Question about frequency response of HB pups...

[freefrog]

I have now hundreds of screenshots showing the frequency responses of various pickups when they are excited by a low impedance coil.

For obvious reasons, these curves can vary according to...
-the capacitance of the cable used;
-the strength of the signal sent to the pickup;
-the volume of the guitar input which collects the signal sent by the pickup;
-the external hardware involved: pots and their value, added wiring and its capacitance, covers...

Most of the time, I can "duplicate" these physical measurements through a 5Spice simulation.

BUT there's ONE thing that I don't understand: why, with some pickups, does the extreme high range rolls off abruptly while its decay remains smooth with other pickups?

To show what I mean, here is a quadruple screenshot showing the response, under variable conditions, of 4 PU's:



The response of the P90 has roughly the shape of a triangle. You would see the same shape if I showed you the response of a Duncan SH55, Gibson Burstbucker or Gibson T-Top.

BUT with the Duncan APH1, SH1 and PG1 tested in my picture above, you can see how the high frequencies die above 12 or 15khz. You'd notice the same thing with a Jackson J50BC, for example...

This phenomenon doesn't appear to depend on magnets (Alnico 5 or 2 as well as ceramic in the examples above). It doesn't appear to depend on baseplates either, nor on wiring insulation, slugs or screw poles, which are roughly the same in vastly different frequency response screenshots but which differ enormously in some other pu's whose response has the same shape...

WHAT is for you the cause of this behaviour (= tendency of some PU's to roll off the extreme highrange), knowing that I've "excited" each pickup from its "screw poles" coil (the frequency response would be different if measured from the "slug poles" bobbin)?

Could it depend on wax potting, for example?

Thx in advance for any documented answer. :-)

[Mike Sulzer]

Are you sure the P-90 does not do the same thing but at a higher frequency, just off your plot? That is what it looks like to me.

Anyway, who cares? Connect a humbucker to a cable and it will peak probably no higher than 4.5 KHz (often a lot lower) and have no significance response up where you are concerned.

The guitar speaker also falls like a brick at about 5 KHz.

[freefrog]

Are you sure the P-90 does not do the same thing but at a higher frequency, just off your plot? That is what it looks like to me.

What you see is due to the hardware involved - mostly the guitar cable used: if I plug the testing gear in itself through such a cable, I notice the same high frequency roll off.

Anyway, who cares? Connect a humbucker to a cable and it will peak probably no higher than 4.5 KHz (often a lot lower) and have no significance response up where you are concerned.

The guitar speaker also falls like a brick at about 5 KHz.

If I've opened this topic, it's because I care! LOL... I like to understand, simply. :-)

Of course, you're right about the filtering effect of humbuckers and guitar speakers.

That said, after almost 10 years of such tests, I can tell you that the "low pass" filtering shown in my screenshot still contributes to the overall "sonic personality" of the mentioned pickups. Hence my desire to understand why they behave like that.

Thx for your answer, BTW.

[Mike Sulzer]

I agree. Understanding is important. Would you describe your set up?

[peskywinnets]

The guitar speaker also falls like a brick at about 5 KHz.

But not everyone uses a guitar speaker - consider the home recordist's signal path...

Guitar->High Quality soundcard->hifi amp->Hifi Speakers..... all capable of well in excess of 5khz.

[Steve Conner]

Sure that is possible, but who actually plays their guitar like that? OK, I do sometimes and have used the effect on tracks.

I can't think what causes the notch, unless it's another resonant mode. A humbucker is two coupled coils each with self-capacitance, so it could have two resonant frequencies.

Freefrog: thanks for sharing your results!

[David Schwab]

If you are using a clean guitar tone, especially with single coils, you often have some very high frequency stuff going on. I had a Bill Lawrence humbucker that was very glassy sounding.

I record bass (and guitar) direct all the time. The high frequency EQ control on my Roland mixer is set at 11.4K (Q=7), and I can hear changes in the signal of both guitars and basses just fine when I change the high EQ, so there are putting out frequencies in that range. But the response does start to drop off after the resonant peak.

[Mike Sulzer]

Steve, does this make sense?

The pickup circuit is a low pass filter which means that it is flat at low frequencies. But when you couple to it from a coil driven with a constant current, you see a response that rises with frequency because dB/dt is increasing with frequency. So that makes sense. But other things do not seem to:

1. Why do all the responses peak at under 2.5 KHz?

2. Why is there no sign of a resonance? (There is a peak, but not the faster rise and then fall that you expect from a resonant circuit near the peak.)

One explanation would be that a fairly high capacirtance guitar cable was used, and the input impedance of the analyzer is loading down the circuit. But there could be other explanations as well.

[freefrog]

I can't think what causes the notch, unless it's another resonant mode. A humbucker is two coupled coils each with self-capacitance, so it could have two resonant frequencies.

Freefrog: thanks for sharing your results!

You're welcome. :-)

I agree about the possible "dual resonance": DiMarzio has patented this idea if my memory serves me and anyway, their HB's do often exhibit a double resonant peak. The HB's of a Gibson Nighthawk, for example, do the same. But precisely, it's a second peak and not a mere abrupt high frequencies roll off like in my screenshots above.

1. Why do all the responses peak at under 2.5 KHz?

2. Why is there no sign of a resonance? (There is a peak, but not the faster rise and then fall that you expect from a resonant circuit near the peak.)

One explanation would be that a fairly high capacirtance guitar cable was used, and the input impedance of the analyzer is loading down the circuit. But there could be other explanations as well.

You're totally right: intentionally, I've used a 900pf guitar cable and purposedly, the circuit is loaded down, by the pots of the guitar + the guitar input which feeds the analyzer.
I could have tested the pu's "directly". I've often done it in the past. But generally don't keep these results because I don't find them meaningful. Testing a PU in a guitar wiring and through a guitar cable shows much more of its "character" IMHHHHHHHHHO.
FWIW, the PU's tested above still roll off the highrange when they are tested out of any circuit. The phenomenon is just easier to see through some guitar wiring/cable/amp input.

To answer to your question: my set up is quite simple and "amateurish". I excite the pu's with a low impedance coil fed by a calibrated soundcard. The guitar signal goes back to the soundcard through a "guitar amp input buffer" (with a 100pf input cap explaining the tight bass in my screenshots) then it feeds the analyzer.

As explained in my very first post on this forum, this method is not pro at all (I'm not a pro) but it gives me pretty consistent results, regardless of the exciter, soundcard and RTA used. And it generally matches the 5Spice analysis that I do in the same time. The only unexplained difference that I notice is the kind of high frequency roll off shown above in the 1st post. :-))

I guess that it has to do with the magnetic behaviour of the PU's, that a 5Spice analysis can't modelize...

[Steve Conner]

Mike, yes, the responses are differentiated, they look like bandpass instead of lowpass. One thing I could never figure out is whether the pickup looks like bandpass or lowpass in a guitar. There's nothing to compare it to. For instance you could say it is bandpass for string displacement and lowpass for string velocity, but which one of those is the true "input" as far as transfer function theory is concerned?

I don't see how that 100pF cap can cause "tight bass", I think it is an accidental differentiation as Mike says.

Two coupled resonant circuits normally just have two peaks, yes. But a humbucker also has stray capacitance to ground at the coil tap. With one coil driven much more than the other, this could end up looking like some crazy notch filter. Lemme places the exciting coil horizontally between the two sets of pole pieces, to excite both coils equal and opposite. Maybe you can try that and see if the notch is still there.

[freefrog]

Mike, yes, the responses are differentiated, they look like bandpass instead of lowpass. One thing I could never figure out is whether the pickup looks like bandpass or lowpass in a guitar. There's nothing to compare it to. For instance you could say it is bandpass for string displacement and lowpass for string velocity, but which one of those is the true "input" as far as transfer function theory is concerned?

I don't see how that 100pF cap can cause "tight bass", I think it is an accidental differentiation as Mike says.

Here is a 5spice analysis of a single coil simulation (= a LRC circuit plugged through a vol + a tone pot to a 1M input; a 1n cap is there to simulate the guitar cable).

Below is the same simulation with a 100pf cap added in series between the single coil sim and the amp input sim…



You see the bassrange: it's flat in the first picture, “tight” in the second one, which looks close to my “real life” tests on guitar PU’s.

Please keep in mind that English is not at all my mother tongue and that I’m not engineer in electronics either. So, I struggle sometimes to express my thoughts. :-)

Two coupled resonant circuits normally just have two peaks, yes. But a humbucker also has stray capacitance to ground at the coil tap. With one coil driven much more than the other, this could end up looking like some crazy notch filter. Lemme places the exciting coil horizontally between the two sets of pole pieces, to excite both coils equal and opposite. Maybe you can try that and see if the notch is still there.

Thx for the "input" :-))
FWIW, I've already tried to put my exciter coil like Lemme does (= perpendicular to the pickup between its two coils) but the results have not been fundamentally different compared to the coil above the screw poles. They widely differ only when I put the exciter directly above the slug poles.
Now, I understand that my test itself could generate as an artifact the effect discussed here. On the other hand, I can tell you that pickups with such a high freq roll off don't sound like those whose response is merely "triangular" looking.
In fact, their highrange roll off can ever be seen on a spectrum analysis if I play the guitar through a wide range preamp without cab.

[Steve Conner]

Oh, the 100pF is in series, not shunt? Well that certainly explains the differentiated look. Why do you use such a small capacitor, do you like the effect? I don't know of any guitar amp that cuts bass so drastically.

I still think the notch is some sort of cancellation caused by phase shifts due to stray and self capacitances. Maybe also magnetic coupling between the humbucker's two coils.

[salvarsan]

I have now hundreds of screenshots showing the frequency responses of various pickups when they are excited by a low impedance coil.

First, thanks for doing all that heavy lifting.

Could you please give more details about your test setup?

I am curious about your signal generator and drive coil.

thx.

[FunkyKikuchiyo]

This is probably a stupid question and overly obvious - but are you doing this through a typical guitar wiring setup? 500k pots and tone caps and all of that? Won't capacitances from there cause this sort of thing...?

[freefrog]

First, thanks for doing all that heavy lifting.

Could you please give more details about your test setup?

I am curious about your signal generator and drive coil.

thx.

In this case, the drive coil is a 16ohm round device (diameter: 15mm) and the "signal generator" is just... the headphones output of a SBLive soundcard (!).

In the past, I have used other coils of different DC resistance, shape and size, as well as various soundcards and RTA's, but it hasn't much incidence on the results.

FWIW: I have an heavier test setup based on a "custom home made" input / output interface, freely designed, built on order and calibrated by a member of my family who draws circuits for a big international brand (on this basis, you'll guess why I don't want to be more precise).
Of course, this heavier system is more accurate... but it's also much more complex to use...
...and its results are not fundamentally different compared to the "quite amateurish" solution described above.
Except a glitch or artifact here and there, the two systems show roughly the same curves for the same pickups so I tend to use the simplest solution. :-))