Hantek 1833C LCR meter, great for pickups

[Joe Gwinn]

What capacitance do you get using "Terman's method"?

Have not yet done that, for lack of a comparison.

BTW, Eddy currents don't influence capacitance.

But they do affect the apparent inductance, which figures into the equation.

[Helmholtz]

[SIZE=18px]

Have not yet done that, for lack of a comparison.

All you need is the PU and a few caps. Comparison is with the meter results.

[Antigua]

It's not surprising that an LCR meter will give inconsistent capacitance readings in general, as the bode plots show some pickups are not purely capacitive above the primary resonance. We don't know why that's the case, but it is the case. The benefit of the Hantek is that it gives multiple test frequencies above the resonance. Outliers can be discarded and an average can be determined.

I'm willing to try the Terman's method, but I would need instructions. I have time to do the test, but not research Terman's method and apply the principle to the application.

[Mike Sulzer]

[SIZE=18px]

Reduced all the R-X data, mathematically converting to Ls (below resonance, about 10 KHz), and Cs (above resonance). The inductance was very stable, and reproduced the Extech values where available. The apparent self-capacitance varies quite a lot when well above resonance, ranging from 120 pF and 174 pF, a ~50% range. I assume that this is due to eddy currents interacting with distributed capacitance somehow, but don't really know. I have plots, but have not figured out how to post them.

AlNiCo magnets do not show a lot of eddy current effects.

Maybe the apparent capacitance varies with frequency because the inductive reactance is still significant except at the highest frequency.

It is interesting that the instrument does not give a value for inductance above 1KHz. Perhaps it thinks that the reading is not accurate.

Poking around the web, I see that some people have successfully updated the firmware. I even found one claim of achieving 1833c functionality from an 1832 just by changing the firmware. The implication is that the hardware is the same, but the programming is different.

[Joe Gwinn]

AlNiCo magnets do not show a lot of eddy current effects.

But there is that nice big sheet of copper-plated mild steel right there, oriented perfectly to support an eddy-current image of the coil winding currents.

Hmm. The pickup coil is thick in the direction perpendicular to the coil plane and baseplate plane, so eddy currents in the baseplate will push back on currents in the coil, which may affect self-capacitance. I need to think about this.

Maybe the apparent capacitance varies with frequency because the inductive reactance is still significant except at the highest frequency.

Not understood. Please expand.

It is interesting that the instrument does not give a value for inductance above 1KHz. Perhaps it thinks that the reading is not accurate.

It turns out that Extech has a similar unit, their model LCR200. In the LCR200 datasheet, they clearly say the their accuracy specs apply only if D (dissipation) does not exceed 0.1 (meaning that Q exceeds ten) - this excludes electric guitar pickups for sure.

The LCR200 users manual also says that if the reactance exceeds 10 Kohms, they switch from SER to PAR mode, which is also crippling for pickups if true.

This is a firmware design issue - the 1833C does seem to measure R (AC resistance) and X (signed reactance) correctly.

Poking around the web, I see that some people have successfully updated the firmware. I even found one claim of achieving 1833c functionality from an 1832 just by changing the firmware. The implication is that the hardware is the same, but the programming is different.

That seems likely, given that the hardware is far harder to change than the firmware.

[Joe Gwinn]

It's not surprising that an LCR meter will give inconsistent capacitance readings in general, as the bode plots show some pickups are not purely capacitive above the primary resonance. We don't know why that's the case, but it is the case. The benefit of the Hantek is that it gives multiple test frequencies above the resonance. Outliers can be discarded and an average can be determined.

Yes.

I'm willing to try the Terman's method, but I would need instructions. I have time to do the test, but not research Terman's method and apply the principle to the application.

As requested by Helmholtz, I will apply Terman's method to my test pickup, and will report the results and document the process.

Your version of Terman's method can use your Bode plot instrument, with the added ability to determine the frequency at which there is zero phase shift in the various resonant peaks.

[Helmholtz]

I'm willing to try the Terman's method, but I would need instructions.

You will need a few caps (say 100p, 220p, 330p, 470p, 680p) and a piece of millimeter paper.

Measure and take down the exact cap values.

Then wire each cap in parallel with the PU and measure the resulting resonant frequencies.

Create a table with cap values, corresponding resonant frequencies ( fn ) and the calculated values of 1/(fn)².

Plot the 1/(fn)² results over the cap values. As long as L is constant, you should get a straight line when connecting the dots.

The PU's self-capacitance is the ("negative") capacitance, where the extrapolated line intercepts the capacitance axis.

[Mike Sulzer]

Not understood. Please expand.

Unless the inductive reactance at a particular test frequency is much larger than the capacitive reactance, both are significant in the measured value. This suggests that you want to use the highest frequency, Or you could attempt to correct approximately the measurement made at a lower frequency by using a low frequency measurement of the inductance and resistance, and removing the impedance of the L and R series combination from the measurement. This is easier to see if you work with admittances rather than impedances since they just add for a parallel combination.

For example, below is a plot of the real and imaginary parts of the impedance of a Seymour Duncan SH2N, a humbucker (which has significant eddy current losses). It has a low frequency inductance of 3.75H and a low frequency resistance of 7215 ohms. If I estimate the C from the Z at 20KHz, I get 62.8pf, which seems low. So from the admittance at 20KHz I subtract the admittance at 20KHz of the L and R combination, convert to Z and compute C. I get 78.1pf, better, but maybe still not good enough. So the next step would be to compute a better value of the inductance at 20KHz (since L is significantly lower at 20KHz than 120Hz), using the new approximation of the C in a similar calculation. That would probably do it. My current computer code uses a different approximation, but I kind of like this iterative one.

[Joe Gwinn]

Unless the inductive reactance at a particular test frequency is much larger than the capacitive reactance, both are significant in the measured value. This suggests that you want to use the highest frequency, Or you could attempt to correct approximately the measurement made at a lower frequency by using a low frequency measurement of the inductance and resistance, and removing the impedance of the L and R series combination from the measurement. This is easier to see if you work with admittances rather than impedances since they just add for a parallel combination.

For example, below is a plot of the real and imaginary parts of the impedance of a Seymour Duncan SH2N, a humbucker (which has significant eddy current losses). It has a low frequency inductance of 3.75H and a low frequency resistance of 7215 ohms. If I estimate the C from the Z at 20KHz, I get 62.8pf, which seems low. So from the admittance at 20KHz I subtract the admittance at 20KHz of the L and R combination, convert to Z and compute C. I get 78.1pf, better, but maybe still not good enough. So the next step would be to compute a better value of the inductance at 20KHz (since L is significantly lower at 20KHz than 120Hz), using the new approximation of the C in a similar calculation. That would probably do it. My current computer code uses a different approximation, but I kind of like this iterative one.

Given that that inductance variation is not linear with frequency (it may be square root though), I think that the easiest approach is iterative search for the zero-phase frequency (orange line), searching the neighborhood of the peak AC resistance (blue line). If there are multiple such peaks, each would get its own iterative search.

I did think of one possible cause for the resonances well above the main resonance, the inductance resonating with stray capacitance to a metallic baseplate or the like.

[Mike Sulzer]

Given that that inductance variation is not linear with frequency (it may be square root though), I think that the easiest approach is iterative search for the zero-phase frequency (orange line), searching the neighborhood of the peak AC resistance (blue line). If there are multiple such peaks, each would get its own iterative search.

I did think of one possible cause for the resonances well above the main resonance, the inductance resonating with stray capacitance to a metallic baseplate or the like.

Yes, that is how I find the resonance: start with a maximum from the blue line, and then fit a short line segment to the imaginary part, centered at the frequency found from the real part, in order to locate the zero crossing accurately. I do not compute the phase, but rather use the rectangular components (real and imaginary).

Why would stray capacitance make a separate peak with a single coil? But strange things can happen when there are two coils in series as with a humbucker.

[Joe Gwinn]

As requested by Helmholtz, I will apply Terman's method to my test pickup, and will report the results and document the process.

My venerable function generator (bought in 2004) broke yesterday, so there will be a delay while I get a new generator. I'm tending towards the Rigol model GD1000Z series.