SkiErg · PM5 · Bluetooth
There is exactly one hard part in this, and it is not the analysis. It is getting every stroke off the machine and onto something a coach can read, without asking an athlete to babysit a phone.
Every SkiErg has a Performance Monitor 5 on it. The monitor watches the flywheel decelerate between strokes and derives force, distance, pace and stroke rate from it. That measurement is good, and it is already happening — the problem has never been the sensor.
The PM5 publishes its data on a Bluetooth service: general status, stroke data, split and interval messages, and the sub-stroke force waveform, all multiplexed onto one channel. It will talk to anything that knows the protocol and is standing close enough.
The app registers the Concept2 profile, finds the monitor, subscribes and decodes each packet as it lands. No phone, no bridge, no second device to keep awake. If there are three ergs in the room, you pick which one is yours.
The session is a real activity file: heart rate, training load, recovery — everything your watch already does for you keeps working, and the session lands in the platform you already use. The stroke stream rides along beside it rather than replacing it.
After the session the watch uploads what it captured, byte for byte. The server normalises it into strokes, splits and force curves — and keeps the original, so when the parser improves, every session anyone has ever recorded gets the improvement without being re-ridden.
Failure mode
Bluetooth at an erg is a hostile environment: a rack of monitors, a hall full of watches, an athlete whose wrist swings a metre every stroke. Connections drop. The rule is that a drop marks a gap in the data and the recording keeps running while it reconnects.
The alternative — a stale number left on screen, presented as live — is the worst thing this product could do. A coach who cannot trust one number stops trusting all of them.
Data handling
Uploaded payloads are never edited in place. Each session records which parser version produced its numbers, and normalisation can be re-run over every session in the archive.
That matters in a sport where the interesting comparison is October against March. If we fix a decoding bug in year two, your year-one tests get fixed too — nobody has to redo a test to keep a season comparable.
What you need
That is the whole list. If you can already see a pace on the monitor, the data we want is being produced.
Works with
Doesn't need
Not sure what your numbers mean yet? Start with the pace calculator or the drag factor guide.
Early access goes out to coaches and to individual skiers, and they get slightly different things. Say which one you are and we will sort it.