An interactive table to present gathered pickup data

[antigua]

I've created a table that allows for the comparison of all of the data I've gathered over the past six months  www.echoesofmars.com/pickup_data/viewer/  

This is only the Stratocaster data I have. I still have to input data for humbuckers, Tele and Gretsch style pickups, and then hopefully Fender bass pickups after that. 

I spent some time looking for a turn key CSV viewer, but came to the conclusion that a hand made viewer could be a lot more powerful, so I created this from scratch. 

I used stratotart's column layout as a starting point, but I moved things around a bit. I'm drawing special attention to the inductance and the loaded resonant peak frequency by making those columns green, as I'm hoping that people who are not so technically inclined will get the idea to pay more attention those figure than DC resistance, the unloaded peak frequency, and other such data that is probably more distracting than it is useful. 

I also put some pickup values in for Lollar and Tonerider based on their own published data. By taking the average capacitance observed from their sets, I can guestimate what the resonant peaks will work out to. Hopefully, if this becomes a useful buying guide, other pickup vendors will see the value in exposing data so that their products can fit somewhere in this table.  Those data rows are distinguished by gray lettering, a mention in the note field, and they have a lot more data gaps than the other rows. 

When you make a selection, the address bar updates with the current config, so you can paste a link that loads that view, for example here is the table sorted by price, or Fender's pickups ranked by inductance.  

Take a look see. I'll wait a big for some feedback, then I'll share this with the web's larger Stratocaster forum. 

[Charlie Honkmeister]

A simple "Thank you"  doesn't seem to be nearly enough.   Finally,  good measurements made with a consistent known methodology.   Great HTML job in being able to not only display,  but sort and search as well.  


Are you interested in having the group members send you more pickups to test over time? 

[antigua]

Jan 14, 2017 13:25:16 GMT -5 Charlie Honkmeister said:

A simple "Thank you"  doesn't seem to be nearly enough.   Finally,  good measurements made with a consistent known methodology.   Great HTML job in being able to not only display,  but sort and search as well.  


Are you interested in having the group members send you more pickups to test over time? 

That would be great. I still have some on hand to test. I'd guess it will be about three months before I'm fresh out at the present rate. 

[JohnH]

Hi Antigua

This is outstanding, thankyou!

The table viewer works perfectly for me. I had been wondering about some simple summary of all the results, but this is much better than I could imagine. But the thanks are mainly for the huge and ongoing body of work behind it. There has never been anything like it. This work is way beyond amateur experimenting, IMO it has an academic level of rigour, repeatability and presentation and would be of great help to students and researchers at all levels.

Suggestions:
Id be interested in a bare csv version too, for use in extracting data from it, post-processing, plotting trends, graphs etc

The loaded res peak and loaded res frequency headers could be clarified that they both apply with 200k and 470pF, and frequency should be hZ

Could there be a column for wire gage (where you know it)? to show why sometimes a stacked or very hot pickup is not as hot as DCR might imply

Most of the capacitance results were based on a 10pF allowance for the rig, but some were 20pF. I've forgotten whether that was a change in the setup or was it an amended assumption?

This thread, or its successor, should be in a sticky. (i'd do it but don't have that function available as a mod)

cheers!

[antigua]

Wire gauge, that's a good one for sure. In many cases, I can figure out the insulation type, but I don't know if that's even worth mentioning or not. I supposed I should include it for completeness. 

For the humbuckers, I'm thinking of only including series data, though I per-coil measurements for most of the ones featuring four conductors. DiMarzio is the only maker that really mismatches the coils to a significant, so I think I'll just note it in the notes, and let people investigate it further if they're interested. In general, if you know the series values for a matched coil humbucker, the coil stats are pretty easy to figure out through simple division. 

The first integrator I made showed a capacitance of 20pF, then I got a newer one from stratotarts that showed closer to 10pF, so that's why they're different. There is probably a +/- 5pF margin for error with that. It has to be noted because the absolute resonant peaks contain that 20pF. It would probably be pretty easy to recompute the absolute resonant peaks with that 20pF factored out. So much to do, so little time. If you want to "enhance" the data, I'll gladly accept it. 

Here is the raw csv www.echoesofmars.com/pickup_data/viewer/strat.csv . The dynamic table code can load any csv, so if you have one of your own and would like it presented like this, I can make it happen. You can just save the html to your computer, for that matter. The csv file path is defined in the object config. I think it would be easier at this point to defer to modeled plots for graphing presentations. Maybe at some point we can work something out to create a web based graphing interface. I might be able to do it with a csv of 6 part models and some help with the math. 

As far as amateurishness goes, I really wish all the bode plot data was available in a common format, so that plotting software could overlay the plots, but there are a handful or reasons why that would be too time consuming to do. I also did some things less that ideal with the first batch of measurements, but I don't have time to go back and redo them. A lot of these pickups are in guitars, and I had to open them up, unsolder the pickups from the the selector switch and everything. 

[stratotarts]

Nice, really nice. I think you mean Loaded Res. Freq. (kHz) w/470pF, not Loaded Res. Freq. (pF) w/470pF.

[JohnH]

I agree with all that. Things like the single coil and parallel results of humbuckers can be referenced just by noting in the table, to show those pickups where such info is available in the full thread post.

It would be neat to have all the graphs plotted interactively from models. There are a few hurdles to jump over for that though.

For one, out of the 69 strat pickup there, I have models for only about 20% of them. So one hurdle is how to efficiently generate a lot more models (and check some of the earlier ones too). Generating them is still only semi-automatic and iterative. Sometimes the routines I used fly off the handle and need manual input. But at least these alnico Strat pickups tend to be the most convergent. Previously I had generated these purely from traces without using the inductance measurement, but recently I've incorporated inductance too following discussions with ms, so in principle, all the date needed is in the numerical values in your table (DCR, L and two pairs of peak frequencies and db values) and doesn't strictly need the graphs except as a check.

Another issue is that the baseline values of each plot are not all at exactly 0db, so some of the peak values need to be adjusted by a small offset. Not a big problem, since this can be taken accurately enough from the graphs.

Generating the plots within a web application would be the last step. Ive been working just with spreadsheets, and what they do including all the options and controls does not translate well. But the core calulations, though lengthy, are quite simple. Given access to the 6 model parts and the other component values, there are about 30 spreadsheet cells that form a calculation. each is just a simple calc with basic +,-,*,/ functions, ending up with a db value given a frequency. Maybe a very stripped down version of just that could be translated into the core of such a graphing tool?

[antigua]

I haven't researched web based graphing options, but I'm pretty certain you just feed it lists of plot points and it handles the rest. How I imagine it would work is we would have a CSV like:

mfg,model,L1,R1,L2,R2,C1,R3
fender,texas special,2.8,6.7,7,188,400
fender,fat 50,3.6,6.1,5,340,123
...

Or we could just tack the values onto the end of the csv I've already created. If you use the same values I collect for L1, R1 and C1, then we'd only need whatever is required to model the eddy currents. 

The javascript would import the CSV, display a list of pickups that can be rendered from the CSV, and then the user selects pickups, the plot points are calculated on the spot and fed to the plotter plugin. We can add nifty things like C and R load easily enough. All I would need help with is the math, for the most part. It would not be as full featured as the Guitar Freak spreadsheet, but all I'm really aiming to do is create a pickup comparison tool. Rather than creating a new viewer, it could even be a subset of the dynamic table, like you'd check check boxes for the pickups you want to plot together, and click "Graph the checked pickups". 

We might be able to simplify things by just neglecting eddy current curves, but Filter'tron style pickups are heavily defined by those curves, unless you just want to pretend they have a hard knee as a close approximation. 

A function that takes in model values and spits back out plot points would be killer. 

[antigua]

Jan 14, 2017 19:12:36 GMT -5 stratotarts said:

Nice, really nice. I think you mean Loaded Res. Freq. (kHz) w/470pF, not Loaded Res. Freq. (pF) w/470pF.

I've fixed the pf -> kHz error, thanks for pointing that out. I'm in the process of adding Filter'tron and Tele style pickups so that I can create tables for those ranges. 

What would you call the general class of pickups that TV Jones offers? They sort of cover "the rest", but then again they are specific in a sense, mostly fitting in the space allotted for a Gretsch Filter'tron. 

[antigua]

Here are some new data sets:

Telecaster sets:
www.echoesofmars.com/pickup_data/viewer/?Type=Telecaster

Filter'tron / TV Jones pickups:
www.echoesofmars.com/pickup_data/viewer/?Type=F%27t%20Sized

The Strat set now loads with this URL www.echoesofmars.com/pickup_data/viewer/?Type=Stratocaster , the original URL above loads everything in one table, which is not especially useful if you're shopping for pickups for a particular type of guitar.

[JohnH]

A function that takes in model values and spits back out plot points would be killer.

Great thought, and Id be happy to contribute to that. I can see how all the maths could work, and its what Ive done in the spreadsheets. But what Im struggling with is the next step of how to present it so it can be coded, or how to actually create the code. Im about 40 years out of date in that area!

The basic numerical activities are:

1.Read in 6 model components values, and a frequency for the first calc.
2.Adapt the C1 and L3 values to new values so they include load L abd C in parallel.
3. A series of steps condensing the full circuit into a simpler voltage divider, in the same way that one would simplify an all-resistor network. This is what in a spreadsheet calc, takes about 30 cells in each row. Simple maths.
4. Calculate the db output (needs a log function, the only non-sumple function needed)
5. Increment frequency and calculate the next result...repeat etc

Within part 3 above, one repeated type of calc is to take the inverse of a complex impedance so it can be added to another. Then invert again to result in a combined parallel impedance. In a spreadsheet, seperate cells are used each time this happens. But in a program code, this could also be a subroutine assuming such facility existed.

What next?

[antigua]

Jan 15, 2017 16:35:29 GMT -5 JohnH said:

A function that takes in model values and spits back out plot points would be killer.

Great thought, and Id be happy to contribute to that. I can see how all the maths could work, and its what Ive done in the spreadsheets. But what Im struggling with is the next step of how to present it so it can be coded, or how to actually create the code. Im about 40 years out of date in that area!

The basic numerical activities are:

1.Read in 6 model components values, and a frequency for the first calc.
2.Adapt the C1 and L3 values to new values so they include load L abd C in parallel.
3. A series of steps condensing the full circuit into a simpler voltage divider, in the same way that one would simplify an all-resistor network. This is what in a spreadsheet calc, takes about 30 cells in each row. Simple maths.
4. Calculate the db output (needs a log function, the only non-sumple function needed)
5. Increment frequency and calculate the next result...repeat etc

Within part 3 above, one repeated type of calc is to take the inverse of a complex impedance so it can be added to another. Then invert again to result in a combined parallel impedance. In a spreadsheet, seperate cells are used each time this happens. But in a program code, this could also be a subroutine assuming such facility existed.

What next?

In most all programming languages, the result of a math operation is assigned to the left, so for example, here is a line of code that calculates for F from the variables L and C (and the constant PI):

F= 1/(2*PI*sqrt(L*C));


L and C would come from the CSV along with the other values needed for plotting, so all I'd really need a is a more complex version of this code that puts all the math together. It can be made up of multiple lines of code, it doesn't have to be just one long equation. 

I said before that I could use a function that takes in the values and returns plot points, but actually I think it would be best to also input a frequency into the function, and then with that, I would simply call the function for every frequency I want to plot. All the function would have to return, therefore, is an amplitude. 

[JohnH]

Ok, I can do that. Probably simplest if I dont call subroutines but just write it out line by line in the same sequence as the spreadsheet.

Along the way, some extra letter variables will be used for intermediate results. Can I also use letter-number variables such as C1, C2 etc?

Ill also provide some test value outputs based on the spreadsheet, so it can be checked.

[antigua]

Jan 15, 2017 17:41:06 GMT -5 JohnH said:

Ok, I can do that. Probably simplest if I dont call subroutines but just write it out line by line in the same sequence as the spreadsheet.

Along the way, some extra letter variables will be used for intermediate results. Can I also use letter-number variables such as C1, C2 etc?

Ill also provide some test value outputs based on the spreadsheet, so it can be checked.

I can probably work with whatever you give me, no need to worry about variable names, but yes, numbered variables are valid so long as the number is not the first character. 

[wgen]

Man, I logged in just to really thank you, even if this is not nearly enough, as someone said before.
Now, this is all really impressive.
I had just finished to read the other thread where some big informations about harmonics sensing came out, even surprisingly enough from some points of view.
Together with all of this, it seems to me that choosing and setting pickups won't be nearly the same thing from now on

[JohnH]

OK, well, maybe it wasn't too hard!

Attached is a text file with my 'code'

I made it by stripping down the excel file that I'm using for model derivation, to just one row. Then I transposed it from a row to a column, and put each variable into a cell with the same reference as I wanted variables, so L1, R1 etc. All the calcs went into variable X1, X2, X3 etc., and the last one is the result.

This meant less risk of getting it wrong, and easier to test.

The spreadsheet version, screenshot attached, has a model entered into the yellow cells for a 6.12k Texas Special. It should give 14.6db at 8750hz and 0db at 400 hz (which I have been using as a zero benchmark)

Then I changed the spreadsheet to display formulae instead of values, and cut and pasted into Notepad. I think I have fixed the spreadsheet formulas where I think they differ from program code, such as Pi() becomes PI etc, but please review, particularly the 'log' statement

I added some imagined input commands just to list the variables, but I'm sure the format of these will need to be changed for whatever program code is being used. But I think it should basically work, and give a db output, given 6 model component values, two load values for R and C, an offset db, and a frequency.

Attachments:

6 part model code 170116.txt (1.26 KB)

[antigua]

Thank you very much. I'll see what I can make happen with this. 

[JohnH]

Lets see how it works out, and once its going, whether you think that TX special model is a good representation. If so, ill post up some more models and then work on others.

A few thoughts:
The 6 model parts are not the same as the measured dcr, L and calculated C, though obviously they are related. But for example, although R1 is not the DCR, it does match when combined with R2 and R3. Damping affects peak frequency, so C1 is not quite the same as the C in a simple model.

The offset value is worth discussing more. I worked it out with a 400hz baseline at 0db, since some of the tests showed some deviation below this. But having made a model, the offset value is almost the same if calculated at 0hz, in which case it is only determined by resistances and could be added to the code instead of being an input variable. But a further question is whether the offsets should all be consistent to 0db, or should they try to allow for overall output levels too? Eg to show a Tx special louder than a Cs69 etc. This could be based on 20Log of inductance for this, but I think it may be closer based on dcr, if scaled for wire cross section and also gauss. Food for thought...

[antigua]

I've updated the table www.echoesofmars.com/pickup_data/viewer/ , it now has PAF's and a few other pickups, wire gauge where it can be determined, and expanded filtering options. 

Here is the latest CSV file  www.echoesofmars.com/pickup_data/viewer/pickups.csv  . With this data, it's also possible to see how closely inductance and DC resistance relate for given classes of pickup, if anyone is interested in taking up that cause. 

I have a mystery to solve also; the capacitances look really low, but the are calculated based on peak resonance and inductance measured at 120Hz. For example, the Seymour Duncan Antiquity, only maths out to 110pF capacitance - and that's with braided hookup wire, which adds about 70pF capacitance all by itself. The resonant peaks of 6.5kHz to 7kHz are rather high for an inductance of 5.1H and 4.2H, that pushes the calculated capacitance numbers down a lot. By contrast, a Bare Knuckle Irish tour middle pickup also has a peak of 6.5kHZ, but only an inductance of 2.6H, and a wopping 210pF capacitance. This makes me suspect that the 120Hz inductance values are too high. If we assume the humbucker has at least as much capacitance as a typical Strat pickup, 150pF, then add the 70pF for the braided hookup cable, that should land well above 200pF. 

Could the fact that the two coils are in two halves account for the lower capacitance, since series capacitance decreases the overall capacitance? 

[ms]

Jan 18, 2017 16:42:27 GMT -5 antigua said:

I've updated the table www.echoesofmars.com/pickup_data/viewer/ , it now has PAF's and a few other pickups, wire gauge where it can be determined, and expanded filtering options. 

Here is the latest CSV file  www.echoesofmars.com/pickup_data/viewer/pickups.csv  . With this data, it's also possible to see how closely inductance and DC resistance relate for given classes of pickup, if anyone is interested in taking up that cause. 

I have a mystery to solve also; the capacitances look really low, but the are calculated based on peak resonance and inductance measured at 120Hz. For example, the Seymour Duncan Antiquity, only maths out to 110pF capacitance - and that's with braided hookup wire, which adds about 70pF capacitance all by itself. The resonant peaks of 6.5kHz to 7kHz are rather high for an inductance of 5.1H and 4.2H, that pushes the calculated capacitance numbers down a lot. By contrast, a Bare Knuckle Irish tour middle pickup also has a peak of 6.5kHZ, but only an inductance of 2.6H, and a wopping 210pF capacitance. This makes me suspect that the 120Hz inductance values are too high. If we assume the humbucker has at least as much capacitance as a typical Strat pickup, 150pF, then add the 70pF for the braided hookup cable, that should land well above 200pF. 

Could the fact that the two coils are in two halves account for the lower capacitance, since series capacitance decreases the overall capacitance? 

The capacitance needs to be based on an effective inductance at the resonant frequency.  This effective inductance is lower than the 120 Hz inductance because of eddy currents. The eddy currents affect both the real and imaginary parts of the impedance.  You can see this on the impedance plots that I have shown here. For example, consider this one:


The yellowish line is the imaginary part of the impedance with the effects of the capacitance removed.  For this humbucker it is significantly below the dashed line, which is what the imaginary part of the impedance would be if the 120 Hz inductance determined it completely.  The resonant frequency is determined by the effective inductance.

The capacitance of a humbucker (assuming no high cap braided cable) should be lower than a strat pickup because the two coils in a humbucker are in series, and so the capacitance should be half that of one of the coils, possibly plus some amount determined by interaction between the coils. 

In any case, measuring the capacitance is not easy.   I think the values I show are good because the effect of the resonant peak appears well removed from the yellowish and green lines.  But I do urge caution on capacitance values. It is not so easy.

[JohnH]

It could be an issue with the basic assumption of 3 simple parts. Its an approximation that goes off by a % in some cases, and the capacitance so calculated has that deviation built into it. Lets review in comparison to some 6 part models, which if not perfect are at least closer.

The kind of issues we are seeing are by analogy, a bit like trying to measure both long and short distances along the the earths surface; without knowing about the earths curveature.

3 part model = flat earth, which is a good assumption for short distances

6 part model = spherical earth, still not a perfect model, and tricky to work out, but better.

[antigua]

Jan 18, 2017 18:17:53 GMT -5 ms said:

Jan 18, 2017 16:42:27 GMT -5 antigua said:

I've updated the table www.echoesofmars.com/pickup_data/viewer/ , it now has PAF's and a few other pickups, wire gauge where it can be determined, and expanded filtering options. 

Here is the latest CSV file  www.echoesofmars.com/pickup_data/viewer/pickups.csv  . With this data, it's also possible to see how closely inductance and DC resistance relate for given classes of pickup, if anyone is interested in taking up that cause. 

I have a mystery to solve also; the capacitances look really low, but the are calculated based on peak resonance and inductance measured at 120Hz. For example, the Seymour Duncan Antiquity, only maths out to 110pF capacitance - and that's with braided hookup wire, which adds about 70pF capacitance all by itself. The resonant peaks of 6.5kHz to 7kHz are rather high for an inductance of 5.1H and 4.2H, that pushes the calculated capacitance numbers down a lot. By contrast, a Bare Knuckle Irish tour middle pickup also has a peak of 6.5kHZ, but only an inductance of 2.6H, and a wopping 210pF capacitance. This makes me suspect that the 120Hz inductance values are too high. If we assume the humbucker has at least as much capacitance as a typical Strat pickup, 150pF, then add the 70pF for the braided hookup cable, that should land well above 200pF. 

Could the fact that the two coils are in two halves account for the lower capacitance, since series capacitance decreases the overall capacitance? 

The capacitance needs to be based on an effective inductance at the resonant frequency.  This effective inductance is lower than the 120 Hz inductance because of eddy currents. The eddy currents affect both the real and imaginary parts of the impedance.  You can see this on the impedance plots that I have shown here. For example, consider this one:


The yellowish line is the imaginary part of the impedance with the effects of the capacitance removed.  For this humbucker it is significantly below the dashed line, which is what the imaginary part of the impedance would be if the 120 Hz inductance determined it completely.  The resonant frequency is determined by the effective inductance.

The capacitance of a humbucker (assuming no high cap braided cable) should be lower than a strat pickup because the two coils in a humbucker are in series, and so the capacitance should be half that of one of the coils, possibly plus some amount determined by interaction between the coils. 

In any case, measuring the capacitance is not easy.   I think the values I show are good because the effect of the resonant peak appears well removed from the yellowish and green lines.  But I do urge caution on capacitance values. It is not so easy.

That makes sense. Can you think of any "rule of thumb" that might help us ball park the actual inductance at resonance, or the actual capacitance, for pragmatic purposes?  Would most humbuckers tend to follow your plot, or should we expect that they will vary widely?

I'll be honest, some of the conjecture I'm doing with numbers to approximate resonances from inductance are borderline sloppy, but I don't feel too bad about it because the pickup companies could, but won't, provide basic specification for their products, and part of me hopes that the potential for misrepresentation of their product might encourage them to be more forthcoming. 

[antigua]

Jan 18, 2017 18:17:53 GMT -5 ms said:

The capacitance of a humbucker (assuming no high cap braided cable) should be lower than a strat pickup because the two coils in a humbucker are in series, and so the capacitance should be half that of one of the coils, possibly plus some amount determined by interaction between the coils. 

I've been messing with LTSpice trying to wrap my brain about how simply separating a coil into two halves can case the overall parasitic capacitance drop become half what it was, but if the modeling is to be believed, that's how it plays out. 

So I measure 110pF for the Antiquity neck. If I subtract 70pF for the cable, that leaves 40pF remaining. Two 80pF coils in series theoretically give you 40pF total. 80pF per each 3.8k 42 AWG coil is not unrealistic for a machine wound coil with a higher tension and uniform layering, so that at least gives me confidence that the numbers are not astronomically wrong, but I agree with your point that the inductance is going to be lower at the resonance than what is measures at 120Hz, though that would mean the capacitance would therefore have to be revised upwards even further to offset the lower inductance.

[ms]

Jan 18, 2017 20:04:39 GMT -5 antigua said:

Jan 18, 2017 18:17:53 GMT -5 ms said:

The capacitance needs to be based on an effective inductance at the resonant frequency.  This effective inductance is lower than the 120 Hz inductance because of eddy currents. The eddy currents affect both the real and imaginary parts of the impedance.  You can see this on the impedance plots that I have shown here. For example, consider this one:


The yellowish line is the imaginary part of the impedance with the effects of the capacitance removed.  For this humbucker it is significantly below the dashed line, which is what the imaginary part of the impedance would be if the 120 Hz inductance determined it completely.  The resonant frequency is determined by the effective inductance.

The capacitance of a humbucker (assuming no high cap braided cable) should be lower than a strat pickup because the two coils in a humbucker are in series, and so the capacitance should be half that of one of the coils, possibly plus some amount determined by interaction between the coils. 

In any case, measuring the capacitance is not easy.   I think the values I show are good because the effect of the resonant peak appears well removed from the yellowish and green lines.  But I do urge caution on capacitance values. It is not so easy.

That makes sense. Can you think of any "rule of thumb" that might help us ball park the actual inductance at resonance, or the actual capacitance, for pragmatic purposes?  Would most humbuckers tend to follow your plot, or should we expect that they will vary widely?

I'll be honest, some of the conjecture I'm doing with numbers to approximate resonances from inductance are borderline sloppy, but I don't feel too bad about it because the pickup companies could, but won't, provide basic specification for their products, and part of me hopes that the potential for misrepresentation of their product might encourage them to be more forthcoming. 

Eddy current effects vary enough so that I think it is a good idea to account for them in the measuring process, that is, account of the effects on the effective inductance.  Many do behave very similarly, of course, but how do yo know if you cannot measure it? 

[ms]

Jan 19, 2017 5:19:27 GMT -5 antigua said:

Jan 18, 2017 18:17:53 GMT -5 ms said:

The capacitance of a humbucker (assuming no high cap braided cable) should be lower than a strat pickup because the two coils in a humbucker are in series, and so the capacitance should be half that of one of the coils, possibly plus some amount determined by interaction between the coils. 

I've been messing with LTSpice trying to wrap my brain about how simply separating a coil into two halves can case the overall parasitic capacitance drop become half what it was, but if the modeling is to be believed, that's how it plays out. 

So I measure 110pF for the Antiquity neck. If I subtract 70pF for the cable, that leaves 40pF remaining. Two 80pF coils in series theoretically give you 40pF total. 80pF per each 3.8k 42 AWG coil is not unrealistic for a machine wound coil with a higher tension and uniform layering, so that at least gives me confidence that the numbers are not astronomically wrong, but I agree with your point that the inductance is going to be lower at the resonance than what is measures at 120Hz, though that would mean the capacitance would therefore have to be revised upwards even further to offset the lower inductance.

One way to look at the series capacitance thing is to first say that series impedances always add.  This is true for Rs, Ls and Cs.  Then, since the impedance of a capacitor varies with the inverse of the capacitance, while the impedances add, the value of the total capacitance must go down.

[JohnH]

I had a look at a couple of examples of capacitances in the table vs modelled capacitances in 6 part models.

The 6.12k Neck Texas special has a total of 128pF in the table, and my model settled on 121pF, very close and it did not rely on any metered values to get there. The Bridge version was also quite close at 173pF vs 161pF.

Pearly Gates N 129pF table, 140pF model
Pearly Gates B 126pF table 145pF model

TV Jones N 111pF table, 110pF model
TV Jones B 131pF table, 153pF model

So the model results show the same tendencies, and are generally within 20pF of the tabulated ones. (Im quoting sum of pickup plus rig, since some of my values have a different 10pF/20pF allowance, which I will fix)

There are several effects at play in the variations. Here are two:

1 Pure damping, represented by a resistive load tends to reduce the maximum peak frequencies relative to no damping. Hence if you were calculating a C value to match the actual maximum, but in a 3 part model, you would over-estimate capacitance to get the peak frequency.

2. Eddy current effects that come with inductive components reduce reactance at higher frequencies in combination with the main inductor. (Call it a reduced effective inductance, but im not convinced tbat is a good description!). Using a measured inductance at 120hz, which is then in reality; effectiveky reduced at high frequency, will lead to an under-estimate if C in the table.

So the factors causing deviation in tabulated C can be in opposite directions, and may or may not compensate. But the table number is still a useful guide or starting point.

[col]

Jan 19, 2017 14:27:19 GMT -5 JohnH said:

I had a look at a couple of examples of capacitances in the table vs modelled capacitances in 6 part models.

The 6.12k Neck Texas special has a total of 128pF in the table, and my model settled on 121pF, very close and it did mot rely on any metered values to get there. The Bridge version was also quite close at 173pF vs 161pF.

And what is the accuracy of your capacitance meter? I seem to recall it being stated here at Gnuts that capacitance meters (unless more expensive) tend to be rather inaccurate.

[antigua]

Jan 19, 2017 15:13:53 GMT -5 col said:

Jan 19, 2017 14:27:19 GMT -5 JohnH said:

I had a look at a couple of examples of capacitances in the table vs modelled capacitances in 6 part models.

The 6.12k Neck Texas special has a total of 128pF in the table, and my model settled on 121pF, very close and it did mot rely on any metered values to get there. The Bridge version was also quite close at 173pF vs 161pF.

And what is the accuracy of your capacitance meter? I seem to recall it being stated here at Gnuts that capacitance meters (unless more expensive) tend to be rather inaccurate.

Modeled results means a meter was not used, but the value was essentially worked out mathematically. 

[col]

Oops. Yeah, I misread John's post.

[JohnH]

I just added more to my last post above...