Journal of Magnetic Resonance 187 (2007) 27–37

www.elsevier.com/locate/jmr

Estimation and measurement of flat or solenoidal coil

inductance for radiofrequency NMR coil design
Jan K. Rainey 1, Jeffrey S. DeVries, Brian D. Sykes *

Protein Engineering Network of Centres of Excellence and Department of Biochemistry, University of Alberta,
419 Medical Sciences Building, Edmonton, Alta., Canada T6G 2H7

Received 3 January 2007; revised 15 March 2007

Available online 6 April 2007

Abstract

The inductance of a radiofrequency coil determines its compatibility with a given NMR probe circuit. However, calculation (or esti-
mation) of inductance for radiofrequency coils of dimensions suitable for use in an NMR probe is not trivial, particularly for flat-coils. A
comparison of a number of formulae for calculation of inductance is presented through the use of a straightforward inductance mea-
surement circuit. This technique relies upon instrumentation available in many NMR laboratories rather than upon more expensive
and specialized instrumentation often utilized in the literature. Inductance estimation methods are suggested and validated for both
flat-coils and solenoids. These have proven very useful for fabrication of a number of new coils in our laboratory for use in static
solid-state NMR probes operating at 1H frequencies of 300 and 600 MHz. Solenoidal coils with very similar measured and estimated
inductances having inner diameters from 1 to 5 mm are directly compared as an example of the practical application of inductance esti-
mation for interchange of coils within an existing solid-state NMR probe.
 2007 Elsevier Inc. All rights reserved.

Keywords: Inductance calculation; Inductance measurement; Solid-state NMR; Flat-coil; Solenoid

1. Introduction of the solid-state NMR field, we will refer to coils with a

circular cross-section as solenoids and those with a rectan-
Typically, solid-state NMR makes use of cylindrical gular cross-section as flat-coils.
solenoid radiofrequency coils. Such a coil is optimal for It is often desirable to employ coils with different dimen-
samples with a circular cross-section. In samples with sions in a given static solid-state NMR probe. Prior to coil
inherent anisotropy, such a circular cross-section may not and sample fabrication, it is useful to determine ranges of
be optimal. For example, samples such as polymer films dimensions which will be functional for a given probe
or solid-supported lipid bilayers will have a relatively small rather than relying on trial and error. As an example, it
thickness in comparison to their length and width. The use may be desirable to test a range of conditions with small
of solenoids with a rectangular cross-section, or so-called amounts of valuable sample, and then to increase the sam-
flat-coils, for samples of this nature provides a maximal fill- ple and coil volume once the optimal sample preparation
ing factor for NMR experiments [1]. Because signal inten- and observation conditions are determined. In our case,
sity is directly proportional to coil filling factor [2], this we have been employing commercial two-channel static
optimizes signal-to-noise ratio. Following the convention solid-state NMR probes. In each of these probes, one chan-
nel is optimized for the 1H-frequency of the spectrometer
*
and its circuitry cannot be easily modified while the second
Corresponding author. Fax: +1 780 492 0886.
E-mail address: [email protected] (B.D. Sykes).
channel is configured to allow switching between different
1
Present address: Department of Biochemistry and Molecular Biology, nuclei by changing capacitors within the channel circuit.
Dalhousie University, Halifax, NS, Canada B3H 1X5. Practically speaking, this means that there is a limited

1090-7807/$ - see front matter  2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.jmr.2007.03.016
28 J.K. Rainey et al. / Journal of Magnetic Resonance 187 (2007) 27–37

range of possible coil geometries which will function at the moshier.net/) and the Excel spreadsheet we provide
1
H frequency for the probe. While a number of formulae (see Section 2.7) calculate this formula, allowing direct
are used for inductance calculation (e.g. [3,4]), these are comparison with the formulae derived herein. To make
typically optimized for coils of much larger dimensions inductance calculations easy to carry out for a wide variety
than those employed for NMR spectroscopy, especially of coil geometries and configurations, we use a fairly
in the case of rectangular cross-sectional coils. This paper fundamental level of electricity and magnetism theory [8].
presents formulae for a priori estimation of working coil Details of the derivation are given in Appendix B. For a
configurations for a given probe. flat-coil approximated as a series of parallel current-carry-
A typical protocol in an NMR laboratory is the testing ing rectangles with dimensions as illustrated in Fig. 1, the
of a new coil’s resonance characteristics directly in the estimate is:
NMR probe using the spectrometer tuning interface. An
4l0
alternate protocol allowing coil characterization without Lffi ðhw½N  1sÞ
relying on the use of an NMR spectrometer is the use of p22 0
a network analyzer or a sweep-generator and oscilloscope
6X B
ub
w
with directional coupler. This allows direct observation of 4 @qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
the resonant frequency response of the probe circuit over n¼lb ð4ðnsÞ þ h Þð4ðnsÞ2 þ w2 þ h2 Þ
2 2

a broad range of frequencies and often facilitates manipu- 13 2

lation of the displayed data more freely. The former case h C7
ties up an NMR probe and may rely on immediate spec- þ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiA5 ð2Þ
2 2 2 2 2
trometer access, which is not always desirable or possible, ð4ðnsÞ þ w Þð4ðnsÞ þ w þ h Þ
while the latter case uses a specialized and costly instru-
ment not available in many NMR laboratories. Alterna- where lb and ub are given by
tively, the inductance of a given coil can be directly
measured using a network analyzer with attached pickup lb ¼ ððN  1Þ=2Þ
loop (e.g. [5,6]) having known resonant properties (Dr. ub ¼ ððN  1Þ=2Þ ð3Þ
P.L. Gor’kov, personal communication). In order to sim-
plify matters, we developed a highly reproducible and for an odd number of turns N, and
cost-effective method of radiofrequency coil characteriza-
lb ¼ ððN  2Þ=2Þ
tion independent of having an NMR spectrometer or probe
immediately available and relying only on access to an ub ¼ ðN =2Þ ð4Þ
oscilloscope, sweep generator, and reflection bridge. This
method uses a straightforward circuit for direct inductance for an even N. The differentiation between Eqs. (3) and (4)
measurement, which we present and use to compare coil keeps the calculation in the centre of a turn closest to the
inductance estimation methods. middle of the coil, which we have found experimentally
to provide a better estimate. It should be noted that a ma-
2. Results and discussion jor simplification during the derivation of this formula,
resulting in the overestimation of inductance, is that the
2.1. Inductance estimation magnetic field throughout the volume of the coil is taken
to be that at the coil centre. In reality, there would be lower
A given channel of an NMR probe acts as an oscillatory magnetic field in regions closer to the windings. This is
tank circuit with an inductor and capacitor in parallel (i.e. compensated for empirically in Section 2.3.
an LC-circuit) [7]. Its resonant frequency (in Hertz), m, will A frequently used approximation for long inductors is
be given by: l0 N 2 A
Lffi ð5Þ
1 1 l
m¼ pffiffiffiffiffiffiffi  pffiffiffiffiffiffiffiffiffi ð1Þ
2p LC 2p LC T
where CT is the capacitance of the variable tuning capacitor
(approximately equal since stray capacitance and self-
capacitance of coil are being neglected) and L is the coil
inductance. Therefore, the ability to calculate coil induc-
tance provides a good estimator of its suitability for use
in a given probe. A good estimator of inductance of a sole-
noidal coil wound on a rectangular former was derived by
Niwa [4]. However, this formula, given in Appendix A,
Fig. 1. Schematic of flat-coil with 5 turns (N = 5), showing definition of
contains a summation of 12 terms based on the coil height (h), width (w), length (l), and turn separation (s = l/(N  1)). The
geometry, and is inconvenient to calculate. Note that the simplification to a series of ideal current-carrying rectangles used for
program ‘‘coil’’ by Moshier (available at http://www. inductance estimate calculation is shown.
 J.K. Rainey et al. / Journal of Magnetic Resonance 187 (2007) 27–37 29

where A is the cross-sectional area and l is the length of the

inductor [3]. For a rectangular coil as defined here, this
gives
l0 N 2 hw
Lffi ð6Þ
½N  1s
The typical flat-coil used for NMR experiments has rela-
tively few turns (4–6 is typical) and a relatively short length
in comparison to its diameter, meaning that Eq. (6) is prob-
ably not a good approximation for most NMR coils. We Fig. 2. Inductance measurement circuit. An inductor L (either a test coil
will use both the idealized Eq. (6) and the Niwa formula with an unknown inductance or a commercial inductor of known induc-
tance) is placed in parallel with variable capacitor CB to create a simple
(Appendix A; Eq. (A.1)) for comparison purposes.
resonant tank circuit. Variable capacitors CA and CC provide proper
A similar analysis to that described above for flat-coils matching and balancing of the resonant circuit.
may readily be carried out for a solenoidal coil with circu-
lar turns (Appendix B), giving the estimate
" ! #2 Appendix C. Using calibrated capacitors CA, CB and
l0 pr6 ½N  1s X ub
1 CC, coil inductance can be determined by measurement
Lffi 2
ð7Þ of the circuit resonance frequency and use of Eq. (1).
4 2 3=2
n¼lb ððnsÞ þ r Þ
Note that this is more cost-effective and feasible for the
where r is the radius of the circular coil, and n, and s retain average NMR laboratory than use of equipment such as
the same meaning as for the rectangular coil and the values a network analyzer (e.g. [12]).
lb and ub are given by Eqs. (3) and (4) for odd and even N, Three simplifying conditions were used during circuit
respectively. (Solenoids are typically reported in terms of analysis. First, resistance of the circuit components was
inner diameter (id) rather than radius; we follow this con- neglected. Second, capacitance arising from the inductor
vention for the remainder of the paper.) The long solenoid or stray capacitance in leads and wires is not taken into
approximation (Eq. (5)) gives the simpler expression account. Third, self-inductance of the capacitors is not
l0 pN 2 r2 included. In order to compensate for these approximations,
Lffi ð8Þ the circuit was calibrated using a number of known induc-
½N  1s
tors. In each case, the measured inductance calculated
which is likely less accurate for the reasons outlined above. using Eq. (1) was greater than the real inductance. This off-
Wheeler developed more accurate estimates for solenoids set was dependent upon the value of CB. Therefore, we cal-
of finite length [9]—in the case of l greater than 0.8r, this ibrated the inductance measurements for each frequency
is given by: extreme of the resonance circuit separately (Fig. 3) and
report the average for each ‘‘measured inductance’’ value.
10l0 pN 2 r2 10l0 pN 2 r2
Lffi ¼ ð9Þ As a whole, we estimate the uncertainty in our measure-
9r þ 10l 9r þ 10½N  1s ments to be 20–30 nH due primarily to factors such as vari-
A different formula is provided by Doty [10] for l greater ations in coil positioning or orientation relative to the
than 0.2r remainder of the circuit and inability to exactly reproduce
the soldering positions for coil legs.
4N 2 r2 ð1  0:2=N Þ
Lffi nH ð10Þ
l þ 1:2r0:9 2.3. Flat-coil inductance results
with r and l given in mm (note that Doty recommends
taking r as inner radius + 20% of wire radius), which is in- The suitability of the flat-coil inductance estimator (Eq.
tended to correct for the helical shape of a real solenoid. (2)) as compared to the long inductor approximation (Eq.
All in all, there is little consensus in the literature as to (6)) and the Niwa formula (Appendix A; Eq. (A.1) was
the most appropriate method to calculate a coil’s induc- examined by fabrication of a number of flat-coils and mea-
tance. The performance of each of these formulae is com- surement of their inductance using the circuit illustrated in
pared below with respect to measured inductances for a Fig. 2 as calibrated using known inductors (Fig. 3). A vari-
variety of solenoidal coils. ety of dimensions were chosen while the number of turns
was kept in the range 4–6 (Table 1), the typical practical
2.2. A simple inductance measurement circuit range of turns for flat-coil NMR. Note that six different
series of coils with nearly constant dimensions but having
Rather than examining changes in NMR probe 4, 5, and 6 turns were examined (series F1-1–F1-3 to F6-
response to test inductance (e.g. [11]), it is much simpler 1–F6-3), three further pairs of coils with 5 and 6 turns were
to make direct use of an LC-circuit. The circuit illustrated examined (F7-1–F7-2 to F9-1–F9-2), and two individual
schematically in Fig. 2 was used, with specifications and coils (F10 and F11) with unique dimensions. These coils
measurement protocol as detailed in Section 4.1 and cover a wide variety of coil height to width ratios. Some
30 J.K. Rainey et al. / Journal of Magnetic Resonance 187 (2007) 27–37

Fig. 3. Calibration of inductance measurement circuit (shown in Fig. 2) using inductors of known inductance (error bar based on tolerance of inductor).
The illustrated regression lines are for the entire range of inductances. Known inductance is plotted against measured inductance. CB in the circuit was
varied between its extreme values (high and low ranges of measurements) for each inductor while CA and CC were left constant, providing two measured
tank circuit resonant frequencies for each inductor. Effective total circuit capacitance (Eq. (13)) combined with the resonant frequency was used to
determine measured inductance value (Eq. (1)). Linear regression (fit lines shown; performed on low or high effective total capacitance measurements
separately) then provides a calibrated measurement of inductance for an unknown inductor examined in the same manner.

comparisons between the inductance properties for flat- side remains the same. To compensate for the different
tened vs. round copper wire and of silver vs. copper wire performance of Eq. (2) with variations in the width to
were also performed. Measured inductances are shown in height ratio, we propose the following as the most useful
Table 1 for each of these 29 coils and measured vs. calcu- estimate of flat-coil inductance:
lated inductances are shown graphically in Fig. 4 (numeri- 8
> 3=7 Eq: ð2Þ if 8 > w=h P 3 and h < 2:5 mm
cal values provided in Supplementary Material). Although >
>
< 4=7 Eq: ð2Þ if w=h P 8 and h < 2:5 mm
major changes to wire diameter, shape and material will
Lflat 
definitely affect coil inductance, we did not notice any sys- >
> 11=14 Eq: ð2Þ if w=h P 3 and h P 2:5 mm
>
:
tematic perturbations in inductance for the direct compar- Eq: ð2Þ if w=h < 3
isons we performed. Our general finding to date has been
ð11Þ
that the use of different wire materials or shapes does not
significantly impact inductance of the typical NMR coil; Note that we are assuming coil dimensions with h < w; if
note, though, that power handling or uniformity of B0 h > w, substitution of h for w and w for h in Eq. (11) should
within the coil may be affected by these types of changes. be carried out. As can be seen in Fig. 4, this estimate per-
With small-volume coils, for example, effects of using forms within 5% of measured inductance for 10/29 coils,
round vs. flattened wire become more pronounced, as dem- within 15% for 18/29 coils, with only 5/29 coils having er-
onstrated by Li et al. [13]. ror of magnitude >25%. Estimates by Eq. (11) that are off
Eq. (2) tends to overestimate flat-coil inductance, with a by >15% range between 11 and 51 nH in magnitude of
typical overshoot in the range of 20–30% and an extreme of error versus measured inductance, with only four cases
120 nH. As mentioned earlier, this general overestimation where this magnitude is >25 nH. (Recall that uncertainty
is most likely to be due to overestimation of the magnetic in our measurement method is estimated at 20–30 nH.) In
field in regions close to the windings of the coil. Eq. (6), a recent paper, Gor’kov et al. provide a measured induc-
in comparison, is seen to provide a much wider range of tance of 80 nH for a 4-turn 8 · 6 mm flat-coil 12 mm
results, where every estimate is an overestimate. Finally, long.[14] Eq. (11) provides an estimated inductance of
Eq. (A.1) provides reasonably good estimates in a number 62 nH, which is a reasonably good estimate well within
of cases, but tends to underestimate inductance by the uncertainty of our method. A scan of the literature pro-
5–30%. Examination of the performance of Eq. (2) vs. vides no further reported inductance values for flat-coils,
coil geometry provides an empirical manner to improve therefore further external validation of Eq. (11) is difficult.
its performance. There is a definite dependence of the area As discussed below, in practice, we find that a flat-coil
and proximity of the current sheets on each pair of sides having an estimated inductance using Eq. (11) on the same
(i.e. top and bottom vs. left and right sides) of the flat-coil order as the estimated inductance of a solenoidal coil
upon estimation accuracy. Note that this can be reduced to (recommended estimate given in Eq. (12)) provides a flat-
the ratio of coil width to height, since the length of each coil which will work in the probe circuit.
 J.K. Rainey et al. / Journal of Magnetic Resonance 187 (2007) 27–37 31

Table 1
Flat-coil dimensions, number of turns, and measured and estimated inductances used to determine suitability of flat-coil inductance estimation formulae
Coil label N w (mm) h (mm) l (mm) w/h Inductance (nH)a
F1-1 6 15.8 5.8 12.2 2.7 273 ± 4b
F1-2 5 16.2 5.6 12.2 2.9 196 ± 0.9b
F1-3 4 15.7 6.0 12.0 2.6 143 ± 3c
F2-1d 6 19.4 4.6 17.0 4.2 237 ± 5b
F2-2d 5 20.2 5.0 16.4 4.0 169 ± 0.2b
F2-3d 4 20.2 5.1 16.4 4.0 144 ± 0.7c
F3-1 6 15.8 3.9 11.8 4.1 197 ± 0.1b
F3-2 5 15.5 3.5 12.25 4.4 141 ± 5c
F3-3 4 15.4 3.9 11.8 3.9 94 ± 0.8c
F4-1 6 15.4 1.9 11.8 8.1 134 ± 2c
F4-2d 6 15.4 1.9 12.2 8.1 111 ± 2c
F4-3 6 15.0 2.2 15.4 6.8 107 ± 2c
F4-4e 5 15.6 1.9 11.7 8.2 95 ± 0.9c
F4-5 5 15.2 2.0 12.0 7.6 88 ± 0.03c
F4-6 4 15.4 1.9 12.5 8.1 60 ± 5c
F5-1d 6 10.9 3.6 12.2 3.0 134 ± 1c
F5-2d 5 11.0 3.5 12.1 3.1 89 ± 0.3c
F5-3d 4 10.8 3.6 12 3.0 53 ± 2c
F6-1 6 9.2 1.8 17.0 5.1 42 ± 0.4c
F6-2 5 9.2 1.8 17.7 5.1 31 ± 2c
F6-3 4 9.2 1.8 15.1 5.1 18 ± 1c
F7-1 6 13.8 1.8 13.7 7.7 83 ± 0.1c
F7-2 5 13.2 1.7 16.4 7.8 54 ± 0.1c
F8-1 6 13.0 6.0 12.6 2.2 246 ± 2b
F8-2d 5 12.4 5.5 12.6 2.3 165 ± 3b
F9-1d 6 11.8 4.6 11.5 2.6 195 ± 0.5b
F9-2d 6 12.3 4.4 13.6 2.8 157 ± 3b
F10 6 20.3 3.8 16.5 5.3 200 ± 2b
F11d 4 12.8 2.8 12 4.6 55 ± 3c
All dimensions (see Fig. 1) are for the interior of the coil except length which is leg-to-leg distance. Unless otherwise indicated, all coils were fabricated with
20 AWG copper wire.
a
Errors are given based upon the paired measurements made at high and low total capacitance for the measurement circuit. Uncertainty in inductance
measurement is 20–30 nH.
b
Inductance calibrated by linear regression over 0–1.0 lH range of inductors (Fig. 3).
c
Inductance calibrated by linear regression over 0–150 nH range of inductors (Fig. 3).
d
Flattened 20 AWG copper wire.
e
19 AWG silver wire.

2.4. Solenoidal coil inductance results Based upon these results, we suggest that Eqs. (7) and
(9) be directly compared at coil diameters of [5 mm,
A total of 24 solenoidal coils with a variety of id’s, with the inductance estimate given by the following
lengths, and numbers of turns were fabricated and induc- expression:
tances were measured using the circuit illustrated in
Fig. 2, as described in Section 4.1. Coil parameters along Lsolenoid 
with measured inductances are given in Table 2. Fig. 5 8
shows measured vs. calculated inductance for: (1) the for- < Eq: ð7Þ if 3=4 Eq: ð7Þ > Eq: ð9Þ
>
Eq: ð9Þ if Eq: ð9Þ > 4=3 Eq: ð7Þ
mula derived in the same manner as for the flat-coil (Eq. >
: Eq: ð7ÞþEq: ½9
(7)); (2) the long-inductor formula (Eq. (8)); (3) Wheeler’s 2
if 3=4 Eq: ð7Þ < Eq: ð9Þ < 4=3 Eq: ð7Þ
inductor of finite length (Eq. (9)); and, (4) Doty’s induc- ð12Þ
tance estimation formula (Eq. (10)). Generally, Eq. (9) pro-
vides an excellent estimate for coils of any id over the range
measured of 1–10 mm. Eq. (7) tends to severely underesti- Note that Eq. (12) will almost always result in Eq. (9) for
mate inductance for diameters of 5 mm and greater, but larger diameters where Eq. (7) provides an underestimate.
performs about equally well to Eq. (9) for the smaller diam- To be certain of the better estimate at diameters of 5 mm
eters. Eq. (10) performs similarly to Eq. (9), but Eq. (9) and above, however, Eq. (9) should be used without refer-
tends to be more consistent with our measurements of coil ence to Eq. (12). Following this estimation procedure, the
inductance. Finally, Eq. (8), which is notably a frequently results shown in Fig. 5 demonstrate that a good estimation
encountered formula for inductance calculation, does not of inductance was provided for all coils tested. Of the 24
provide a consistent inductance estimate in the regime of coils tested, 11/24 were predicted at <5% error and 19/24
NMR solenoidal coils. at 15% error or better, with the remaining 5 coils at
32 J.K. Rainey et al. / Journal of Magnetic Resonance 187 (2007) 27–37

Fig. 4. Measured vs. calculated inductance for a series of 29 flat-coils (Table 1). Measurements were carried out using the circuit shown in Fig. 2 as
calibrated in Fig. 3; calculations were carried out using the indicated inductance estimation formulae. The solid line indicates the exact agreement between
calculation and measurement.

26% error or better (magnitudes of error for these 5 coils would be completely ineffective in this estimation, since
were 5–25 nH). the inductance ranges it estimates are 3· the real values.
Therefore, we have found Eq. (11) to be extremely useful
for selection of the number of turns and fine-tuning of coil
2.5. Practical testing of flat-coils dimensions without the need for building and testing of
numerous coils.
For a Chemagnetics Apex DRNS Wideline static probe
(Varian Associates, Fort Collins, CO) operating at
300 MHz for the 1H channel, the two solenoidal coils pro- 2.6. Examining solenoids with similar inductance but
vided by the manufacturer were a 5 mm coil (5-2, Table 2; different inner diameter
estimated inductance (Lest)  113 nH) and a 10 mm coil
(10-1, Table 2; Lest  170 nH). Coils F9-2 (Lest  208 nH), Use of NMR coils with different diameters is common
F4-2 (Lest  135 nH), and F4-4 (Lest  99 nH) were for solution-state as well as solid-state NMR. In cases
designed for, and work with, this probe. (Note that values where sample is limited, going to smaller coil diameter
of Lest are given for the required lead-to-lead length of assists by increasing concentration through the decreased
11.75 mm for the probe, as opposed to the lengths volume required to fill the coil (see e.g. [15] for extensive
reported in Tables 1 and 2.) In general, each of these discussion). Using a Varian T3 PISEMA static solid-state
three flat-coils tunes and matches very satisfactorily in NMR probe operating on an INOVA 600 MHz spectrom-
the probe circuit with the same capacitor ranges for vary- eter (Varian Associates, Palo Alto, CA), we have compared
ing heteronuclei as the corresponding solenoidal coil. As performance of four coils shown in Fig. 6: 5-4 (6-turn,
might be anticipated, coil F-8 has a slightly different 5.0 mm id, Lobs  55 nH, Lestimate 62 nH), 3-1
tuning range than the 10 mm solenoid provided by the (9-turn, 3.2 mm id, Lobs  56 nH, Lestimate 56 nH), 2-1
manufacturer, but has still performed well with all (14-turn, 1.9 mm id, Lobs  66 nH, Lestimate 49 nH), and
solid-supported oriented lipid bilayer samples we have 1-2 (23-turn, 1.1 mm id, Lobs  56 nH, Lestimate 49 nH).
examined to date. Each coil has similar measured and estimated inductance,
In the case of a Varian/Chemagnetics T3 narrowbore but different diameter and number of turns, allowing direct
HX PISEMA probe (Varian Associates) operating at a comparison of performance in the NMR probe of coils
1
H-frequency of 600 MHz, the two coils provided by the with different diameter. Using neat water as a sample in a
manufacturer were a 5 mm solenoid (5-4, Table 2; tube of the maximum possible diameter, we compared
Lest  62 nH) and a 4-turn flat-coil (F5-3; Lest  57 nH). applied RF-field strength (90 pulse width; cB1) and inte-
In order to increase the allowable sample width with a grated signal area per millimol of water. Note that peak
flat-coil, we estimated dimensions that would provide a area for identically acquired and processed spectra pro-
similar inductance (F11; Lest  68 nH); this coil indeed vides a direct, linewidth independent measure of sensitivity,
provides very similar tuning and matching performance since we observed identical noise amplitude for all four
in comparison to coils F5-3 and 5-4. Note that Eq. (6) coils. Since tube wall thickness becomes increasingly
 J.K. Rainey et al. / Journal of Magnetic Resonance 187 (2007) 27–37 33

Table 2 solid-state NMR probe for all measurements which makes

Solenoid dimensions and number of turns alongside measured inductances direct comparison of coil capabilities possible and demon-
used to determine suitability of solenoidal inductance estimation formulae
strates the utility of <10 lL volume coils in a standard
Coil label N id (mm) l (mm) Inductance (nH)a probe.
b
10-1 5 10.3 14.4 161 ± 3c
8-1 4 8.3 5 122 ± 0.2 2.7. Quick-reference guide to practical coil design for an
8-2 4 8.3 10.2 84 ± 1
6.5-1 6 6.5 11.6 127 ± 0.9
existing probe
6.5-2 5 6.6 12.0 76 ± 0.3
6.5-3 4 6.6 11.2 48 ± 0.4 The following process is our recommended design pro-
5-1d 6 5.0 1.8 224 ± 0.4c tocol for introducing new coils into an existing probe. In
5-2b 8 5.3 14.4 129 ± 3 order to make this process rapid for any laboratory, we
5-3 6 4.9 10.2 72 ± 0.5
5-4 6 5.0 12 55 ± 3
have produced a spreadsheet for Microsoft Excel allowing
5-5 5 5.0 11.2 45 ± 0.8 side-by-side comparative calculation of inductance esti-
5-6 4 4.9 11.9 25 ± 2 mates for flat- and solenoidal radiofrequency NMR coils
3.5-1 10 3.5 12 76 ± 2 of many geometries based on the equations recommended
3.5-2 8 3.5 12 50 ± 1 in this paper (freely available at http://structbio.bio-
3.5-3 6 3.5 11.0 31 ± 3
3.5-4 5 3.5 11.5 19 ± 2
chem.dal.ca/jrainey/).
3-1 9 3.2 12 56 ± 0.1
2.5-1 12 2.4 15 50 ± 2 (1) Calculate estimated inductances of existing coil(s)
2.5-2 10 2.4 12 36 ± 1 (Eqs. (11) and (12)), making note appropriate capac-
2-1e 14 1.9 12.6 66 ± 3 itor combinations for any exchangeable capacitors in
2-2e 17 1.9 21.2 44 ± 1
2-3e 15 1.9 20.5 33 ± 3
the probe.
1-1f 25 1.0 12 58 ± 0.4 (2) Optimize number of turns, length, and cross-sectional
1-2f 23 1.1 12 56 ± 0.1 dimensions of new coil based on sample size (Eq. (11)
Unless otherwise indicated, all coils were fabricated with 20 AWG copper or (12)). Knowing multiple useable inductance values
wire. (i.e. coils that operate with different sets of exchange-
a
Errors are given based upon the paired measurements made at high able capacitors) may be useful at this point, since a
and low total capacitance for the measurement circuit. Uncertainty in given sample may be more readily wrapped in a coil
inductance measurement is 20–30 nH. Unless otherwise noted, induc-
designed to have one inductance vs. another. How-
tance calibrated by linear regression over 0–150 nH range of inductors
(Fig. 3). ever, one should also take into special consideration
b
19 AWG silver wire. that the coil does not self-resonate [8,16] at or near
c
Inductance calibrated by linear regression over 0–1.0 lH range of the desired operating frequency of the probe as this
inductors (Fig. 3). will affect probe tuning and efficiency.
d
32 AWG copper wire.
e (3) Fabricate coil and test its performance in the probe
22 AWG copper wire.
f
26 AWG copper wire. using a sweep generator and oscilloscope, the spec-
trometer tuning circuitry, or other standard method.
crucial with smaller diameter coils, the peak area per
molecule normalized by the filling factor of the coil was
compared. Fig. 6 shows these measures for each of the four 3. Conclusions
coils tested as labelled with nominal id’s of 1, 2, 3 and
5 mm. Although desirable for allowing study of diverse sam-
Fig. 6 demonstrates a direct–linear relationship between ple types by a given NMR probe, the calculation of
the cB1 and sensitivity. Note that this is an inverse relation- inductances for radiofrequency flat-coils is not trivial.
ship to coil diameter (for coils of similar inductance), dem- We present a comparison of a variety of inductance esti-
onstrating an 10 fold-increase in sensitivity for a 1 mm mation formulae with measured inductances obtained
coil vs. a 5 mm coil if filling-factor is taken into account. using a straightforward inductance measurement method
This is on the same order of magnitude as the increases relying upon instrumentation available in many NMR
observed with microflow NMR, where smaller sample vol- laboratories rather than more specialized, and costly,
umes (1.5 lL nominal observe volume) are employed instruments often employed in the literature. Estimation
[15]. Note that in this latter case, a customized probe is methods are presented which have proven very useful
used due to the requirements of liquid flow through the for fabrication of new coils in our laboratory under a
coil. In work on microcoils, an 10–12-fold increase in sig- variety of cases in two different static solid-state NMR
nal-to-noise has been observed for a decrease in diameter probes operating at 1H frequencies of 300 and 600
from 1 mm to 100 lm [11,12]. At these dimensions, skin MHz. A direct comparison of the performance in an
effects become increasingly important, however, and sig- NMR spectrometer of solenoidal coils with very similar
nal-to-noise increases logarithmically with decreased measured and estimated inductances but different inner
diameter rather than linearly. Here, we use the same static diameters is also presented.
34 J.K. Rainey et al. / Journal of Magnetic Resonance 187 (2007) 27–37

Fig. 5. Measured vs. calculated inductance for a series of 24 solenoidal coils (Table 2). Measurements were carried out using the circuit shown in Fig. 2 as
calibrated in Fig. 3; calculations were carried out using the indicated inductance estimation formulae. The solid line indicates the exact agreement between
calculation and measurement.

Fig. 6. Coil sensitivity in form of observed peak area for water samples in shown solenoidal coils of given nominal inner diameter. All coils demonstrated
identical noise amplitude. Solid line is linear fit to uncorrected area per mmol while dashed line is for area per mmol corrected for coil filling factor.

4. Experimental iable capacitors (250 WVDC, Johanson Manufacturing

Corporation, Boonton, NJ, USA) (CA and CC; ranges
4.1. Inductance measurement 0.8–12.1 pF and 0.8–11.9 pF) and in parallel with a third
(CB; range 0.8–12.2 pF) to create a simple matched tank
The circuit illustrated schematically in Fig. 2 was used, circuit. A 50 X load terminated the circuit at the ground
where the coil was placed in series with a pair of 5202 var- plane to ensure proper loading of the circuit. An SG-677/U
 J.K. Rainey et al. / Journal of Magnetic Resonance 187 (2007) 27–37 35

sweep generator (Texscan Corporation, Indianapolis, inductance coils used linear regression results from the
IN, USA) was used to provide a frequency scan over the entire data set.
range 1 MHz–1.2 GHz and the circuit response (tuning/
matching/Q performance) was measured through a Acknowledgments
ZFDC-10-2 directional coupler (Mini-Circuits, Brooklyn,
NY, USA) using a model 475A oscilloscope (Tektronix, Thanks to Nic Shaw for helpful discussions. This work
Beaverton, OR, USA). An alternative configuration made was funded by the Canadian Protein Engineering Network
use of a 1–400 MHz sweep generator (Model 1061, Wave- of Centres of Excellence. J.K.R. is grateful for postdoctoral
tek, Indianapolis, IN, USA). Upon insertion of a new coil fellowships from the Alberta Heritage Foundation for
into the circuit, the capacitance of CB was set to both min- Medical Research, the Natural Sciences and Engineering
imum and maximum values. Capacitors CA and CC could Research Council of Canada, and the Canadian Institutes
be varied at each value of CB to achieve a well resolved of Health Research Strategic Training Program in Mem-
and matched maximum in the frequency scan, which pro- brane Proteins and Cardiovascular Disease; B.D.S. is sup-
vides the resonant frequency of the circuit. In practice, ported as a Canada Research Chair in Structural Biology.
we found that keeping CA and CC at their maximum capac-
itance provided a good match. The range of each capacitor Appendix A. Niwa formula for flat-coil inductance
and its rate of variation over that range was calibrated indi-
vidually using a model 820 BK capacitance meter (B&K Niwa’s formula for a solenoid of N equally spaced turns
Precision Corporation, Yorba Linda, CA, USA) prior to wound over a length of l on a rectangular former of height
assembly of the circuit. Approximate, total effective capac- h and width w is given by Grover [4] as:
itance (Ceff) of the circuit is given by 
wh 1 l w 1 l h 1
CACB þ CBCC þ CACC L ¼ 0:008N 2 sinh1 þ sinh1 
C eff  ð13Þ l 2h l 2w l 2
CA þ CC  2  
h l 1 w 1 w2
 1 2 sinh qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  1  2 sinh1
as developed in Appendix C. Coil inductance can then be l h 2 2 2 l
l 1 þ h =l
determined approximately given an observed resonance
frequency using Eq. (1). To compensate for systematic er- h 1h w 1w h
 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  sinh1  sinh1
rors between measured and actual inductance, the circuit 2 2 l h 2 l w
l 1 þ w2 =l
was calibrated using a range of 27 inductors of differing 0 1
specifications with overlapping inductance ranges (6 IM-2
Bp wh C
series molded inductors in the range 150 nH–1 lH 10% tol- þ @  tan1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiA
erance, Vishay Intertechnology, Malvern, PA, USA; 8 2 2 2
l 1 þ g2 =l
CM252016 series chip inductors in the range 10–150 nH sffiffiffiffiffiffiffiffiffiffiffiffiffi  sffiffiffiffiffiffiffiffiffiffiffiffiffiffi
10% tolerance and a 1 lH inductor 10% tolerance, 11 1 l2 g2 1 g2 1 l2 1 l 2 w2
þ 1þ 2 1 þ  1 þ
CI201210 series chip inductors in the range 12–220 nH 3 wh l 2 l2 3 wh 3 wh l2
5% tolerance, and one CM160808 chip inductor at 47 nH sffiffiffiffiffiffiffiffiffiffiffiffiffi
   
5% tolerance, Bourns, Riverside, CA, USA). All inductors 1 w2 1 l2 h2 1 h2
 1  1þ 2 1
were chosen such that their self-resonant frequencies were 2 l2 3 wh l 2 l2
well above the resonant frequencies being measured.  
1 l g 3  w3  h3
We found different performance of the circuit at each þ ðA:1Þ
6 wh l2
extreme of capacitor CB. In each case, the measured induc-
tance arising from Eqs. (1) and (13) was greater than the where g2 = h2 + w2. Note that based on similarity of terms,
real inductance. As shown in Fig. 3, each circuit resonance it is possible that the fourth term should have a l/w multi-
measurement made at the high-frequency end (low capaci- plier, however we were unable to obtain the original Niwa
tance value of CB) vs. the low-frequency end provided a paper from 1918 [17] to verify this; comparing calculated
significantly greater difference between measured and real inductances for the two forms of the formula, addition of
inductance. Therefore, we calibrated the inductance mea- this l/w multiplier reduces inductance estimates by
surements for each frequency extreme of the resonance cir- 0.1–1.1 nH.
cuit separately. Specifically, linear regression of measured
vs. actual inductance values was used to provide an expres- Appendix B. Derivation of inductance estimates
sion for corrected value of inductance. As this will be
dependent upon the circuit in question, this is not discussed As mentioned in the main text, we use a fairly funda-
in detail. It is the average of these two corrected values that mental level of electricity and magnetism theory [8]. To
is reported as the ‘‘measured inductance’’ for all coils begin with, we consider the magnetic field at an arbitrary
examined. For all coils measured with inductance below point in space, P, caused by a conductor. This is defined
150 nH, linear regression results from the calibration by the law of Biot and Savart. For an infinitesimally short
over the range of 0–150 nH were employed; all higher segment of conductor of length dl, a vector can be defined
36 J.K. Rainey et al. / Journal of Magnetic Resonance 187 (2007) 27–37

in the direction of the current, dl. The field at point P where y varies between 1/2h and 1/2h. Integrating Eq.
induced by a given conductor segment is dependent on (A.3) with Eqs. (A.7) and (A.8) substituted in over
the distance from segment dl to P, r, and the direction of y = 1/2h to 1/2h gives:
the vector between the conductor and P, ^r. The contribu- 0 1
tions of each conductor segment to the total magnetic field l0 I B h C
at P, BP, are: Bleft ¼ Bright ¼ @qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiA ðA:9Þ
p
ð4z2 þ w2 Þð4z2 þ w2 þ h2 Þ
l0 Idl  ^r
dBP ¼ ðA:2Þ
4p r2 where reversal of the sign of I applies for the left versus
right side, giving identical values for each side. The total
with magnitude given by: value of the magnetic field arising from the contributions
l0 Idl sin / of a given rectangular turn of the coil separated from point
jdBP j ¼ ðA:3Þ P by distance z along the coil centre-line is then given by
4p r2
the summation of Eqs. (A.6) and (A.9) multiplied by 2 to
where l0 is the permeability of free space, I is the current in give contributions for each of the four sides of the
the conductor, and / is the angle between dl and ^r. rectangle:
For clarity, the terminology used to refer to features of a 0
flat-coil in the following derivation is illustrated in Fig. 1. 2l0 I B w
To allow the integration to be independent of the number Bturn ðzÞ ¼ @qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
p
of turns in the coil, we simplify the geometry of the coil ð4z2 þ h Þð4z2 þ w2 þ h2 Þ
2

such that we have a series of parallel rectangles separated 1

from each other by distance s. Conductor width is also h C
þ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiA ðA:10Þ
taken to be nonexistent and skin effects [9,17] are not taken 2
ð4z2 þ w2 Þð4z2 þ w2 þ h Þ
into account. The magnitude of the magnetic field caused
by one turn (i.e. one rectangle of conductor) of the coil If the turns are separated by a uniform distance s (Fig. 1)
at some arbitrary distance z along the centre-line of the coil the total magnetic field at the centre of the coil is given
may be determined by substitution of the following equa- by the following sum:
tions into Eq. (A.3). First, for the top and bottom 0
conductors:
2l0 I X B
ub
w
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi B¼ @qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 p n¼lb
z2 þ ðh=2Þ ð4ns2 þ h2 Þð4ns2 þ w2 þ h2 Þ
sin / ¼ ðA:4Þ 1
r2
h C
and þ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiA ðA:11Þ
2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2 ð4ns2 þ w2 Þð4ns2 þ w2 þ h Þ
2 2
r2 ¼ ðx  1=2wÞ þ z2 þ ðh=2Þ ðA:5Þ
where lb and ub are given by
lb ¼ ððN  1Þ=2Þ
where x varies between 1/2w and 1/2w. Integrating Eq.
(A.3) with Eqs. (A.4) and (A.5) substituted in over ub ¼ ððN  1Þ=2Þ ð3Þ
x = 1/2w to 1/2w gives:
for an odd number of turns N, and
0 1
l0 I B w lb ¼ ððN  2Þ=2Þ
C
Btop ¼ Bbot ¼ @qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiA ðA:6Þ ub ¼ ðN =2Þ ð4Þ
p 2 2 2 2 2
ð4z þ h Þð4z þ w þ h Þ
for an even N. These different limits keep the calculation in
Note that the direction of the current is reversed in top ver- the centre of a turn closest to the middle of the coil.
sus bottom while the integration direction is also reversed We now make use of the fact that the total energy sup-
in top versus bottom giving an identical solution for each. plied by a current increase from 0 to I to an inductor with
Second, for the left and right conductors: inductance L is given by:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi E ¼ 1=2LI 2 ðA:12Þ
z2 þ ðw=2Þ2
sin / ¼ ðA:7Þ while the energy stored per unit volume of an inductor is:
r2
B2
and e¼ ðA:13Þ
2l0
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2
2
r ¼
2
ðy  1=2hÞ þ z2 þ ðw=2Þ
2
ðA:8Þ We now get an approximation of the total energy of the
coil at current I by multiplying Eq. (A.13) by the volume
 J.K. Rainey et al. / Journal of Magnetic Resonance 187 (2007) 27–37 37

of the rectangular coil, which can be equated to Eq. (A.12) following expression for the total effective capacitance
giving: of the circuit:
B2 ðhw½N  1sÞ CACB þ CBCC þ CACC
1=2LI 2  ðA:14Þ C eff  ð13Þ
2l0 CA þ CC

Solving for L and substituting Eq. (A.11) for B provides an

estimation of coil inductance purely in terms of the number Appendix D. Supplementary data
of turns of the coil and its dimensions:
Supplementary data associated with this article can be
4l found, in the online version, at doi:10.1016/j.jmr.
L ffi 20 ðhw½N  1sÞ
p2 0 2007.03.016.

6X B
ub
w References
4 @qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2
n¼lb ð4ðnsÞ þ h2 Þð4ðnsÞ þ w2 þ h2 Þ
[1] B. Bechinger, S.J. Opella, Flat-coil probe for NMR spectroscopy of
132
oriented membrane samples, J. Magn. Reson. 95 (1991) 585–588.
h C7 [2] R. Freeman, A Handbook of Nuclear Magnetic Resonance, Addison
þ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiA5 ð2Þ Wesley Longman, Harrow, Essex, England, 1997.
2 2 2
ð4ðnsÞ þ w2 Þð4ðnsÞ þ w2 þ h Þ [3] M.T. Thompson, Inductance calculation techniques—part I: classical
methods, Power Control and Intelligent Motion 25 (1999) 40–45.
A similar analysis may readily be carried out for a solenoi- [4] F. Grover, Inductance Calculations, D. Van Nostrand Company,
New York, 1946.
dal coil with circular turns. The total magnetic field at the
[5] J.C. Ginefri, E. Durand, L. Darrasse, Quick measurement of nuclear
centre of the coil is given by magnetic resonance coil sensitivity with a single-loop probe, Rev. Sci.
! Instrum. 70 (1999) 4730–4731.
l0 Ir2 X
ub
1 [6] A. Leroy-Willig, L. Darasse, J. Taquin, M. Sauzade, The Slotted
B¼ ðA:15Þ
2 n¼lb ððnsÞ2 þ r2 Þ3=2 cylinder—an efficient probe for NMR imaging, Magn. Reson. Med. 2
(1984) 20–28.
where r is the radius of the circular coil, and n, and s retain [7] M.H. Levitt, Spin Dynamics—Basics of Nuclear Magnetic Reso-
nance, Wiley, New York, 2001.
the same meaning as for the rectangular coil and the lb and
[8] H.D. Young, University Physics, eighth ed., Addison-Wesley, Read-
ub are given by Eqs. (3) and (4) for odd and even N, respec- ing, MA, 1992.
tively. Substitution into Eq. (A.15) provides the following [9] H.A. Wheeler, Formulas for the skin effect, Proc. I.R.E. 30 (1942)
estimate of inductance 412–424.
" ! #2 [10] F.D. Doty, Probe design and construction, in: D.M. Grant, R.K.
l0 pr6 ½N  1s X ub
1 Harris (Eds.), Encyclopedia of NMR, John Wiley, New York, 1996,
Lffi 2
ð7Þ pp. 3753–3762.
4 2 3=2
n¼lb ððnsÞ þ r Þ [11] D.A. Seeber, R.L. Cooper, L. Ciobanu, C.H. Pennington, Design and
testing of high sensitivity microreceiver coil apparatus for nuclear
magnetic resonance and imaging, Rev. Sci. Instrum. 72 (2001) 2171–
Appendix C. Circuit capacitance calculation 2179.
[12] T.L. Peck, R.L. Magin, P.C. Lauterbur, Design and analysis of
microcoils for NMR microscopy, J. Magn. Reson. B 108 (1995)
If we make the simplifying assumption that the induc- 114–124.
tance measurement tank circuit (Fig. 2) has no resistance, [13] Y. Li, A.G. Webb, S. Saha, W.W. Brey, C. Zachariah, A.S. Edison,
the total impedance (ZT) at a given frequency x of the cir- Comparison of the performance of round and rectangular wire in
cuit contains only imaginary components: small solenoids for high-field NMR, Magn. Reson. Chem. 44 (2006)
 1 255–262.
1 1 [14] P.L. Gor’kov, E.Y. Chekmenev, R. Fu, J. Hu, T.A. Cross, M. Cotten,
Z T ¼ Z CA þ þ þ Z CC W.W. Brey, A large volume flat coil probe for oriented membrane
Z CB Z L
proteins, J. Magn. Reson. 181 (2006) 9–20.
 1
1 1 1 [15] D.L. Olson, J.A. Norcross, M. O’Neil-Johnson, P.F. Molitor, D.J.
¼ þ jxC B þ þ ðA:16Þ Detlefsen, A.G. Wilson, T.L. Peck, Microflow NMR: concepts and
jxC A jxL jxC C capabilities, Anal. Chem. 76 (2004) 2966–2974.
[16] L.D. Wolfgang, C.L. Hutchinson (Eds.), The ARRL Handbook for
When the circuit is at resonance, Eq. (1) applies (note
Radio Amateurs, 16th ed., The American Radio Relay Leauge,
that x is in units of rad/s, as opposed to in Hz) and Newington, CT, 1991.
the total impedance will be zero. Rearrangement of [17] Y. Niwa, Unknown, Researches of the Electrotechnical Laboratory
Eq. (A.16) under the resonance condition provides the 73 (1918).