mimosa wrote:That's interesting, and makes sense up to a point. But what about a stick that used to work, and doesn't any more? My thinking is that a stray byte or two in a sound or text file may pass unnoticed, but not when it comes to booting an OS from the stick.
- A flash stick, which worked for a time and now is dead?
There are several reasons. The mechanical stress during insertion and removal may have caused a fatigue fracture on an important solder or strip conductor. This can be done in the vicinity of the plug or the memory controller. Corrosion of solder joints can also be the cause. For example, when unsuitable because permanently aggressive, soldering flux was used, the board would have to be very thoroughly cleaned. That will increase the price and therefore cleaning does not occur. You never know what dilapidated company carried out the soldering. Higher humidity then promotes the decomposition. Suitable soldering flux is passive at ordinary temperature. Boards must then not have to be cleaned. Also brand manufacturers produce faulty goods, especially when trying to reduce manufacturing costs. Remember to the scandal with bloated and leaking electrolytic capacitors on motherboards and in power supplies.
- A flash stick, which worked perfectly a time and now shows errors?
The above reasons also fit this, only that they are not final. A loose connection.
More likely is a memory controller with internal defect. In more modern sticks of average quality, the memory chips should be able to endure at least 100000 to 1 million write operations per memory cell. Of course, there are also early failures. Therefore, a defective memory chip is conceivable as the cause. A softened power supply can be a cause in question, too. Whether memory controller or memory chips are defective, this is testable.
On a 2 GiB flash stick taken together two to three ISO images would fit. If you write now an ISO image on the stick with "dd", then still about 1/2 to 2/3 of stick memory is free. Can one now boot that image and install it, everything is good. But when later can not be booted from this stick or the installation process dies or the installation performed is faulty, then perhaps memory cells have lost their content. The organization of the data flash memory controller may be damaged, also. A test may be as follows. One writes the same image to the stick with "dd" again - there is still enough unused space for an image file. The flash memory controller puts the image on the available unused memory. The place for the previous image is released, but not used again until all other unsued space was once occupied (wear leveling algorithms).
If the stick has a file system and are music, photo, video and movie files on it, then some faulty memory cells have little or no noticeable effects. For photo files, you realize it still best when image content changes or the viewer refuses representation. In movies you see mostly only error messages in log files that indicate faulty frames. That there could be such an incident, if played from a file on a storage device:
http://forum.salixos.org/viewtopic.php? ... 833#p39400. If it is only a file, you can continue to use the stick. To do this, renames the corrupted file and then copy the file to the stick again from an intact source. The kept broken file blocks the use of the underlying defective memory cells.
The power supply of the computer can be responsible for the misconduct of the sticks, too. The power supply may already become weak. The electrolytic capacitors in the power supply and on the mainboard can be progressed dried out (bad ripple values). A new graphics card in the computer, with higher power consumption may degrade the values of the power supply and the mainboard. All that deteriorates the function of the stick. A stick with deteriorated values of electrical charges in some memory cells may then deliver incorrect data. This is testable. Does the stick work on a new computer or on a power supply powered USB hub?
The program F3 - an alternative to h2testw - can test USB mass storage devices on Linux and Mac OS X to find errors and fakes. With h2testw one can afford it under Windows too.
F3 - an alternative to h2testw, by Michel Machado
http://oss.digirati.com.br/f3/
Something to read:
https://fixfakeflash.wordpress.com/2010 ... A0machado/
https://flashdrivefacts.wordpress.com/2 ... it-drives/
http://encyclopedia.thefreedictionary.com/bit+rot
For German language readers only:
http://www.heise.de/download/h2testw.ht ... ntare=alle