The Series Capacitor Topology: Node Labels and Core Math
When you wire capacitors in series, the total capacitance drops, but the total voltage rating increases. The most critical concept for bench work and circuit design is that the voltage across capacitors in series is inversely proportional to their individual capacitance values. Unlike resistors in series (where the largest resistor drops the most voltage), the smallest capacitor in a series string will endure the highest voltage.
To visualize this, let us define a standard two-capacitor series topology with three distinct nodes:
- Node A: The positive terminal of the DC voltage source (V_total), connected to the positive lead of Capacitor 1 (C1).
- Node B: The floating junction node connecting the negative lead of C1 to the positive lead of Capacitor 2 (C2).
- Node C: The ground reference (0V), connected to the negative lead of C2 and the negative terminal of the voltage source.
Because capacitors in series share the same charge (Q = C × V), and Q is constant across the string, the voltage distributes according to the inverse capacitance ratio. The exact formula to find the voltage across C1 is:
V_C1 = V_total × [ C2 / (C1 + C2) ]
If C1 and C2 are perfectly matched (e.g., both 100µF), the voltage splits evenly. If C1 is 100µF and C2 is 50µF on a 300V bus, C1 drops 100V and C2 drops 200V. This inverse relationship dictates why you must carefully match capacitance values or use active/passive balancing in high-voltage designs.
Design Walkthrough: Building a 400V DC Bus Snubber
Let us move from theory to the workbench. Suppose you are repairing a variable frequency drive (VFD) or building a motor controller, and you need a 50µF snubber capacitor rated for a 400V DC bus. You check your inventory and find you only have standard 100µF, 250V aluminum electrolytic capacitors (e.g., United Chemi-Con KXG or Cornell Dubilier 380LX series).
Wiring two 100µF 250V caps in series yields the target 50µF total capacitance (100 × 100 / 200). Ideally, the 400V bus splits into 200V across each capacitor, safely below their 250V rating. However, you cannot wire them directly without balancing resistors. Aluminum electrolytics have significant and highly variable leakage currents. If C1 leaks less current than C2, C1 will charge to a higher voltage, potentially exceeding 250V, venting its electrolyte, and failing catastrophically.
Step-by-Step Breadboard and Verification Test
Before soldering this into a high-voltage PCB, validate the node voltages on the bench using a low-voltage proxy.
- Discharge and Prep: Short both capacitors with a 100Ω power resistor to ensure they start at 0V. Do not use a screwdriver; the di/dt spike can damage internal foil.
- Wire the Topology: Connect C1 positive to Node A (V+ rail). Connect C1 negative to C2 positive at Node B. Connect C2 negative to Node C (GND rail).
- Install Balancing Network: Solder a 100kΩ 1W resistor in parallel with C1 (Node A to Node B) and a second 100kΩ 1W resistor in parallel with C2 (Node B to Node C).
- Low-Voltage Proxy Test: Apply 12V DC to Node A (relative to Node C). Using a digital multimeter (DMM), probe Node B. You should read exactly 6.0V (±0.2V). This confirms the resistors are forcing an even split.
- Ramp and Monitor: If testing with a variable DC supply, ramp the voltage to 50V, then 100V. Measure Node B at each step. The voltage at Node B should always track at exactly 50% of V_total. If it deviates by more than 5%, one of your capacitors has a severe internal leakage fault and must be binned.
Behavior Matrix: What Happens When Component Values Shift
Understanding failure modes is what separates a hobbyist from a reliability engineer. The table below maps exactly how the voltage across capacitors in series reacts to component degradation and catastrophic faults.
| Condition / Fault | Total Capacitance | Voltage Across C1 | Voltage Across C2 | Physical Result on the Bench |
|---|---|---|---|---|
| Nominal (Matched) | 50µF | 200V (50%) | 200V (50%) | Stable operation; resistors dissipate 0.4W each. |
| C1 Capacitance Drops (Aging) | Decreases | Increases (>50%) | Decreases (<50%) | C1 runs hotter due to higher ESR; bleeder resistors prevent overvoltage if sized correctly. |
| C1 Fails OPEN | 0µF (Circuit broken) | Floats to leakage ratio | Floats to leakage ratio | DC current stops. Circuit loses snubbing function. Node B voltage becomes unpredictable. |
| C1 Fails SHORT | Equals C2 (100µF) | 0V | Full V_total (400V) | Catastrophic cascade. C2 is subjected to 400V (rated for 250V). C2 will overvolt, vent, or explode within milliseconds. |
| C1 Leakage Spikes | 50µF (Nominal) | Decreases | Increases | Without bleeder resistors, C2 overvolts. With 100kΩ bleeders, the 2mA bleeder current masks the microamp leakage spike, saving C2. |
Series vs. Parallel: Why Choose Series for High Voltage?
When designing power stages, you must choose between series and parallel topologies based on your bottleneck: voltage headroom or energy storage (capacitance).
Choose Series When: Your bus voltage exceeds the maximum rated voltage of available capacitors. High-voltage DC links (like 400V or 800V EV traction inverters) almost exclusively use series strings of lower-voltage capacitors because manufacturing a single 1000V, high-µF aluminum electrolytic is physically massive and cost-prohibitive. Series wiring also inherently reduces equivalent series inductance (ESL) if the physical layout is optimized, which is critical for high-frequency switching nodes.
Choose Parallel When: Your voltage is well within the rating of a single component, but you need high ripple current handling or massive energy storage (e.g., a 12V car audio amplifier needing 5 Farads). In parallel, the voltage rating of the bank is strictly limited by the lowest voltage-rated capacitor in the group, while capacitance and ripple current capability sum linearly.
Frequently Asked Questions
Why is the voltage across capacitors in series inversely proportional to their capacitance?
This is rooted in the fundamental physics of charge storage, defined by Q = C × V. When a DC voltage is applied to a series string, the same displacement current flows through all components, meaning every capacitor stores the exact same charge (Q). Since Q is constant, if you rearrange the formula to V = Q / C, it becomes mathematically obvious that as capacitance (C) gets smaller, the voltage (V) required to hold that identical charge must get larger. For a deeper mathematical proof of series and parallel capacitor networks, refer to the All About Circuits DC textbook chapter on capacitors.
Do I need balancing resistors for MLCC ceramic capacitors in series?
Generally, no. Multi-Layer Ceramic Capacitors (MLCCs) have exceptionally high insulation resistance (often >10,000 MΩ·µF) and virtually zero steady-state DC leakage current compared to aluminum electrolytics. If you series two 10µF 50V X7R ceramics to handle a 90V rail, the voltage will split based purely on their capacitance tolerance (which is usually ±10% to ±20%). However, if you are using high-K dielectrics like Y5V in high-temperature environments where insulation resistance degrades, or if the circuit requires rapid, predictable discharge for safety compliance, adding high-value bleeders (e.g., 10MΩ) is still recommended.
What happens to the voltage across capacitors in series if one fails short?
A shorted capacitor acts as a piece of wire. Its voltage drop immediately falls to 0V. Consequently, 100% of the total bus voltage is instantly transferred to the remaining capacitor(s) in the string. If you have two 250V capacitors in series on a 400V bus, and C1 shorts, C2 is suddenly hit with 400V. Because 400V vastly exceeds C2's 250V dielectric breakdown limit, C2 will rapidly draw massive current, boil its internal electrolyte, and vent or rupture. This cascading failure mode is why high-reliability designs use active voltage balancing ICs or overvoltage crowbar circuits alongside passive resistors.
How do I measure the voltage across a floating series capacitor node safely?
Measuring Node B (the junction between series capacitors) requires care. A standard digital multimeter has an input impedance of 10MΩ. If your balancing resistors are 1MΩ or higher, the DMM itself will act as a parallel resistor, pulling Node B toward ground and giving you a falsely low voltage reading. Always use a DMM with a 10MΩ impedance for circuits with low-value bleeders, or use the 'Low-Z' (LoZ) mode if your meter supports it to bleed off ghost voltages. If you need to view the AC ripple voltage on Node B with an oscilloscope, never connect a standard grounded scope probe. The probe's ground clip will short Node B directly to earth ground, instantly destroying the capacitor and your scope. You must use a high-voltage differential probe or isolate the oscilloscope.
Can I put different value capacitors in series to get a specific voltage split?
Yes, you can intentionally use mismatched capacitance values to act as a capacitive voltage divider, forcing a specific AC or transient voltage split. However, for DC biasing, this is highly discouraged. The DC voltage will still attempt to split inversely, but the differing leakage currents of mismatched parts will fight the capacitive division, causing the node voltage to drift unpredictably over time and temperature. For DC applications, always use matched capacitance values and rely on resistors to set the DC node voltage. For further reading on capacitive dividers and AC behavior, consult the Electronics Tutorials guide on series capacitors.
