Circuit for Guitarfreak

[frets]

Hi Guys,

I’m posting this in a new thread as I promised to put up a circuit. John can decide if he wants to move it to a PM.

Anyway, I got this circuit through a friend given my interest in voicing filters. It is most complex in terms of components and before I build it, I would like to analyze it. I also thought it would be a good one to learn the aspects of Guitarfreak.

[Deleted]

I know I'm on Linux and some things are a pain in the butt (if I'm allowed to say but)

I've not really use GuitarFreak but I know it's a great calculator
And I tend to use spreadsheet to make circuits .. so is some thing I would like to learn to use

[JohnH]

Hi Frets, thanks for posting this.

Actually, its not something that can be properly modelled in GF as it currently exists. The reason is that its using the two coils of a pickup, and putting components between them that affect one coil differently to the other. In the current GF , you can model simple combinations of pickups and coils in series and parallel, and out of phase, but then those are combined into one equivalent pickup before they meet the tone shaping circuitry. So there's no way to put for example, components across one coil only, out of two. GF will do things like Strat circuits with series and parallel switching, then leading to one tone and volume circuit. On a Les paul, it can model either pickup, or both pickups provided that both sets of knobs are set the same.

To do these circuits where coils or pickups are tone-shaped differently, it needs a doubling and rewrite of the code so that each coil can have its own analysis. This would even apply to a Les Paul, where both pickups are engaged and one is turned down. This is on my 2-do list, but its probably nearer to being a bucket list!

But if you're interested in this circuit, we can build a SPICE model to investigate it

On GuitarFreak, did my answer on the other thread yesterday help? You can reply here about it and if it needs more Ill start another thread

[frets]

Hi John,
You know my ADD, I didn’t see that post until just now (when your referenced it).  Okay, I’ll try that out and see if I can do it in Guitarfreak.  I don’t want anybody spending time trying to Spice out that circuit I posted.  It’s not all that important, I just wanted to know what you told me in the post I did not see.

Geez, do I make any sense?😸😁

As always, thanks John!!

[sumgai]

We don't really need to look much further than the switching logic to see what's happening here.

On the left side of the switch, the three positions are:

a) center = neither cap in series with the coil;
b) one of the caps in series with the coil;
and
c) the other cap in series with the coil.

On the right side, we see:

a) center = series link goes nowhere;
b) one of the cap/resistor combos in series with series link;
and
c) the other cap/resistor combo in series with the series link.

In finality, we see that in the center position, nothing happens to the signal, we have a standard humbucker. When the switch is down, we have the (presumably hot) coil in series with a cap, and the series link going to ground through a combination of a cap and resistor. When the switch is up, we have a different value cap in series with the 'hot' coil, and a different combo of cap and resistor in series with the series link.

The tricky part for the two non-center switch positions is, the 'not-hot' coil is still in play.   While that coil is going from ground (raw ground as shown) to ground via the switch, it's still linked in to the 'hot' coil. The frequencies that didn't get through the filter (the combined cap and resistor) and out to ground are still available for going into the 'hot' pup's negative lead.  In effect, what was subtracted from the 'hot' pickup (those frequencies that were sent to ground via the series link) are now replaced by the remaining frequencies coming from the 'not-hot' coil.  Essentially, there should be little overlap or 'notching out' of anything, but the origin of the frequencies, coming from two physically separated coils, that should make for some interesting tones.

At this point, the pickup is still able to buck hum, but I foresee that it may not be as quiet as when the switch is in the center position, due to the filtering action placed upon the series link.

And the magic question is, how will it sound. That's up to the user/player/modder/owner, but if my name were frets, I'd certainly give this a go!

HTH





sumgai

[JohnH]

As I see it, this has a middle switch position that is a simple humbucker. The outer positions each do a variation on partly shunting one coil, plus they put the whole signal in series with a cap, or cap//resistor, to cut some bass probably. Maybe these settings will make a somewhat single-coil-like sound. Id expect that personally I might prefer not to feed through the series caps (ie without the components on the left side), and then I'm seeing it as a couple of variations on part bypass of a humbucker, and I know that can sound pretty good, better than a full coil cut.

[Yogi B]

Mar 26, 2022 15:39:20 GMT -5 JohnH said:

Actually, its not something that can be properly modelled in GF as it currently exists. The reason is that its using the two coils of a pickup, and putting components between them that affect one coil differently to the other. In the current GF, you can model simple combinations of pickups and coils in series and parallel, and out of phase, but then those are combined into one equivalent pickup before they meet the tone shaping circuitry.

I'd hesitate to say "equivalent", there are already components that should affect each coil differently: the series resistance & inductance (R1, L1), when pickups are combined in parallel; and everything else (C1, Ld, Rd, Rl), when combining in series. In most cases GF's 'averaged coil' is a pretty close approximation especially in parallel combinations — but in part that's because all guitar pickups are somewhat similar.

I don't suppose there's a way to add a pickup model that is composed of completely arbitrary voltage & impedance?

---

Semi-related issue this has reminded me of...

With the standard pickup models we have a single capacitance between the ends of the coil. With one coil or multiple parallel coils this is fine because one end of the coil(s) is grounded thus the winding capacitance and capacitance to ground are in parallel — however, this is not true of any but the bottommost coil in a series stack.

I suppose the crux of the matter is whether the capacitance between the coil and the baseplate/cover/polepieces/shielding is significant. Yeah, it'll be swamped by cable capacitance and even unloaded I doubt it'd make an audible difference, but I'm curious if it's so small as to be practically invisible on a plot as well.

---

JohnH, sumgai: are we just ignoring the inductor? I mean, at 10,000uH = 10mH it definitely is ignorable (being around at least 20× too small to have any significant effect, in the presented configuration). So, maybe one of us should clue frets in on that fact .

Also I'm 99% sure that an equivalent switching scheme can be done with only an ON/ON/ON DPDT — though this is also likely a good candidate for being a premature optimization as it relies upon the fact that the impedance of the 2200pF cap is larger than that of the 3300pF cap + 330k resistor (hint).

[JohnH]

Hi Yogi

GF does what it does using the same classical theory presented in any electronics class, or modelled in a Spice sim. So in principle anything can be added, including a pickup with arbitrary voltage and complex impedance, so long as it can be expressed as 4 numbers (voltage, resistance, reactance, and phase), at each frequency step. But, it gets to a point of diminishing returns, as compared to a Spice model, although GF can model position, pick position and string vibration too.

On the way it models pickups, it just uses a reasonable 6 part model, which gets pretty close while being not too hard to derive from tests that Antigua does. When combining pickups, I think it's equally accurate for multiple puckups of the same type, or of the same series such as within a B, M, N set with different winds but the same design. When combining different pickups in series or parallel, it's less accurate but still useful enough for its main job of helping to think about the circuitry. It can't currently do pickups where each has different components across them.

The last big addition was getting the string geometry and vibrations in, and without those any combination output can never be right. The next step is multiple pickups explicitly modelled instead of combined. That's a big step and probably a do-over of the spreadsheet.

[sumgai]

Mar 28, 2022 5:39:46 GMT -5 Yogi B said:

.... are we just ignoring the inductor?

While I didn't mention it specifically, I didn't intend to pretend it's not there at all. My descriptions were meant to be kept short, 'cause you know how I like to get all wordy and stuff.

Nonetheless, the signal path is affected in the same manner - if the signal flows through the series link for one coil, and a frequency cutoff occurs, then the opposite action also occurs for the other coil. I still contend that a combination of lows from one coil and highs from the other coil will be the resultant tone. As John notes, it should sound somewhat like a single coil, but with interesting side effects. (My words, not his.) Where things might get really interesting would be if this circuit was installed on a guitar with an SS arrangement, similar to a Jaguar, Jazzmaster, Mustang, etc.

HTH



sumgai

[frets]

Sumgai😻😻😻,

Thank you for you thorough analyses of this circuit. I do plan on building it as I found the switches for it. I just have not had time yet.

I am finally getting a grip on Guitarfreak. It’s extremely helpful.

[Yogi B]

Mar 28, 2022 7:38:55 GMT -5 JohnH said:

On the way it models pickups, it just uses a reasonable 6 part model, which gets pretty close while being not too hard to derive from tests that Antigua does. When combining pickups, I think it's equally accurate for multiple puckups of the same type, or of the same series such as within a B, M, N set with different winds but the same design. When combining different pickups in series or parallel, it's less accurate but still useful enough for its main job of helping to think about the circuitry. It can't currently do pickups where each has different components across them.

I was thinking along the lines of, say, a simpler version of partial coil bypass, consisting of only a capacitor. Since the additional cap is directly in parallel to C1 it could just be added to it. Then, even though we'd have the same circuit topology, GF's result would be far less accurate than usual since the value of C1 would be much larger than for any typical pickup.

GF does what it does using the same classical theory presented in any electronics class, or modelled in a Spice sim. So in principle anything can be added, including a pickup with arbitrary voltage and complex impedance, so long as it can be expressed as 4 numbers (voltage, resistance, reactance, and phase), at each frequency step.

What if each of those 'numbers' is not constant, but instead has a complicated relationship to frequency? Specifically I'm wondering about externally calculating the Thevenin equivalent circuit for both of frets's partial bypasses then feeding those to GF instead of the six-part model.

Mar 28, 2022 11:12:42 GMT -5 sumgai said:

Mar 28, 2022 5:39:46 GMT -5 Yogi B said:

.... are we just ignoring the inductor?

While I didn't mention it specifically, I didn't intend to pretend it's not there at all. My descriptions were meant to be kept short, 'cause you know how I like to get all wordy and stuff.

Nonetheless, the signal path is affected in the same manner - if the signal flows through the series link for one coil, and a frequency cutoff occurs, then the opposite action also occurs for the other coil.

Opposite but not necessarily equal, the inductor will affect the bypassed coil at a much lower frequency than the non-bypassed coil. And at only 10mH I reckon both of these frequencies will be too high to have any significant audible effect. (At 20kHz the impedance of 10mH is only about 1.2K)

A more straight forward example would be with just a simple bypass resistor, a 10k resistor: for the bypassed coil, that's an order of magnitude less than than the typical load (10k ≪ 125k); Whereas for the non-bypassed coil the addition of the resistor will do a lot less (for most frequencies, 10k isn't significantly less the the impedance of the bypassed coil).

Mar 28, 2022 5:39:46 GMT -5 Yogi B said:

Also I'm 99% sure that an equivalent switching scheme can be done with only an ON/ON/ON DPDT

This:

(I've drawn the switch as a DP3T so as to be clearer about what's connected in each position, but it is just emulating a DPDT ON/ON/ON)

[JohnH]

I was thinking along the lines of, say, a simpler version of partial coil bypass, consisting of only a capacitor. Since the additional cap is directly in parallel to C1 it could just be added to it. Then, even though we'd have the same circuit topology, GF's result would be far less accurate than usual since the value of C1 would be much larger than for any typical pickup

That'd be OK, but, since there's only one cap in each model, theres no way to preserve how it woudl interact with one coil only if it was a cap bypass on one coil. And if the two coils were modelled as separate pickups, the again the cap acting on one coil woudl get smooged across both when they combine.

GF does what it does using the same classical theory presented in any electronics class, or modelled in a Spice sim. So in principle anything can be added, including a pickup with arbitrary voltage and complex impedance, so long as it can be expressed as 4 numbers (voltage, resistance, reactance, and phase), at each frequency step.

What if each of those 'numbers' is not constant, but instead has a complicated relationship to frequency? Specifically I'm wondering about externally calculating the Thevenin equivalent circuit for both of frets's partial bypasses then feeding those to GF instead of the six-part model

This could have some legs, at least for looking at one pickup at a time, with various bypasses tested within that pickup.

Is that something you would be interested for us to develop together? if so, Ill look into GF and identify any necessary protocols for such a set of data. First thoughts:

GF currently is set up to calculate one horizontal row for each frequency, and its based on each semitone of a chromatic scale based on A440, from 55hz up to above 10khz

For the Thevenin equivalent representing a partly bypassed pickup, you would resolve each complex result into a voltage value and a real and imaginary impedance which will vary at each frequency.

When Im modelling the whole pickup, I start with a voltage of nominally 1 and the whole shebang is a long series of voltage dividers. At the end, I have a final table of voltages to convert to dbs for plotting.

For what you would do, you'd need to have a different voltage at each frequency to go with the impedances. Then that would be fed into the rest of GF. This could be generated in a separate linked spreadsheet page, with all the parameters to control it, or it could be just a table of fixed numbers made with some different code. Doing it as an excel page within the GF spreadsheet would allow changes to be instantly seen. We may not need to use phase.

If you are doing all that, might as well add a few R and C and maybe L components that you can dial in or out, around the pickup coils.

A toggle on the front page would bring in this new data in place of the results of the 6-part models

[Yogi B]

Mar 29, 2022 2:34:30 GMT -5 JohnH said:

Is that something you would be interested for us to develop together?

Sure.

I've set up a proof of concept spreadsheet, which I'll shortly link you to via PM...

[JohnH]

Great thanks! I'll check it out.