24 Jan
2006
24 Jan
'06
8:13 a.m.
At 01:15 AM 1/24/2006, Gooijen, Henk wrote:
I know, I know! And I was a bit disappointed that he didn't grind
down the top of the failed chip to do a proper repair. :-)
Of course, he can regain the extra point by giving an explanation
of why swapping a chip is allowed under the proper interpretation
of the official repair rules.
- John
Tony Duell
wrote:
And the board basically worked. The display at boot-up was sane.
Suspecting a RAM problem, I put in a complete new set of
4164s -- no change.
Hmmm, sounds like swapping to me ... ;-) 
24 Jan
24 Jan
2:02 p.m.
At 01:15 AM 1/24/2006, Gooijen, Henk wrote:
I know, I know! And I was a bit disappointed that he didn't grind
down the top of the failed chip to do a proper repair. :-)
What failed chip? All the RAMs were good...
Tony
Duell wrote:
And the board basically worked. The display at boot-up was sane.
Suspecting a RAM problem, I put in a complete new set of
4164s -- no change.
Hmmm, sounds like swapping to me ... ;-)
Of course, he can regain the extra point by giving an explanation
of why swapping a chip is allowed under the proper interpretation
of the official repair rules.
Err, when you know the replacement part is good, when you know the
replacement part is directly equivalent to the old part, and when the old
part cannot be repaired?
-tony
6:09 p.m.
New subject: grinding down chips was Re: QX10 graphics board
And I was a bit disappointed that
he didn't grind
I've been dying to ask this question. Can you
actually learn something (hopefully a whole lot!)
about a chip if you actually did this??? What if there
was some old chip for which there is no documentation.
If, given the availability of the proper equipment
(surface grinder?), you were able to take off say
.0001" of material at a time, or thereabouts ;), would
you have the ability to photograph it, and have
something in the way of a working schematic?
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
down the top of the failed chip to do a proper
repair. :-) 
26 Jan
26 Jan
10:09 p.m.
New subject: grinding down chips was Re: QX10 graphics board
On 1/24/06, Chris M <chrism3667 at yahoo.com> wrote:
I've been dying to ask this question. Can you
actually learn something (hopefully a whole lot!)
about a chip if you actually did this??? What if there
was some old chip for which there is no documentation.
If, given the availability of the proper equipment
(surface grinder?), you were able to take off say
.0001" of material at a time, or thereabouts ;), would
you have the ability to photograph it, and have
something in the way of a working schematic?
I don't know if features would be immediately identifiable, but with
some work it should be possible. I know of a company that was
attempting to build silicon debugger using a high speed image
intensified camera to watch photons emitted by the transistors as they
pass current. Should work for reverse engineering... Given a couple
years and $500,000 I could probably build you one.... Then again I
would imagine that the device is patented.
Eric
11:24 p.m.
New subject: grinding down chips was Re: QX10 graphics board
On 1/24/06, Chris M <chrism3667 at yahoo.com> wrote:
I've been dying to ask this question. Can you
actually learn something (hopefully a whole lot!)
about a chip if you actually did this??? What if there
was some old chip for which there is no documentation.
If, given the availability of the proper equipment
(surface grinder?), you were able to take off say
.0001" of material at a time, or thereabouts ;), would
you have the ability to photograph it, and have
something in the way of a working schematic?
Aren't there chips (i.e. crypto) that are designed to be destroyed
if they are disassembled? I don't know what they do to the package,
but if I had an IC with an EPROM containing my crypto keys, I'd want
the chip to self-destruct during an attempt to pry open the package
:).
--
"The Direct3D Graphics Pipeline"-- code samples, sample chapter, FAQ:
<http://www.xmission.com/~legalize/book/>
Pilgrimage: Utah's annual demoparty
<http://pilgrimage.scene.org>
27 Jan
27 Jan
12:21 a.m.
New subject: grinding down chips was Re: QX10 graphics board
On 1/26/2006 at 9:24 PM Richard wrote:
Aren't there chips (i.e. crypto) that are designed
to be destroyed
if they are disassembled? I don't know what they do to the package,
but if I had an IC with an EPROM containing my crypto keys, I'd want
the chip to self-destruct during an attempt to pry open the package
I don't know about grinding, but the guy at MIT who figured out the Xbox
innards did something of the sort (regardless, it's a great read):
http://www.xenatera.com/bunnie/proj/anatak/xboxmod.html
The operative term is "decapsulating"; if you google on it, you can find
some interesting stuff, like:
http://www.smtinfocus.com/technical_papers_list.html
Cheers,
Chuck
12:27 a.m.
New subject: grinding down chips was Re: QX10 graphics board
I don't know about grinding, but the guy at MIT
who figured out the Xbox
innards did something of the sort (regardless, it's a great read):
A guy I went to school with does this for Motorola. He gets chips back
that are failing in the field and dissects them to find out what went
wrong.
I think the actual process involves shaving rather than grinding.
William Donzelli
aw288 at osfn.org
10:36 a.m.
New subject: grinding down chips was Re: QX10 graphics board
William Donzelli sez...
I did something similar as a kid at BNR (now Nortel) at their Corkstown facility.
I had purchased a whack of "unmarked but guaranteed to be the same chip"
I don't know about grinding, but the guy at
MIT who figured out the Xbox
innards did something of the sort (regardless, it's a great read):
A guy I went to school with does this for Motorola. He gets chips back
that are failing in the field and dissects them to find out what went
wrong.
I think the actual process involves shaving rather than grinding. from the old MIL (Microsystems International Limited)
at a surplus place,
and mentioned to one of the engineers at BNR that I'd like
to know what they were.
The process was that they put it into a high temperature high-oxygen-content
chamber and basically "eroded" the plastic from around the chip. Over the
course of a few hours, the eroded plastic was simply brushed away,
finally exposing the chip. We were then able to read the part number off
the die -- ML741 :-)
Cheers,
-RK
--
Robert Krten, PARSE Software Devices +1 613 599 8316.
Realtime Systems Architecture, Consulting, Books and Training at www.parse.com
Looking for Digital Equipment Corp. PDP-1 through PDP-15 minicomputers!
5:11 a.m.
New subject: grinding down chips was Re: QX10 graphics board
At 09:21 PM 1/26/06 -0800, you wrote:
On 1/26/2006 at 9:24 PM Richard wrote:
Diamond slitting saw should also turn up some useful information.
Joe
Aren't there chips (i.e. crypto) that are
designed to be destroyed
if they are disassembled? I don't know what they do to the package,
but if I had an IC with an EPROM containing my crypto keys, I'd want
the chip to self-destruct during an attempt to pry open the package
I don't know about grinding, but the guy at MIT who figured out the Xbox
innards did something of the sort (regardless, it's a great read):
http://www.xenatera.com/bunnie/proj/anatak/xboxmod.html
The operative term is "decapsulating"; if you google on it, you can find
some interesting stuff, like:
http://www.smtinfocus.com/technical_papers_list.html
Cheers,
Chuck
3:10 a.m.
New subject: grinding down chips was Re: QX10 graphics board
Aren't there chips (i.e. crypto) that are designed
to be destroyed
if they are disassembled? I don't know what they do to the package,
but if I had an IC with an EPROM containing my crypto keys, I'd want
the chip to self-destruct during an attempt to pry open the package
:).
I read a brilliant article on the encryption boards used in banking. The
actual clever bits are encased in epoxy, and powered by batteries on the
board. If you remove the batteries, it loses the contents of memory (and
all keys). If you attempt to cut the package open, you cut through a
layer consisting of many turns of resistance wire. If yout cut it, or
bypass bits, the resistance changes and it clobbers the memory. If it
gets too cold, or too hot, again it clobbers the memory. X-raying it is
right out, too.
Gordon.
3:48 a.m.
New subject: grinding down chips was Re: QX10 graphics board
gordonjcp at gjcp.net wrote:
I read a brilliant article on the encryption boards
used in banking. The
actual clever bits are encased in epoxy, and powered by batteries on the
board. If you remove the batteries, it loses the contents of memory (and
all keys). If you attempt to cut the package open, you cut through a
layer consisting of many turns of resistance wire. If yout cut it, or
bypass bits, the resistance changes and it clobbers the memory. If it
gets too cold, or too hot, again it clobbers the memory. X-raying it is
right out, too.
Does it have "U R SCREWED" etched on the die? :)
Doc
11:17 a.m.
New subject: grinding down chips was Re: QX10 graphics board
I read a brilliant article on the encryption boards
used in banking. The
actual clever bits are encased in epoxy, and powered by batteries on the
board. If you remove the batteries, it loses the contents of memory (and
all keys). If you attempt to cut the package open, you cut through a
layer consisting of many turns of resistance wire. If yout cut it, or
bypass bits, the resistance changes and it clobbers the memory.
I just had a similar experience with a Compaq RAID array. The drives
were encoded soas to only work with the controller, attempting to read
them with another controller and/or removing that controiller from the
system disrupted the configuration so that the disks contents were
effectively rendered unreadable, even on reassembly. Six big cache
batteries on the controller showed that the state of the array was keyed
in to and bound in with it and 'cutting apart' the components broke the
security.
Securing data on disks is still a big problem today.
Not the case it seemed with this system.
John A.
5:10 a.m.
New subject: grinding down chips was Re: QX10 graphics board
At 09:24 PM 1/26/06 -0700, you wrote:
On 1/24/06, Chris M <chrism3667 at yahoo.com> wrote:
Not that I ever heard of. A friend of mine used to failure analysis for
an aerospace contractor and he routine cut ICs apart to examine and test
them. Many of them were examined for things like ESD, overheating and
chemical exposure damage that didn't cause immediate failures but could do
damage and cause failures later. The ICs frequently passed operational test
even after they were cut open.
Also several months ago someone on this list mentioned that he worked in
a shop that had a wire bonding machine and he frequently took "failed" ICs
and replaced the failed bond wire and keep on using them. He said that some
ICs had had the bond wires replaced numerous times.
I don't know what they do to the package,
I've been dying to ask this question. Can
you
actually learn something (hopefully a whole lot!)
about a chip if you actually did this??? What if there
was some old chip for which there is no documentation.
If, given the availability of the proper equipment
(surface grinder?), you were able to take off say
.0001" of material at a time, or thereabouts ;), would
you have the ability to photograph it, and have
something in the way of a working schematic?
Aren't there chips (i.e. crypto) that are designed to be destroyed
if they are disassembled? but if I had an IC with an EPROM containing my crypto
keys, I'd want
the chip to self-destruct during an attempt to pry open the package
:).
You'd probaly better with battery backed RAM that would lose it power
(and contents) when it was removed from the circuit. Or else redisign and
use a PAL or some form or programmable logic that has a security fuse to
prevent it from being read.
Joe
--
"The Direct3D Graphics Pipeline"-- code samples, sample chapter, FAQ:
<http://www.xmission.com/~legalize/book/>
Pilgrimage: Utah's annual demoparty
<http://pilgrimage.scene.org>
7 p.m.
New subject: grinding down chips was Re: QX10 graphics board
(and contents) when it was removed from the circuit.
Or else redisign and
use a PAL or some form or programmable logic that has a security fuse to
prevent it from being read.
Oh come on. PALs are about as secure as a cardboard front door.
If you're going to allow unpackaging the chip, then remember that PAls
used the old fusible-link technology. And those links will be in the top
metalisation layer. You could see which ones were intact and which were
blown with a microscope. And I guess if you could open up a new example
of the same make and type of PAL without damaging it, you could blow the
links one at a time and keep on looking at the chip to work out which was
which.
But why go to all that trouble? Even if the security fuse is blown, it's
possible to reverse-engineer a PAL using well-known techniques of
applying various inputs and looking at the outputs. With a PAL there is
no way to have a hidden variable. Any flip-flop in the chip (wether part
of thr 'R' of a registered PAL, or made from the AND/OR matrix) will
appear on one of the pins (this is not true of GALs, unfortunately). So
with a PAL it's relatively easy to work out what's going on.
-tony
4:08 p.m.
New subject: grinding down chips was Re: QX10 graphics board
OK genius, you suggest something that NO ONE can break! I'm sure the
CIA, FBI, MI-5 and the Mafia would all love to know about it.
Joe
At 12:00 AM 1/28/06 +0000, you wrote:
(and contents)
when it was removed from the circuit. Or else redisign and
use a PAL or some form or programmable logic that has a security fuse to
prevent it from being read.
Oh come on. PALs are about as secure as a cardboard front door.
If you're going to allow unpackaging the chip, then remember that PAls
used the old fusible-link technology. And those links will be in the top
metalisation layer. You could see which ones were intact and which were
blown with a microscope. And I guess if you could open up a new example
of the same make and type of PAL without damaging it, you could blow the
links one at a time and keep on looking at the chip to work out which was
which.
But why go to all that trouble? Even if the security fuse is blown, it's
possible to reverse-engineer a PAL using well-known techniques of
applying various inputs and looking at the outputs. With a PAL there is
no way to have a hidden variable. Any flip-flop in the chip (wether part
of thr 'R' of a registered PAL, or made from the AND/OR matrix) will
appear on one of the pins (this is not true of GALs, unfortunately). So
with a PAL it's relatively easy to work out what's going on.
-tony
9:19 p.m.
New subject: grinding down chips was Re: QX10 graphics board
So, you think
that the NSA has cracked ROT13?
Maybe NSA, but certainly not the FBI (IMOHE).
9:41 p.m.
New subject: grinding down chips was Re: QX10 graphics board
On 1/28/2006 at 6:19 PM Fred Cisin wrote:
How do they do with content that has been run through ROT13 TWICE?
oooo...that would be tough for them. But they'd swagger around and claim
that they had the best people for the job. They once (and for all I know,
may still have) had their hearts set on automated forensics. Dump your
hard disks, floppies and CD's in this magic machine and it would
automagically dig up all the dirt. No experience necessary.
They put the project out for bids and got really peeved when we didn't bid
on it. I told them that their goal was unrealistic and that they'd
constantly lag current technology. Instead they got one of the DC-area
defense contractors to do the work. 400% over budget, they still had
nothing. Ended up scrapping the whole project, but they still had a chip
on their shoulder (I didn't say "I told you so", honest!). We should have
probaby bid on the thing, but I had to much of a conscience.
You know, that 20-year old hangover called Star Wars is STILL being funded
(this year to the tune of $12B). Defense contractors have made a bundle
and still can't hit the side of a barn with the thing.
I can't talk about the spooks. Don't ask.
Cheers,
Chuck
So, you think that the NSA has cracked ROT13?
Maybe NSA, but certainly not the FBI (IMOHE).
7:37 a.m.
New subject: grinding down chips was Re: QX10 graphics board
OK genius, you suggest something that NO ONE can break! I'm sure the
CIA, FBI, MI-5 and the Mafia would all love to know about it.
Hmmm... A RAM-based FPGA (something like a Xilinx XC3000 or XC4000
series), battery-backed supply, configured, and then the configuration
memory removed should keep you guessing...
Removing the fevice from the board would lose the configuration, and I
doubt if you could remove the encapsulation without losing it either.
Actually, reverse-engineering an FPGA from the configuration data is a
non-trivial task as the format of said data is undocumented.
-tony
6:36 p.m.
New subject: grinding down chips was Re: QX10 graphics board
Joe R. wrote:
OK genius, you suggest something that NO ONE can
break! I'm sure the
CIA, FBI, MI-5 and the Mafia would all love to know about it.
That bit's easy - producing something that nobody can break but that
authorised viewers can still decrypt is the tricky part ;-)
7:01 p.m.
New subject: grinding down chips was Re: QX10 graphics board
In message <3.0.6.16.20060127101019.441734dc at pop-server.cfl.rr.com>
"Joe R." <rigdonj at cfl.rr.com> wrote:
You'd probaly better with battery backed RAM
that would lose it power
(and contents) when it was removed from the circuit. Or else redisign and
use a PAL or some form or programmable logic that has a security fuse to
prevent it from being read.
Problem is, SRAM has this annoying tendency of storing data for a really long
time if it's frozen beforehand.
It also tends to "remember" data bits that have been stored in the memory
cells without any writes for a long time.
Both of these "behavioural artifacts" have been documented extensively. It
is, however, a minute past midnight and I honestly can't remember where I
read about this...
--
Phil. | Acorn RiscPC600 SA220 64MB+6GB 100baseT
philpem at dsl.pipex.com | Athlon64 3200+ A8VDeluxe R2 512MB+100GB
http://www.philpem.me.uk/ | Panasonic CF-25 Mk.2 Toughbook
... Monday is the root of all evil!
7:08 p.m.
New subject: grinding down chips was Re: QX10 graphics board
Problem is, SRAM has this annoying tendency of storing
data for a really long
time if it's frozen beforehand.
It also tends to "remember" data bits that have been stored in the memory
cells without any writes for a long time.
Magnetic media also has a pretty good "memory", even after many, many
erasings and rewrites. I seem to remember you can get the "history" of a
magnetic area on a disk by looking at the very edges of that area, as it
builds up like growth rings in a tree.
William Donzelli
aw288 at osfn.org
7:19 p.m.
New subject: grinding down chips was Re: QX10 graphics board
In message <Pine.SUN.4.20.0601281905410.2120-100000 at osfn.org>
William Donzelli <aw288 at osfn.org> wrote:
Magnetic media also has a pretty good
"memory", even after many, many
erasings and rewrites. I seem to remember you can get the "history" of a
magnetic area on a disk by looking at the very edges of that area, as it
builds up like growth rings in a tree.
Solution: Melt the magnetic media down into a lump. For hard drives, use
thermite. For magtapes, a nice hot fire - the hotter the better :)
I'd love to see some TLA recover data from a hard drive that's been melted
down into a misshapen lump of metal :)
"What is that.. thing.. on the table?"
"Modern Art, Officer."
--
Phil. | Acorn RiscPC600 SA220 64MB+6GB 100baseT
philpem at dsl.pipex.com | Athlon64 3200+ A8VDeluxe R2 512MB+100GB
http://www.philpem.me.uk/ | Panasonic CF-25 Mk.2 Toughbook
... Shh! Be vewy quiet, I'm hunting wuntime errors!
7:23 p.m.
New subject: grinding down chips was Re: QX10 graphics board
On 1/28/2006 at 7:08 PM William Donzelli wrote:
Magnetic media also has a pretty good
"memory", even after many, many
erasings and rewrites. I seem to remember you can get the "history" of a
magnetic area on a disk by looking at the very edges of that area, as it
builds up like growth rings in a tree.
As I recall, it's a long hard process. I'm trying to recall the original
paper by some guy in Switzerland using a modified STM. I think he said he
could retrieve about 1000 overwritten bits per hour.
--Chuck
7:54 p.m.
New subject: grinding down chips was Re: QX10 graphics board
In message <200601281623530975.3A404C18 at 10.0.0.252>
"Chuck Guzis" <cclist at sydex.com> wrote:
As I recall, it's a long hard process. I'm
trying to recall the original
paper by some guy in Switzerland using a modified STM. I think he said he
could retrieve about 1000 overwritten bits per hour.
Neat, but I doubt any TLA has that much time to burn on the recovery of a
HDD.
I've seen a few crypto apps turn up with "plausible deniability" type
features. You basically encrypt two different versions of the file with two
different encryption keys. One key gets you, say, a shopping list or a photo
of a family pet while the other gets you some leaked source code or other
incriminating stuff. The theory is, you can't tell that the "hidden" file
is
in there unless you have the key that goes with it. Well, in theory at least.
--
Phil. | Acorn RiscPC600 SA220 64MB+6GB 100baseT
philpem at dsl.pipex.com | Athlon64 3200+ A8VDeluxe R2 512MB+100GB
http://www.philpem.me.uk/ | Panasonic CF-25 Mk.2 Toughbook
... A seminar on Time Travel will be held two weeks ago
29 Jan
29 Jan
5:51 a.m.
New subject: grinding down chips was Re: QX10 graphics board
Philip Pemberton wrote:
In message <3.0.6.16.20060127101019.441734dc at
pop-server.cfl.rr.com>
"Joe R." <rigdonj at cfl.rr.com> wrote:
That's why the encryption cards banks use for ATMs detect when they get
below a certain temperature, and overwrite the RAM.
Gordon
You'd probaly better with battery backed
RAM that would lose it power
(and contents) when it was removed from the circuit. Or else redisign and
use a PAL or some form or programmable logic that has a security fuse to
prevent it from being read.
Problem is, SRAM has this annoying tendency of storing data for a really long
time if it's frozen beforehand.
It also tends to "remember" data bits that have been stored in the memory
cells without any writes for a long time. 
27 Jan
27 Jan
2:18 a.m.
New subject: grinding down chips was Re: QX10 graphics board
I've been dying to ask this question. Can you
actually learn something (hopefully a whole lot!)
about a chip if you actually did this???
There are actually successful companies in just that business:
http://www.chipworks.com/
Be sure to click on the "Silicon Art Gallery" link at the top. For a while
Chipworks had a free CD with other interesting photos, I'm not sure if it's
still available.
Getting at the die isn't all that complicated - it's called
"depackaging"
or "decapsulation" in the trade. For ceramic and PGA packages with a metal
lid, it's generally pretty easy to just pry the lid off. For plastic
chips, dripping fuming nitric acid on the package eats it away but leaves
the silicon and bonding wires in tact.
There are labs that do this sort of work on a regular basis:
http://www.mefas.com/Decapsulation.htm
http://www.dpems.com/
Be sure to read bunnie's story of how to skirt the "security" features of a
PIC microcontroller by sticking a tiny piece of electrical tape(!) on the die.
http://www.bunniestudios.com/wordpress/?page_id=40
Modern, high end chips (P4, Opteron, Cell, etc.) are pretty hard to get at
though. Rather than having bounding wires at the edge of the chip, they
have tiny solder balls scattered across it. The chips are soldered/glued
face down directly to the package interconnect. I'm sure it takes some
serious wizardry to pry them loose again.
[Why yes, I am a chip collector...how could you tell?]
Cheers,
jp
http://www.saccade.com/
6:33 a.m.
New subject: grinding down chips was Re: QX10 graphics board
At 03:09 PM 1/24/06 -0800, you wrote:
Absolutely!
What if there
And I was a
bit disappointed that
he didn't grind
I've been dying to ask this question. Can you
actually learn something (hopefully a whole lot!)
about a chip if you actually did this??? down the top of the failed chip to do a proper
repair. :-) was some old chip for which there is no documentation.
The silicon dies usually have the manufacturer's name and model number
on them. Somewhere on-line is a micro-photogragh of an 1101 (or one of the
other EARLY Intel ICs). You can clearly read the model number on it. You
do need a GOOD microscope for this kind of stuff. I just recently got one
and I've started checking some ICs but I'm no expert. However I have a
couple of friends that are. One did failure analysis for Litton Laser and
another works for a company that supplies unmounted silicon die (ICs
without the package) and other custom ICs to industry and the military. If
you get REAL serious you can get VERY EXPENSIVE, tiny probes and actually
probe internal signals on the die.
If, given the availability of the proper equipment
(surface grinder?), you were able to take off say
.0001" of material at a time, or thereabouts ;),
Have you ever tried to open an IC that has a lid? It's not difficult.
I've dropped boards and had them hit on the lid and pop it off and leave
the rest of the IC intact.
would
you have the ability to photograph it, and have
something in the way of a working schematic?
Photograph it yes but a schematic is MUCH more involved since all dies
have multiple layers. It's sort of like trying to make a schematic of a
multilayer circuit board when all you can see is the top layer.
Joe
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
12:24 p.m.
New subject: grinding down chips was Re: QX10 graphics board
On 1/27/2006 at 11:33 AM Joe R. wrote:
Photograph it yes but a schematic is MUCH more
involved since all dies
have multiple layers. It's sort of like trying to make a schematic of a
multilayer circuit board when all you can see is the top layer.
Ditto that--and recall that we're talking here about an old graphics
controller; no crypto and (relatively) very old technology. I'd rather
haul out the logic analyzer to figure out funcationally what the chip was
doing and work from there. Often, old house-branded chips turn out to be
some commodity product with a few special mods. Much of the design world
is evolutionary, rather than revolutionary--it's safer.
Cheers,
Chuck
7:07 p.m.
New subject: grinding down chips was Re: QX10 graphics board
On 1/27/2006 at 11:33 AM Joe R. wrote:
If we're talking about the Epson QX10, then there's no problem at all.
The later version of the board (which is what's in both my machines)
contaisn a 7220 (data sheet available), 16 RAMs, a bit of TTL and an
Epson gate array. But the technical manual shows schematics for that _and
for an earlier version_. That doesn't have the gate array, it has more
TTL and a character generator EPROM.
There must be some differences between the 'extra' TTL and the gate array
(IIRC there's a latch on a bus in one version that's not in the other),
but they are sufficiently similar that you could recreate a replacement
for the gate array based on the schematics in the manual. The only thing
you wouldn't have is the exact font of the character generator, but I am
sure you could create something that was readable.
A bigger problem in maintaing the QX10 are the 2 hybrids in the SMPSU
(but IIRC part-scheamtics without component values are in the manual) and
the 2 ASCIS on the floopy drive. And you'd have probllems swapping out
the complete drive as this is a 1/3rd height unit (not to mention the
fact that you'd want to keep the original one since it's an interesting
voice-coil unit).
-tony
Photograph it yes but a schematic is MUCH more
involved since all dies
have multiple layers. It's sort of like trying to make a schematic of a
multilayer circuit board when all you can see is the top layer.
Ditto that--and recall that we're talking here about an old graphics
controller; no crypto and (relatively) very old technology. I'd rather 
6:31 p.m.
New subject: grinding down chips was Re: QX10 graphics board
On Tue, 24 Jan 2006 15:09:33 -0800 (PST)
Chris M <chrism3667 at yahoo.com> wrote:
At one point I had a lot of house-marked ceramic chip packages
that I thought were DRAM but couldn't be certain of. Since they
were in cerdip packages, it was easy to use a cold chisel to
split one of them open. And I used a microscope to look at the
die, and sure enough, it was marked as a 4116 DRAM part. I used
the rest of the parts as 4116s. This would be MUCH more
difficult with plastic parts.
And I was a
bit disappointed that
he didn't grind
I've been dying to ask this question. Can you
actually learn something (hopefully a whole lot!)
about a chip if you actually did this??? What if there
was some old chip for which there is no documentation.
If, given the availability of the proper equipment
(surface grinder?), you were able to take off say
.0001" of material at a time, or thereabouts ;), would
you have the ability to photograph it, and have
something in the way of a working schematic?
down the top of the failed chip to do a proper
repair. :-)
28 Jan
28 Jan
7:42 a.m.
New subject: grinding down chips was Re: QX10 graphics board
At one point I had a lot of house-marked ceramic chip
packages
that I thought were DRAM but couldn't be certain of. Since they
were in cerdip packages, it was easy to use a cold chisel to
split one of them open. And I used a microscope to look at the
die, and sure enough, it was marked as a 4116 DRAM part. I used
the rest of the parts as 4116s. This would be MUCH more
difficult with plastic parts.
Ayone else remember the 'Optic RaM', a poor-man's image sensor.
From what I remember they were a normal DRAM chip with
a quartz lid to
the package (it's been suggested you could take a cerdip 4116
or 4164,
knock off the top and replkce it with a window).
The idea was you focussed an image onto the chip. Stored 1's in all
locations, waited a bit and read out the contents. Those cells which had
been exposed to enough light would read out as 0's. Do it again with
different waiting times (note, once you've read out a pattern, you've
lost the charge stored in the cells, you have to start again, you can't
read the chip repeatedly without filling it with 1's), combine the
resulting bit patterns to get a sort-of grey scale image.
Disadvantages were the fact you had to capture the same image several
times to get that grey scale, and the fact that there was an insensitive
strip down the middle of the die where the address logic, etc was.
-tony
7:47 p.m.
New subject: grinding down chips was Re: QX10 graphics board
In message <m1F2pPI-000IyHC at p850ug1>
ard at p850ug1.demon.co.uk (Tony Duell) wrote:
Ayone else remember the 'Optic RaM', a
poor-man's image sensor.
Oh yeah - the Micron IS32 OpticRAM...
From what I remember they were a normal DRAM chip with
a quartz lid to
the package (it's been suggested you could take a cerdip 4116 or 4164,
knock off the top and replkce it with a window).
You can - there's a file called "kuckuck.zip" floating around that contains
schematics and documentation (in German IIRC).
I'd love to get a few DRAMs to play with. ISTR the "kuckuck" file I
mentioned
calls for a NEC D4164 in CERDIP packaging with a gold cover - apparently you
can put a soldering iron onto the cover then just push it off with a knife
(it's soldered on).
--
Phil. | Acorn RiscPC600 SA220 64MB+6GB 100baseT
philpem at dsl.pipex.com | Athlon64 3200+ A8VDeluxe R2 512MB+100GB
http://www.philpem.me.uk/ | Panasonic CF-25 Mk.2 Toughbook
... Only in your dreams, Commander. ? Troi
8:19 p.m.
New subject: grinding down chips was Re: QX10 graphics board
On 1/29/2006 at 12:47 AM Philip Pemberton wrote:
> From what I remember they were a normal DRAM chip
with a quartz lid to
> the package (it's been suggested you could take a cerdip 4116 or 4164,
> knock off the top and replkce it with a window).
I recall an early scanner that made no bones about doing just this. From
what I recall, you charged the DRAM with ones or zeroes (don't recall
which), exposed it to light and read it looking for discharged cells. You
could do gray-scaling with several consecutive reads.
This was back in the days of 4116/4164 and IIRC, it mattered whose DRAMs
you used. Intels cells decayed pretty quickly, but the NEC parts would
easily hold a bit for a few seconds.
Cheers,
Chuck
24 Jan
24 Jan
7:03 p.m.
On Tue, 24 Jan 2006 19:02:42 +0000 (GMT)
ard at p850ug1.demon.co.uk (Tony Duell) wrote:
Or, to be historically correct: when there's a repair depot to
ship the old part to, in a company where 'field repair' is
discouraged because the same 'problem' in the assembly will be
troubleshot time after time by every technician in the field, and
a permanent 'fix' will never make it into the new design because
the people in the development lab next door to the repair depot
who watch rework/repair trends never find out about the problem.
At 01:15 AM 1/24/2006, Gooijen, Henk wrote:
I know, I know! And I was a bit disappointed that he didn't grind
down the top of the failed chip to do a proper repair. :-)
What failed chip? All the RAMs were good...
Tony
Duell wrote:
And the board basically worked. The display at boot-up was sane.
Suspecting a RAM problem, I put in a complete new set of
4164s -- no change.
Hmmm, sounds like swapping to me ... ;-)
Of course, he can regain the extra point by giving an explanation
of why swapping a chip is allowed under the proper interpretation
of the official repair rules.
Err, when you know the replacement part is good, when you know the
replacement part is directly equivalent to the old part, and when the old
part cannot be repaired?
-tony
6821
days inactive
6826
days old
36 comments
17 participants
participants (17)
-
allain＠panix.com -
ard＠p850ug1.demon.co.uk -
aw288＠osfn.org -
cclist＠sydex.com -
chrism3667＠yahoo.com -
cisin＠xenosoft.com -
doc＠mdrconsult.com -
gordonjcp＠gjcp.net -
jfoust＠threedee.com -
julesrichardsonuk＠yahoo.co.uk -
korpela＠ssl.berkeley.edu -
legalize＠xmission.com -
pdp11＠saccade.com -
philpem＠dsl.pipex.com -
rigdonj＠cfl.rr.com -
root＠parse.com -
sastevens＠earthlink.net