[NTLK] Brainwipe after replacing ROM (was: Jim's dead Newton)

[fg2]

OK, let's get rid of some assumptions first. 

Whenever I replace a ROM board, I remove all power from the Newton, let it
sit for a while, replace the ROM board, and then connect the Newton to
power. My assumption is that everyone will do it the same way since it isn't
a good idea to rip a board out of another board while said other board is
still powered. Silvain, when you replace a ROM board, do you do it the same
way and in the same order? If not: What do you do differently?

> Result: after more than 12 tries, each time I switch between two 2x00 US
ROMs, the Newton warns me at startup that the Brain Wipe has happened.

Since "Brain Wipe" isn't part of our Newtons' vocabulary: What EXACTLY does
the Newton say at startup?
 
> As I'm doubtful, I obviously put a few notes and pkg on each ROM before
changing it

Just for the record: You can't put anything on a ROM board. As Matthias
pointed out, its two chips are mask-programmed and can't be written to. Your
notes and packages would end up in the internal Flash RAM on the mainboard,
unless there's a memory card in the Newton and the Newton is configured to
store new data there. 

>  In any case, physically they are identical, below the link to see the two
ROM in question:
>
http://message-pad.net/Download/Documents/Newton-Pictures/ROM-NEWTON-2x00.jp
g.

I agree. Those two look utterly identical. And they're the same as those I
used for the test described below.

> I absolutely do not want to contradict what you say

Actually, contradicting yours truly is something everybody SHOULD want. And,
by all means, do. If mankind weren't in the habit of contradicting each
other occasionally, we'd still be living on trees, waiting for the next
mammoth to pass by so we can finally have some dinner. 

> but I would just like to know why I have a different result than yours.

So would I.

What I find interesting is your report that the Newton will actually warn
you that a brainwipe is about to happen. Or did I get that wrong? Because
I've never had such a warning even when using DIFFERENT ROMs. Normally, in
such a case the Newton will notice the change immediately after power-up,
get rid of all user data including OS updates, and then power up as if it
has just left the factory. If your 2000 actually warns you before this
happens, I'm more than interested in getting my hands on your two ROM boards
plus a Watson ROM board for a couple of days. 

I'm not surprised that the 2000 will recognize the Watson ROM as being
different. Normally one would use some kind of error-detecting code to
detect accidental changes to the ROM data, like e. g. a CRC (cyclic
redundancy check) checksum. These values are of course different for
different language Newton 2x00 ROMs, since their data is different. They're
also different for Newtons and eMates. I'd be surprised if they weren't
different for Newtons and Watsons, too, which is why I'd expect this kind of
reset to work with a Watson ROM, too.

Be it as it may: I just

a) Connected Frankenstein 1, a somewhat crippled Newton 2000, to a power
adapter

b) Brainwiped it (Pressed reset and power switch, held power switch,
confirmed two times)

c) Ran it through its Welcome routine

d) Made Extras the backdrop in order to change a system setting

e) Wrote a "Hello World" note

f) Removed all power

g) Put in a different ROM with the same language

h) Connected the AC adapter. The Newton brought up the expected pen
calibration routine, reported that it had been reset because all power was
removed, and that was it. Extras was still the backdrop, and my note was
still there.

However, thought I, Monsieur Pilet normally knows what he is talking about.
So I shook the other five ROM boards out of my bag, made sure they were all
US type boards like those Sylvain shows in his picture, and repeated steps f
to h another five times. 

Three times I got the expected result. Two of the boards didn't work at all,
which I found surprising. A quick check under my microscope showed that they
both had the same problem: Extremely poor soldering. So poor that the pins 1
and 2 of ROM chip U17 didn't make contact with their soldering pad at all.
There was white residue between all pins, and I found tiny balls of
soldering tin prowling between the pins. After cleaning both boards with
alcohol, removing all stray tin, applying a liberal dose of expensive
Chip-Quik flux to all pins and re-soldering every single pin of the two ROM
chips, both boards worked hunky dory. Just like the other four.

So what do we learn here? 

1) Six out of six boards yielded the result I expected. If you replace a
MP2000 ROM board with another MP2000 ROM board that has the same language,
the Newton (well, at least MY Newton) will NOT be reset and all user data
and OS updates will be kept.

2) If you're trying to fix a dead Newton by replacing the ROM board, and the
Newton remains dead, you should not assume that the ROM board is innocent.
Assumptions. How I hate them! I mean, how likely is it that out of the six
ROM boards that have found their way into my bag over the last two decades,
two boards, boards that have obviously not been tampered with by a third
party, show EXACTLY the same soldering problems? 

3) From now on Frank will not only check the mainboards of dead Newtons
under his microscope, but also their ROM boards. 

4) There must be something different on Sylvain's side of the problem.
Either this is a communication problem, or Sylvains findings have something
to do with the Watson ROM having been in the Newton 2000.

Cheers

Frank

-- Newton software and hardware at http://www.pda-soft.de