[PROJECT] A Dozen Chargers in the Lab

By Sean Sullivan on 19 Jan 2026

A Dozen+ Chargers in the Lab: Apple is Okay, but Not the Best

all chargers — 12 of the ~two dozen we tested

Have you ever wondered if there are ‘good’ and ‘bad’ chargers? Is there really a difference between a $5 brick you get at an airport and a $30 one you order direct from a name brand? As it turns out, cheaply made chargers can operate far differently from name-brand ones, and can sometimes be dangerous to use.

This work was done first by one Google engineer Ken Shirriff many years ago. It was at the top of Hacker News at the time, and has been reposted a couple of times. Ken made the case that there is absolutely a difference between the best and worst chargers, and that poorly built chargers have the ability to not only degrade battery life but cause other strange issues like touchscreen glitches.

Fast forward six years and it still looks like there is no real authoritative source that tests big batches of chargers like this. My friend Spencer joined along, being a mechanical engineer, and we started bouncing ideas back and forth until this project eventually came to life. He helped complete a lot of the work below - thank you, Spencer!

Here’s the synopsis of everything below:

GaN chargers are among the best you can buy today, and the marketing around them is real.
Cheap chargers can also perform well, so the price you pay for a charger doesn’t always correlate to its performance.
Common brands like Apple generally do test favorably, but some smaller brands actually do better.
There is an extreme amount of variation even within the third party brands. Some third-party produce both high-quality and extremely low-quality chargers, and you’d have no way of knowing without testing them.

Ken Shirriff's original charger teardown article from 2012

What we tested

To get as much out of this project as possible, we tested a lot of chargers. Here’s the full list:

Anker ASPD053a USB C
Anker 637 Magnetic Charging Station (USB A port and USB C port)
AOHI GaN 65W AOC-C019
AOHI GaN 100W AOC-C018
Amazon Kindle 9W PS57CP
Dealworthy 20W PYSD-20PCQ-1CU
SHARGEEK 35W Retro S035U
Apple HomePod Mini A2305 (Same as Apple 20W Retail charger)
Apple 30W Retail A2164
Eero 15W C110011
Anker 100W A121B GaN Flagship
Anker Nano 45W A121D GaN Flagship
Spigen PowerArc ArcStation Pro 30W GaN PE2013UJ
Apple 96W Retail A2166
VEBNER 65W PD65U55
Govee 10W BI12T-050200-BdUU
Counterfeit AT&T 10W ‘NA010050020’
Counterfeit Chinese Wall Adapter 10.5W WP2UWR
Counterfeit Samsung 10W EP-TA20JBE

There was also a Macbook Pro M1, an iPhone 17 Pro, and an Anker 621 MAG GO 5000mAh Battery thrown into the mix. These aren’t chargers, so I’ll keep them out of the list above, but they are included in the various results below.

Each charger was tested on several dimensions: Advertised power versus actual power, power quality, and overcurrent.

Advertised power versus actual power tells you how well the charger can maintain its intended voltage while putting out maximum current.
Power quality means the amount of noise that appears when viewing the charger with an oscilloscope. This noise comes in many forms, but the general idea is that the flatter, the less noise.
Overcurrent shows how a charger behaves when a device tries to pull too much current. This is achieved by connecting a charger to an electronic load and having that load request more and more current until the device shuts off. It helps show how well-regulated a charger is, and if it is capable of protecting itself during load.

Testing Setup

Testing USB C chargers is a little complex. Modern chargers rely on the USB C Power Delivery (PD) standard, which generally specifies five voltages they can choose from: 5V, 9V, 12V, 15V, 20V. There are modern variants of this that complicate things even more, such as Programmable Power Supply (PPS) and Adjustable Voltage Supply (AVS), but I won’t get into those. For this project, we only looked at chargers according to the USB C PD spec. USB C PD also specifies that devices should be within +/- 5% of their advertised voltage in order to achieve certification (this is oversimplifying, but the general rule), so we followed that rule when rating them.

With this in mind, our setup consisted of a few different pieces:

An oscilloscope, specifically the Rigol DS1054Z
An electronic load
A USB C Power Delivery decoy board
A voltmeter
Plenty of electrical wires and extra USB cables

The process involved connecting the charger into a power source, and connecting the other end to a USB C Decoy Board, and connecting that decoy board to an electronic load. The oscilloscope probes tapped in right at the decoy board [A1].

The decoy board was needed because the USB C PD Spec needs the load device to choose a voltage. Otherwise the charger won’t send anything. In the chargers Ken tested, most often they’d just send 5V regardless, so this setup was a little more complicated.

A note about resistance

One of the things we learned during this project was that resistance could be artificially skewing the measurements. We did a couple of things to minimize this, so it shouldn’t be a big deal.

Note

To account for this, we ensured each test was conducted with my voltmeter directly connected to the charger's output terminal. If you're interested, the equivalent resistance of the voltmeter during testing was 6.67 mΩ, and the entire system (cable + breakout board + voltmeter) was 170 mΩ. Again, this resistance doesn't apply to any of the numerical measurements we have, since we connected directly to the charger output. However, we are honestly proud how minor this resistance is for a testing setup, considering USB C cables themselves are allowed to carry a resistance of up to 166 mΩ per the USB C PD spec.

Rating system for Oscilloscope graphs

Noise and Switching Ripple

After conducting the oscilloscope tests, we rated each charger from 1 to 5, using lightning bolts as a visual. These ratings consider two separate factors: overall high-frequency noise (Vpp/spikes) and switching ripple.

Noise includes the more chaotic high-frequency artifacts, large vertical spikes, and overall ‘fuzziness’ in the output. By measuring the overall peak-to-peak (Vpp) voltage, you can rank the true noise.
Switching Ripple refers to the high-frequency peaks (e.g., 50KHz, 100KHz) caused by the charger’s internal oscillator continually charging and discharging. This differs from Ken’s methodology, as he used the word Ripple specifically to mean the 120Hz line-frequency fluctuations caused by household AC power.
Most of the chargers look incredibly dense and ‘thick’ compared to the ones Ken examined. I believe this is because my scope has a memory depth of 24 million points, while the most I saw on Ken’s graphs were around ~50,000 equivalent points of memory depth. I believe the inclusion of this raw high-frequency data at every zoom level is the most honest way to display these chargers.

For noise, the bolt ratings loosely map to measured peak-to-peak voltage noise on the oscilloscope:

1 bolt: over ~250mV
2 bolts: under ~250mV
3 bolts: under ~160mV
4 bolts: under ~100mV
5 bolts: under ~80mV

These are just general guidelines. I used my judgment for each charger based on everything we could see, including the shape of the waveform and the FFT spectrum, not just the raw mV reading.

Old setup with oscilloscope — My original setup with the oscilloscope (before I started using USB C Decoy boards)

The Results

Scroll through the cards and click any one to expand the full oscilloscope graphs. Each card shows the worst graph that we could reproduce for that charger, assuming typical power usage.

Note

We learned during testing that all devices showed similar noise profiles at different voltages, so we stuck with lower voltages for these screenshots. Most screenshots below were taken when the charger was giving 9V.
Within each voltage, we cycled through each charger's range of allowable current, and usually selected the worst graph. If the charger showed noise at the extreme low or high end of current, but no issues for the vast majority of measurements, we'd take the less severe graph. This is because what you're seeing should approximate the noise your device would actually experience during charging, not some hypothetical rating you will never actually see during normal usage.

A few more details on noise and ripple

They look confusing at first glance, but you don’t need to be an electrical engineer to understand these next graphs.

Yellow Line (Voltage Trace): This represents the actual voltage being delivered to your device. The horizontal axis represents time, and the vertical axis represents voltage. In a perfect world, this would be a single, razor-thin, perfectly flat horizontal line. Any deviation from this is an imperfection.
Purple Line (FFT Spectrum): You’ll see this purple mountain range near the bottom of many graphs. This is a Fast Fourier Transform (FFT). It takes all the imperfections from the yellow line and sorts them by frequency. A large spike here would mean there is a significant amount of energy at that specific frequency.
Noise vs. Switching Ripple: This describes the imperfections on these graphs. It’s important to look at them separately:
- Switching Ripple is the presence of a main, high-frequency spike (and its harmonics) on the FFT. This is caused by the natural charging and discharging cycle of the charger’s internal circuit. AC Chargers will switch on and off many times per second to effectively create something that is DC current. For that reason, if there’s a significant peak at 90KHz and nowhere else, for instance, we’d say the charger has its primary switching ripple at 90KHz. If the charger has peaks at 90KHz and 180KHz, that is the switching ripple and its harmonic at 180KHz.

Note

Ken used the term ripple to mean Line Ripple, to describe the lower 120Hz fluctuations caused by the AC wall input; we use Switching Ripple to focus purely on the oscillator.

Noise includes any imperfections on the voltage graph measured in peak-to-peak voltage (Vpp). There are many different kinds of noise, including high-frequency vertical spikes, fuzzy blur, and general jitter. Switching ripple is one component of this.
‘Fuzzy’ Blur: If the yellow line looks incredibly thick or “fuzzy,” it means the charger is constantly shifting its switching speed back and forth (frequency dithering) to dynamically smear its noise.
Good vs. Bad: A high-quality charger produces a thin, flat yellow line with minimal action on the purple FFT graph (ideally just one sharp needle representing its switching ripple). A poor-quality charger will have a thick, fuzzy yellow line (lots of jitter/noise), obvious rolling waves (line ripple), and many aggressive, evenly spaced harmonic spikes on the purple FFT graph (severe switching ripple).

Oscilloscope Graphs

Macbook Pro M1

Overall rating: ⚡︎⚡︎⚡︎⚡︎⚡︎

The MacBook Pro had some of the cleanest power of the devices we tested. This is pretty unsurprising since it's just pulling DC current from its internal battery, and it doesn't have to chop up AC power like the chargers do. There's also no high frequency switching noise. This makes sense, since it doesn't have to switch thousands of times a second to step down AC voltage into something manageable.

AOHI GaN 65W Charger

Overall rating: ⚡︎⚡︎⚡︎⚡︎⚡︎

The best AC power adapter. Very low noise, very low ripple. The only noticeable peak in the FFT is the 125KHz switching frequency. I believe this charger has a new technology in them that allows it to use different switching frequencies depending on how much power is requested, so the switching frequency may change with time. But all in all, the best power supply we saw.

Anker 621 MAG GO 5000mAh Battery

Overall rating: ⚡︎⚡︎⚡︎⚡︎⚡︎

No switching noise or ripple seen here. The one small issue in testing was that it measured 4.74V, just barely under the 4.75V USB limit. The power quality itself is excellent so it stays at a high rating.

Spigen PowerArc ArcStation Pro

Overall rating: ⚡︎⚡︎⚡︎⚡︎⚡︎

We were originally skeptical of this one, but it turned out very well. This was marketed as one of the first GaN chargers, having been released sometime back in 2023. It's using a ~140KHz switching frequency, which seems to be the only source of switching ripple on the 50us graph. I zoomed out and there was a tiny amount of additional noise on the FFT, but not enough to be truly noticeable. The zoomed-out graph looks pretty similar to the one in this image, so I'm just showing the zoomed in one here.

Eero 15W Charger

Overall rating: ⚡︎⚡︎⚡︎⚡︎

For such a random charger, one that you can't even purchase yourself, it is surprisingly high quality. It comes included with Eero devices, and looks to be custom-built just for them. Not sure if Amazon/Eero bought this one off the shelf and rebranded it or it's their own circuit design. Some switching ripple exists, as you can see on the bottom graph, but it does a great job at filtering out most other noise. It also has a sawtooth waveform, so there is noise, but it is pretty small.

AOHI GaN 100W Charger

Overall rating: ⚡︎⚡︎⚡︎⚡︎⚡︎

This one performed very well. From our research, this is a GaN charger that uses some components that are so new they have never been used in consumer chargers before. This charger should really be considered a flagship for how much effort they put into it. It's a little disappointing that the power signature isn't better. It is good, but for how bleeding-edge this charger is supposed to be, it performs comparably to chargers half its size and price. The really cool thing about this charger is that it supports PPS and has the ability to step the voltage in small increments, when needed by the device. We didn't test this, but it does have that capability. One truly interesting thing about this charger was how it managed its voltage. As the current approached the 2-3A range, the voltage actually increased, topping out at 12.36V when I pulled 3A. This is the first charger I tested that dynamically increased its voltage to account for resistance in the charging system.

Anker 637 Magnetic Charging Station 2 - USB C Port

Overall rating: ⚡︎⚡︎⚡︎⚡︎⚡︎

This is a ball-shaped charging station that offers USB A, USB C, and wireless charging outputs. While about 4 years old, it delivers quite clean power. The voltage trace is fairly tight with only minor noise, so nothing alarming. As you can tell, it's only the test of the USB C port. We didn't bother testing the noise of the USB A port since they sit on the same circuit board (so likely share the same voltage regulation), and also most people will not be using the USB A port on this. One other interesting note about this charger: During overcurrent testing, it happily provided far above its rated current. It's probably because the charger supports 1) 5v 3a, 2) 9v 3a, 3) 12v 3a, 4) 15v 3a, 5) 20v 3.25a As you can see, it will actually go above 3 A at the highest voltage. They probably just didn't think anyone would actually use over 3 A at the lower voltages.

Counterfeit Chinese Wall Charger (USB A Only)

Overall rating: ⚡︎

This was another charger that tested surprisingly well. It cost less than $10 and has very low high-frequency noise and ripple. There is a bit of high-frequency noise in the 50us graph, but the voltage is pretty stable. The only issue is that it was 0.5V below the 5V threshold during testing, which puts it out of spec for a USB Charger. It gets 1 bolt because of this.

Apple 96W Macbook Charger

Overall rating: ⚡︎⚡︎⚡︎

Good, but should be better for how expensive this charger is. The total noise is low at 92mV, but there is some noise visible within those intermittent peaks when you go into the lower timeframe. There's also high-frequency ripple with plenty of harmonics. It also delivered less voltage than advertised, going down to 8.63V when we'd try to pull 3A @ 9V. Overall rating is a 3 because of that high-frequency noise and the voltage sag. Not bad but should be better for $100.

Apple 30W Retail Charger

Overall rating: ⚡︎⚡︎⚡︎⚡︎

Another decent Apple charger. At 5 W (5V, 1A), it has a pretty messy power signature with a lot of noise, but the voltage graph flattens out to this nice, smooth line at 7.5 W and above. That probably happens because nobody at Apple expected people to use this charger for 5 W of power, being USB C. We also tested this 7.5 W in different forms: 5V 1.5A, and also 15V 0.5A. In both 15V and 5V modes, it needed to achieve a grand total of 7.5W before the noise graph smoothed out. so 15V ~0.5A, and 5V ~1.5A. I displayed the 7.5 W+ graph because in normal usage you will almost never encounter that noise, since Apple devices almost always pull more than that during normal charging.

Anker ASPD053a

Overall rating: ⚡︎⚡︎⚡︎

Not that impressive. The voltage is noticeably noisier than a lot of the other chargers. The frequency spectrum also shows ripple with harmonics at roughly 90KHz intervals.

iPhone 17 pro

Overall rating: ⚡︎⚡︎⚡︎

We gave it 3 bolts to be nice, there's nothing great about using the iPhone 17's USB C port as a charger.

Shargeek Macintosh Charger

Overall rating: ⚡︎⚡︎⚡︎

For being nearly $30, and only rated for 27W, this charger isn't doing that well. It looks well-regulated in the sense that the waveform is repeatable and standard, but the peaks are so jagged that it is delivering a lot of noise.

Govee Power Supply

Overall rating: ⚡︎⚡︎

As expected, not great. This Govee gets two bolts since it delivered roughly what it advertised in terms of voltage and current, but it looks noisy.

Dealworthy Charger

Overall rating: ⚡︎

Dealworthy is a Target brand, and it's designed to be low cost. You can see just how cheaply it's been made by these graphs. Lots of noise, poor regulation, and a lot of switching ripple with harmonics all over the FFT. Another interesting thing is that they appear to be smearing the switching frequency by doing some kind of frequency dithering. It's operating at 40KHz, but we suspect it is going slightly above and below that frequency to spread out the massive amount of noise it makes.

Counterfeit AT&T

Overall rating: ⚡︎

Bad. A quick glance on the 5ms timeframe and the voltage almost looks normal, but zooming in reveals a waveform dense with high-frequency noise and severe peaks. There is also meaningful ripple in the lower timeframe. The voltage also measured 5.38V, so it is technically out of the acceptable range for a USB charger. 1 lightning bolt.

Anker A121B 100W Charger

Overall rating: ⚡︎⚡︎

This is a "GaN" flagship charger released by anker sometime in late 2025. It's better than its 45W cousin (the Anker A121D), but this charger isn't that great. It does push out the advertised voltage and current, but it does so with a wall of noise. Zooming in to the smaller timeframe shows voltage peaks every few microseconds. What a strange design from Anker, who claim this to be a GaN charger. It is clearly not made well.

Amazon Kindle Charger

Overall rating: ⚡︎

Another awful charger. Considering this comes as the standard Kindle charger, it is a poor showing from Amazon. The voltage is wide and noisy, with a chaotic high-frequency spectrum. It also measured 5.30V, well above the 5.25V USB ceiling and therefore out of spec for a USB charger. If the graphs weren't bad enough, the bad voltage would give it 1 bolt.

Apple HomePod Mini Charger

Overall rating: ⚡︎

Disappointing, especially given that this ships directly from Apple. The voltage is thick and irregular, and shows signs of both low-frequency and high-frequency ripple at a glance. On the bottom of the screen, the frequency spectrum shows elevated noise across the board. Its noise is at 308mV peak-to-peak and it ranks among the worst chargers in this entire test. This test was surprising. We also happened to have two of these chargers, one from a HomePod and one from an iPhone of years past. This charger is actually the same as the Apple 20W [you can buy directly from their website](https://www.apple.com/shop/product/mwvv3am/a/20w-usb-c-power-adapter). And while you could, I have no idea why you would want to. $20 can get you a much better charger than this.

Vebner GaN 65W Charger

Overall rating: ⚡︎

Being marketed as a GaN charger, this is another disappointment. The voltage has a lot of noise, 332mV of it, and the frequency spectrum is elevated across the board. The voltage looks a little more concentrated than it does on the Apple HomePod Mini chart, suggesting to me that this charger might actually be a little better. But in terms of absolute noise, it's worse.

Anker A121D 45W iPhone 17 charger

Overall rating: ⚡︎

Is this even GaN??? Easily the most disappointing charger of the bunch. Anker just released this (late 2025) and couldn't help themselves from marketing just how great it was. The only good thing about this charger is that it outputs its advertised voltage and current. Aside from that, it's awful. Noise everywhere, and constant high-voltage spikes that you can see in the lower timeframe. Also, at that lower timeframe, it's unclear what the charger is actually doing. It seems to be constantly sending out various voltages, somewhat resembling a sawtooth but also just resembling noise. Sidenote: you might point out that earlier we mentioned the AOHI GaN chargers have the ability to change their switching frequency as needed. You may be saying, how is that any different from what you are saying Anker did here? It's a valid question. The difference is that Anker clearly had too much noise at the switching frequency of the charger, at ~90KHz, and they needed to spread that noise out. So they employ a variable switching frequency but only vary it by 20-50KHz. Constantly moving the switching frequency up and down, so the noise doesn't make any one peak so high that it fails regulatory testing. On the other hand, AOHI has no noise to begin with, so when they change the switching frequency, it's only happening once (I suspect only when the requested voltage changes). They don't need to constantly switch up and down, because their aim is not to spread the noise out.

Counterfeit Samsung

Overall rating: ⚡︎

The worst one of the bunch, but that would be expected with a counterfeit charger. The original researcher (Ken Shirriff) doubled his scale to show the graph of his worst charger, but we wanted to evaluate all the chargers on the same scale, so you can see how bad some of these really are. On this image, it's tough to figure out what this charger is even trying to do. It provided 4.4V when 5V was requested, so it doesn't even pass USB spec. Another issue was overcurrent. This charger allowed the current to rise nearly 1A past its rated upper limit. This 10W charger was pumping out nearly 14W before it finally shut off. None of the other chargers did this.

←← BETTER WORSE →→

Oscilloscope Graphs - Results Table

The DC items in here, like the Macbook Pro, are colored in grey to avoid conflating them with the AC chargers. They’re just included here for reference.

Charger	Noise in mV	Noise Rating (Vpp and Spikes)	Switching Ripple Rating (FFT Harmonics)
Macbook Pro M1	24mV	⚡︎⚡︎⚡︎⚡︎⚡︎	⚡︎⚡︎⚡︎⚡︎⚡︎
Anker 621 MAG GO 5000mAh Battery	36mV	⚡︎⚡︎⚡︎⚡︎⚡︎	⚡︎⚡︎⚡︎⚡︎⚡︎
Spigen PowerArc ArcStation Pro	56mV	⚡︎⚡︎⚡︎⚡︎⚡︎	⚡︎⚡︎⚡︎⚡︎⚡︎
AOHI GaN 65W Charger	60mV	⚡︎⚡︎⚡︎⚡︎⚡︎	⚡︎⚡︎⚡︎⚡︎⚡︎
Eero 15W Charger	60mV	⚡︎⚡︎⚡︎⚡︎⚡︎	⚡︎⚡︎⚡︎
AOHI GaN 100W Charger	72mV	⚡︎⚡︎⚡︎⚡︎	⚡︎⚡︎⚡︎⚡︎⚡︎
Anker 637 Magnetic Charging Station 2 - USB C	76mV	⚡︎⚡︎⚡︎⚡︎⚡︎	⚡︎⚡︎⚡︎⚡︎
Counterfeit Chinese Wall ChargerUSB A Only	84mV	⚡︎⚡︎⚡︎⚡︎	⚡︎⚡︎⚡︎
Apple 96W Macbook Charger	92mV	⚡︎⚡︎⚡︎⚡︎	⚡︎⚡︎⚡︎⚡︎
Apple 30W Retail Charger	96mV	⚡︎⚡︎⚡︎	⚡︎⚡︎⚡︎
Anker ASPD053a	128mV	⚡︎⚡︎⚡︎	⚡︎⚡︎⚡︎
iPhone 17 pro	136mV	⚡︎⚡︎⚡︎	N/A
Shargeek Macintosh Charger	152mV	⚡︎⚡︎⚡︎	⚡︎⚡︎⚡︎
Govee Power Supply	164mV	⚡︎⚡︎	⚡︎⚡︎
Anker A121B 100W Charger	200mV	⚡︎⚡︎	⚡︎⚡︎
Dealworthy Charger	212mV	⚡︎⚡︎	⚡︎⚡︎
Anker A121D 45W iPhone 17 charger	244mV	⚡︎	⚡︎⚡︎
Counterfeit AT&T	260mV	⚡︎	⚡︎⚡︎
Amazon Kindle Charger	288mV	⚡︎	⚡︎⚡︎
Apple HomePod Mini Charger	308mV	⚡︎	⚡︎⚡︎
Vebner GaN 65W Charger	332mV	⚡︎	⚡︎
Counterfeit Samsung	468mV	⚡︎	⚡︎

Expected/Actual Output

Note

Our setup was not robust enough to handle testing at the full 65W or 100W that certain chargers are rated for. That is why we tested each charger at one of its max current/voltage configurations. Say Charger A is called a "65W" charger on the box. Charger A supports:

20V, up to 3.25 A (to reach max power of 65W)

And it also supports:

5V, up to 3 A
9V, up to 3 A
12V, up to 3 A
15V, up to 3 A

In this case, we'd do base level testing to see if we can get the charger to output its max power for a second or two. Then, for the oscilloscope traces, we'd drop the device down to something like 9V and cycle through all its current levels. This avoids overheating our breakout board. As stated earlier, all chargers showed similar noise profiles at their different voltages, so we're confident that the results you see below are reliable.

The Actual Voltage column is highlighted green if the measured voltage is within ±5% of the nominal output voltage (the USB-C PD spec limit) and red if it falls outside that window in either direction. For example, a 5V output must stay between 4.75V and 5.25V.

Expected/Actual Output Table

The DC items in here, like the Macbook Pro, are colored in grey to avoid conflating them with the AC chargers. They’re just included here for reference.

Charger	Attempted To Pull (W)	Actual Voltage (V)	Actual Current (A)	Actual Power (W)	Passes Spec?
Eero 15W Charger	15 (5v, 3a)	4.76	2.91	13.85	✓
Shargeek Macintosh Charger	27 (9v, 3a)	8.9	3.0	27.0	✓
Apple 30W Retail Charger	30 (20v, 1.5a)	20.04	1.42	28.45	✓
Apple HomePod Mini Charger	19.98 (9v 2.22a)	8.90	2.23	19.85	✓
Vebner GaN 65W Charger	27 (9v, 3a)	8.64	3.04	26.26	✓
Govee Power SupplyUSB A Only	10 (5v, 2a)	5.04	2.02	10.18	✓
Counterfeit AT&TUSB A Only	10 (5v, 2a)	5.38	2.01	10.81	✗
Counterfeit Samsung	10 (5v, 2a)	4.4	2.0	8.9	✗
Amazon Kindle Charger	9.36 (5.2v, 1.8a)	5.30	1.81	9.59	✗
Counterfeit Chinese Wall ChargerUSB A Only	10.5 (5v, 2.1a)	4.5	2.13	9.6	✗
Dealworthy Charger	20 (12v, 1.67a)	11.97	1.57	18.79	✓
Anker 621 MAG GO 5000mAh Battery	12	4.74	2.43	11.54	✗
iPhone 17 pro	4.5	4.81	1.12	5.39	✓
Macbook Pro M1	15 (5v, 3a)	5.05	3.03	15.30	✓
Anker ASPD053a	27 (9v, 3a)	8.88	3.02	26.82	✓
Apple 96W Macbook Charger	45 (15v, 3a)	15.08	3.02	45.54	✓
AOHI 100W GaN Charger	36 (12v, 3a)	11.82	3.02	35.52	✓
Anker 637 Magnetic Charging Station 1 - USB A portUSB A	12 (5v, 2.4a)	5.13	2.42	12.41	✓
Anker 637 Magnetic Charging Station 2 - USB C port	36 (12v, 3a)	11.79	3.02	35.65	✓
Spigen PowerArc ArcStation Pro	27 (9v, 3a)	8.91	3.02	26.91	✓
Anker A121D 45W iPhone 17 Charger	45 (15v, 3a)	14.82	3.03	44.90	✓
Anker A121B 100W Charger	45 (15v, 3a)	14.99	3.02	45.26	✓
AOHI GaN 65W Charger	36 (12v, 3a)	12.13	3.02	36.63	✓

Overcurrent Testing

The last piece of Ken Shirriff’s testing we wanted to replicate was overcurrent. Naturally, since devices get to decide how much current they want to pull, it’s up to the charger to make sure the device doesn’t pull too much current. Chargers generally have two strategies to deal with this: 1) shut off completely, or 2) reduce the output voltage. Option 1 is preferred in most cases. Option 2 is really just a consequence of not doing option 1. There is a third scenario, where the charger can somehow increase the current at the same voltage, but this would probably mean overheating it.

Note

During overcurrent testing, we set the electronic load to keep pulling more and more current until the charger shut off. While Ken was able to display the relationship between voltage and overcurrent on beautiful time-based graphs, we lack that recording capability. So our results are very simple.

In case you're interested, all my voltage graphs were relatively tight during overcurrent testing, and we only had one charger (Counterfeit Samsung) that completely blew the test. The others did a great job of keeping their voltage within a reasonable range, even during overcurrent. For that reason, we would say the values below are the true values of the voltage and current of the device at their shutoff period.

Ken current — A good example of Ken's overcurrent graphs. I wasn't able to recreate his setup like this.

Ken current 2 — An example of a bad charger Ken tested. While I wasn't able to capture graphs like this and save them, what I can say is that my graphs all looked like the earlier one, not this one (with the exception of counterfeit samsung)

The ideal value here is having a voltage equal to 5V or 9V. You don’t want 3 or 4 volts, or 7 or 8 volts, because that suggests the charger allowed more current out than it should have.

We don’t have overcurrent results for all chargers in the table below. This will be updated as more test results become available.

Overcurrent Results Table

Adapter	Difference between advertised Current and Current at shutoff	Difference between advertised Voltage and Voltage at shutoff	Difference between advertised Power and Power at Shutoff (W)
Vebner GaN	0.42	0.13	2.5
AOHI GaN 100W	0.6	0.13	3.5
Anker A121B 100W GaN Charger	0.35	0.12	2.2
Anker A121D 45W iPhone 17 Charger	0.44	-0.09	1.9
Anker 637 Magnetic Charging Station 2 - USB C port	0.7	-0.22	7.6
Apple HomePod Mini (20W)	0.42	-0.11	1.7
Apple 30W Retail	0.17	-0.14	0.4
Apple 96W	0.46	-0.37	1.1
Amazon Kindle	0.1	-0.4	-0.2
Spigen PowerArc ArcStation	0.22	-0.27	0.2
AOHI GaN 65W Charger	0.33	-0.3	0.7
Eero 15W Charger	0.36	-0.31	0.8
Counterfeit Chinese Wall Charger	0.23	-0.57	-0.2
Counterfeit Samsung	0.81	-1.2	0.64

A mixed bag of results from overcurrent. The Kindle is the most disappointing one in my opinion, because it shut off so quickly at just 0.1 A above its rated current, and somehow its voltage still dropped. However, the worst offenders are from the Apple 96W, Spigen PowerArc, Eero, and the counterfeit Chinese wall charger. All had voltage drops that suggest they go below what would be considered USB standards.

A new update: I re-ran the tests one more time with a brand new multimeter and updated the table. Upon reviewing, the Spigen, Apple 96W, Counterfeit Chinese wall charger, AOHI GaN 65W, and Amazon Kindle, and Eero are all starting to break past the +/- 5% voltage threshold for USB certification. I re-tested them under the oscilloscope, and found some interesting results: while most of these chargers offered relatively focused voltage traces, usually ranging by about 0.1V, the counterfeit Samsung was the one outlier. This charger had such a noisy voltage trace that, near shutoff, it oscillated between 3.7 V, 0, and 8.5 V. This dangerous behavior was truly unique and not present in *any of the other chargers I tested. When your phone or laptop attempts to pull too much current, this charger will start sending massive shockwaves of voltage.*

—

Conclusion

Wow, that was a lot. Evaluating chargers seemed quite difficult at first, but this whole project quickly turned into something exciting. What I liked most about this is that you don’t need to be an electrical engineer to see the difference between good and bad chargers. The oscilloscope really does make it obvious. Things get confusing when you go past the ‘what’ and try to start understanding the why. Why certain peaks appear in the FFT, why some scope traces look as bad as they do, or why a charger measured fine at 1A but looked terrible at 2.5A. I don’t think Spencer or myself have great answers to these.

The practical takeaway, in my opinion, is that most name-brand chargers of today are unlikely to damage your phone battery. Since modern devices filter noise internally, and the majority of these chargers stayed within that USB spec of ±5% voltage tolerance, they’re unlikely to be a problem. Just avoid excessively cheap, no-name chargers and you should be fine.

I hope you enjoyed reading. This really was a pleasure to put together. Car chargers are next on my list, and I suspect that may be a story of its own. If you have any suggestions to make, corrections to add, or just want to share something you found interesting, feel free to reach out and let me know. Contact is on the About page.

-Sean

Last thing: a huge shout out to both Ken and Spencer! I could not have done this without either of them.

Sean Sullivan
A Dozen Chargers in the lab

Appendix

[A1] Tip: If you want to do testing of your own, always use the spring ground clip. Using the default alligator ground clip will lead to extremely high noise ratings.

#technology