Ripple and power quality
What the small wobble on a charger's output is, how it is measured properly, and what it does — and doesn't — tell you
Ripple is the small, rapid variation left on a charger's DC output after conversion from mains power, usually quoted in millivolts peak-to-peak (mVp-p). Lower ripple at a stated load and bandwidth is a sign of tight regulation. It is not a battery-health figure: your iPhone's own charging circuit conditions incoming power before it reaches the battery, and research on lithium-ion cells has found little or no added ageing from high-frequency ripple. A ripple number means little without its test conditions.
In 30 seconds
- Every switching charger has some ripple. It comes from chopping mains power at high speed and smoothing it. "Zero ripple" is not a physical possibility for this kind of supply. 1
- Ripple is measured in millivolts, typically tens of millivolts on a 5–9 volt output. Apple's 20W adapter measured 15–47.2 mVp-p across loads in ChargerLAB's 2023 test. 2
- The number depends on how it's measured — bandwidth, probe technique, where on the cable, what load. Two figures from different benches are not comparable. 314
- It is not a battery-life figure. No credible evidence shows ripple within USB Power Delivery tolerances shortens phone battery life; heat and long periods at full charge matter more. 56
- ERONIX will publish Fermi20's ripple at every power level and load, with method, when lab testing is complete.
In three minutes
What ripple is
A USB-C charger turns 230V alternating current into a few volts of direct current. Inside, it rectifies the mains, stores energy, then switches it through a small transformer tens or hundreds of thousands of times a second, and smooths the result with capacitors. The smoothing is never perfect. What remains on the DC output is ripple: a small repeating variation, plus sharp switching spikes at the moments the circuit switches. 1
Two kinds matter:
- Switching ripple repeats at the converter's switching frequency (tens to hundreds of kilohertz). It is small and fast.
- Low-frequency ripple follows the mains (100 Hz after rectification of a 50 Hz supply) and shows how well the charger holds its output steady against the dips in its own internal supply.
Two ways to report it
- Peak-to-peak (mVp-p) — the gap between the highest and lowest points of the waveform. It captures spikes and is the most commonly quoted figure.
- RMS (root mean square) — an average of the variation's energy. It is lower than peak-to-peak and less sensitive to brief spikes.
A good report gives both, and says which.
Why the method changes the answer
An oscilloscope will happily "see" noise that isn't really on the output: radio pickup on a long ground lead, high-frequency energy beyond what matters, or voltage drop in the test cable. Power-supply makers therefore follow a common practice 314:
- Limit bandwidth to 20 MHz on the oscilloscope, so radiated pickup doesn't inflate the reading.
- AC-couple the input, so only the variation is displayed, not the 5 or 9 volts underneath.
- Use tip-and-barrel probing — the probe tip on the output and its ground barrel pressed straight onto ground, no long ground clip.
- Place a 0.1 µF ceramic capacitor in parallel with a 10 µF capacitor across the output at the measurement point, representing the input capacitance of a real device.
- Measure at the cable end, under a stated electronic load, at each voltage the charger offers.
Skip any of these and the reading can be several times higher or lower. That is why ERONIX does not place its numbers on the same chart as figures from another lab. A comparison is valid only on one bench, with one method.
What USB Power Delivery allows
USB Power Delivery specifies a tolerance band around each nominal voltage. STMicroelectronics' application notes tabulate this as the maximum ripple allowed: ±250 mV at 5 V and ±450 mV at 9 V — that is ±5% of nominal. 7 We found no separate peak-to-peak ripple limit for fixed-voltage sources in primary specification text; treat the ±5% band as the outer envelope that output voltage, including ripple, must stay within. Well-designed chargers sit far inside it — Apple's 20W measured 47.2 mVp-p at its worst point, about a tenth of the 9 V allowance. 27
A reference: Apple's 20W adapter
ChargerLAB measured an Apple 20W USB-C Power Adapter, model A2940 (made by Salcomp), in September 2023 2:
| Output condition | Ripple (mVp-p) |
|---|---|
| No load | 15–18 |
| 5 V, 3 A | 28 and 34.4 (at the two AC inputs tested; the source does not label which is which) |
| 9 V, 2.22 A | 47.2 and 36.8 (same caveat) |
Single unit, ChargerLAB's method; bandwidth and probing details are not published in the text. ChargerLAB rated the adapter "the one you can't go wrong with". These figures describe that unit on that bench and are shown for context only; they are not compared with any ERONIX figure.
Does ripple affect my battery?
The cell inside your iPhone does not see the charger's output directly. Power passes first through the iPhone's own charging stage — a regulating circuit with its own input capacitance and switching frequency — which conditions what reaches the battery. 89 Research has tested the question directly on lithium-ion cells: a 2019 peer-reviewed study comparing cells aged with converter-induced high-frequency ripple against conventionally tested cells found "little to no further influence of current ripple"; a 2022 dissertation from the same group concluded that a severe impact of ripple on battery reliability is unlikely and that cycle depth mattered far more. 56
What does age a battery is well documented: heat, and time spent at high charge. That is why iPhone slows charging past 80% and offers an 80% Charge Limit. 1011
So ripple is a fair measure of how well a charger is engineered. It is not a reason to fear for your battery, and ERONIX does not sell it as one.
For engineers
E1. Ripple versus switching transients
Analog Devices distinguishes ripple — the residual AC component coherent with the switching frequency — from transients, the high-frequency ringing at switching edges. Transients carry little energy, and their peak-to-peak amplitude depends heavily on setup and layout of the measurement; a 20 MHz limit removes most of the high-frequency content, and the band-limited time-domain result agrees with frequency-domain measurement. 1
E2. Recommended measurement specification (ERONIX)
| Parameter | Setting |
|---|---|
| Measurement point | USB-C plug end of a defined reference cable (type, length and resistance recorded) |
| Termination | 0.1 µF X7R ceramic ∥ 10 µF low-ESR at the probe point 34 |
| Probe | 1× or low-attenuation passive probe, tip-and-barrel, or 50 Ω coax with DC block 1 |
| Oscilloscope | 20 MHz bandwidth limit; AC coupling; ≥ 10 switching cycles plus ≥ 2 mains half-cycles per capture |
| Load | Calibrated electronic load, constant-current mode, 0/10/25/50/75/100% of each PDO's maximum current |
| Rails | 5 V and 9 V, plus every additional rail or PPS point the unit advertises |
| AC input | 230 V 50 Hz (India), 100 V 50 Hz and 115 V 60 Hz (low-line), 264 V 50 Hz (high-line) |
| Thermal state | Cold (after ≥ 2 h unpowered at ambient) and warm (after 60 min at full load) |
| Reported | mVp-p and mV RMS per point; sample count; worst case across 3 samples; instrument models and calibration dates; uncertainty |
| Optional | FFT of the captured trace to separate switching-frequency and 100 Hz components |
Full procedure: 19_LAB_TEST_PROTOCOL.md §5.
E3. Interpreting other labs' figures
lygte-info.dk, for example, reports both RMS and peak-to-peak under its own fixture and full bandwidth — its 30W test of a UGREEN CD127 reported 437–684 mVp-p alongside 29–48 mV RMS, while a Baseus BS-EU905 showed 90–216 mVp-p. 1213 These are different products, different power classes and a different method from ChargerLAB's; the spread shows why method must travel with every number. They are cited here only to make that point.
E4. What lower ripple can indicate
Lower ripple at a given load and bandwidth generally reflects adequate output filtering (capacitance and ESR), a stable control loop, and good layout. It says nothing on its own about isolation, protection behaviour, thermal design, transient response or long-term capacitor health. ERONIX treats ripple as one row in a quality profile alongside efficiency, no-load draw, transient response, temperature and protection tests.
Myths
"A good charger has zero ripple." Every switching supply has some. A specific, low, conditioned number is the honest claim.
"Ripple damages iPhone batteries." The evidence doesn't support this for compliant chargers; the phone's charging stage sits in between. 56
"Lower ripple means faster charging." Charging speed depends on the power level agreed and the iPhone's demand, not on ripple.
"Any two ripple figures can be compared." Only if measured with the same bandwidth, probing, load and cable. 31
Questions people ask
What is a safe ripple level for a USB-C charger? USB Power Delivery's tolerance band is about ±5% of nominal voltage — ±250 mV at 5 V and ±450 mV at 9 V, as tabulated by STMicroelectronics. Well-designed chargers typically measure tens of millivolts, far inside it. 72
Can ripple cause touchscreen problems while charging? Touch glitches while charging are usually linked to electrical noise reaching the device through poorly designed or counterfeit chargers. We have not found published evidence quantifying this for compliant chargers, so we make no claim about it. If touch behaves oddly on one charger and not another, stop using that charger.
How can I check my charger's output at home? A USB-C power meter shows voltage, current and the power levels on offer, which is useful for spotting a charger that won't negotiate 9 V. Measuring ripple properly needs an oscilloscope and the method above; inexpensive meters cannot show it.
Will ERONIX publish its ripple figures? Yes — at each voltage and load, with instrument, bandwidth, cable and sample count, on the test report page once complete.