Why this measurement is difficult to get right
A test bench is a measurement system rather than a single instrument, and its validity depends on the weakest link in the chain. A bench built around a reference of unknown traceability will produce confident numbers that cannot be defended, and the failure only becomes visible when a customer asks for the evidence behind them.
The practical sequence is the reverse of what a catalogue encourages. Define the quantity and the range first, then choose a reference several times more accurate than the tolerance being judged, and only then arrange the signal path so that nothing in it degrades the reference. Selecting an instrument first and designing the bench around it usually means buying more accuracy than the rest of the chain can preserve.
The third component is the record. A bench that cannot associate a reading with a serial number or a batch is not producing evidence, it is producing numbers. Whether that matters depends on what the bench is for, but if it matters at all it has to be designed in, because traceability added afterwards is usually a manual step that someone eventually skips.
What has to be measured
- Generating a known electrical, pressure or temperature value for comparison.
- Reading back the output of the device under test at the accuracy the judgement requires.
- Conditioning and isolating signals between the device and the measuring instrument.
- Recording results against a serial number or batch so the reading can be traced back.
- The ambient conditions of the bench itself, where temperature or humidity affects the reference.
Which instrument measures what
| Measurement | What the reading has to establish | Instruments commonly specified |
| Sourcing electrical signals | That a known value of the required accuracy is presented to the device under test rather than merely a plausible one | Electrical and multifunction calibrators |
| Sourcing pressure | A pressure the device under test can be judged against, including any static head in the connection | Pressure calibrators and test pumps |
| Sourcing temperature | A stable temperature at the sensor, held long enough for the device to reach it | Temperature calibrators and dry blocks |
| Reading the device output | What the device under test actually produces, independently of what was applied to it | Digital multimeters and panel meters |
| Signal isolation and conversion | That the measuring instrument is not loaded or disturbed by the device under test | Industrial automation and control devices |
| Recording and traceability | Which reading belongs to which unit, which is what turns a bench into evidence | Data acquisition and monitoring |
The ratio between the reference and the tolerance being judged, not the absolute accuracy of the reference, is what makes a bench valid.
Narrowing the choice for a calibration loop
- Write the uncertainty budget before selecting hardware, so the target is defined rather than discovered.
- Set the ratio between the reference and the tolerance being judged - a working ratio of four to one is common, and a larger one raises cost quickly.
- Decide whether calibration happens on site or in a workshop, which is the difference between portable and bench instruments.
- Confirm what documentation the bench has to produce, since that decides whether recording and certification functions are required.
- Establish the signal path in full, including any isolation, conversion or long cable run that could load the measurement.
- Check the ambient conditions around the bench, because a reference specified at a controlled temperature may not hold its figure in a workshop.
- Decide how the reading is associated with a serial number or batch, and who does it.
- Set the recalibration interval for the reference, and note who will perform it.
Where this application ends and the next begins
| Adjacent application | What separates it from building a test bench |
| Electrical parameter testing | That work measures a device in operation, while a bench applies a known value and asks what the device makes of it - the reverse direction. |
| Product quality inspection | Inspection judges a physical article against a drawing tolerance, whereas a bench judges whether a measuring instrument tells the truth. |
| Motor & drive testing | That is a performance test on a machine using instruments assumed to be honest, while a bench is what makes that assumption defensible. |
| Energy efficiency testing | Consumption over a cycle is measured in the field; a bench is a controlled environment and is not where an operating cycle is evaluated. |
| HVAC commissioning | Commissioning measures an installed system against its design intent, and it depends on instruments whose traceability was established elsewhere. |
What to send with the enquiry
- The quantities and ranges the bench has to cover.
- The tolerance being judged, and the ratio the customer wants between it and the reference.
- Whether the bench is permanent or portable.
- What documentation the bench has to produce for each unit tested.
- Whether the device under test is a complete instrument or a component.
- Ambient conditions at the bench, including temperature variation.
- Whether signal isolation or conversion is needed between device and instrument.
- How readings are associated with a serial number or batch, and by whom.
- Quantity required and the date it has to be in use.
- The recalibration interval and documentation the customer will accept.
