Searching \ for '[EE]: recover data from damaged eeprom?' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/mems.htm?key=data
Search entire site for: 'recover data from damaged eeprom?'.

Exact match. Not showing close matches.
PICList Thread
'[EE]: recover data from damaged eeprom?'
2002\10\08@123620 by Mike Mansheim

flavicon
face
I have a board that had a 16F876 and a 24LC01 eeprom on it.  Something
happened to the board that damaged both the processor and the eeprom.
If the eeprom is moved to an otherwise good board, SCL is low all the
time.  Is there any way to recover the data stored in this chip?
Thanks for any help.

--
http://www.piclist.com hint: To leave the PICList
spam_OUTpiclist-unsubscribe-requestTakeThisOuTspammitvma.mit.edu


2002\10\09@031059 by Vasile Surducan

flavicon
face
On Tue, 8 Oct 2002, Mike Mansheim wrote:

> I have a board that had a 16F876 and a 24LC01 eeprom on it.  Something
> happened to the board that damaged both the processor and the eeprom.
> If the eeprom is moved to an otherwise good board, SCL is low all the
> time.  Is there any way to recover the data stored in this chip?

 The same like used by the russians when they have copied the 8080 a
looong time ago:
 use a very thin and delicate file and a good microscope...
 then on a mathematics copy-book draw x and o as you see on the
microscope

 :)

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2002\10\09@163124 by Peter L. Peres

picon face
On Wed, 9 Oct 2002, Vasile Surducan wrote:

*>On Tue, 8 Oct 2002, Mike Mansheim wrote:
*>
*>> I have a board that had a 16F876 and a 24LC01 eeprom on it.  Something
*>> happened to the board that damaged both the processor and the eeprom.
*>> If the eeprom is moved to an otherwise good board, SCL is low all the
*>> time.  Is there any way to recover the data stored in this chip?
*>
*>  The same like used by the russians when they have copied the 8080 a
*>looong time ago:
*>  use a very thin and delicate file and a good microscope...
*>  then on a mathematics copy-book draw x and o as you see on the
*>microscope

Must have magic microscopes them Russians, to see electrical charge in
EEPROM transistors buried under their access lines and transistors. The
files must have been excellent too.

Peter

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2002\10\09@235047 by Wagner Lipnharski

flavicon
face
Peter L. Peres wrote:
{Quote hidden}

Now lets try to be honest.  Nobody in common sense will use a tunneling or
deep probe microscope to duplicate a 8080... think about it... how
difficult is to design a microprocessor?

Think seriously about it.

Even you can do it pretty easily. What is the worse part?  To have the way
to produce the wafers, right? not the design.
So, you can be very much sure, that if somebody else have a way to produce
silicon wafers, much before they will have somebody to design silicon
microchips.

I am not defending or attacking no one, but in someway is ridiculous to
think that someone who has ability to produce a wafer will have no ability
to design the chip.  It is the same as to think that someone is very expert
in producing a car, engine and everything else, but dumb enough to not know
how to design it.

You can design a microcontroller in 30 minutes... wanna bet?

You just need two EPROMS, a SRAM, an oscillator and few logic gates.

You assemble the logic gates to do left/right shift, add, compare and
simple logic packages.
Connect the RAM to feed and to be fed by the logic gates.

Build two binary sequence counter (multiple 74HCT93).

One binary counter will scan the first eprom for the program. The program
eprom output will gate directly the addresses of the second binary counter.
This second counter will scan particular addresses on the second eprom.
This second eprom contains the steps necessaries to the logic packages to
operate, its "data" output will effectively turn on gates, fed data, open
and close path for add, shift, move, etc, all controlled by the second
eprom sequence of events.

Install few I/O ports around, and implement more and more instruction
routines at the second eprom. In time, you will have more instructions than
the Pentium IV.

In a board no larger than 5x6 inches you can have a heavy processor, just
built with gates, counters and eproms. That's it.

Now, convert all of this to wafer design, you can launch your own processor
at the market, without using a microscope to steal any idea around.

VV46NER

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2002\10\10@010557 by Russell McMahon

face
flavicon
face
> > *>  The same like used by the russians when they have copied the 8080
> > a *>looong time ago:

> Now lets try to be honest.  Nobody in common sense will use a tunneling or
> deep probe microscope to duplicate a 8080... think about it... how
> difficult is to design a microprocessor?
>
> Think seriously about it.
>
> Even you can do it pretty easily. What is the worse part?  To have the way
> to produce the wafers, right? not the design. ...............

I'm certainly not wanting to get into an argument here, but it has always
been my understanding that in the earlier days of microprocessors there was
indeed quite a lot of copying at a fairly low level. Nowadays you can hire a
few of the better graduates (or some of someone else's better engineers
:-) ) and be well on the way to rolling your own.

I think the point is that *in its day* the 8080 (or the 4004 / 8008 / ...)
was fairly leading edge and as it was also the market leader and ensuring a
very high level of computability was essential. I have read some extremely
interesting accounts of what was involved in getting early processors to
meet desired specs. One was the 80C86 from Intel themselves - the insides of
this in many places bore no relationship to the NMOS part (or to something
based on sound engineering practices IMHO :-) ). Another was a book ?"The
soul of a machine" or something like that about the Data General ?Nova
computer development where they were trying to beat the DEC VAXs on at least
a few instructions. They did, but the kluges required were immense. Neither
of these examples involved copying other people's designs directly but the
amount of non-straight-forward engineering to get the desired result was
immense. In the case in question, being able to copy an existing part for an
IC design would probably be a great help.



       Russell McMahon

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\10@014557 by Roman Black

flavicon
face
*>On Tue, 8 Oct 2002, Mike Mansheim wrote:
*>
*>> I have a board that had a 16F876 and a 24LC01 eeprom on it.
Something
*>> happened to the board that damaged both the processor and the
eeprom.
*>> If the eeprom is moved to an otherwise good board, SCL is low all
the
*>> time.  Is there any way to recover the data stored in this chip?


Hi Mike, it may be possible to do with little effort,
but with no guarantee of success. :o)

I may be mistaken <grin re recent posts> but SCL on the
24LC01 does not pull down internally or need to pull down
internally in order to operate.

The fact that yours is low all the time (even with a
pull-up resistor) seems to indicate that the silicon is
fused internally on that one pin. This is a very common
fault situation, and probably points to the internal
protect diode for that pin.

You can un-fuse the damaged section of the silicon
with a reasonable chance of success. I have seen this
happen on a few occasions with TV set jungle chips, one
pin has shorted to gnd internally, but after some minutes
of testing etc with excess current into that pin it
un-fuses and returns the chip to some level of functionality.

Since your eeprom chip has only fused one pin, it may be
only a very small part of the chip damaged, like the pin
protect diode or just a tiny molten blob on one of the
semiconductor "tracks".

You MAY be able to fix it by doing this;
* don't power up the chip
* connect it's Vdd pin to gnd via a 1k resistor
* connect gnd to gnd
* inject controlled current into the bad pin to gnd
* slowly increase current over time until it un-fuses
* never exceed 5v, to protect any working body diodes

I suggest measuring the fault resistance of the bad pin,
small silicon normally faults to 50 ohms or thereabouts.
Then start injecting the controlled current. A variable bench
supply and series resistor will do it you if you don't have a
variable current supply. Start at maybe 20mA and increase
in 20mA steps, and with each step wait a few minutes.

Use an ammeter, with a voltmeter on the fault pin.
As you increase the current the fused silicon area will
dissipate the bulk of the power and get hot, and as it
gets very hot it's resistance will rise, shown by a
marked increase in voltage on the fault pin as the current
is held constant. Increasing current a bit further should
be enough to cause the fused area to become open circuit.
Maybe try mounting the chip vertical with that pin at the
bottom, so any molten silicon etc will run off away from
the chip?? Maybe tapping the chip during the final stage??

Once it has un-fused you can test the chip, if it was just
a body diode it may be fixed. Slowly increasing the current
over time reduces the chance of blowing the gold leadwire,
which has a low resistance compared to the fused silicon
area.

I like to do destruction testing before using semis in
critical designs. Have you ever put a glass diode under
a microscope and watched the changes in the diode chip and
gold wire as it starts to cook? Much fun on a rainy
afternoon. :o)
-Roman

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\10@021122 by Vasile Surducan

flavicon
face
Wagner, unfortunately it seems you must improve a little your humour
sense. The stress you may have in your capitalist world does not help
anybody...and blinded you to see it was an inocent joke.

regards,
Vasile
http://www.geocities.com/vsurducan


On Wed, 9 Oct 2002, Wagner Lipnharski wrote:

{Quote hidden}

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\10@054854 by Mike Singer

picon face
Failed to understand Vasile's message, but Wagner's
point of view was quite clear to me:
Why
"to duplicate a 8080... think about it... how difficult is to
design a microprocessor?"
"You can design a microcontroller in 30 minutes... wanna bet?"

Now it is easy, after a "moment of truth" has come.
But in 197X it didn't seem to be so (the posting was
"the russians when they have copied the 8080 a looong
time ago").
There were soviet microprocessors non the worse then
Z80, but microprocessor alone does mean nothing.

Correct me if I'm wrong :-) , since 197X were my school years.

Mike.

Vasile Surducan wrote:
> Wagner, unfortunately it seems you must improve a little your humour
> sense. The stress you may have in your capitalist world does not help
> anybody...and blinded you to see it was an inocent joke.
.
> > Now lets try to be honest.  Nobody in common sense will use a
tunneling or
> > deep probe microscope to duplicate a 8080... think about it... how
> > difficult is to design a microprocessor?
> >
> > Think seriously about it.
.
> > You can design a microcontroller in 30 minutes... wanna bet?

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\10@092705 by Vasile Surducan

flavicon
face
Mike, there are three big barriers between you and me:
-- language
-- age
-- geo-political position

At least one of the olders canadian PIC-listers which have known the
importance of the 8080 in that times have understood me. :)

And, maybe you didn't know, but the K580...( russian clone of 8080 ) was
partially built exactly like I was joking in my first mail. On that days
the embargo against socialist countries was a little crude like it's now.
( if I'm asking a free Microcip CD, or a free sample, I will never got it,
this is just a not understandable example for the youngest piclisters... )
About Wagners ability of designing an 8080 in 30 minute, I'm glad about
his skills, but he forgot to tell us he need another 6 monts to test it
and solve just a few from dozens of bugs... :)


regards,
Vasile
http://www.geocities.com/vsurducan


On Thu, 10 Oct 2002, Mike Singer wrote:

{Quote hidden}

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\10@100522 by mike

flavicon
face
Probably not. The usual failure mode is that the input protection
diode fails short.
If you REALLY REALLY want to try to get the data, you could try to
blow the diode out with short high current low voltage pulses but I'd
guess this is more likely to blow the bond wire than the diode.
On Tue, 8 Oct 2002 11:23:27 -0500, you wrote:

>I have a board that had a 16F876 and a 24LC01 eeprom on it.  Something
>happened to the board that damaged both the processor and the eeprom.
>If the eeprom is moved to an otherwise good board, SCL is low all the
>time.  Is there any way to recover the data stored in this chip?
>Thanks for any help.

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\10@103311 by Wouter van Ooijen

face picon face
> Now lets try to be honest.  Nobody in common sense will use a
> tunneling or
> deep probe microscope to duplicate a 8080... think about it... how
> difficult is to design a microprocessor?

Not that difficult (at least *now*), but what they wanted might havew
been a uP that could run all the 8080 software, which is much more
difficult to design from scratch!

Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\10@111327 by Dale Botkin

flavicon
face
On Thu, 10 Oct 2002, Vasile Surducan wrote:

> About Wagners ability of designing an 8080 in 30 minute, I'm glad about
> his skills, but he forgot to tell us he need another 6 monts to test it
> and solve just a few from dozens of bugs... :)

He is (and we are) also "standing on the shoulders of giants".  I doubt
any of us could have done so in 1974.  The 8080 was pretty advanced for
the time.

Dale

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\10@114940 by Wagner Lipnharski

flavicon
face
Vasile Surducan wrote:
{Quote hidden}

-----


The interesting point is that exactly in 1972 I had 17 yrs old, I did it. I
built a discrete 8 instructions microprocessor, add, add with carry, shift
left, shift right, load, store, complement and branch on condition, using
two 2708 eproms,  a 10 Watts winter heater TTL board, running at the bullet
speed of 1MHz.  The longest instruction took 12 bytes at the "instruction
execution" second eprom. The design was done in less than 30 minutes over a
restaurant table, two days to wire up, 5 days to program - 2708 programmer
assembly included - ah, no compilers, just paper and pencil. I may have the
original design in some box around.  3 years later all was replaced by a
brand new Z80 I bought when traveling to Toronto. That was done not by
skills, but by interest and guts to try.  You can be sure that if it would
at least flag a 6 months time to solve few of the bugs, I would be dealing
with chicken eggs today. I am trilled that you think something like that
could take 6 months to debug... are you in eggs business already?  ;)

By the way, I am from Brazil (to the ones who ran away from school; Brazil
is the huge country at the right side of the South America - almost the
same size of US), we are not exactly 100% capitalist - we use to say we
are - but our minimum wage is around $100, that is not a pretty sign of
capitalism.  As for the *humor* part; We will be serving Portuguese 10
minutes language lessons during the 4pm coffee break - please check in with
the pretty girl at the end of the hall.  First give-away lesson; "Buenos
Dias" is far from being Portuguese... and Argentina is not our capital.
Tomorrow's lesson; "Why our beach bronze tan model body girls in Brazil
does not (never did) wear hairdo with bananas and pineapples", and
Saturday's special: "The truth that lies beneath the controversial
Brazilian word -camisinha-".

Cheers,
VV46NER
--------

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\11@022411 by Peter L. Peres

picon face
On Wed, 9 Oct 2002, Wagner Lipnharski wrote:

*>Peter L. Peres wrote:
*>> On Wed, 9 Oct 2002, Vasile Surducan wrote:
*>>
*>> *>On Tue, 8 Oct 2002, Mike Mansheim wrote:
*>> *>
*>> *>> I have a board that had a 16F876 and a 24LC01 eeprom on it.
*>> Something *>> happened to the board that damaged both the processor
*>> and the eeprom. *>> If the eeprom is moved to an otherwise good
*>> board, SCL is low all the *>> time.  Is there any way to recover the
*>> data stored in this chip? *>
*>> *>  The same like used by the russians when they have copied the 8080
*>> a *>looong time ago:
*>> *>  use a very thin and delicate file and a good microscope...
*>> *>  then on a mathematics copy-book draw x and o as you see on the
*>> *>microscope
*>>
*>> Must have magic microscopes them Russians, to see electrical charge in
*>> EEPROM transistors buried under their access lines and transistors.
*>> The files must have been excellent too.
*>>
*>> Peter
*>
*>Now lets try to be honest.  Nobody in common sense will use a tunneling or
*>deep probe microscope to duplicate a 8080... think about it... how
*>difficult is to design a microprocessor?
*>
*>Think seriously about it.

I am thinking. I was born in one of them 'behind the iron curtain'
countries and it took me only a short time to think about this.

Do not look for logic in what those people did at the time. Bright minds
from up above told hem what to do and they did it. The fact that that
certain social system is not famous for good management in general should
ring a *loud* bell.

They DID copy the 8080 and the Z80 and I think one of the 68xx Motorola
chips. But I doubt they used microscopes and such.

*>I am not defending or attacking no one, but in someway is ridiculous to
*>think that someone who has ability to produce a wafer will have no ability
*>to design the chip.  It is the same as to think that someone is very expert
*>in producing a car, engine and everything else, but dumb enough to not know
*>how to design it.

;-) You should check out who exactly makes (physically) cars and other
much higher tech implements. I am not saying that what you are saying does
not make sense, it makes perfect sense. However major parts of the world
do not.

Producing wafers is very different from producing workable microprocessor
systems and both are different from successfully marketing them in
products or per se. The East had good technology and people in some areas
but the management and planning glue that was supposed to make it work in
the bigger picture was rotten.

*>You can design a microcontroller in 30 minutes... wanna bet?

No. I drew a micro that was to be implemented with TTL chips (4 bit) just
for fun in 1986 or so. I never tried to build it, it would not have been a
great micro (say a quarter of a 16C54 on 4 bits) and I doubt I could have
sold it or made something out of it except sink tens of hours into it.

*>You just need two EPROMS, a SRAM, an oscillator and few logic gates.
*>
*>You assemble the logic gates to do left/right shift, add, compare and
*>simple logic packages.
*>Connect the RAM to feed and to be fed by the logic gates.

...

You don't really need this. All you need is a Mealy (or was it Moore - the
one with the latch in the feedback) automaton with wide EPROM and that's
your entire CPU minus the registers, which you would not use anyway, just
use a SRAM and an address counter with provisions to load it from SRAM
output. This kind of core can be used to run microcode or RISC style (one
giant lookup table in the EPROM). For 8 bits the size is manageable, you
can probably fit it in 1M of 8bit EPROM or EEPROM. 4bits is a joke to
implement like this (16 operations on 4 x 4 bits plus a few more on single
4 bits would fit in a single 64k EEPROM). More than 8 bits causes table
size explosion.

It is interesting to simulate these designs using C on a larger computer
(PC). Also you can reprogram new instructions on the fly if you allow the
microcode EEPROM to be writable by the system. DDCF ? ;-)

The real problems will begin when you will want to add interrupts. Because
that's when your deterministic state machine will likely stop being
deterministic unless you are very very very clever and careful about these
things.

Peter

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2002\10\11@022413 by Peter L. Peres

picon face
On Thu, 10 Oct 2002, Dale Botkin wrote:

*>On Thu, 10 Oct 2002, Vasile Surducan wrote:
*>
*>> About Wagners ability of designing an 8080 in 30 minute, I'm glad about
*>> his skills, but he forgot to tell us he need another 6 monts to test it
*>> and solve just a few from dozens of bugs... :)
*>
*>He is (and we are) also "standing on the shoulders of giants".  I doubt
*>any of us could have done so in 1974.  The 8080 was pretty advanced for
*>the time.

Yes, and if it would have used just 1-2 more support chips besides the two
it mandatorily required you could have called it a herd of microprocessors
(to avoid having to call it a slice or something).

The 8085 was a definites step up, it is the conceptual parent of the 8051
(with which it is almost pin compatible) and of the 8088 in minimal mode.
The 8085 is still around, it gets used in some medical equipment because
is has a FDA approval afaik.

Peter

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2002\10\11@032713 by Vasile Surducan

flavicon
face
Wagner Lipnharski <.....wagnerKILLspamspam@spam@USTR.NET> written:

> By the way, I am from Brazil (to the ones who ran away from school; Brazil
> is the huge country at the right side of the South America - almost the
> same size of US), we are not exactly 100% capitalist - we use to say we
> are - but our minimum wage is around $100, that is not a pretty sign of
> capitalism.  As for the *humor* part; We will be serving Portuguese 10
> minutes language lessons during the 4pm coffee break - please check in with
> the pretty girl at the end of the hall.  First give-away lesson; "Buenos
> Dias" is far from being Portuguese... and Argentina is not our capital.
> Tomorrow's lesson; "Why our beach bronze tan model body girls in Brazil
> does not (never did) wear hairdo with bananas and pineapples", and
> Saturday's special: "The truth that lies beneath the controversial
> Brazilian word -camisinha-".
>
 Heee-heeh !
 So you're "just a little" latin eh ?
Thank you very much telling me about where BraZil is, I need it a lot !
:) also the capital :), BTW there is the same beautifull Rio wiew from
the Crist's head ?  If you haven't realised yet I'm not an american (
pretty bad...) , then maybe you must know your capital is not Rio de Janeiro,
neither Sao Paolo, but Brasil. I forgive you, let see if you know where is
Romania and the Romanian capital... maybe Ukraina ?

 best luck with your 30 minutes built in CISC microprocessors :)

 all the best,
 Vasile

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2002\10\11@034006 by Russell McMahon

face
flavicon
face
> You don't really need this. All you need is a Mealy (or was it Moore - the
> one with the latch in the feedback) automaton with wide EPROM and that's
> your entire CPU minus the registers,

The original Apple II floppy disk controller, named an IWM, was implemented
from bipolar proms and latches. Worked surprisingly well at a time when
"real" FDD controllers were rather dear . IWM, by the way, stood for
"Integrated Woz Machine" after ?Steve Wozniak, its designer.


           RM

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2002\10\11@113928 by Roman Black

flavicon
face
Mike Harrison wrote:
>
> Probably not. The usual failure mode is that the input protection
> diode fails short.
> If you REALLY REALLY want to try to get the data, you could try to
> blow the diode out with short high current low voltage pulses but I'd
> guess this is more likely to blow the bond wire than the diode.


Which is exactly why I went to the trouble of
posting a long post on how to do it without damaging
the wire. I've blown up a lot of silicon (both for
fun and for work) over the years. :o)
-Roman

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


More... (looser matching)
- Last day of these posts
- In 2002 , 2003 only
- Today
- New search...