Meet Vibromera.com — our new international website. Visit Vibromera.com →

Engineering worksheet – source controlled

Vibration Acquisition Worksheet

Record already selected acquisition settings and calculate their transparent timing, DFT spacing, overlap and raw-payload arithmetic. The worksheet does not choose machine-specific settings or create an ISO acceptance plan.

Explicit settings onlyNo automatic FmaxNo ISO alarm limits
This is not a measurement-plan generator. Select the sensor, mounting, point, direction, quantity, descriptor, bandwidth, filters, window, averaging method, route interval and acceptance criteria from the applicable machine standard, manufacturer instructions and a competent procedure. This page performs arithmetic only after those choices are documented.

Controlled acquisition record

All fields are required. Decimal point and decimal comma are accepted for frequency fields. Integer fields accept ungrouped digits only. No result is calculated until explicit submission.

Identify the asset, point and direction; the worksheet does not select them.
Use a traceable document or controlled setup identifier.
Used only for the mathematical Nyquist precondition check.
Enter the actual saved-data rate, not an assumed analyzer bandwidth.
Positive integer from 2 to 10,000,000.
Number of equal-length records in the ensemble.
1 to N. Overlap is derived as 100(1 – H/N)%.
Each point/direction/configuration requiring a separate recording.
Simultaneous channels affect payload, not sequential acquisition time.
Use the actual storage representation; 4 bytes is not assumed.
Document what the arithmetic cannot validate.
Required confirmations
No result until the form is submitted.
Sequential signal-time lower bound
P times the ensemble signal duration; excludes setup, settling, movement, instrument processing and operator time.
One record duration
T = N / fs.
DFT bin spacing
Delta f = fs / N = 1 / T; not resolving power.
Nyquistova frekvencia
fs/2 is a mathematical marker, not usable instrument bandwidth.
Record overlap
Derived from the exact integer hop H.
Unique ensemble samples
N + (M – 1)H per simultaneous channel.
Ensemble signal duration
Unique sample count divided by fs.
Raw payload lower bound
Exact raw sample bytes only; excludes headers, metadata and processed results.
Sampling precondition
Checks only whether fs is strictly greater than 2finterest.
Arithmetic is not adequacy. Being below the Nyquist marker does not verify sensor response, mounting, signal conditioning, anti-alias attenuation, analyzer usable bandwidth, window leakage, dynamic range, transient capture, order tracking or diagnostic sensitivity. Obtain those facts from the actual measurement chain and procedure.

Equations and boundaries

Record and DFT arithmetic

T = N / f_s [s]
Delta f = f_s / N = 1 / T [Hz]
f_Nyquist = f_s / 2 [Hz]

N is the number of uniformly spaced samples in one record and fs is the sample rate in samples/s. Delta f is the spacing between unpadded DFT bins. It is not a guarantee that two nearby components can be resolved; window shape, record content, signal-to-noise ratio and analysis method also matter. Zero padding can add displayed frequency points without improving observation-time resolution.

Overlap and ensemble duration

overlap = 100(1 – H/N)%
N_unique = N + (M – 1)H
T_ensemble = N_unique / f_s
T_sequential,min = P T_ensemble

H is the integer hop between record starts, M is the number of averaged records and P is the number of sequential acquisition setups. A simultaneous channel does not add sequential signal time. This uses the common acquisition sample-count convention N/fs; the elapsed interval from the first sample instant to the last sample instant is one sample interval shorter.

Raw sample payload lower bound

bytes_raw,min = P C N_unique B

C is the number of simultaneously saved channels and B is the actual bytes stored for each sample of each channel. The exact integer result excludes file headers, timestamps, metadata, padding, compression behavior, tachometer channels, spectra, cross-channel products and other processed data. It is therefore a lower bound, not a file-size prediction.

Necessary sampling precondition only

f_s > 2 f_interest

The worksheet flags whether the entered settings satisfy this strict mathematical precondition. Passing it is not sufficient evidence that the selected frequency band is valid. A realizable acquisition requires a compatible sensor and signal chain plus anti-alias filtering and the instrument’s specified usable bandwidth.

What was removed from the former generator

Confirmed unsafe shortcuts removed: machine-type multipliers for Fmax, automatic three-direction point counts, a rounded frequency bucket, approximate BPFO = 0.4 x rolling-elements x shaft frequency, fixed four-byte storage, and generic ISO 10816 Group 2 alarm bands. None had enough input data or source-controlled scope to support the generated recommendation.
  • Bearing characteristic frequencies require bearing geometry and contact angle. The official SKF formula for BPFO contains rolling-element count, rolling-element diameter, pitch diameter and contact angle; a universal 0.4 multiplier is not an exact bearing calculation.
  • Four records with 75% overlap use a hop of N/4, so the unique ensemble count is N + 3N/4 = 1.75N, not 2.5N.
  • ISO 10816 has multiple machine-specific parts and has been replaced progressively by the ISO 20816 series. A generic four-zone table cannot be attached to every asset or measurement quantity.
  • Derived parameters may be produced from the same acquired channels. Multiplying acquisition time by a list of displayed parameters overstates signal time unless each parameter truly requires a separate acquisition setup.

Standard and source record

ZdrojStatus checked 16 Jul 2026What it supports hereWhat it does not support
ISO 13373-1:2002, Edition 1Published; confirmed in 2024, stage 90.93.Official scope covers general vibration condition-monitoring measurement and data-collection procedures, including parameters, transducers, locations, attachment, operating conditions and signal conditioning.The public abstract does not prescribe the numeric settings entered here or justify an automatic machine plan.
ISO 13373-2:2016, Edition 2Published; systematic review underway, stage 90.20 at the audit date.Official scope covers processing, analysis and presentation of vibration data and notes that some techniques are application-specific.The public abstract does not establish this worksheet’s arithmetic as an ISO formula or select Fmax, averaging or alarms.
ISO 17359:2018, Edition 3Published; confirmed in 2023, stage 90.93.Official scope provides general procedures for setting up condition-monitoring programmes for machines.It does not make a generic web form a complete programme or acceptance procedure.
ISO 20816-1:2016, Edition 1Published; stage 90.92, revision in progress.General framework for measurement and evaluation of machine vibration.It does not justify universal ISO 10816 alarm bands, point directions or an automatic measurement plan.
NI, Spectrum Analyzer MeasurementsOfficial manufacturer technical guide; accessed 16 Jul 2026.DFT bins and the inverse relationship between record duration and frequency-bin spacing.Bin spacing alone does not prove component separability or acquisition adequacy.
NI, Bandwidth and Nyquist Sampling TheoremOfficial manufacturer technical guide; accessed 16 Jul 2026.Sampling must exceed twice the highest frequency component and anti-alias filtering is required.The Nyquist marker is not the specified usable bandwidth of a real instrument chain.
SKF, Spectrum Analysis – The key features of analyzing spectra, CM5118 ENOfficial manufacturer PDF; BPFO formula page visually checked.Bearing characteristic frequencies require bearing geometry and contact angle; BPFO is not determined by element count and shaft speed alone.It does not validate the former 0.4 approximation or generic ISO severity limits.
NEEDS_LICENSED_SOURCE: the full normative texts of ISO 13373-1:2002, ISO 13373-2:2016, ISO 17359:2018 and ISO 20816-1:2016 are not licensed in this audit record. The page uses their official public scopes only. No clause number, acquisition setting, alarm threshold or normative formula is invented.

Worked arithmetic example

For fs = 5120 samples/s, N = 4096, M = 4, H = 1024, P = 12 sequential setups, C = 2 simultaneous channels and B = 4 bytes/sample: T = 0.8 s, Delta f = 1.25 Hz, fNyquist = 2560 Hz, overlap = 75%, Nunique = 7168 samples, Tensemble = 1.4 s, sequential signal-time lower bound = 16.8 s, and raw payload lower bound = 688,128 bytes (672 KiB). A 1000 Hz frequency of interest is below the mathematical Nyquist marker, but that statement alone does not verify usable bandwidth.

The example is reproducible arithmetic derived from the displayed equations. It is not a recommended analyzer setup or an ISO acceptance example.

Planning checks outside this worksheet

  • Confirm the applicable machine/product standard, edition, part, machine scope, support, operating state, quantity, amplitude descriptor, points, directions and evaluation band.
  • Confirm sensor type, transverse sensitivity, calibration status, mounting resonance, cable/IEPE requirements, environmental limits and full measurement-chain uncertainty.
  • Confirm analog anti-alias response, usable analyzer bandwidth, sample-rate definition, decimation, AC/DC coupling, high-pass and low-pass filters.
  • Confirm window, amplitude correction, spectral units, one-sided/two-sided convention, DC/Nyquist-bin handling, average type, overlap/hop behavior and whether records are continuous.
  • For bearing frequencies, use exact bearing geometry and actual contact angle or verified manufacturer data; record slip assumptions and uncertainty where relevant.
  • Set alarms only under the exact applicable standard or controlled baseline procedure. Do not transfer ISO zones between different machine types, measurement quantities, descriptors or frequency bands.

Často kladené otázky

Does Delta f equal frequency resolution?

It is the unpadded DFT-bin spacing. Practical ability to separate components also depends on window response, observation time, signal-to-noise ratio, component amplitudes and the analysis method. Zero padding does not create new observation-time information.

Why enter hop in samples instead of only overlap percent?

The exact integer hop defines record start locations and avoids hidden rounding. Overlap percent is then derived from H and N.

Why is the time result a lower bound?

It includes only the signal duration for sequential setups. Sensor placement, route travel, settling, speed changes, checks, instrument processing, retransmission and operator work are excluded.

Why does channel count not multiply time?

Channels recorded simultaneously share the same time interval. They multiply the raw data payload. If a channel or configuration must be acquired later, count it as another sequential setup P instead.

Does passing the Nyquist check prove the setup is correct?

No. It checks only one mathematical necessary condition. The sensor, mounting, signal conditioning, anti-alias filter and instrument bandwidth must support the required band, with adequate uncertainty and dynamic range.

Can this worksheet calculate BPFO or ISO alarm limits?

No. Bearing frequencies need actual bearing geometry and operating conditions. Alarm limits need the exact applicable standard or a controlled baseline procedure. The former approximations were removed rather than relabeled.

Scientific audit revision: 16 July 2026. English source page; translations are reviewed only after the source is verified.
Categories:

WhatsApp
Balanset-1A - 1975 €Spýtajte sa inžiniera