# Errata in IEC 61260 and EN 61260

> One page of the phonometry documentation, as Markdown. Index: https://docs.jmrp.io/phonometry/llms.txt

Canonical: https://docs.jmrp.io/phonometry/reference/errata/iec-61260/
Language: en
Alternate: https://docs.jmrp.io/phonometry/es/reference/errata/iec-61260/index.md
Version: phonometry 4.0.0
Install: pip install "phonometry[full]"
Author: José Manuel Requena Plens (https://orcid.org/0000-0003-1250-6212)
Published: 2026-10-28
Updated: 2026-11-16T21:36:14Z
License: MIT (https://github.com/jmrplens/phonometry/blob/main/LICENSE)
Summary: Errata in IEC 61260 and EN 61260: 6 defects in the printed text, with the evidence and what phonometry does about each.

---

These are the 6 entries of the [errata registry](https://docs.jmrp.io/phonometry/reference/errata/) on IEC 61260-2:2016, IEC 61260-3:2016, IEC 61260-1:2014 and EN 61260:1995, in the order the registry lists them. The registry's introduction says how each claim is checked against the printed page and what the status of an entry means.

## IEC 61260-2:2016 / IEC 61260-3:2016, Annex A, Formula (A.2) (the coefficient of the frequency terms left unsquared)

- **Location:** Annex A (informative), A.1.3, Formula (A.2), printed p. 19 of
  IEC 61260-2:2016 and printed p. 16 of IEC 61260-3:2016, and the worked
  example of A.3.5 on printed p. 21 and printed p. 18 respectively. Both parts
  print the same annex.
- **The print:** Formula (A.1) is the sum of the squared sensitivity
  coefficients of Formula (17) of IEC 61260-1:2014 times the squared standard
  uncertainties, and Formula (A.2) is offered as its simplification,

  $$
  u_{L_\mathrm{c}} = \left[u_{L_\mathrm{in}}^2 + \left(\frac{10}{\ln(10)}\right)^2\left(\frac{u_{T_\mathrm{sweep}}}{T_\mathrm{sweep}}\right)^2 + \left(\frac{10}{\ln(10)}\right)^2\left(\frac{u_{T_\mathrm{avg}}}{T_\mathrm{avg}}\right)^2 + \left(\frac{10}{\ln(f_\mathrm{end}/f_\mathrm{start})\times\ln(10)}\right)\times\left[\left(\frac{u_{f_\mathrm{end}}}{f_\mathrm{end}}\right)^2 + \left(\frac{u_{f_\mathrm{start}}}{f_\mathrm{start}}\right)^2\right]\right]^{1/2}\ \mathrm{dB},
  $$

  in which the first two coefficients carry the exponent 2 and the third does
  not. A.3.5 then works an example with $u_{L_\mathrm{in}} \approx 0{,}042$ dB,
  $T_\mathrm{sweep} = T_\mathrm{avg} = 20$ s with 0,05 s and 0,02 s,
  $f_\mathrm{end} = 50\,000$ Hz with 5 Hz and $f_\mathrm{start} = 0{,}5$ Hz
  with 0,05 Hz, and prints $u_{L_\mathrm{c}} \approx 0{,}057$ dB, an expanded
  uncertainty of 0,115 dB, and 0,128 dB with a display of 0,1 dB resolution.
- **The problem:** the derivative of Formula (17) with respect to
  $f_\mathrm{end}$ is $10/[\ln(f_\mathrm{end}/f_\mathrm{start})\ln(10)\,f_\mathrm{end}]$,
  and (A.1) squares it with every other coefficient. Left unsquared, the
  coefficient is 0.377 for the example's sweep of five decades, and its
  frequency term is $0.0038\ \text{dB}^2$ instead of $0.0014\ \text{dB}^2$. As printed, (A.2) gives
  $u_{L_\mathrm{c}} = 0.075$ dB, an expanded uncertainty of 0.150 dB and
  0.161 dB with the display; with the square it gives 0.057 dB, 0.115 dB and
  0.128 dB, the three values both parts print. Both are computed with the
  $u_{L_\mathrm{in}}$ the example derives, 0.0416 dB from a 0.1 dB display
  resolution and a 0.03 dB constancy, not with the 0.042 dB it rounds it to,
  which would give 0.058 dB. The unsquared form is also not in $\text{dB}^2$, so it
  changes with the unit of level.
- **Evidence:** a recomputation of the printed example both ways. Verified on
  PDF page 21 (printed p. 19) and PDF page 23 (printed p. 21) of
  IEC 61260-2:2016, and on PDF page 18 (printed p. 16) and PDF page 20
  (printed p. 18) of IEC 61260-3:2016. The consolidated
  IEC 61260-2:2016+AMD1:2017 (Edition 1.1) prints the same Formula (A.2) on
  PDF page 23 (printed p. 19) and the same 0.057 dB, 0.115 dB and 0.128 dB on
  PDF page 25 (printed p. 21): the amendment adds only the TEM-waveguide
  option of IEC 61000-4-20 to the field-immunity test, and the defect stands
  in the current edition.
- **Library behaviour:**
  [`swept_level_uncertainty`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/filters/time_invariance.py)
  squares the coefficient, as (A.1) does, and reproduces 0.057 dB, 0.115 dB
  and 0.128 dB; three rows of the conformance report pin them, and
  `test_formula_a2_as_printed_would_not_reproduce_its_example` in
  [`tests/filters/test_time_invariance.py`](https://github.com/jmrplens/phonometry/blob/main/tests/filters/test_time_invariance.py)
  holds the printed form to the 0.075 dB it gives.
- **Status:** unreported.

## IEC 61260-3:2016, B.2.6 (an averaging period that ends 6 s to 7 s after the sweep passes 6.3 Hz)

- **Location:** Annex B (informative), B.2.6, printed p. 20.
- **The print:** "The averaging period will end 6 s to 7 s after the sweep
  frequency is equal to the lowest midband frequency, 6,3 Hz."
- **The problem:** the example's own settings end it 18 s to 19 s later. B.2.2
  sweeps from 0.01 Hz to 1 MHz, eight decades, in 30 s, which is 3.75 s per
  decade, and B.2.3 averages for 30 s from a start 0.5 s to 1.5 s before the
  sweep. The sweep reaches 6.3 Hz
  $3.75 \lg(6.3/0.01) = 10.5$ s after it starts, which is 11.0 s to
  12.0 s into the averaging, and the averaging ends at 30 s. B.2.3 checks the
  same settings at the other end and gets them right: the sweep is at 398 kHz
  to 736 kHz when the averaging ends. IEC 61260-2:2016 prints the same example
  and, in its own B.2.6, "18 s to 20 s".
- **Evidence:** a recomputation from B.2.2 and B.2.3. Verified on PDF page 21
  (printed p. 19) and PDF page 22 (printed p. 20) of IEC 61260-3:2016, and on
  PDF page 25 (printed p. 23) of IEC 61260-2:2016.
- **Library behaviour:** none needed. The sentence argues that the impulse
  response of the lowest filter has died away when the averaging ends, which
  18 s makes stronger, and no number of the example depends on it.
  [`verify_time_invariance`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/filters/time_invariance.py)
  sizes the averaging from the slowest pole of the bank it is given.
- **Status:** unreported.

## IEC 61260-2:2016, 7.2.4.3 and Formula (3), with IEC 61260-1:2014, 5.16 (the summation difference taken both ways round)

- **Location:** IEC 61260-2:2016, 7.2.4.3 and Formula (3) on printed p. 11 and
  7.2.4.5 on printed p. 12; IEC 61260-1:2014, 5.16, printed p. 21.
- **The print:** 7.2.4.3 defines $\Delta P_j(\Omega_i)$ as "the difference
  between the input signal level minus the reference attenuation and the
  level of the summed output signals" and gives

  $$
  \Delta P_j(\Omega_i) = 10\lg\left[10^{-0{,}1\,\Delta A_{j-1}} + 10^{-0{,}1\,\Delta A_j} + 10^{-0{,}1\,\Delta A_{j+1}}\right]\ \mathrm{dB}.
  $$

  5.16 of IEC 61260-1 sets the limits "for the difference between (a) the
  level of the input signal minus the reference attenuation and (b) the level
  of the sum of the time-mean-square output signals from adjacent filters":
  +0,8 dB and −1,8 dB for class 1, +1,8 dB and −3,8 dB for class 2. 7.2.4.5
  applies those limits to $\Delta P_j(\Omega_i)$ "calculated according to
  formula (3)".
- **The problem:** with $\Delta A = A - A_\mathrm{ref}$ and
  $A = L_\mathrm{in} - L_\mathrm{out}$, Formula (3) is the level of the summed
  outputs minus the input level less the reference attenuation, (b) minus (a).
  The words of both parts name (a) minus (b), the same number with the other
  sign. The limits are not symmetric, so the two are different tests: a set
  whose adjacent outputs sum 1.0 dB above the input fails class 1 by
  Formula (3) and passes by the words, and one whose outputs sum 1.0 dB below
  it passes by Formula (3) and fails by the words.
- **Evidence:** a comparison of the sentence with the formula it introduces.
  Verified on PDF page 13 (printed p. 11) and PDF page 14 (printed p. 12) of
  IEC 61260-2:2016, and on PDF page 23 (printed p. 21) of BS EN 61260-1:2014.
  The consolidated IEC 61260-2:2016+AMD1:2017 (Edition 1.1) prints the same
  words and Formula (3), 7.2.4.3 running from PDF page 15 (printed p. 11) to
  PDF page 16 (printed p. 12) with Formula (3) and 7.2.4.5 on the latter; the
  amendment does not touch 7.2.4, and the conflict stands in the current
  edition.
- **Library behaviour:**
  [`verify_filter_class`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/filters/compliance.py) applies the
  limits of 5.16 to Formula (3) as printed, as 7.2.4.5 instructs. The
  verdicts it gives the default Butterworth banks do not depend on the
  reading: at 48 kHz the octave bank sums from −0.06 dB to +0.69 dB and the
  one-third-octave bank from −0.25 dB to +0.76 dB, both class 1 either way.
  Those it gives a Chebyshev I or elliptic bank do: their neighbouring
  outputs overlap and sum up to +3.01 dB above the input, which is no class
  by Formula (3) and class 2 by the words. Their relative attenuation reads
  no class already, so the overall class of such a bank is none either way.
  `test_the_summation_is_what_tones_through_the_bank_read` in
  [`tests/filters/test_pattern_evaluation.py`](https://github.com/jmrplens/phonometry/blob/main/tests/filters/test_pattern_evaluation.py)
  holds Formula (3) to what the running bank reads.
- **Status:** unreported.

## EN 61260:1995, 3.2, NOTE 3 (the base-two octave ratio labelled base-ten)

- **Location:** clause 3.2, "octave ratio", NOTE 3 and equation (2), printed
  p. 4; clause 3.7, printed p. 5.
- **The print:** NOTE 1 of 3.2 reads "This standard permits two options,
  designated base-ten and base-two, for determining an octave-band, or
  fractional-octave-band, frequency ratio." NOTE 2 reads "For base-ten
  systems, $G_{10} = 10^{3/10}$" as equation (1), and NOTE 3 reads "For
  base-ten systems, $G_2 = 2$" as equation (2).
- **The problem:** NOTE 3 gives the base-two ratio, $G_2 = 2$, under the name
  of the other system, so the clause names the base-ten system twice and the
  base-two system never. Clause 3.7 reads equation (2) the intended way: "G
  represents an octave frequency ratio calculated according to equation (1)
  for base-ten systems or (2) for base-two systems". NOTE 3 should read "For
  base-two systems".
- **Evidence:** read on PDF page 8 (printed p. 4) and PDF page 9 (printed
  p. 5) of BS EN 61260:1996, which is EN 61260:1995 (IEC 61260:1995)
  incorporating amendment A1, the copy dated "© BSI 23 August 2002".
- **Library behaviour:** no change required. The filter banks and
  [`verify_filter_class`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/filters/compliance.py) with
  `edition="1995"` use the base-ten ratio $G = 10^{3/10}$ of equation (1), the
  one NOTE 4 prefers, and nothing in the library reads equation (2) by its
  label.
- **Status:** unreported.

## EN 61260:1995, 3.15 (the normalizing constant printed with a stray factor of ten)

- **Location:** clause 3.15, "normalized effective bandwidth", printed p. 6;
  equation (14) of 4.5.2, printed p. 11.
- **The print:** 3.15 defines the normalized effective bandwidth as an
  integral of the ratio of the output to the input time-mean-square signal
  and reads "the ratio of time-mean-square signal is normalized by
  multiplying by a constant equal to 10 $10^{0,1A_\mathrm{ref}}$ where
  $A_\mathrm{ref}$ is the reference attenuation, in decibels; symbol
  $B_\mathrm{e}$".
- **The problem:** a filter of attenuation $A$ (3.12) passes the ratio
  $10^{-0.1A}$ of the time-mean-square signals; multiplied by
  $10^{0.1A_\mathrm{ref}}$ it becomes $10^{-0.1\,\Delta A}$, the integrand of
  equation (14),
  $B_\mathrm{e} = \int_0^\infty 10^{-0.1\,\Delta A(f/f_\mathrm{m})}\,\mathrm{d}(f/f_\mathrm{m})$.
  The constant as printed carries a second "10" in front of the power, ten
  times $10^{0.1A_\mathrm{ref}}$, which would make the effective bandwidth of
  the definition ten times that of equation (14) and the filter integrated
  response of 4.5.1 10 dB above it, outside every class of 4.5.3. The
  constant should read $10^{0.1A_\mathrm{ref}}$.
- **Evidence:** read on PDF page 10 (printed p. 6) of BS EN 61260:1996, which
  is EN 61260:1995 (IEC 61260:1995) incorporating amendment A1, the copy dated
  "© BSI 23 August 2002", against equation (14) on PDF page 15 (printed
  p. 11) of the same copy, which integrates $10^{-0.1\,\Delta A}$ with no such
  factor. ANSI S1.11-2004, which carries the same definition as its 3.15,
  prints the constant as $10^{0.1A_\mathrm{ref}}$ with no second ten, read on
  PDF page 13 (printed p. 3) of the copy incorporated by reference in
  49 CFR 227.
- **Library behaviour:** no change required.
  [`verify_filter_class`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/filters/compliance.py) with
  `edition="1995"` integrates equation (14) by the trapezoidal rule of
  equation (16), on the relative attenuation itself;
  `test_equation_16_integrates_a_flat_response_to_its_span` and
  `test_equation_16_on_an_ideal_band` in
  [`tests/filters/test_filter_class_1995.py`](https://github.com/jmrplens/phonometry/blob/main/tests/filters/test_filter_class_1995.py)
  pin it.
- **Status:** unreported.

## EN 61260:1995, 4.9 and 5.8.3, equation (19) (the summation difference taken both ways round)

- **Location:** 4.9, printed p. 12; 5.8.3 and equation (19), printed p. 16;
  5.8.5, printed p. 17.
- **The print:** 4.9 reads "the difference between (a) the level of the input
  signal minus the reference attenuation and (b) the level of the sum of the
  time-mean-square output signals from various filters of specified filter
  bandwidth shall not exceed ±1,0 dB; +1,0 dB, –2,0 dB and +2,0 dB, –4,0 dB
  for classes 0, 1, and 2 instruments, respectively". 5.8.3 defines "the
  difference $\Delta P(f_i)$ between the level of the input signal minus the
  reference attenuation and the level of the summed output signals" by

  $$
  \Delta P(f_i) = 10\lg\left[10^{-0{,}1\,\Delta A_{j-1}} + 10^{-0{,}1\,\Delta A_j} + 10^{-0{,}1\,\Delta A_{j+1}}\right]\ \mathrm{dB},
  $$

  equation (19), and 5.8.5 holds "the difference $\Delta P(f_i)$ calculated
  according to equation (19)" within the tolerances of 4.9.
- **The problem:** with $\Delta A = A - A_\mathrm{ref}$ (equation (8)) and the
  filter attenuation $A$ the input level minus the output level (3.12),
  equation (19) is the level of the summed outputs minus the input level less
  the reference attenuation, (b) minus (a). The words of 4.9 and 5.8.3 name
  (a) minus (b), the same number with the other sign. The class 0 tolerance
  is symmetric and reads the same either way; those of classes 1 and 2 are
  not, so there the two are different tests: a set whose outputs sum 1.5 dB
  above the input fails class 1 by equation (19) and passes by the words,
  and one whose outputs sum 1.5 dB below it passes by equation (19) and fails
  by the words. IEC 61260-1:2014 and IEC 61260-2:2016 carry the same
  conflict over (see the entry on IEC 61260-2:2016, 7.2.4.3 and Formula (3)),
  and so does ANSI S1.11-2004: its 4.9 words the difference the same way,
  and its informative Annex D defines it in the same words in D.8.3,
  computes it by equation (D4), the sum of equation (19), and holds that to
  the tolerances of 4.9 in D.8.5.
- **Evidence:** a comparison of the sentences with the equation they
  introduce. Verified on PDF page 16 (printed p. 12), PDF page 20 (printed
  p. 16) and PDF page 21 (printed p. 17) of BS EN 61260:1996, which is
  EN 61260:1995 (IEC 61260:1995) incorporating amendment A1, the copy dated
  "© BSI 23 August 2002", with 3.12 on PDF page 9 (printed p. 5) and
  equation (8) on PDF page 10 (printed p. 6). ANSI S1.11-2004 read on PDF
  page 16 (printed p. 6) for 4.9 and PDF page 30 (printed p. 20) for D.8.3 to
  D.8.5, in the copy incorporated by reference in 49 CFR 227.
- **Library behaviour:**
  [`verify_filter_class`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/filters/compliance.py) with
  `edition="1995"` applies the tolerances of 4.9 to equation (19) as
  printed, as 5.8.5 instructs. The verdicts it gives the default Butterworth
  banks do not depend on the reading: at 48 kHz the octave bank sums from
  −0.62 dB to +0.69 dB and the one-third-octave bank from −0.66 dB to
  +0.76 dB, both inside class 0. Those it gives a Chebyshev I or elliptic
  bank do: their neighbouring outputs overlap and sum up to +3.01 dB above
  the input, which is no class by equation (19) and class 2 by the words.
  Their relative attenuation reads no class already, so the overall class of
  such a bank is none either way.
  `test_the_4_9_limits_bound_equation_19_as_printed`
  in
  [`tests/filters/test_filter_class_1995.py`](https://github.com/jmrplens/phonometry/blob/main/tests/filters/test_filter_class_1995.py)
  pins the reading.
- **Status:** unreported.
