Deriving values

Overview

A dashboard often needs a value the device does not send directly: a calibrated reading, an average of two channels, a running total, or a value a script produces with no device frame at all. Serial Studio has five mechanisms for this. They overlap, so this page says which one fits which job.

Mechanism	Use it for	Runs	Edition
Value transform	Calibrating, converting, or filtering one dataset's own value	Once per frame, per dataset	Pro
Computed dataset	A new channel built from other datasets, such as an average or a ratio	Once per frame	Pro
Variables	Constants and state shared between datasets: calibration tables, integrators, flags	Read and written by transforms and scripts	Pro
Control Loop	Driving the device and reacting to data over time, outside the frame cycle	On its own thread, paced by `delay()`	Free; Variables it uses are Pro
`dashboardTick()`	Values a script owns, with no device frame, that must also be recorded	When the Control Loop calls it	Free; Variables it uses are Pro

Start at the top of the table and stop at the first row that does the job: each row costs more setup than the one before it.

Value transform

A value transform rewrites one dataset's value after parsing and before display and export. The cheapest form is an Expression, one arithmetic line with no script engine:

v * 0.01 + 273.15

Lua and JavaScript transforms define transform(value) and can keep state between frames:

local avg = nil
function transform(value)
  avg = (avg == nil) and value or (0.9 * avg + 0.1 * value)
  return avg
end

See Dataset Value Transforms.

Computed dataset

A computed dataset has no frame index. Its transform produces the whole value, typically from other datasets' final values:

function transform(value)
  return (datasetGetFinal(10) + datasetGetFinal(11)) / 2
end

A transform that never reads value makes its dataset computed automatically when the project loads; the editor greys out the Frame Index field and says why. Transforms run in project-tree order, so place a computed dataset after the datasets it reads. See Computed datasets.

Variables

Variables tables hold named values that parsers, transforms and scripts share. Constant variables are set in the Project Editor and fixed for the session; computed variables are written with tableSet() and keep their last value:

function transform(value)
  local total = (tableGet("energy", "total") or 0) + value * 0.1
  tableSet("energy", "total", total)
  return total
end

Use Variables for calibration constants that change per deployment, and for state that more than one dataset reads. See Variables.

Control Loop

The Control Loop is one setup() / loop() script per project, running beside the frame pipeline. It suits work that is not tied to a single frame: wake-up handshakes, polling, watchdogs, and values that depend on time rather than on arriving data. It reads and writes Variables with tableGet() and tableSet(). Transforms only re-run when a frame arrives, so after writing variables while the device is silent, call refreshDashboard() to update the view. See Control Loop.

dashboardTick()

When a script owns the values, for example a simulation that writes Variables while the device sends nothing useful, call dashboardTick() from the Control Loop. It re-runs the transforms and publishes the result to the dashboard and to the enabled exports (CSV, MDF4, Historian, MQTT), so the script's values are recorded like device data. Make the frame parser return an empty list in this setup; otherwise every value is exported twice. See Variables tables.