Formats
Four instruments, four export paths, and the same three traps in each: which layer you exported, which units you got, and whether the step markers agree with what the current actually did.
Every data workflow starts the same way: somebody exports from the instrument's closed format and hopes nothing important was lost on the way. Usually something is — and you cannot see it in the finished file.
Neware.ndax and .nda | Newer BTS versions write .ndax, older ones .nda. Export from the BTSDA client, picking the cycle range you need and the layer you actually want: record-level rows are the raw measurement, while step-level and cycle-level rows have already been aggregated by the instrument. Capacity columns are usually cumulative within a step and reset at step boundaries — check that on your own file, because the behaviour depends on the BTS version. Current may be in A or mA, and the sign convention for charge is not universal. |
Arbin.res and MITS Pro | Arbin writes an Access-format database holding several tables. Channel-normal is the record-level measurement and is the one analysis needs; the statistics tables are already reduced and that reduction cannot be undone. Exports carry both a cumulative test time and a time within the current step: integrating on the wrong one is a quiet, common error, because both columns are monotonic and both look plausible. Cycle index follows the schedule, not what the current actually did. |
MaccorMacView exports | The structure is straightforward; what bites is units and step handling. Current and capacity units depend on how the test was configured, and headers do not always state them unambiguously — a file in mA and a file in A look identical apart from a factor of a thousand, and the shape of the curve does not change. Schedules can end a step on several conditions, so segment from the current rather than from an expected pattern. |
BioLogic.mpr to .mpt | The text export opens with a long header block describing the technique and its parameters. Generic CSV readers treat that header as data unless you skip it explicitly, which yields a file that parses without error and is completely wrong. Keep the header: it is the closest thing to a machine-readable protocol any of these instruments will give you. Column names carry their units in the name itself. |
Verify
Only record-level data lets you recompute anything, and the file does not say which layer it is. If a colleague hands you a file, establish this before you start.
There is no single standard for either. Getting the current sign wrong makes capacity, efficiency and dQ/dV meaningless rather than merely inaccurate.
Cycle and step indices come from the schedule; the current comes from the cell. When they disagree, the cell is right.
Start
The free check takes ninety seconds and will tell you which of the three traps applies to your file.