New: the AI firewall watches the whole platform and explains each incident in plain words. See how it works

Machine downtime tracking from a run signal to a downtime log

A state timeline is not a downtime log. How to go from a run signal to a list of stops with reasons, short-stop rules and planned versus unplanned time.

Machine downtime tracking turns a run signal into a log of stops: when each stop began, how long it lasted, why it happened, and whether it was planned. Read one reliable run signal per machine, record its edges with time stamps, set a minimum stop length, give every longer stop a reason code, and keep planned and unplanned time apart.

I build dashboards for plants. A state timeline looks good on a wall screen, but at the end of a shift the lead wants a list of stops with reasons.

How do you calculate machine downtime from PLC signals?

Downtime is the sum of stop durations in a period, and each stop runs from the moment the run signal drops to the moment it comes back. You need the edges of the run signal with their time stamps.

The best signal is a run bit the PLC already sets, read without changing its program, as covered in PLC data logging. If the gateway polls the signal, each edge time is only as exact as the polling interval: with a 10-second poll, a stop can start up to 10 seconds before it is seen. Reporting on change, with the time stamp taken where the signal is read, avoids most of that error.

How short a stop should count as downtime?

Pick a minimum stop length for each line and count shorter stops separately as short stops, with their number and total time, instead of dropping them. Forty stops of 20 seconds cost more than 13 minutes, and nobody fills in a reason for each.

I set the threshold near the time an operator needs to clear a jam without leaving the station, and I apply the same idea the other way: a restart that runs only a few seconds before the next stop is merged into one stop. Without that, one jam can fill several log rows.

Hermetry connects field devices to live dashboards, rules and alerts: IoT telemetry and SCADA in one platform. In its flows, counter, statistics and watchdog blocks keep a memory per device: a Delta block writes the difference from the previous reading and, if set, the milliseconds since it; a Counter counts messages over a sliding window of up to 7 days; a Watchdog sends a timeout when a device stops sending. Results stored with Save telemetry become derived values, charted, summarised and alarmed on like any other key.

Planned vs unplanned downtime: why keep them apart?

Planned downtime is decided in advance, such as breaks, scheduled maintenance, planned changeovers or no orders; unplanned downtime is everything else. Mixing them hides the stops the shift could have prevented behind the ones it chose.

I keep the reason list short. I start with two levels, a group (mechanical, electrical, material, quality, waiting, planned) and a few reasons in each, and let the operator pick one when a stop passes the minimum length. Stops without a reason stay in the log as unassigned, and their count is the first number I put on the shift board.

What should a shift downtime log contain?

Each row is one stop, with enough fields that the shift lead can explain it the next morning without asking anyone.

FieldExample
Start and end10:42:05 to 10:58:40
Duration16 min 35 s
MachineFiller 2
Group and reasonMaterial: no caps
Planned or unplannedUnplanned
Coded by and commentOperator, line B: cap hopper empty

Totals per reason, sorted from largest, show where the shift lost time. For the part counts that belong next to the stops, see production tracking with machine states and counters and the manufacturing page.

Frequently asked questions

How do you handle a stop that runs across a shift change?

Split it at the shift boundary so each shift reports its own part, and keep one stop identifier so the parts can be joined again. The reason code belongs to the whole stop, and the next shift can see that it inherited a running stop.

Can downtime be tracked on machines without a PLC?

Yes. A current sensor on the main motor or a counter on the output conveyor, read by a gateway, gives a usable run signal. Calibrate the current threshold against the machine idling and running, because an idle motor still draws some current.

Who should enter the downtime reason?

The operator who cleared the stop, at the time of the stop, from a short list on the line’s screen or a tablet. I would not rely on reasons typed in hours later. The shift lead then reviews only the unassigned stops instead of every row.

Share

See your field data live this week

Start a trial in your region with your first devices, or book a demo and bring your own use case.