Free · nothing uploaded
The fastest way to find out whether your data is usable, and it costs nothing. Your browser reads the file on your own machine, works through it and shows the result: nothing needs to be sent anywhere and there is nothing to sign up for. A typical export takes about ninety seconds.
Drop a CSV or TSV from your instrument here
or
Your browser reads the file. Nothing is uploaded and no account is needed.
Large files are read as a sample — enough for screening.
What it looks at
The first four checks run on any file with numbers and a timestamp, from a sensor log to a rig export. The last two appear only when current, voltage and time are found in the file — that is, on cycler data.
We look for steps that go backwards, repeat or jump. While the time column is wrong, capacity, energy and efficiency are all off by an amount nobody can put a number on.
Empty cells and non-numeric junk, counted per column rather than averaged away. If somebody later fills those places with zeros, every integral shifts.
Fifty or more identical values in a row. Usually a stuck sensor, or a logger that carried on writing after the channel dropped out.
Changes far outside the normal spread of that same channel. A real switch of regime and a sensor glitch look identical, so we show them to you rather than deleting them on your behalf.
We work out the segments from the current alone, so you can compare them with your own step markers. If the two disagree, one of them is wrong.
Just the derivative, with no smoothing at all. This plot alone tells you whether you need cleaning or your data was fine to begin with.
What the check does not do. Whether the noise can be separated from the signal, whether a model can be built and what a cleaned dQ/dV would look like — those are questions for the engine; a browser cannot answer them. But the results will tell you whether they are worth asking.
Step two · by volume
You get two things: the cleaned data, and a full list of what changed in it. The second file is what lets you answer a reviewer later.
By the amount of work: how many rows, how many columns and what you asked for. The total depends on size: a base plus the cells actually processed. The minimum top-up is €15. There is one balance per account and it can be spent on anything: cleaning, dossiers, compiles, or keeping a stream filtered. The figures on this site are indicative; your cabinet works out the exact, binding amount for your job and shows it before anything runs.
If the file cannot be parsed, we do not charge. If it turns out your data was already clean, we do not charge either. Inventing an improvement that is not there is not something we will do.
Cycler exports have their own route. Cycles, capacity, efficiencies and dQ/dV peak integrity are beyond a general-purpose cleaner — it knows nothing about them. See the battery workflow →