Test and measurement instrumentation
When the measurement is the product, the specification is an error budget — and tempco, thermal EMF and leakage usually consume more of it than the converter does.
Evidence
What we can substantiate
Everything below is work we have actually done, described at a level that respects the client’s confidentiality. If you need detail, ask under NDA.
- Error budget written before the parts are chosen Initial tolerance is usually the smallest term. Tempco over the operating range, reference drift over years, amplifier offset and self-heating typically dominate — a 0.1% resistor at 100 ppm/°C drifts six times its own tolerance across 60 °C. Buying a better converter while the reference drifts is money in the wrong place.
- Kelvin sensing wherever a connection resistance is in the path Contact and trace resistance sit directly in series with the measurement unless the sense path is separated from the force path. Four-wire connection planned at the schematic, and the sense junction placed where it actually defines the measured point.
- Guarding and leakage control on high-impedance nodes Above a few hundred megohms, board leakage across a humid or contaminated surface parallels your input. Guard rings driven at the node potential, careful use of solder mask, and in extreme cases a PTFE standoff rather than a PCB trace.
- Thermal EMF managed in the sense path Every dissimilar-metal junction is a thermocouple, and tens of microvolts of gradient-driven offset is easy to generate and hard to explain. Symmetrical layout so junctions cancel, and heat sources kept away from the sense path.
- Isolated front ends for floating and common-mode-hostile measurements Measuring across a shunt at a high common-mode potential needs a barrier, and the barrier's own leakage and CMTI become part of the error budget.
- Test fixtures designed alongside the product Bed-of-nails and functional test hardware built with the board, with probe access planned rather than found, and a fixture that holds calibration between shifts.
Common questions
Straight answers
What actually limits precision in practice?
Rarely the converter. Temperature coefficient usually dominates over any realistic operating range — a 0.1% resistor with 100 ppm/°C drifts six times its own tolerance across 60 °C, so paying for tighter initial tolerance without tighter tempco buys nothing. After that it is grounding, self-heating and thermal EMF at dissimilar-metal junctions. Write the budget down and the right place to spend money becomes obvious.
My readings drift for twenty minutes after power-on. Why?
Self-heating reaching equilibrium. Components dissipating power warm themselves and their neighbors, and every tempco in the signal path moves until the thermal gradients settle. It is not a fault, it is a design characteriztic — and it can be reduced by lowering dissipation in the sense path, arranging symmetry so gradients cancel, or accepting a specified warm-up time. Which of those is right depends on how the instrument is used.
Can you help with the calibration and verification side?
We design for it — accessible calibration points, stable references, documented adjustment procedures and fixtures that hold. Accredited calibration itself belongs with a 17025 laboratory, and we will design the interfaces they need rather than leaving them to improvise.
Got a board to design — or one that won’t boot?
You talk to the engineer who would do the work. Reply within one business day, and we’ll sign your NDA before you go into detail.