Pickup Testing and Methodology Questions

[Josh]

Hi,

I recently got into messing with (making) pickups. I've read up quite a bit about how people are testing, and adding things in to make their measurements appear "more realistic."

That's where the whole thing seems messed up to me.

The first problem is that that test rig will never exactly match every guitar, so what's the point? It seems to me like it's just skewing the numbers arbitrarily.

The second problem is that measuring any DUT is already modifying the DUT. Adding arbitrary values via another DUT to the original DUT doesn't guarantee realistic comparisons, it simply requires you to continue using those same modifications to make future comparisons.


It would make more sense to me to do this:

Measure the pickups directly with your LCR meters and/or scopes.
Measure the same pickups in an actual guitar circuit.
(If you want to compare, then also measure with the pickup testing contraptions.)

Do this for a number of pickups. Measure each pickup in circuits using 250K pots, then in circuits using 500K pots. You can also measure the values of the test pickguard without pickups, and see how that compares to measuring the pickups in the pickguard, minus the pickups' independent measurements. 

From there you should be able to do some math and figure out what the average change for each pickup type is in circuit (and if the type makes any difference).

Assuming we can get valuable statistical data from those tests, then you can skip all the extra stuff and just measure the pickups directly. Then anybody sharing the data can compare directly also. Finally, you can simply apply the math to see what the difference might be in circuit.

I would think having this information should remove the need to do anything but test the pickups directly. That would make comparing information simpler as well. If every user makes their own test rig, you can assume pretty confidently that there will be a variance from one rig to the next. 

Has anybody tried anything like this yet? What do you think?

Thanks,
Josh

[stratotarts]

Josh,

First of all, I think you should really check out the GuitarFreak program. It performs a lot of the math that you mention, and incorporates many rounds of alignment with theoretical and empirical measurements, such that it has become very good at estimating the actual in circuit behaviour of any pickup, given its test measurements.

Also, many years have passed since the test apparatus in use, has any appreciable load on the DUT. The V5.8 integrator, for example, places about the same load on a pickup under test, as a short length of connecting wire. So that objection is mainly rhetorical in nature. We are talking macro scale electronics, not quantum mechanics.

Overall, what comes to my mind, is that what makes this forum unique, is the adherence to the presentation of the results of actual experiments, wherever theoretical ideas are put forth. If you find the current state of experimentation lacking, the best thing to do is to do some experimental work and post the results here, so we don't get into purely theoretical or rhetorical discussions.

In fact, most of your doubts have already been entertained and dealt with here on this forum. It's just hard to dig up sometimes because the ideas and projects presented here go back many years. Also some of the procedures you suggest, are already in use by many people. But, your main idea, which is to allow the interpretation of in circuit performance exclusively from test data, can be and often is done using GuitarFreak.

Once someone gets a good grasp of the circuit behaviours, it is even feasible to quicky mentally estimate the final in circuit implications of some test result.

[Josh]

Awesome, thank you! I'll check that out. A spreadsheet doing all the extra work sounds nice. 😉

Everything I found searching for info was a little old (2018ish), it was hard to find anything relevant newer than that. 

That sounds interesting with the newer version of the Integrator, but is there a real benefit to it if the spreadsheet is already able to correctly work with the raw pickup data?

I did do a few pickup tests, I can share the data in another post. It's direct to pickup testing though.

Thanks,
Josh

[Josh]

Maaaan, I still don't love the idea of having anything besides the pickup being tested directly...but I love projects and test gear, and the integrator circuit looks kinda fun. 🤣

[stratotarts]

Not sure what you mean, "tested directly".

[Josh]

Jun 4, 2023 16:38:04 GMT -5 stratotarts said:

Not sure what you mean, "tested directly".

Probes direct to the pickup & exciter, rather than introducing other things into the test circuit. But I ordered a board to make the Integrator anyway. TEAS strikes again.

[ms]

I think there are two kinds of measurements that are necessary:
1.  Impedance versus frequency.  You can make a device to do this that does not significantly affect the measurement. (It takes a resistor and a recording interface.)
2. Frequency response using an exciter coil.  I prefer to measure carefully the input impedance of the device that the pickup is connected to  and then compute the open circuit response.  For me the  preferable device is a recording interface.

[stratotarts]

It's been a while since I looked at the passive (simple) measurement procedure so I looked up the thread. The circuit with the test coil and 10x probe is the one you're using, I assume? That should be okay. The integrator closely duplicates a 10x probe in its input circuit, to make it possible for people who don't have a scope to do it (at the moment it now appears that the majority of people are using scopes anyway...). The response would be the same as a scope, when the integrate/bypass switch on the integrator is set to "bypass".

It is with the impedance measurement that some conversions have to be made to the resulting data, because it can't be done without some resistive load. That is the one I was worried about but I guess you are doing the other one with the test coil.

Since you're asking about the plot, it looks to me like some interpolation has missed the top of the peak, causing it to be chopped off. Perhaps you need to adjust the settings to capture more data points.

[antigua]

Jun 3, 2023 19:07:39 GMT -5 Josh said:

Hi,

I recently got into messing with (making) pickups. I've read up quite a bit about how people are testing, and adding things in to make their measurements appear "more realistic."

That's where the whole thing seems messed up to me.

The first problem is that that test rig will never exactly match every guitar, so what's the point? It seems to me like it's just skewing the numbers arbitrarily.

The second problem is that measuring any DUT is already modifying the DUT. Adding arbitrary values via another DUT to the original DUT doesn't guarantee realistic comparisons, it simply requires you to continue using those same modifications to make future comparisons.

There are usually two measurements, loaded and unloaded, loaded is with 470pF and 200k ohms parallel added, and unloaded is measured without anything added. I guess you're saying that 470pF and 200k are arbitrary, it is, but what matters is that it's consistent.

The 200k resistance is lower than with a Les Paul of two 500k pots, and lower than a Strat or Tele with two 250k pots (when a single pickup is selected that also has a tone control), so as said, you can sort of mentally adjust the values based on the guitar you plan to use. So if you see a Q factor of "2", you know that in a Gibson it will be a little higher, and in a Fender it will be a little lower. 

The 470pF represents a guitar cable,  and of course they come in all lengths, so you have to look at the outcome, suppose it's 3kHz, you have to think of it as a range, where a lower C cable will get you maybe 3.3kHz and a really long cable can drop it way down to 2kHz, but at least you know when comparing pickups that there's a 470pF baseline. 

One of the problems of using a real pick guard to test is that potentiometers are notorious for having imprecise resistances, they tend to be up to 10% off of the nominal value. And of course, you'd have to use a real guitar cable, and there again, there's probably no good candidates for a "standard guitar cable", even if you suppose a standard 10 foot length, the capacitance per foot for guitar cables varies by a lot. 

So maybe what has gone unsaid is that even when two guitarists compare the same pickups, they're hearing something very different from one another, but when people talk about what they hear, it's all very subjective, and at best you get an average consensus. There's still a lot of interesting observations to come out of it though, like why is a JB called a bright pickups by so many guitarists,  even though it has a loaded peak of 2kHz? Why is a Little '59 described as being "PAF-like" if it has a resonant peak that is 1kHz below that of a typical PAF clone? I'm happy to say that we've gathered so much information in the past few years that lots of hypothesis can be made about these things, where it wasn't possible before due to a lack of data to draw from.  

[Josh]

Jun 6, 2023 17:18:28 GMT -5 antigua said:

There are usually two measurements, loaded and unloaded, loaded is with 470pF and 200k ohms parallel added, and unloaded is measured without anything added. I guess you're saying that 470pF and 200k are arbitrary, it is, but what matters is that it's consistent.

The 200k resistance is lower than with a Les Paul of two 500k pots, and lower than a Strat or Tele with two 250k pots (when a single pickup is selected that also has a tone control), so as said, you can sort of mentally adjust the values based on the guitar you plan to use. So if you see a Q factor of "2", you know that in a Gibson it will be a little higher, and in a Fender it will be a little lower. 

The 470pF represents a guitar cable,  and of course they come in all lengths, so you have to look at the outcome, suppose it's 3kHz, you have to think of it as a range, where a lower C cable will get you maybe 3.3kHz and a really long cable can drop it way down to 2kHz, but at least you know when comparing pickups that there's a 470pF baseline. 

One of the problems of using a real pick guard to test is that potentiometers are notorious for having imprecise resistances, they tend to be up to 10% off of the nominal value. And of course, you'd have to use a real guitar cable, and there again, there's probably no good candidates for a "standard guitar cable", even if you suppose a standard 10 foot length, the capacitance per foot for guitar cables varies by a lot. 

So maybe what has gone unsaid is that even when two guitarists compare the same pickups, they're hearing something very different from one another, but when people talk about what they hear, it's all very subjective, and at best you get an average consensus. There's still a lot of interesting observations to come out of it though, like why is a JB called a bright pickups by so many guitarists,  even though it has a loaded peak of 2kHz? Why is a Little '59 described as being "PAF-like" if it has a resonant peak that is 1kHz below that of a typical PAF clone? I'm happy to say that we've gathered so much information in the past few years that lots of hypothesis can be made about these things, where it wasn't possible before due to a lack of data to draw from.  

Thanks for the detailed reply!

It's awesome that everybody here has a consistent comparison based on those arbitrary values, however...

What about trying to compare to data from manufacturers? Are they adding the same, or some other arbitrary load, or testing the pickups directly? I ordered a board to make the Integrator because it looked like it would be entertaining, I'll see how the measurements look with it.

From another post I made, I compared my direct to pickup measurements:

Hemi Neck
Lace website:
Resistance: 7.0k
Peak Frequency: 2750
Inductance: 3.5 henries

My TH2830 LCR: Ls: 3.69066H @ 100Hz

Hemi Bridge
Lace website:
Resistance: 10.0k
Peak Frequency: 2600
Inductance: 4.6 henries

My TH2830 LCR: Ls: 4.99706H @ 100Hz

My measurements were series inductance at 100Hz with kelvin probes connected directly to the pickup. Other frequencies, or using parallel inductance didn't come nearly as close to Lace's measurements.

I haven't seen many bode plots from manufacturers to compare with.

Thanks,
Josh

[timtam]

Jun 6, 2023 19:05:23 GMT -5 Josh said:

What about trying to compare to data from manufacturers? Are they adding the same, or some other arbitrary load, or testing the pickups directly?
....

I haven't seen many bode plots from manufacturers to compare with.

Almost no manufacturers provide bode plots, and IIRC those few that do don't specify their measurement conditions.

guitarnuts2.proboards.com/thread/9834/pickup-manufacturers-publish-bode-plots

So there's currently not really any manufacturer-supplied data to compare to, standardized or otherwise. The manufacturers seem mostly interested in maintaining the marketplace as is, where people are forced to buy pickups with little useful information on how they will sound. If that is to change, it will probably have to come from buyer pressure. Unless maybe one big manufacturer takes the lead in providing full specifications on their pickups, and it catches on with buyers, and the other manufacturers are forced to catch up. Until then, the bulk of the measurements have been done by enthusiastic 'amateurs'.

Others know the history better than me, but the move to adopt a standardized protocol with 470pF/200 kOhm loading seems to stem back to Helmuth Lemme ....

www.buildyourguitar.com/resources/lemme/

("The first version of this article was published in the American magazine "Electronic Musician" (edited by Craig Anderton), December 1986, p. 66 - 72.")

www.planetz.com/wp-content/uploads/2014/10/Pickup_Measuring_Technique.pdf

While Lemme's Pickup Analyser device has apparently been spotted at some pickup manufacturers, others like PRS seemingly developed their own bode plot rigs/protocols (with or without knowledge of what others were doing).

guitarnuts2.proboards.com/thread/9868/prs-muses-on-pickup-analysis

I do think there is something to be said for adding a typical "Gibson" and "Fender" resistance loading to the protocol. While furthering our understanding of pickup physics will often require the single standard loading for all pickups, also being able to show bode plots with loadings more akin to particular guitar types should make it easier to convince players of the importance of bode plots.

[antigua]

Jun 6, 2023 19:05:23 GMT -5 Josh said:

It's awesome that everybody here has a consistent comparison based on those arbitrary values, however...

What about trying to compare to data from manufacturers? Are they adding the same, or some other arbitrary load, or testing the pickups directly? I ordered a board to make the Integrator because it looked like it would be entertaining, I'll see how the measurements look with it.

From another post I made, I compared my direct to pickup measurements:

Hemi Neck
Lace website:
Resistance: 7.0k
Peak Frequency: 2750
Inductance: 3.5 henries

My TH2830 LCR: Ls: 3.69066H @ 100Hz

Hemi Bridge
Lace website:
Resistance: 10.0k
Peak Frequency: 2600
Inductance: 4.6 henries

My TH2830 LCR: Ls: 4.99706H @ 100Hz

My measurements were series inductance at 100Hz with kelvin probes connected directly to the pickup. Other frequencies, or using parallel inductance didn't come nearly as close to Lace's measurements.

I haven't seen many bode plots from manufacturers to compare with.

Thanks,
Josh

I don't think arbitrary is the right word, it's more like a standard, and standards are often arbitrary, but the fact of them being a standard is more meaningful than the fact of if being arbitrary, like the precise length of a foot in the English system... who's foot? The 470pF capacitive load was previously used by Helmuth Lemme as far back as 2009. I'm sure 470pF was chosen because it's a standard cap value that comes close to a 10 to 15 foot guitar cable, which is a typical length.

As far as pickup manufacturers go, it seems clear that they prefer to just sell pickups based on description of how the pickup might sound in your guitar. About the most they provide spec-wise is DC resistance and in a few cases, inductance. I suppose a lot of guitarists wouldn't understand what a bode plot is, but speaker manufacturers like Celestion and Jensen provide lots of specs and plots. If ever there was evidence that there is a lot of snake oil in how pickups are marketed, that would be it, IMO. 

Fender provides inductance values for a lot of their pickups, but often the values they state are flat out wrong, suggesting inductance values that aren't believable for a given pickup. In the case of Lace, if they don't tell us how they measured the inductance, we can't know, so it just has to be regarded as semi-reliable information. Your inductance measurements don't seem to be overly deviated from Lace's, looks like its within 10% 

[antigua]

Fishman created this plot below in order to show potential customers how their Fluence model accurately modeled the EQ curve of vintage and active pickups...




and interestingly if you see how the humbucker lands around 2kHz, and the Strat 3kHz to 4kHz, it's likely they used a 470pF load, or something close to it, otherwise the frequencies would have a larger spread. The Q factors depicted don't make sense though, under almost no circumstances will a Gibson type humbucker have a higher Q than a Strat pickup. 

In this case I don't think Fishman was trying to be informative so much as just provide evidence that they took a technical approach to recreating the voicing of passive pickup, probably the same in the case of PRS. 

[Josh]

Looking at that plot, there's definitely something goofy with my test setup. I typically see dB in that range on these plots. However, with my test setup, I'm seeing in the 40dB range without the 100Ω on the exciter, or in the 4dB range with the 100Ω on the exciter. 

I built the Integrator circuit but, unfortunately, Mouser didn't have the variable capacitor anymore, and I didn't notice, so I had to order one from digikey.