Convert TDMS to Parquet
Drop a .tdms file here, with its .tdms_index if you have one.
The file is read in this browser tab. It is never uploaded.
Not supported: DAQmx raw data, ExtendedFloat and FixedPoint channels, arrays of dimension other than 1. They are reported, and the rest of the file stays readable.
Why Parquet
A CSV file stores numbers as text: types are lost and a double must be printed in full to come back intact. Parquet keeps each column's type, compresses well and is read in one line by pandas and most data tools.
What the file contains
- One Parquet file per group, one column per channel, with the channel's own type: integers stay integers,
DoubleFloatstays a 64-bit float. - A time column in seconds from the waveform properties, and an optional timestamp column in UTC from
wf_start_time. - Channels shorter than the others are completed with nulls, never with zeros.
Every export is checked in the tests: each file is read back by pandas and compared, value by value, with what the npTDMS library reads from the same TDMS file.
Reading it
import pandas as pd
df = pd.read_parquet("measurement_Group.parquet")Size of the tool
The Parquet writer is a separate module of 183 KB compressed, downloaded the first time a Parquet export is started, then kept for offline use.