless of it's capacitance or resistance.

[trevorus]

As Glass Snuff said, changing the value of the pots will "open up" the sound a bit. More clarity, more punch. I tend to like 500K's with humbuckers. They tend to keep them from being dark or muddy. 300K isn't a ton different from 250K, especially with the tolerances of the resistor tracks. Humbuckers always sounded a little dull with a 250K volume IMO.

[trevorus]

Noise from cables is usually due to poor construction. Particularly the shielding. High quality braided shield has worked the best for me. It could also be the dielectric insulator is causing noise especially when moved. Good quality cable is pretty noise free regardless of it's capacitance or resistance.

Oddly enough, one of my nicest, clearest cables has some physical noise. It seems that the construction of the shield and conductor arrangement can contribute to this. I used to use it with a G&L bass I had, and when I clicked it to active mode, the noise was non-existant. Seems like the load at the guitar end has a lot to say about this sort of thing.