Skip to content

Sensing & Instrumentation

Every number in the system starts somewhere physical.

Most operational reporting problems are presented as software problems and turn out to be measurement problems. The number nobody trusts is the one somebody keys in from a clipboard; the metric that cannot be improved is the one nothing observes. This is the work of deciding what to instrument, choosing sensing that answers the question at the accuracy the decision actually needs, and getting its output onto the network in a form something else can read.

Every reading, to a calibrated instrument
TraceableEvery reading, to a calibrated instrument
Every point, against the decision it serves
JustifiedEvery point, against the decision it serves
Year two against year one
ComparableYear two against year one

Sounds like

You might recognise one of these.

  • We are paying someone to walk the floor twice a shift and write numbers on a clipboard.

  • We cannot tell you our real downtime, only what people remembered to report.

  • The machines have data on them and none of it reaches anything we can query.

  • We were quoted a sensor package and we have no idea whether it is the right one.

What this includes

The work, specifically.

Not every engagement needs all of it. This is the range we cover and what each part is actually for.

  • Measurement planning

    What question is being asked, what would answer it, and what accuracy that answer actually requires. Over-specifying is the most common and most expensive error here: a decision that turns on a five-percent difference does not need a laboratory-grade instrument at ten times the cost per point.

  • Sensor selection and siting

    Choosing the instrument and, just as importantly, where it goes. Placement determines what a reading means, and a correctly calibrated sensor in the wrong position produces confidently wrong data forever.

  • Reading equipment you already own

    Before specifying anything new, we check what is already talking. A machine with a serial port and a manual is usually cheaper to read than to replace, and this step routinely removes hardware from the quote.

  • Signal conditioning and calibration

    Power, isolation, noise, drift and a calibration schedule somebody can actually keep. The parts of an install that decide whether year-two data is still comparable to year-one data.

  • Environmental hardening

    Wash-down, cold store, vibration, dust and temperature range as design inputs rather than as failures discovered in month three.

What you get

Deliverables, not documents.

  • A measurement plan naming each point, its purpose and its required accuracy
  • A bill of materials with the reasoning for each choice and the alternatives rejected
  • Installation and commissioning notes an electrician or millwright can follow
  • Calibration procedure and schedule
  • Verified readings arriving at a system that can store them

Shapes

How this usually runs.

  1. Feasibility and measurement plan

    2–3 weeks

    What can be measured, to what accuracy, at what cost per point, and whether the answer justifies the install. This phase regularly ends in a recommendation to instrument less than was asked for.

  2. Pilot install

    4–8 weeks

    One line, one asset or one site instrumented end to end and running against the real system, with the readings validated against a known reference.

  3. Rollout

    Ongoing

    The remaining points, with calibration and spares handled as a process rather than as a series of favours.

Tooling

What we build it with.

No tool here was picked because it was new. Where we do reach for something novel, it is in one place, for a stated reason, and it is written down.

Sensing
  • 4–20 mA and 0–10 V transmitters
  • Thermocouples and RTDs
  • Current transformers
  • Vision and optical counting
  • Load cells
Acquisition
  • Industrial I/O modules
  • Linux on ARM
  • ESP32
  • Raspberry Pi CM
Delivery
  • MQTT
  • Time-series storage
  • Sparkplug B

Questions

Sensing, honestly.

  • Usually some of them, and it is the first thing we check. Existing instrumentation that is calibrated and correctly sited is the cheapest data available to you, and a surprising amount of it is already producing readings nothing is listening to.

  • Less accurate than most quotes assume. Accuracy is priced steeply and the right level is set by the decision the number supports, not by what is available. We state the required accuracy per point in the plan so you can see what each increment costs you.

  • You do — the measurement plan, the bill of materials, the firmware and the configuration, on the same terms as every other engagement here.

Next step

Tell us what’s breaking.

Forty-five minutes, no charge, no deck. We’ll tell you what we’d do, what it would likely cost, and whether you should be building this at all.