One Capacitor, Two Readings: What Series and Parallel Mean on an LCR Meter
Add a 1 kΩ resistor in parallel with a 100 nF film capacitor. An LCR meter can now report roughly 353 nF as its series capacitance and 100 nF as its parallel capacitance at 1 kHz. Both readings describe the same impedance. Repeat at 10 kHz and use the impedance to check which property of your circuit actually changed.
Connect the capacitor and the removable load
An inductance, capacitance and resistance meter (LCR meter) applies a small alternating signal and reads the voltage and current together. The Hioki IM3536 reports the impedance magnitude and phase and computes capacitance using equivalent circuit models. Here the capacitor is a nonpolar Vishay Roederstein MKT1820410014, metallized polyester film, 100 nF ±5%, 100 V direct current (DC), with 10 mm lead pitch. The manufacturer’s ordering table gives the 100 nF, 100 V bulk part with tolerance digit 5, then gives digit 4 for the ±5% version. Use a Vishay MRS25000C1001FCT00 1.00 kΩ ±1% metal-film resistor for the shunt, a second path between the same nodes. Measure the loose resistor in ohms with a digital multimeter before adding it to the fixture.
Use the Hioki 9140-10 1 m four-terminal probe. Its high-current and high-sense contacts, HCUR and HPOT, meet at node HI. Its low-sense and low-current contacts, LPOT and LCUR, meet at node LO. Set the bare capacitor on a small insulating fixture between two short rigid terminals, each at least 0.3 mm in diameter for reliable probe contact. Clip onto the terminals with both contacts on each side seated. Connect one resistor lead to LO. Its other lead goes to HI only when the link is in place. Keep the probe positions and lead dress fixed between cases. The LCR source is the only supply to this disconnected fixture.
Set up the instrument before changing the link
- Use the normal LCR measurement mode and automatic impedance ranging. Select no DC bias and leave low-impedance high-accuracy mode off. With the fixture disconnected, select constant-voltage (CV) mode and set 0.100 V root mean square (RMS), the effective alternating voltage. This version of the meter can display the actual terminal voltage,
Vac. In normal mode its nominal internal source resistance is 100 Ω, so the open-circuit voltage setting would generally give a different terminal voltage. At nominal values the shunted impedance magnitude is about 847 Ω at 1 kHz and 157 Ω at 10 kHz, so the current in the shunted fixture is around 0.118 mA or 0.636 mA at 0.100 V. The resistor dissipates about 10 µW, far below its 0.6 W rated dissipation at 70 °C under the data sheet conditions. - Set the cable length to 1 m before correction. Follow the 9140-10 open and short procedures at the contact geometry of the fixture, For open correction close the tips of each probe and keep the two clips separated by the width of the test fixture. For short correction clamp the two clips to the supplied short bar. Enable both corrections at 1 kHz and 10 kHz, using all-frequency correction or repeating spot correction for each frequency before its two cases. Remove the correction short before connecting the fixture. If you move the probe contacts or change the cable, redo correction.
- Select four display parameters: impedance magnitude |Z|, impedance phase angle θ, series capacitance Cs and parallel capacitance Cp. Read
Vacfrom the separate monitor. Keep any error or out-of-range indication with the original observation and resolve it before using the number. Check that the meter reached CV and thatVacstays within 90–110 mV for this exercise. - At 1 kHz, record three settled readings with the link open and three with it closed. Repeat both at 10 kHz. Pause triggering before changing the link at the separate shunt terminals. Continue after the connection is stable, keeping the selected port, capacitor and fixture the same. Record the displayed units for each quantity and the resistor’s measured resistance.
Convert both displays back to the same impedance
Impedance Z describes how a part opposes an alternating current, including the timing difference between voltage and current. In polar form the meter gives the magnitude in ohms and phase θ in degrees. Write Z = Rs + jX, where j represents a 90° rotation, Rs is the series resistance and X is the reactance, or the imaginary component of impedance. For a capacitive reading X is negative. At angular frequency ω = 2πf, Cs = −1/(ωX).
Admittance Y = 1/Z describes how easily the signal current flows. Write it as G + jB. The real component G is conductance, while the imaginary component B is susceptance. Each is expressed in siemens. Because the two physical branches have the same voltage, their admittances add. The parallel model has Rp = 1/G and Cp = B/ω. A resistor across the capacitor ideally adds 1/Rshunt to G and adds zero to B.
The dissipation factor D is the ratio G/B for this capacitive network. Inverting Y gives X = −B/(G2 + B2). Substitution into the series formula gives Cs = Cp(1 + D2). Both equivalent circuits describe one frequency point. The series numbers computed at 1 kHz should not be carried to 10 kHz as fixed components.
| Freq. (kHz) | D | Cs (nF) | Cp (nF) | Rs (Ω) |
|---|---|---|---|---|
| 1 | 1.592 | 353.3 | 100.0 | 716.96 |
| 10 | 0.159 | 102.5 | 100.0 | 24.70 |
At 1 kHz, B ≈ 0.628 mS while the resistor adds 1.000 mS to G. At 10 kHz the same G is small compared with B ≈ 6.283 mS, so the readings converge. With no shunt, film loss is small and Cs and Cp should already be close. The manufacturer lists low loss at both frequencies for this film family. No nominal table value in this article is a measured result.
Check the actual readings
Fill the blank observation sheet with three rows in each condition. Use hertz, ohms, degrees, nanofarads and volts, stripping the unit suffixes. Use bare or shunt as prefilled, with repeat numbers 1–3. Run python3 lcr_models.py readings.csv --shunt-ohm 1000, replacing 1000 with your measured resistor value. If you omit the option, the check uses the nominal 1000 Ω.
The script reconstructs both capacitances from |Z| and θ and compares them with the displayed values. These are consistency checks on outputs of the same meter. They do not establish its absolute accuracy. It also checks that the shunt adds roughly 1/R to the measured conductance without materially changing susceptance. Only then does it report the ratio between the loaded and bare series readings and the equivalent resistances. The reading spread, percentage margins and drive window are checks for this fixture.
Exit 0 means the observations support the comparison. Exit 1 marks an unresolved setup or reading problem. Exit 2 marks a file or CSV parser error. If the series and parallel displays disagree with the converted impedance, first check the units, the selected parameters and the sign of the angle. If adding the resistor changes B or barely changes G, check whether a test lead moved, the resistor was placed on the wrong nodes or correction is no longer valid. Keep the recorded observation with its error rather than substituting the ideal values.
Predict what a 2 kΩ shunt would show
At 1 kHz replace the 1.00 kΩ shunt in the ideal model with 2.00 kΩ. What should the parallel and series capacitance labels read?
Answer
Cp remains 100 nF. Doubling the shunt resistance halves its conductance, so D becomes about 0.796. The resulting series display is 100 nF × (1 + 0.796²), or about 163 nF. The provided checker is written for the 1.00 kΩ experiment; this question uses the equations directly.
Full Python analyzer
Download the standalone code or use the experiment archive with the diagrams, blank worksheet, synthetic example and tests. The tests check the software with calculated data. Fill the observation sheet with your own instrument readings.
Complete source for lcr_models.py
#!/usr/bin/env python3
"""Validate readings from the 100 nF film capacitor plus 1 kΩ shunt experiment."""
import argparse
import cmath
import csv
import math
from pathlib import Path
from statistics import median
import sys
HEADER = ("frequency_hz", "state", "repeat", "z_ohm", "theta_deg", "cs_nf", "cp_nf", "vac_v")
FREQUENCIES = (1000, 10_000)
CASES = tuple((f, state) for f in FREQUENCIES for state in ("bare", "shunt"))
class FixtureError(ValueError):
"""The supplied observations do not resolve the experiment."""
def require(ok, location, problem):
if not ok:
raise FixtureError(f"{location}: {problem}")
def read_records(path):
rows = {case: {} for case in CASES}
with Path(path).open(newline="", encoding="utf-8-sig") as stream:
reader = csv.reader(stream, strict=True)
require(tuple(next(reader, [])) == HEADER, "CSV", "unexpected header")
for line, row in enumerate(reader, 2):
label = f"line {line}"
require(line <= 13 and len(row) == len(HEADER), label, "unexpected CSV row")
frequency, state, index = row[:3]
require(frequency in ("1000", "10000") and state in ("bare", "shunt")
and index in ("1", "2", "3"), label, "unknown frequency, state or repeat")
group = rows[int(frequency), state]
require(index not in group, label, "repeat appears twice")
try:
group[index] = tuple(float(value) for value in row[3:])
except ValueError as exc:
raise FixtureError(f"{label}: supply numeric readings") from exc
records = {key: list(group.values()) for key, group in rows.items()}
return records
def parallel_impedance(frequency, capacitance_nf, parallel_ohm):
return 1 / complex(1 / parallel_ohm, math.tau * frequency * capacitance_nf * 1e-9)
def summarize(records, frequency, state):
label = f"{frequency} Hz / {state}"
require(len(records) == 3, label, "need three complete repeats")
omega = math.tau * frequency
results = []
for fields in records:
mag, theta, series_nf, parallel_nf, v = fields
require(all(math.isfinite(x) for x in fields), label, "non-finite reading")
require(.090 <= v <= .110, label, "terminal voltage outside the 90–110 mV screen")
require(-90.05 <= theta <= -10, label, "phase outside expected capacitive quadrant")
require(50 <= mag <= 2500 and 75 <= series_nf <= 1000 and 75 <= parallel_nf <= 145,
label, "readings outside expected fixture envelope or printed units")
phase = math.radians(theta)
z = cmath.rect(mag, phase)
y = 1 / z
reconstructed_cs = -1e9 / (omega * z.imag)
reconstructed_cp = 1e9 * y.imag / omega
require(abs(series_nf - reconstructed_cs) <= .012 * reconstructed_cs
and abs(parallel_nf - reconstructed_cp) <= .012 * reconstructed_cp,
label, "displayed Cs or Cp disagrees with the polar impedance models")
results.append((mag, theta, series_nf, parallel_nf, y.real, y.imag))
mags, angles, css, cps, gs, bs = zip(*results)
require(max(mags) - min(mags) <= .010 * median(mags)
and max(angles) - min(angles) <= .35
and max(css) - min(css) <= .012 * median(css)
and max(cps) - min(cps) <= .012 * median(cps),
label, "repeated measurements are not stable")
return median(css), median(cps), median(gs), median(bs)
def analyze(records, shunt_ohm):
require(math.isfinite(shunt_ohm) and 970 <= shunt_ohm <= 1030,
"fixture", "shunt must be near 1 kΩ")
results = {}
for f in FREQUENCIES:
cs0, cp0, g0, b0 = summarize(records[f, "bare"], f, "bare")
cs1, cp1, g1, b1 = summarize(records[f, "shunt"], f, "shunt")
require(92 <= cp0 <= 108 and abs(g0 / b0) <= .040,
f, "bare film capacitor reference is outside this fixture screen")
delta_g = g1 - g0
require(abs(delta_g * shunt_ohm - 1) <= .07,
f, "conductance change does not support the stated shunt")
require(abs(b1 - b0) <= .030 * b0 and abs(cp1 - cp0) <= 3,
f, "susceptance or parallel capacitance changed with the shunt")
ratio = cs1 / cs0
d = g1 / b1
z_parallel = parallel_impedance(f, cp1, 1 / g1)
results[f] = {"bare_cp_nf": cp0, "shunt_cs_nf": cs1, "shunt_cp_nf": cp1,
"cs_ratio": ratio, "d": d, "delta_cp_nf": cp1 - cp0,
"estimated_shunt_ohm": 1 / delta_g,
"derived_series_ohm": z_parallel.real,
"derived_parallel_ohm": 1 / g1}
require(results[1000]["cs_ratio"] >= 3.0 and 0.98 <= results[10_000]["cs_ratio"] <= 1.08,
"comparison", "frequency dependence of series display was not resolved")
return results
def main(argv=None):
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument("readings_csv")
parser.add_argument("--shunt-ohm", type=float, default=1000,
help="measured shunt resistance, otherwise nominal 1000")
args = parser.parse_args(argv)
try:
results = analyze(read_records(args.readings_csv), args.shunt_ohm)
except (OSError, UnicodeError, csv.Error) as exc:
print(f"ERROR: {exc}", file=sys.stderr)
return 2
except FixtureError as exc:
print(f"UNRESOLVED: {exc}", file=sys.stderr)
return 1
print("f_Hz bare_Cp shunt_Cs shunt_Cp Cs/Cs0 shunt_D shunt_Rs_est shunt_Rp_est added_R_est")
for f, item in results.items():
print(f"{f:5d} {item['bare_cp_nf']:9.3f} {item['shunt_cs_nf']:9.3f} "
f"{item['shunt_cp_nf']:9.3f} {item['cs_ratio']:8.4f} {item['d']:8.4f} "
f"{item['derived_series_ohm']:12.3f} {item['derived_parallel_ohm']:14.3f} "
f"{item['estimated_shunt_ohm']:12.3f}")
print("Capacitance in nF; resistance in ohms. This is a fixture comparison.")
return 0
if __name__ == "__main__":
sys.exit(main())
Related measurements and primary sources
- Series capacitor voltage sharing observes two physical capacitors after a step. The ceramic capacitor bias experiment measures a change in effective capacitance under DC bias.
- Hioki IM3536 user manual (manufacturer PDF) for measurement parameters (pp. 41–44), CV mode (pp. 53–56) and open/short correction (pp. 104–116). The manual’s appendix on series and parallel equivalents explains why the two display models differ. Hioki support and current English manual lists the latest revision.
- Vishay MKT1820 data sheet for construction, loss and the 100 V ordering table. Vishay MRS25 data sheet for resistance code, tolerance and dissipation.