Update: A patch for this issue is currently being tested.
Adding myself to Xennex's list:
Not affected by "2 page auto-close problem" with .11, .12
Xennex1170 - SanDisk, Class 4, 8GB µSDHC
taming - Lexar, Class 2, 4GB µSD
JD Gumby - generic, 4GB
novou - generic (Nokia), ?? - it's cheap, 2GB µSD
markemark Transcend, Class 4, 16GB µSDHC, TS16GUSDHC4
Psyke - San Disk, ??, 2GB µSD
SF&F Reader - Kingston, Class 4, 4GB µSDHC
moorejl57 - generic, ??, 512MB µSD
The Terminator - Sandisk, ??, 8GB µSDHC
"2 page auto-close problem" with .11, .12
register512 - Patriot Memory, Class 10, 16GB µSDHC #PSF16GMCSDHC10
markemark - Transcend, Class 6, 8GB µSDHC, #TS8GUSDHC6
viduliya -
3x PNY, Class 4, 2GB µSD
p_a_smith - Dane Elec, ??, 2GB µSD
viktor_haag - Adata, Class 10, 8GB µSDHC
moorejl57 - generic (Borders), ??, 4GB µSD
Tangent - Lexar, ??, 4GB µSDHC
Wild fantastical conjecture: Some cards need more current to run. (?) (
sometimes due to more speed & capacity?)
If they're under-supplied they screw up/reset
(a reset looks like an eject & insert, which triggers a home-screen return. RE:
markemark, that return is by design AFAICT)
Perhaps .11 reduced the current serving the µSD card & controller?
(maybe indirectly, through more intensive processes, eg: better page layout algos)
Can someone with this issue pan around on a webpage (or do something else to draw current) and see if it occurs?
Full disclosure: I have no experience with embedded systems or electrical eng., I just mess with computers.
Even more insane: What if the Kobo is caching µSD epubs in ram/onboard flash and then putting the card & controller to sleep to save power? (issue: not enough sleep current supplied to keep stability)
Viduliya's post seems to support the sleep hypothesis, their cards only display the issue if you wait 1min.+ (sleep timeout?) between page turns. Though if the card wakes up every page turn, it means they
aren't caching the data from the card.
tl;dr
FJames summarized this better than me.