AC and DC Withstand Records: Why Equal Peak Voltage Is Not Equivalence

1. Establish exactly what is being substituted

Identify the original test and the requested replacement in separate columns. Include the circuit revision, connected nodes, conditioning, waveform, voltage convention, duration and decision rule. An AC instrument photograph and a pass label do not establish that the required DC sequence was performed, even when both instruments can display the same numerical voltage.

Distinguish a diagnostic comparison from a conformity decision. Engineering can study how a known assembly responds to two waveforms, but that study does not itself authorize replacing an approved acceptance method. The governing equipment requirement, component construction and responsible review determine which evidence is acceptable. Do not infer permission from the available tester or from a convenient conversion factor.

2. Match the voltage convention before comparing numbers

For an ideal sinusoid with no DC component, peak voltage is the RMS value multiplied by the square root of two. Thus an assumed 100 V RMS sinusoid reaches approximately +141.42 V and −141.42 V. A constant +141.42 V DC waveform matches its positive maximum but never reverses polarity.

This deliberately low numerical example illustrates the convention only; it is not a proposed acceptance voltage. A distorted AC waveform or one with a DC offset does not necessarily follow that simple RMS-to-peak relation. Record the actual waveform or an appropriate verified characterization when the difference matters. Peak-to-peak voltage, peak magnitude and RMS are three different quantities and must not share an unlabeled voltage field.

3. Calculate the alternating current that exists without a conductive defect

A capacitance carries current continuously under alternating excitation, even in the ideal lossless limit. For sinusoidal voltage its RMS current is 2πfCVrms. That load remains during the AC hold; it is not simply the initial charge associated with a DC voltage rise.

For an assumed 10 nF capacitance at 100 V RMS, the capacitive current is about 0.3142 mA at 50 Hz and 0.3770 mA at 60 Hz. Increasing frequency by 20% increases this ideal current by 20% without changing insulation resistance. The source must maintain the required waveform under the actual combined specimen, cable and fixture load. Failure to reach the requested voltage is a different event from a correctly applied stress followed by specimen breakdown.

4. Do not transfer the current threshold without its measurand

The total AC current can include an in-phase contribution and a quadrature capacitive contribution. In a simple linear parallel resistance-capacitance model these components combine vectorially. Their RMS magnitudes do not add arithmetically, and the ratio Vrms divided by total current is an impedance magnitude rather than an insulation resistance.

Using the 50 Hz example, add an assumed in-phase current of 0.1000 mA. Total current becomes approximately 0.3297 mA, from the square root of 0.3142² plus 0.1000². Dividing 100 V by that total gives approximately 303 kΩ, although the assumed resistive branch is 1 MΩ. The in-phase current may itself include dielectric loss, so even a phase-separated reading is not automatically the steady DC conduction result.

5. Keep polarity history and stack response in the review

A sustained DC exposure and a reversing field need not distribute voltage identically through a multilayer insulation system. Capacitive division, finite conductivity and interfacial charge describe different limiting behavior. An identical external peak does not establish an identical local field throughout a fired dielectric, protective coating and neighboring interface.

Retain the actual material sequence and the time or frequency conditions used to interpret it. Do not turn this distinction into a blanket claim that AC is always harder or DC is always easier. The relevant weakness may be a bulk layer, a contaminated surface, an edge or an attached component, and the method must address that specific construction. Connected suppression or sensing components also require explicit inclusion or isolation decisions.

6. Include the final electrical state, not only the hold interval

The DC method can leave the specimen and fixture charged after its stress interval ends. For an ideal capacitance, stored energy is one half of capacitance times voltage squared. The assumed 10 nF at 141.42 V contains approximately 0.000100 J; multiplying capacitance by ten multiplies that energy by ten, while doubling voltage multiplies it by four.

Those arithmetic relationships do not define a safe handling threshold or a discharge time. Real fixtures can contain additional capacitance and dielectric recovery behavior. Use qualified personnel, rated test arrangements, protective enclosure or interlocks and the actual instrument's approved discharge and verification procedure. Preserve the same control after an interrupted or failed test. A completed software step or a passing judgment is not permission to touch an unverified electrical boundary.

7. Identify which missing field prevents reuse of the record

A reusable report needs the conditions that distinguish the two methods. Resolve missing information with the original record or an approved additional evaluation rather than filling it with an assumed equivalence.

8. Deliver a method-compatibility decision with explicit scope

The engineering outcome should state whether the existing record directly satisfies the named requirement, whether an explicitly permitted substitution is supported, or whether a different evaluation is needed. Keep the reason tied to waveform, construction and method rather than to a preference for one kind of instrument.

Provide the supplier and system team with the exact tested boundary and any unresolved difference. If additional comparisons are approved, retain actual voltage and current behavior instead of collecting only a second pass label. This makes later changes to fixture capacitance, connected electronics or layer construction reviewable without recreating the ambiguity that prompted the comparison.