[NTLK] Pretec Cards and Newt Absolved of Reset Accusations

From: Bob Carls Dudney <kosmicdollop_at_saber.net>
Date: Sat Aug 22 2009 - 11:17:01 EDT

Wow! Thank you James, PCBman, and Dan very much
for the detailed exposť on flash cards and resets
(key points quoted and appended below).

Your thorough reports and discussion reveal
James' AMD card died because its critical
attribute memory eeprom chip (flash needs some
ROM =-O ) died either on the card's last
successful mount in the Newton (long before the
freeze due to Clarity bugs), or during the
restart when the card failed to mount. (Memory
terms explained in appended quotes.) Although
obviously at the end of its lifespan*, the eeprom
was still intact enough for Newt to read the
critical CIS attribute info (see quotes) and
mount the card during the last successful restart.

                *perhaps shortened by physical damage
                  PCBman noted

Per my inexpert understanding of these
revelations, a more detailed analysis is appended
below how a worn out (or weakened) flash card
that actually expired on mounting could continue
functioning normally until next mount attempt
while failing to give its owner any sign of
impending disaster, and then raise false
suspicions damage was caused by restarting with
the card inserted. Any such suspicions are
evidently unfounded in fact:

This new understanding added to other flash card
realities below affirms my dozen mishap-free
years disregarding warnings to remove cards
during restarts with several Pretec cards as main
data stores in multi-hour-every-day Newton use.

Evidently no facts support alleged threats to
flash cards, including Pretecs, from leaving a
card inserted in Newt during restarts of any
sort. -- But not a brain wipe: I would not risk a
card on that.

I much look forward to any
corrections/clarifications from list members
better informed than non-technician/non-engineer
me.

-- And, of course, James', mine, and countless
others' tragic experiences confirm the need for
backing up all computer data. (I've twice lost
Newton data, both due to corruption, not to any
hardware failure, including Pretec (yet: no flash
card lasts forever, obviously.))

While Pretec may be the only manufacturer to
issue warnings about leaving a card inserted
during restarts, that neither proves Pretec cards
are vulnerable to restarting, nor that they are
the only brand of card with this highly
questionable/theoretical?/evidently nonexistent
issue.

My very faded memory of discussions around a
decade ago is the NewtonTalk consensus was there
were no significant risks of leaving a card
inserted during a restart, hence I have not
ejected my primary 20 Meg Pretec card prior to
many thousands of restarts (mostly due to
discharged batteries). (On the other hand, one
obviously ought to be cautious about ejecting a
card from a frozen or otherwise unresponsive
Newt. It's of course also wise to wait at least a
couple minutes (90 seconds minimum?) before
restarting an unresponsive Newt (with or without
busy icon visible) to see if it's just catching
up with tasks that well might include writing to
stores.)

Further, since I know of no other warnings to
eject flash memory cards prior to restarting any
other computer or electronic device, it seems
all-the-more reasonable to conclude there is no
reason to do so with Newtons.

I've not taken the time to search, but if there
seems a preponderance of reports on Pretec over
other card brands having problems after restarts,
that may just be sampling bias. The several times
I searched for an additional flash card, the best
prices I could find were Pretec cards from Tima
Scientific (also on lowest voltage cards, what I
suspect many others also preferred.) It would
thus not surprise me Newts hold more Pretec cards
than other brands.

It also seems the main issue in card
longevity/durability is the quality of its chips.
My faded memory is Pretec got its chips from
Intel. I also suspect no more than two or three
different manufacturers actually made/make the
chips the numerous brands of cards contain (at
least during Newton's heyday).

Summing up: James' and Dan's August* thread
further dispels evidently unfounded fears of
damaging any brand of card by
restarting/rebooting with a card inserted. I know
of no evidence simply restarting can cause a
flash card problem. (Presuming one has waited
sufficiently long after a freeze for any
unfinished card writes to complete. -- If data
loss occurs from unknowingly restarting during
data writing, the fault obviously isn't alleged
but false vulnerability of cards to simply
restarting, although an uninformed operator could
jump to that erroneous conclusion.))

                                  *both meanings

Again, many thanks to you three for the info extracted below.

And, again, I look forward to any needed corrections by fellow Newtoneers.

B C D

Detailed analysis EEPROM failure of James' card:

As I understand PCBman's explanation, the
attribute memory eepROM chip is only accessed
when a card is mounted (as during a restart). If
so, the last successful CIS attributes read may
have been the final death blow to the eeprom,*
but such a demise would not be evident until a
subsequent mount or restart when Newt next
attempted but could not read the inaccessible or
corrupted CIS attributes. In the interim such a
mortally wounded card would appear intact to Newt
and its user.

Or the final failed attempt to mount could have
dealt the knock-out. Whether the former or the
latter (or the combination of both), clearly one
(or the combination) of those final two CIS read
attempts were more than the eeprom chip could
take. Newt was thus no longer able to mount nor
access any data on the card, as PCBman so very
well explained.

*My understanding is ROM chips tolerate extremely
high but finite numbers of reads

During Aug09, James Wages and Dan wrote:
|
>my dead 32MB linear flash card was NOT Pretec!
>It was AMD D-series 5v linear
>after Clarity freeze, removing battery pack, then powering up
>message said:
>
>"There is a problem with this card. (Newton cannot recognize this type of
>card.)"
>-/-/-/-/-/-/-/-/-/-/-/-/-/-/-/-/-/-/-/
>
>PCBman full diagnostics on dead AMD card:
>
>The problem is the card information structure
>(CIS) in attribute memory is corrupted.
>Newton cannot read the CIS and won't go any further because it
>doesn't know what you have just plugged in!
>When the card is inserted the REG pin is pulled low
>so memory accesses are from attribute memory. Once the info has been
>read and parsed, the REG pin is pulled high and memory accesses now
>apply to the common memory space.

> Did PCBMan suggest the reset caused the failure?

>No.

====================================================================
The NewtonTalk Mailing List - http://www.newtontalk.net/
The Official Newton FAQ - http://www.splorp.com/newton/faq/
The Newton Glossary - http://www.splorp.com/newton/glossary/
WikiWikiNewt - http://tools.unna.org/wikiwikinewt/
====================================================================
Received on Sat Aug 22 11:18:15 2009

This archive was generated by hypermail 2.1.8 : Sat Aug 22 2009 - 12:30:00 EDT