DC-Bias Capacitance Drop in Class 2 MLCCs: Sourcing Checks Before Substitution

For design engineers and sourcing specialists comparing MLCC offers, the capacitance value printed on a datasheet is a starting point, not the usable figure. In Class 2 dielectrics such as X5R and X7R, the barium titanate ceramic loses permittivity when a DC electric field is applied. A part marked 10 µF at 0 V can deliver between 4 and 6 µF at its rated DC voltage, depending on case size, voltage rating, and the vendor's dielectric formulation. This bias-dependent behavior is documented in manufacturer application notes, but it rarely appears on a sourcing datasheet as a headline parameter. Two parts with identical nominal values can leave the same circuit running at very different margins once installed.

The Real Cost Drivers Behind Unreported DC-Bias Loss

The cost impact of DC-bias loss shows up in three places. One: oversizing. When the design team does not have bias data, the common workaround is to specify a larger capacitance or a higher voltage rating. A 10 µF/10 V X7R in an 0805 case, delivering roughly 5 µF at 5 V, is frequently replaced with a 22 µF/16 V part in a 1206 case to recover the intended effective value. That change increases unit cost two to three times and consumes extra board area.

Two: field returns and debug time. In a buck converter input decoupling network, a 10 µF part that becomes 4 µF at 12 V raises the input ripple and can shift the loop response. The bench technician re-tests the removed part at zero bias, reads 10 µF, and the failure is attributed to other blocks in the circuit; the actual contributor is a condition that exists only on the energized board. Debugging time erodes engineering budget while the root cause stays hidden.

Three: batch-to-batch inconsistency. The DC-bias curve is sensitive to dielectric grain size and dopant concentration in a production lot. Two lots from the same manufacturer can differ by 5–8 percentage points in effective capacitance at the same bias point; differences between vendors are often wider. When the operating point sits close to the design margin, these lot variations push a fraction of the assembled units outside the acceptable performance band.

Comparing Offers on Equal Terms

Comparing quotes on a fair basis requires fixed measurement conditions:

• Bias voltage: compare effective capacitance at the operating voltage of the application, not at the part's rated voltage. If the circuit runs at 6.3 V, ask for the value at 6.3 V even when the part is rated 16 V.
• AC measuring level: Class 2 MLCCs are commonly measured at 1 kHz with 1 Vrms, though some vendors use 0.5 Vrms. Keep this consistent across offers because the AC level changes the reading slightly.
• Temperature: most DC-bias curves are plotted at 25 °C. At an operating temperature above 60 °C, the combined effect of bias and temperature is larger than either effect plotted alone, so the separate curves do not cover the real condition.

When a supplier does not provide a bias curve, request one for the part under consideration. Without it, the part can still be accepted, but the acceptance burden moves to your incoming inspection plan. When curves are available, check the percentage remaining at the operating point. A difference of 10 percentage points between two offers at the same bias is a valid reason to test both parts in your own lab.

A quick screening relationship helps before the full curve is available: a part rated at least two times the applied voltage typically retains 85–95% of its nominal capacitance at the operating point. A part operated near its rated voltage often retains 50–70%. Use this ratio to shortlist candidates, then confirm with the actual vendor curve.

Incoming Inspection Checklist

The following table is a practical starting point for incoming inspection of Class 2 MLCCs used as cross-reference or second-source items:

For a 16 V-rated part operated at 6.3 V, a capacitance ratio below 0.85 warrants investigation. For a 10 V-rated part operated at 9 V, a ratio around 0.6 is in line with expectation — the question is whether the design accounted for it.

Negotiation and Stocking Advice

Negotiation leverage sits on the effective value, not the nominal mark.

• Ask the supplier to confirm the effective capacitance at your operating voltage and include it in the acceptance criteria. A statement such as "10 µF nominal; 6.8 µF at 6.3 V typical" gives you a testable specification. A supplier that only repeats the nominal value leaves the bias behavior open to interpretation.
• Specify a minimum capacitance at bias for the incoming batch. This places the acceptance test on the condition the part actually experiences in service.
• For stocking, two parts with different voltage ratings can substitute for each other when the effective capacitance at the fixed operating voltage falls into the same band. A 16 V-rated and a 25 V-rated part may both be acceptable for a 12 V rail; carrying the higher-rated part as a substitute across multiple designs reduces SKU count at the cost of slightly higher unit price.
• Avoid stocking high-voltage-rated parts as a blanket fix. When the applied bias is low, the capacitance ratio stays above 0.9, and a higher-rated part is unnecessary cost. Match the rating to the design team's voltage derating policy.

When a MLCC enters end-of-life, the replacement search starts with the effective capacitance at the operating point. A legacy 10 µF/16 V part that delivered about 6 µF on a 5 V rail can be replaced by a modern 10 µF/10 V part that delivers around 7.5 µF at 5 V. The new part satisfies the margin, but the case size and price difference should be weighed against the surplus capacity it brings.