Engineering Note — Where Accuracy Class Really Comes From
Across the market, DC dividers reach accuracy classes from roughly 0.01% to 3% and AC dividers from about 0.2% to 10%. That span is not a price-tier story; it is a design story. Three parameters move the scale factor: the temperature coefficient of the resistive elements (standard high voltage resistors hold better than 30 PPM per °C, and matched coefficients matter more than the absolute figure), the voltage coefficient (resistance falls as applied voltage rises, so a divider calibrated at one voltage and used at another carries a built-in non-linear error), and characterisation of every assigned measurement range rather than a single point. A divider is accurate because its components are matched and its ratio is documented — not because of where it was built. That is the basis on which a DEMIKS divider should be weighed against any alternative on a shortlist.
Read the table that way and the answers are straightforward. DC laboratories are honest work for a resistive divider; an AC withstand bay calls for a capacitive divider; and a laboratory that refuses to own three instruments where one will do should look hard at the RC divider. The low-damped capacitive divider sits outside this three-way comparison, because impulse measurement is a different problem — covered in the divider range above. None of this is academic: matching the divider type to the waveform is what keeps loading error, frequency response and corona behaviour inside the limits your test certificate claims.