Absolute nonsense.
Whether the file system is FATxx or NTFS, the circumstances that cause
AutoChk (the automatic check on startup) are identical, namely...
1) A bit is set to indicate file system writes were interrupted
This is the most common cause, and happens after Windows crashes,
resets, fails to shut down properly, or is switched off or reset by
the user. When this happens, files that were being written will be
incomplete, and the next startup detects the relevant status bit and
"fixes" the file system. This does NOT recover data, which is:
- typically damaged by the "fix"
- made to look as if it is no longer corrupted
Also, cues that could be used for manual repair are lost.
Sometimes you will see this even after what appears to be a proper
shutdown. Unless this is (2), (3) or (4), this would most likely be
due to an ATX power off that occurs before the data is writen from the
hard drive's internal cache RAM to the actual platters.
2) A different bit is set to indicate a disk access failed
This indicates impending hard drive failure, and once again the
relevant bit (a different one this time) will cause AutoChk to "fix"
things. This time a full surface scan is done, which takes a long
time. Note that no previous bad exit is required to trigger this form
of AutoChk; any failed disk access at any time during the last Windows
session will cause this, even if you did shut down properly.
If this is the cause, then attempting to convert the file system from
FATxx to NTFS is one of the worst things you can do!
3) You did a ChkDsk /F or /R on a volume that was in use
C: is *always* in use, which means that whenever you try to do a
ChkDsk /F or /R (to "fix" file system, and do that plus a surface
scan, respectively), you will be advised that the system is set to
perform this operation on the next boot. AFAIK, this is done by
setting the relevant bits as per (1) or (2).
4) You have explicitly set ChkDsk to run on startup
This can be ChkDsk in the startup axis, or set as a Task to be run on
startup or on account logon.
See
http://cquirke.mvps.org/ntfs.htm
Number 1 myth: "NTFS doesn't suffer data corruption on bad exits" (or
even more absurdly, "NTFS doesn't suffer data corruption").
All that NTFS' transaction rollback does, is revert to the last good
metadata (i.e. discarding any changes that were in progress). The
"user" data (i.e. the actual data within the file's cluster chain) is
not preserved at all, so the outcome is as bleak as for FATxx, except
there are no low-level recovery tools for NTFS.
The most accurate diagnostic instrument
in medicine is the Retrospectoscope