The notion that placing the device on the shelf for a week does some magical thing to it is junk science, unless your shelf happens to be in the beam path of an X-ray, gamma ray, or other ionizing radiation source that is causing physical changes to the integrated circuits inside the device.
The OBi 1x0 devices consist of the following subsystems, which communicate with each other via an industry-standard SPI (serial) interface:
- The custom System On a Chip (SOC), which contains the CPU and several I/O bus interfaces, which communicate with
- An Etron SPI CMOS synchronous DRAM module/memory controller
- A Winbond SPI serial flash memory module/controller, which has its own, internal DC-DC voltage regulator to power flash memory write operations, etc.
- A Silicon Labs ProSLIC FXS module, which also has its own, internal power supply to support ring, off-hook and on-hook voltage generation
- The 110 also has a Silicon Labs FXO module, to support the POTS interface
The input from the external power supply (wall wart) is connected to a typical integrated circuit-based DC-DC regulated step-down power supply subsystem, to produce the 3.3VDC power used by the chips listed above. Yes, there are a few electrolytic capacitors on the board. These serve as a filter cap for the main power supply, and some small smoothing caps for the other chips. No, these capacitors don't hold a charge for weeks; they are quickly drained by the attached components when source power is removed, and no, unlike a PC, there is no large, main power supply creating the multiple different voltages used by the flash and SLIC subsystems.
There was a well-documented "capacitor rot" issue during the 1990s, which was caused by a improperly-formulated, stolen electrolyte recipe used by some unscrupulous manufacturers. Proprietary additives in the water-based electrolyte were omitted, resulting in corrosion. That problem impacted millions of manufactured circuit boards, such as PC motherboards, resulting in product failures and recalls. At the time, I managed a huge PC repair contract, for tens of thousands of Dell and other PCs, which contained the bad caps, which resulted in thousands of motherboard replacements. That problem was eliminated after the counterfeit electrolyte was discovered, and this was years before any OBi devices were manufactured. In that mess, it wasn't Dell's fault; it was their motherboard vendor's purchase of the bad caps.
In general, however, all electrolytic capacitors have a limited, finite lifespan, and they may degrade after 5-10 years of use, depending on operating environment and materials used. It's to be expected that all electronic products, from all manufacturers, containing electrolytic caps, will eventually experience errors. It's one reason you now see PC motherboard manufacturers advertising the high quality of their
solid (not electrolytic) capacitors.
If you bought an OBi 110 years ago, and it's now 2016, it's conceivable that its electrolytic caps could be degraded, but there will be no visible damage as there was to the defective caps of the 1990s, and the cap would need to be electrically-tested to determine if it was bad -- not something worth doing on a low-cost CE device.
Flash memory devices also have finite lifespans. Modern SSDs, for example, have built-in firmware to manage read and write errors, and to identify and flag bad cells, and use overprovisioned (spare) cells instead. The much-less-sophisticated SPI flash module in the OBi products isn't as robust in its ability to be "self-healing".
It's possible that a device could experience one or more flash memory write errors, and then, eventually, be able to recover and write successfully. Eventually, that module will fail again, just like a SSD that starts being flagged by SMART will eventually die.
The fact that a device "has been operating for X years since the failure" doesn't mean anything, since the flash memory module is not continuously writing its entire capacity, over and over in that period. Writes only occur when you do a firmware update (many writes to many blocks of cells), or update the device's settings (a few writes to some blocks of cells), or the OBi logs to its call history file (a few writes at a time).
Bottom line: yes, a corrupted firmware image would cause the green/red behavior, and yes, it may be possible to recover from it, but a device that repeatedly fails with firmware image X, and then magically works with a rewrite of that same, exact image a week later, will likely fail again at some future date, due to degraded hardware components. It's up to you, the user, to decide if you'd rather wait for a hard-failure, or to preemptively replace the product.
It would be nice if Obihai published MD5 values for its firmware images, but the average Joe User is barely able to figure out how to perform a manual update, let alone be tech-oriented enough to validate the file's MD5.
