Power spectral density (PSD) calculator from CSV

The flat part of the plot sits at about 0.01 g²/Hz, the level the signal was made with, and its Grms is the square root of that level times the 180 Hz width of the band. The tone at 250 Hz is a sharp peak, the first row of the peak table next to the plot; a band around it gives the tone’s RMS, its amplitude divided by √2, and so does the table’s Grms of the peak (0.3537 g). To read the peak in dB, tick Show in dB in the dialog.

The same in Python

With the sample file downloaded, these lines give SciPy’s answer for the same settings (skiprows=[1] skips the units row):

import numpy as np, pandas as pd
from scipy import signal

x = pd.read_csv("megarows-sample-vibration.csv", skiprows=[1])["Bridge_Deck_Accel"].to_numpy()
f, p = signal.welch(x, fs=1000, window="hann", nperseg=4096, noverlap=2048)
band = (f >= 20) & (f <= 200)
print(f"Grms {np.sqrt(np.trapezoid(p, f)):.3f} g")
print(f"20-200 Hz: mean level {p[band].mean():.4f} g2/Hz")
k = np.argmax(p)
print(f"peak {f[k]:.2f} Hz  PSD {p[k]:.3f} g2/Hz")

They print:

Grms 1.389 g
20-200 Hz: mean level 0.0100 g2/Hz
peak 250.00 Hz  PSD 0.342 g2/Hz

We ran them on 8 October 2026 with NumPy 2.1.3, SciPy 1.14.1, pandas 2.2.3 and Python 3.12. A test runs the same calculation through MegaRows on every change to the site and checks that its numbers are the ones on this page, and that they match SciPy’s to about 12 significant digits. Download the sample file (CSV, 425 KB).

The settings, and what they change

Segment length and frequency step

The frequency step is the sample rate divided by the segment length: 1,000 ÷ 4,096 = 0.244 Hz for the sample. Longer segments give finer steps but fewer segments to average, so a noisier PSD; shorter segments give a smoother PSD with coarser steps. The result plot says how many segments were averaged.

Overlap

Each segment starts where the last one is half done (50 % overlap), so more segments fit in the signal and the windowed ends of one are the middle of the next. 0 % keeps segments apart; more than 75 % adds little.

Window

Every segment is tapered by a window before its spectrum is taken. Hann is the usual choice for random vibration. Flat top reads the size of a tone between two frequency steps most accurately; Blackman leaks least to nearby frequencies; rectangular (no taper) keeps the finest detail but leaks the most.

Density or spectrum scaling

Density divides by the frequency step, in unit²/Hz: the level of random vibration stays the same whatever the segment length, and the area under it is the mean square, so its square root is the Grms. Spectrum gives the power in each step, in unit²: a pure tone reads its RMS squared (0.125 g² for the sample’s 0.5 g tone), but random vibration reads lower as the steps get finer.

dB view

Show in dB plots 10·log10 of the PSD, with 1 unit²/Hz as 0 dB, so levels many decades apart fit on one linear axis.

Units: g²/Hz from g

The PSD takes its unit from the channel. Give each column its unit in a units row under the header, as the sample does ([s] and [g]), and the PSD comes out in g²/Hz and the Grms in g, the units of a random-vibration test profile. A signal in m/s² gives (m/s²)²/Hz, in Pa gives Pa²/Hz, and a signal without a unit gives a PSD per Hz.

• The highest frequency is half the sample rate: a file logged 1,000 times a second shows up to 500 Hz, so a profile up to 2,000 Hz needs at least 4,000 readings a second.
• From g to m/s²: multiply the signal by 9.80665 first (Math › Scale and offset); the PSD then grows by 9.80665².

Grms of the whole range and of a band

The Grms is the root of the area under the PSD: the overall level of a random vibration. The result plot shows it over every frequency, from 0 Hz to half the sample rate. Type a band into its From and To boxes, such as 20 to 200 Hz, and it shows the Grms of that band instead; the edges are read between frequency steps, so a band need not start on one. Set Grms band in the dialog to open the plot with the band already set.

The Grms of each peak

The peak table next to the plot lists the highest peaks of the PSD, highest first, with the Grms of each: the root of the area under the peak, from the dip on its left to the dip on its right. For a tone that is close to its RMS. A peak counts when it stands at least 3 dB above the dips either side; copy the table into a spreadsheet or download it as CSV.

Checked against SciPy

Private by design: nothing is uploaded

Vibration test data is often confidential. This tool reads your file and calculates the PSD in your browser, so the file and its name never leave your device. It works on files with millions of rows: the PSD uses every sample of the channel, and the practical limit is your device’s memory.