Flakey wifi router; oscilloscope screencaps of psu

[brew918]

I've got an wifi router that has become unreliable over time. I've attached screencaps of my DS1054Z probing the 12vDC power supply: is this normal?  Could these voltage spikes be causing stability problems?

New to ee, eevblog and oscilloscopes.

[tunk]

Difficult to say.
Is it measured with a load?
Don't think many (any?) routers run on 12V internally,
they will have internal DC-DC converter(s) that delivers
e.g. 3.3V etc. Some time ago I had to replace some
capacitors in a 5 port switch.

Edit:
Maybe you could open the router and post some photos?

[peteru]

Where are you measuring? On the DC connector? With load?

You'll probably get a better idea of whether this is a problem if you can find the first DC-DC converter in the router and examine the output of that.

[wraper]

IME often this happens because electrolytic capacitors in the router died. I don't think those pulses are nearly enough to hinder normal operation of router.

[brew918]

Okay I think this is the internal converted voltage ~3.3v. It's a test probe point right after a large inductor (says 150) close to the dc supply input.

Does this look healthy?  The second image is an alternate 12v switching power supply (wall wart) with no known problems.

[peteru]

Seems OK. In fact, the second power supply seems to have more noise/ripple than the first.

Have you tried completely reflashing the firmware. I've seen a number of embedded devices that end up with flipped bits in flash memory over the course of a few years. The usual fix is to take note of existing configuration (screenshots work well), restore factory defaults, reflash the firmware image and then restore factory defaults again. After that, reconfigure the device from scratch - do not restore backed up configuration.

[MathWizard]

That looks pretty normal to me, the spikes on the 12V from a standard 12V wallpack, such as a forward converter with probably a DIP-8 off-mains/line controller IC

[brew918]

I'll try the factory reset. The flipped bits theory makes sense.  These things are basically computers with operating systems running on super low end hardware, a lot more complex than microcontrollers.

[peteru]

A factory reset alone is unlikely to be sufficient as it will only erase the user settings most of the time. Any corrupted code and system data will stay as it is.

You need to get a hold of the firmware image, erase the flash and apply the new firmware. Think of this process as a format of the HDD and a reinstall of the operating system.

[wraper]

A factory reset alone is unlikely to be sufficient as it will only erase the user settings most of the time. Any corrupted code and system data will stay as it is.

Just update the firmware via normal process. Usually barely anything from the old data will be left (mentioned below).

You need to get a hold of the firmware image, erase the flash and apply the new firmware.

Bad idea since flash also contains serial number, model/hardware revision, MAC address and may contain WIFI modem calibration data.

[peteru]

You need to get a hold of the firmware image, erase the flash and apply the new firmware.

Bad idea since flash also contains serial number, model/hardware revision, MAC address and may contain WIFI modem calibration data.

Of course you would not erase the entire chip. Doing so on most consumer devices is tricky, even if you get access to the bootloader. We often go into a significant amount of effort to prevent end users from bricking their devices. Even the "factory mode" is often a one shot affair, where we erase the software that the factory uses during the manufacturing process as the device reaches the end of that process. Once that is done, the only way to alter the factory information/flash partitions is using an in-circuit flash programmer or similar "drastic" measures.

What I was talking about is the fact that in order to reprogram those flash partitions that contain the firmware image, each block of flash will typically be erased, tested, remapped if the erase fails, programmed, tested, remapped if the programming failed. Thus, reflashing firmware on a router will typically not only "re-install" the software, but also go through an erase and verification cycle to ensure that the new image is stored in good blocks.

[wraper]

What I was talking about is the fact that in order to reprogram those flash partitions that contain the firmware image, each block of flash will typically be erased, tested, remapped if the erase fails, programmed, tested, remapped if the programming failed. Thus, reflashing firmware on a router will typically not only "re-install" the software, but also go through an erase and verification cycle to ensure that the new image is stored in good blocks.

I don't think that any sort of remapping is possible on the vast majority of consumer routers since they use SPI NOR flash.

[peteru]

The devices that I work on have a bootloader that uses UBI. As long as the first couple of erase blocks where the bootloader lives are functional, the rest of the flash blocks can be remapped as required.