Ohmforge Oscilloscope Capture Analyzer

What it is for

Most oscilloscopes can save a waveform as CSV, but the file is just thousands of numbers. This tool reads that export and gives you the measurements you would read off the instrument, and a few it may not have: true RMS over whole periods, period jitter, 10–90 % rise and fall time averaged over every edge, and a spectrum with THD, SNR, SINAD and ENOB.

It understands the CSV layouts of Rigol, Siglent, Keysight and Tektronix scopes, plain "time, channel" tables (also with semicolons and decimal commas), logic-analyzer CSV and WAV files. With a WAV file a sound card works as a two-channel scope. It does not read binary formats such as .wfm or .bin, and it cannot add bandwidth your scope did not capture.

Worked example

The first example is a Rigol-style export of a 1 kHz square wave between 0 and 3.3 V with 25 % duty, sampled at 125 kSa/s. The tool reports 1 kHz, 3.366 V peak-to-peak and 1.605 V RMS, a duty cycle of 25 % (250 µs high, 750 µs low) and a 10–90 % rise time of 32 µs, averaged over 20 edges. Top and base come out as 3.3 V and almost exactly 0 V, so the small noise only adds 0.914 % overshoot.

The same run also computes a THD of 84.6 % and warns that the signal is far from a sine. That is expected: THD describes how much a waveform differs from a sine, so for a square wave it measures its shape, not a fault.

The third example is a Keysight-style CSV of a 1 kHz sine with 1 % second and 0.5 % third harmonic, analysed with the flat-top window. The tool finds a THD of 1.125 % (calculated value: 1.118 %), the second harmonic at −39.94 dBc and an SNR of 62.2 dB.

How to read the result

The measurement table has one column per channel. Top and Base are the two flat levels of a pulse. When a signal has no flat levels, such as a sine, the maximum and minimum are used instead. Frequency, duty cycle and pulse widths come from crossings of the 50 % level. Cycle RMS is taken over whole periods only, so it does not depend on where the capture happens to start.

The spectrum plot marks the fundamental (f1) and its harmonics. Levels are in dBV, which means dB relative to 1 V RMS. The distortion table gives THD, THD+N, SNR, SINAD and ENOB. If THD is missing, the record has too few periods for the FFT size and window you chose, and the warnings say so.

Tips and common mistakes

• Capture at least 10 periods. Short records give a coarse spectrum and leave THD empty.
• Use Blackman–Harris for distortion and flat top for accurate amplitudes. Rectangular only works when a whole number of periods fits the FFT.
• If noise causes extra crossings, raise the hysteresis. If the frequency reads too high, noise is the usual cause.
• Use Analyse from/to to measure one burst or to skip the time while a circuit starts up.
• For a WAV file, enter the volts at full scale if you know it. Sound cards usually block DC and roll off above about 20 kHz.
• A rise time close to your scope's own rise time (about 0.35 divided by its bandwidth) shows the scope's limit, not your circuit's.

FAQ

My file has no time column. What now?

Enter the sample rate. The tool then builds the time axis from it. Rigol files with a Start and Increment column and sigrok files with a "Samplerate" comment need nothing extra.

Why does RMS differ from my multimeter?

The tool measures true RMS including DC, and RMS AC with the DC part removed. Many cheap meters show only the AC part and assume a sine.

How accurate is THD?

With the Blackman–Harris or flat-top window and enough periods, the THD of a test sine comes out within a few thousandths of a percent. Noise that falls on a harmonic's frequency counts as distortion, so very small THD values read slightly high.

Can I compare the capture with a simulation?

Yes. In the Ohmforge workspace, open Instruments → Capture analyzer…, then the "Compare with simulation" tab. It overlays the capture with the transient result for a net you choose.