Weird problem I ran into on this 16 GB flash disk. I was using it to transfer large 4+ GB multipart rar archives with individual file size of 100 MB. The files copied from computer to USB stick and vice versa with no problem, however there were CRC errors in 3 or more of the 100 MB files out of say 40. I could recopy these files and the CRC check would pass but then fail on other files later. Over time the drive would show symptoms of actual corruption, like missing files or files with odd names as if the File Allocation Table had been corrupted. These problems would go away for awhile after a reformat of the drive. The whole thing acted like there were bad sectors on the disk - if I were to compare it to a hard drive. Yesterday I used a program to make an empty 16 GB file which I put into transferred to the drive in 10, 50, and 100 MB multipart archives. Upon testing the rar I had CRC errors on the 100 MB files, but good on the 10 and 50 MB. I redid the 100 MB test and it passed. I checked the amount of disk space used after initial format, after completely filling the drive then erasing the individual files and after all tests passed. Keep in mind I formatted the drive as FAT32 but had the trouble on NTFS as well.
After initial format: 8,192 bytes used
After filling and emptying drive: 16,384 bytes used
After all tests passed: 155,648 bytes used
Based on this it appears that 139,264 bytes are no longer available to be used. This would be typical of a hard disk - where the controller automatically marks blocks defective and does not reuse them unless the drive is reformatted.
I wouldn't say the problem is fixed but it is interesting. After all of the hard work I found a program called H2Test would probably have automated the tests I performed. Ultimately I am looking to get an RMA to return the drive for a new one from the manufacturer. I sort of self solved this problem but input is welcome.
237 views
Usually answered in minutes!
I was messing with this again because the drives corrupt themselves over time and found some more interesting information. Apparently there are known problems with the Prolific flash controller for the PL-2528 chipset used in the A-Data PD16 series. One such problem is that it takes up to 15 seconds for a computer to recognize the drive. This makes it difficult to use in general but especially in Linux. It turns out that Prolific has a program to flash the PS-2528 controller. I can't tell you anything about it as the instructions I've found are in Chinese and info is hosted on a Russian website. That said, I did find the file "Prolific_PL-2528_MPTool_CM_M2209.rar" and clicked the "Start" tab from within the program. It said it partitioned, formatted, and upgraded the drive firmware. The drive is now recognized by windows in about 5 seconds instead of 15. I'm hoping this fixes the corruption issues as well. If not it might be worth paritioning the drive into 4 equal parts (four drive letters) as the corruption may be related to passing a boundary from one 4096 segment to another. That's just a guess.
The PL-2528 update definitely helped but I was still seeing CRC errors until I reformatted the drive in XP disk management as NTFS with clusters (allocation unit) size of 512. The CRC seemed to be related to some kind of byte boundary on a large cluster size 16 or 32Kb.
At 512 I've filled and emptied the drive dozens of times without incident, where before I could only do it a couple of times before having to reformat.
×