This powerbank pushes smartphone to reboot. Using oscilloscope to know what's going on

A nice little powerbank was handed to me one day - owner did not need it, and I thought it may serve as a always-in-pocket backup. Not that my phone needed this due to the fact it was modified to have three batteries embedded instead of one. But, as an electronic gadget fan, there was no chance to skip on this nifty little powerbank.

Everything was great until a phone was connected to it to get some charge. But instead the phone rebooted. Thinking that was some glitch, the phone was reconnected to the powerbank. Just to reboot again. Probably, that was the reason the previous owner wanted to get rid of this powerbank.

Rather than guessing, let`s use an oscilloscope and a programmable electronic load to solve this mystery and find out what is going on.

What a Clean Power Supply Looks Like

Let`s start with a known-good example. A high-quality charger or power bank should provide a flat line of stable DC voltage. No fluctuation, no ripple, no spikes, no dips - a straight horizontal line when viewed on an oscilloscope in DC mode.

In AC mode, which filters out the baseline DC voltage and highlights voltage noise, the trace should also remain perfectly flat with zero volts of noise. Any deviation suggests high-frequency noise.

In our control test with a reliable charger, the voltage remained steady. Ripple was exactly zero, as one should expect under no load. The device behavior remained predictable.

USB Specification Context

The official USB 2.0 specification Chapter 7 Electrical section 7.3.2 page 178 (https://www.usb.org/document-library/usb-20-specification) defines the allowable voltage range for VBUS

:

• High-power ports (basically, all ports existing today): 4.75 V to 5.25 V
• Low-power ports (these are non-existent today): 4.40 V to 5.25 V

Although the spec doesn`t mention the word "ripple", all ripple and noise must occur within those voltage boundaries. That gives us about +/-250 mV for high-power ports. Most quality designs aim for ripple under 5%, which is exactly 250 mV at 5V to avoid compatibility and stability issues.

Testing the Problem Power Bank

The problematic power bank didn't look suspicious at first. Standard USB-A port, single with probably quality enough 18650 cell inside, clean outer casing. No signs of damage or poor build quality. But the oscilloscope told a different story.

With no load connected, you would expect voltage output in a form of a flat line, as it was shown earlier. Instead, the oscilloscope shows voltage spikes. That`s already abnormal - there should be no DC fluctuation. Especially there should be no voltage spikes when no load is applied.

As the load increased, the ripple grew worse. At 0.2 to 0.3 A, the output started to show sinusoidal-like behavior, which is not what you want from a regulated DC supply.

At around 0.5 A, the waveform became even more unstable - spikes stacked on top of other spikes. The voltage was no longer predictable, and the oscilloscope trace clearly showed chaotic behavior.

This level of instability explains why a smartphone would reboot when connected. Power Management ICs (PMICs) inside phones are sensitive to supply voltage fluctuations, and large ripple can easily trip undervoltage protection or cause internal regulators to reset.

Measuring Noise in Oscilloscope AC Mode

Switching the oscilloscope to AC coupling allows us to isolate high frequency alternating voltage ripple by removing the DC component. On a good power source, the ripple signal remains flat or barely noticeable, as you’ve seen earlier.

On the problematic power bank, even at no load, the ripple was around 100 mV. As load increased to 0.3 A, ripple approached 200 mV. Once the load exceeded 0.8 A, ripple decreased - again highlighting the strange behavior of this particular unit.

Such ripple levels almost reached the informal industry threshold of 250 mV at 5V and clearly fall outside best practices for USB power sources.

Unexpected Behavior Under High Load

What made this power bank unusual is that its performance drastically improved as the load increased.

At around 65% of its output capacity, ripple dropped, and the waveform began to resemble a clean DC signal. This is the opposite of what most power supplies do - typically, ripple increases as the load rises.

This inverse behavior suggests a weak or poorly implemented feedback loop in the voltage regulator circuit or insufficient filtering. The design likely lacks adequate compensation and filtering at lower loads. When load increases, the control loop stabilizes somewhat, but this is corresponds to a short period of overall charging cycle.

Most of the time device will be charged with high ripple. Most mobile devices don’t draw high continuous high current during normal operation. After charging to about 80%, which takes a short amount of time, the device switches to CV mode with current dropping from max to zero. This stage takes several time longer than the initial charging to 80%.

During the CV stage, loads of 100…500 mA are typical, and that’s exactly the range where this power bank performs worst.

Analyzing Likely Cause: Weak Circuit Design

The root issue likely lies in a bare-minimum boost converter design:

• Poor or missing LC output filter
• Inadequate compensation in the feedback loop
• Inexpensive switching controller IC
• High ESR capacitors, or none at all
• Bad PCB layout and ground noise

No amount of casing quality or battery branding can fix poor schematic design.

Final Thoughts

This is a reminder that the quality of a USB power source isn’t about how it looks and feels - it’s about the behavior under load. If ripple is high and unpredictable, the power source is not safe for use with sensitive electronics.

If a power bank causes phones to reboot or glitch, the problem may not be the phone. Unstable voltage due to excessive ripple or poor design can trigger all kinds of system-level failures.

Before trusting any unknown power bank—especially a cheap or free one—run it through basic electrical tests. If the ripple exceeds 250 mV under moderate load, it’s not suitable for powering modern electronics.