Every Adapter Pays a Standby Tax

Unplug any phone charger and the voltage on its input capacitor doesn't vanish instantly. It decays over a second or two — that's the discharge resistor doing its safety job. But the resistor can't tell the difference between "plugged in and charging" and "plugged in and idle."

So it burns power continuously, from the moment of first plug-in to the day the adapter dies. The 53mW figure assumes a 1MΩ resistor at 230VAC — typical for a 90W notebook adapter's X-capacitor.

From what we see across Shenzhen lots (2025–2026), the discharge resistor is the last unexamined line item in no-load budgets. Designers optimize the controller's standby, shrink the feedback network — and leave a resistor burning a tenth of the whole budget.

Why the Resistor Is There at All

The X-capacitor sits across the live and neutral lines to filter conducted EMI. It's a genuine capacitor — it stores charge. Unplug the adapter and that cap holds up to the peak line voltage, waiting for someone to touch the plug.

Safety standards close the loop. IEC 62368-1 (the standard that replaced 60950/60065 for IT and AV gear) requires X-capacitors above 0.1µF to discharge to a safe voltage within a defined window after unplug.

The resistor guarantees that discharge. A 1MΩ bleeder across a 330nF cap gives roughly a 0.33s time constant — comfortably inside the window. And critically, the passive resistor keeps working under a single fault, which the standard also demands.

The Math Behind the 53mW

The loss is pure P = V²/R. At 230VAC with a 1MΩ bleeder: 230² / 1MΩ ≈ 53mW, continuously. At 120VAC (US), the same resistor burns ~14mW — the tax is smaller but still present.

But you can't just raise the resistance to cut the loss. Bigger R means slower discharge. For a 330nF X-cap needing discharge under the standard's window, the practical ceiling is around 3MΩ — which still burns ~17.6mW at 230VAC.

The loss scales with the EMI filter you need. Bigger X-cap for conducted-emission margin = bigger standby tax.

The 53mW That Breaks a 0.1W Budget

US DOE Level VI (mandatory since February 2016) caps no-load power at 0.1W for single-voltage AC-DC adapters up to 49W. The EU's ErP Lot 7 Tier 2 allows 0.3W (≤51W class). Either way, a 1MΩ bleeder at 230VAC consumes over half the US allowance — 53% of 0.1W.

That's the uncomfortable part: the discharge resistor alone can fail a no-load compliance test, before the controller, before the feedback, before anything the designer optimized.

At 120VAC the same design passes with room to spare — which is why the problem hides in the EU market and surfaces as a certification surprise when a US-designed adapter ships to 230V countries.

The Fix: Switch the Resistor Out When AC Is Live

How do you kill the tax without breaking the safety path? The CAPZero-2 approach is brutally simple: put a high-voltage switch in series with the discharge resistor, and drive it from the line itself.

AC present = switch open, resistor carries no current, loss drops to under 5mW. AC removed = the IC detects it within ~30ms and closes, and the resistor discharges the cap exactly as before.

The switch is a two-terminal IC (CAP200DG-class) with an internal 1000V MOSFET, no external bias rail, and no ground pin. It sits in series with the bleeder like a component, not a subsystem.

The discharge behavior is unchanged — the safety engineer sees the same RC discharge path, now with the resistor guaranteed present by two independent pin pairs per terminal. The no-load budget gains back the 48mW the resistor was burning.

The Design Win Nobody Talks About: Bigger X-Caps

There's a second payoff hiding in the same swap. EMI filter design is a trade: bigger X-capacitance shrinks the differential-mode inductor you need — but the standby tax used to cap how big you dared to go.

Remove the tax and the trade disappears. Designers can take a 1µF X-cap instead of 330nF, shrink the inductor, and hold the same EMI margin at lower cost — while the discharge IC burns under 5mW either way. The standby budget stops being the ceiling on filter performance.

For EU-market adapters and appliance boards, that's the quiet design win. Per the CAPZero-2 datasheet's operating range (100nF to 6µF), one part number covers essentially the whole filter space.

When the Tax Is Not Worth Collecting

What about the designs where the tax doesn't matter? The honest counter-case: a multi-kilowatt power supply's bleeder loss is a rounding error next to its main-stage losses. A 1MΩ resistor at 53mW against a 2kW load is 0.0026% — the IC's BOM cost doesn't earn its keep there.

And below ~100nF of X-capacitance, the discharge requirement is usually met by a small fixed resistor anyway. The CAPZero-class part pays off in the middle: regulated no-load budgets, meaningful X-caps, EU-market volumes. That's adapters, chargers, appliances, and displays — not server PSUs.

From our experience sourcing for EU-market power designs (2025–2026), the part lands where the compliance engineer sits in the room during the BOM review. If no-load power is on the test plan, the discharge resistor is on the table.

Frequently Asked Questions

Q1: Is the 53mW really worth chasing?

A: On a DOE Level VI adapter, yes — it's over half the 0.1W no-load budget. The controller and feedback might draw 40mW combined; the discharge resistor alone can decide pass or fail. Cutting it to under 5mW is the single cheapest compliance win on the board.

Q2: Why not just use a bigger resistor?

A: Because discharge time scales with R×C. A 330nF X-cap with a 10MΩ bleeder has a 3.3s time constant — likely outside the standard's discharge window. The practical ceiling is ~3MΩ for that cap, and it still burns ~17.6mW at 230VAC.

Q3: Does the discharge IC change the safety story?

A: No — the RC discharge path is unchanged when the plug comes out. The IC closes within ~30ms of AC removal and the resistor does exactly what it did before. The single-fault story is arguably cleaner: two independent pins per terminal mean a lifted pin can't open the path.

Q4: What voltage rails does CAPZero-2 work on?

A: The part is self-powered from the line and rated for a 1000V drain. The datasheet validates operation down to industrial 18-24VAC rails and covers X-caps from 100nF to 6µF. No external bias rail, no ground pin.

Q5: Is this only for adapters?

A: No — anything with an X-cap across the mains and a no-load requirement. Appliance control boards, displays, LED drivers, and industrial low-voltage rails all carry the same bleeder tax. The 53mW case is the adapter because that's where the regulation is tightest.

Q6: What does the retrofit look like for an existing design?

A: Two resistor slots become three parts. Split the existing bleeder into R1 and R2 and drop the CAPZero-2 IC between them — D1 (pins 2/3) on one side, D2 (pins 6/7) on the other. Total resistance stays in the 150kΩ to 7.5MΩ window, RC < 1s is preserved, and the 5V rail never needs to know. Layout change: one small SOIC-8 footprint and a trace reroute.