The Infamous Bad Data Problem

[Tom Droege <tdroege@veriomail.com>]

As you all know, I have been plagued with a "bad data" problem.  It is most 
likely caused by the data from the memory card being one byte off.  That is 
what it looks like.

Whenever it happens I take care to note what might have happened.  This 
evening while testing TOM2, the bad data appeared suddenly.  It happened 
just after I interrupted the program while it was in the middle of running 
Rob's Download program.  After that, it read consistently bad data.

I then shut down windows properly.  It shut down as if it was doing it 
properly.  I turned off power, and when I turned it back on, it said that I 
had not turned off the computer properly and did the memory scan 
thing.  Now it read out the memory card properly.

OK, it looks to me like Windows is stashing an odd byte 
somewhere.  Possibly it is a random thing on turn on and it thinks that 
there is an odd byte to read and does so.  The result is that sometimes 
when I turn on the system reads out the bytes in correct sequence and 
sometimes it is one byte off.

OK, I accuse my hardware first.  But I have tried everything.  It is 
beginning to really look like there is something in Windows that gets one 
byte off on the I/O port, and then often remembers that on turn 
on.  Barf!  This may be hard to believe.

Now that several of you are trying to run Mark IVs and with different 
operating systems, it will be interesting if you see this problem.  The 
symptom is an image with -32k to +32k range and a sigma of large.  The data 
looks like big time noise.  It suddenly comes and is hard to get rid 
of.  Usually a few restarts of Windows will do it.  Sometimes it takes all 
night.

Tom Droege