Full 68040RC CPU, was: Re: hi

[Jeff Walther]

Date: Fri, 3 Oct 2003 15:23:30 -0600
Subject: Re: hi

My Q610 also has an LC.  Apparently the FPU emulation is not stable
enough to warrant use of any *nix on the US versions of the Q610?
I would like to put a full '040 in mine also.  Any suggestions where to
pick one up?
I've seen a few of them in the Apple section on Ebay.  I think in the 
Vintage section, but perhaps in the Motherboard section.  They also 
turn up occasionally in the Electronic Components area (under 
Business and Industrial).

In many cases it's cheaper just to buy a Q700 or Q650 motherboard and 
salvage the chip from it.   The shipping is more, but you have this 
nice pin protector surrounding the chip...

Jeff Walther



--
Quadlist is sponsored by  and...
Small Dog Electronicshttp://www.smalldog.com   | Enter To Win A |
-- Canon PowerShot Digital Cameras start at $299   |  Free iBook!   |
 Support Low End Mac 

Quadlist info:  
 --> AOL users, remove "mailto:";
Send list messages to:  
To unsubscribe, email:  
For digest mode, email: 
Subscription questions: 
Archive: 
Using a Mac? Free email & more at Applelinks! http://www.applelinks.com

Re: SCSI Manager 4.3 and the Q900/950?

[Derek Morton]

Determining the needed speed of the SRAM does not really work in that 
fashion...  although it is a common mistake people make.  What you need 
to keep in mind is that to run at zero wait states the memory must get 
the data on the bus within the (zero wait state) timing requirements of 
the CPU.  Remember that any time for decoding, processing and signal 
propagation must be included in the equation.  I don't recall what I 
ended up getting, but I think it might have been around 10 or 12 nSec.  
With crafty design you can do a sort of pre-fetch and simply hold the 
outputs off until decoding is complete.  This allows for slower devices 
to be used, but at a tremendous power cost (as all devices must be 
enabled all the time).

Is the zero wait states based on the speed rating of the SRAM or are 
there other specifications that you must take into account?   50 MHz 
is a 20ns period, so are you looking at 20ns SRAM?
I found some right angle connectors...  And only one tube at that!  I 
think I may end up using a secondary connector to bring the board back 
upright...  but there is the extra capacitance to consider .  
One of the advantages of being in the design industry is that I have a 
lost of resources to draw from and can readily get samples of many 
parts.

Did you find some?  Or did you decide on a work around of some kind? I 
had a similar problem when I was looking for the CPU connector for the 
x500 PMacs.  AMP has about 300 in stock, but their minimum order is 
1008 at $9 each and all the distributers are out of stock.   But for 
$9072 they'll run off another batch of 1008 for me...
The code will copy the ROM to the SRAM (very easy routine) and then I 
can modify it since it is in RAM.  To substitute the RAM version for 
the ROM version I can have the GALs disable the memory manager for any 
ROM calls.  This is the same thing needed to make the SRAM run as zero 
wait state memory (as opposed to a write through cache).

So you need to write some kind of utility to redirect the ROM reads to 
your SRAM card, plus add/modify the actual ROM code.   That's the part 
that is a complete mystery to me.   I wouldn't know where to begin, 
except at the beginning of several years of study of Mac programming.
There are a couple of board houses I deal with at work.  I am sure that 
I can get somebody to the board for me...  But with tooling charges it 
will run me around $400.00 per design to get five flats (I will 
probably put both boards on the same flat and likely three or four 
steps of each).  We have the facilities for doing prototypes at work, 
but the masked outputs from the board house are so much nicer.  :-)

I agree that a ROM SIMM is much better than replacing the motherboard 
ROMs.  It's just that getting the .050" board  seems to be difficult 
these days.  Will you be etching your own boards?
The 53CF96 is one change...  Let's just say that a few PPC 8100 Macs 
made the ultimate sacrifice for my creation.  I also swapped out the 
Ethernet Controller with a 33MHz SONIC.  This allows me to up the I/O 
bus frequency...  unfortunately to do that I also need to modify some 
of the timing registers in the controllers...  Hence the ROM mods.  It 
must be done in the ROM to be able to boot from devices on the built in 
busses, as the timing will fall way out of spec otherwise.  Since I had 
to make these ROM mods, I thought it would be nice to add a few extra 
changes.

Did you put in a 53CF96 for the SCSI chip or something more advanced? 
The CF would give you the potential for Fast SCSI instead of just 5 
MB/s and I think it's pin compatible with the existing 53C96. However, 
the part is very hard to find and rather expensive ($25/chip)--at 
least it was a couple of years ago.   There was a lot of 10 or so of 
them on Ebay in the last year or so.   Did you pick those up, if this 
is the chip you're using?

What did you do to the ethernet controller?
I already have the PROMs (1 set), but I need to get everything else 
done before I burn them.  The two GAL programs are done, so this winter 
I hope to get the layouts done (after review and verification).  With 
any luck I will have a lickity-split 950 this coming spring.

HEY!  What do you think about the idea of adding a ROM socket to the 
SRAM board?  It could be used to have a ROM based system (7.xx or maybe 
even 8.x).  Can you say Slot $E declaration ROM?  Kind of like a...  
What?  Classic?  4 MB would probably do it...  Trick the system into 
thinking the ROM is a drive...  It could probably even be flash...  
Hmmm...

Derek

--
Quadlist is sponsored by  and...
Small Dog Electronicshttp://www.smalldog.com   | Enter To Win A |
-- Canon PowerShot Digital Cameras start at $299   |  Free iBook!   |
 Support Low End Mac 

Quadlist info:  
 --> AOL users, remove "mailto:";
Send list messages to:  
To un

Re: Daystar Cache Speed

[Derek Morton]

The GAL in question was one of two used in a computer design and 
performed address decoding.  I know the device was properly programmed 
and I know the person who soldered it on the board (normally they are 
soldered during the reflow process, but this was a change to an 
existing board)...  After this individual (who is not particularly 
adept at soldering) placed the parts on we had a 25+% fallout rate...  
When I removed the devices in question and soldered them on myself the 
system worked as expected.  I concluded that the problem lay in 
overheating during the installation process.
How do you determine if it was overheated?  Or was that in the case 
where you know what hte contents should be, and if what you read out 
doesn't match what it was programmed with, then you know it was 
overheated?
I have never heard of a universal Quad package programmer...  And it 
seems the mechanical complexity would be make the design of such a 
beast quite daunting.  A hard wired socket (to a DIP) should work if 
you want to build one.

I think there's a "universal" PLCC adapter for this programmer which 
does all the PLCC sizes up to 84 pins, but I'll need to double check. 
Another possibility is to wire wrap a PLCC through-hole socket to 
straight pin headers and tell the programmer that it's looking at the 
DIP equivalent to the GAL.  Trade time for money...
I use the quad power Lattice GALs, and if I remember correctly the 
quarter power 16V8 draws 55mA or just over 1/4 watt.  If you use the 
standard (Bipolar) device you would be at 1 Watt...  And this does not 
include any drive current.  This may not seem like much power, but that 
power in a plastic device which is less than 1/4 sq. in. it will cause 
it to get quite warm.

So it is normal for them to be warm bordering on hot in operation? The 
chip in question was a Lattice GAL16V8.
This sounds like a recipe for disaster to me...  At least for 
non-destructive removal.  The spec is probably (I am not at work with 
my data books) something like 3-5 seconds at temperature, and I suspect 
you are keeping the device at temperature for much longer.

I use Chip Quik.   It is an alloy one melts into the existing solder 
which lowers the melting point.   If one forms a bead on each edge of 
the chip which has pins, one can heat all the pins on that edge. Then 
by moving the soldering pencil back and forth between the beads (four 
sides on a PLCC) eventually you get all four sides hot enough to come 
loose at the same time.   With the melting point lowered by the Chip 
Quik, I've found this safe (for the chip) and effective, until my 
experience with the GALs.  But I don't really know for certain that 
they were heat damaged.
Honestly, I would have to give this some thought...  a 16V8 can have up 
to (I believe) 17 total inputs...  but then it would have no outputs.  
The biggest problem is if the device program reads the outputs back in 
as inputs.  This allows internal latching which would be frustrating to 
figure out.  The first step would be to figure out which pins are being 
used as outputs and which are being used as inputs...  Then slowly go 
through the combinations keeping note of any odd behavior (such as 
latching).  A good piece of information to have before you begin is 
what the GAL pins are wired to and the function of those signals 
(address, data, chip enable, etc.).

How would you do this?  I was thinking, set up on a Proto-board with a 
5V supply, momentary contact switch for the clock, switches to the 
inputs and LEDs on the outputs.  Then run through combinations (256 
for an 8 input GAL) and do some repeats of the same inputs to see if 
the output changes to test for whether there's a state machine 
implemented.  Then (if it's just combinational) derive the logic using 
a Karnaugh map or software that solves a Karnaugh map, though I 
haven't any such software.
Derek

--
Quadlist is sponsored by  and...
Small Dog Electronicshttp://www.smalldog.com   | Enter To Win A |
-- Canon PowerShot Digital Cameras start at $299   |  Free iBook!   |
 Support Low End Mac 

Quadlist info:  
 --> AOL users, remove "mailto:";
Send list messages to:  
To unsubscribe, email:  
For digest mode, email: 
Subscription questions: 
Archive: 
Using a Mac? Free email & more at Applelinks! http://www.applelinks.com

Re: SCSI Manager 4.3 and the Q900/950?

[PeterH5322]

>AMP has about 300 in stock, but their minimum order 
>is 1008 at $9 each and all the distributers are out of stock.

Call the AMP rep and ask for an engineering sample.

Don't go overboard. One or two will usually be provided with no problem. 
More usually raises suspicion.

I usually ask for two because with fine-line parts one could be damaged 
in shipment, anyway.

Samples are pulled from inventory and then manually bagged, thereby 
leading to possible handling damage, not to mention damage during 
shipment.

-- 
Quadlist is sponsored by  and...

 Small Dog Electronicshttp://www.smalldog.com   | Enter To Win A |
 -- Canon PowerShot Digital Cameras start at $299   |  Free iBook!   |

  Support Low End Mac 

Quadlist info:  
  --> AOL users, remove "mailto:";
Send list messages to:  
To unsubscribe, email:  
For digest mode, email: 
Subscription questions: 
Archive: 

Using a Mac? Free email & more at Applelinks! http://www.applelinks.com

Re: SCSI Manager 4.3 and the Q900/950?

[Jeff Walther]

Date: Wed, 1 Oct 2003 21:22:52 -0500
From: Derek Morton <[EMAIL PROTECTED]>

I am looking at
making a memory card which allows the system to run full bore at zero
wait states (RAM and ROM).  Think of it as a cache card on steroids.
There are a couple of problems, one of which is that the memory is not
cheap and I have already dropped a bundle getting a stick of high speed
SRAM (0 wait states up to 50 MHz and 8 MB worth if I remember
correctly).  This surely will not be sufficient for any real-world
application, but it will show the maximum performance of the system.
That is a very cool idea.  And a very expensive idea.   I've been 
watching Ebay for over a year for some inexpensive SRAM and little 
has turned up.   The prices from distributers are scary. 
Fortunately for me, I don't need anything as fast as you do.

Is the zero wait states based on the speed rating of the SRAM or are 
there other specifications that you must take into account?   50 MHz 
is a 20ns period, so are you looking at 20ns SRAM?

My current plan is to make the PDS card first (those connectors were
IMPOSSIBLE to find
Did you find some?  Or did you decide on a work around of some kind? 
I had a similar problem when I was looking for the CPU connector for 
the x500 PMacs.  AMP has about 300 in stock, but their minimum order 
is 1008 at $9 each and all the distributers are out of stock.   But 
for $9072 they'll run off another batch of 1008 for me...

and I will have the card (once programmed) replace the
ROM on boot.
So you need to write some kind of utility to redirect the ROM reads 
to your SRAM card, plus add/modify the actual ROM code.   That's the 
part that is a complete mystery to me.   I wouldn't know where to 
begin, except at the beginning of several years of study of Mac 
programming.

 make a ROM SIMM.
I agree that a ROM SIMM is much better than replacing the motherboard 
ROMs.  It's just that getting the .050" board  seems to be difficult 
these days.  Will you be etching your own boards?

I have been doing electronic system design for years (it is my
profession) and while the ROM mods do present a challenge, I feel
confident it is quite doable.
I always believe these things are doable.  The trick is having the 
skills to do it.  It's very neat that you do.

As to my current board mods...
I have replaced the SCSI and ethernet controllers.  The new features
are simply waiting for some code to release their potential (along with
new crystals).
Did you put in a 53CF96 for the SCSI chip or something more advanced? 
The CF would give you the potential for Fast SCSI instead of just 5 
MB/s and I think it's pin compatible with the existing 53C96. 
However, the part is very hard to find and rather expensive 
($25/chip)--at least it was a couple of years ago.   There was a lot 
of 10 or so of them on Ebay in the last year or so.   Did you pick 
those up, if this is the chip you're using?

What did you do to the ethernet controller?
Thanks for the info on the ROM SIMM pin-out.
You're welcome.   It was tiny compared to the other stuff you're 
doing.   I made a small error.   Replacing the ROM chips on the 
motherboard with 2 Mb chips wouldn't work, as they're 4 Mb chips.   I 
was thinking in terms of pinout and you can't get a 4 Mb chip in a 40 
pin package, so my brain settled on 2 Mb.   I doubt this mistake 
would cause you any trouble even if it went unnoticed, becuase you 
won't be replacing the motherboard chips, but it's best if I mention 
it.

Jeff Walther

--
Quadlist is sponsored by  and...
Small Dog Electronicshttp://www.smalldog.com   | Enter To Win A |
-- Canon PowerShot Digital Cameras start at $299   |  Free iBook!   |
 Support Low End Mac 

Quadlist info:  
 --> AOL users, remove "mailto:";
Send list messages to:  
To unsubscribe, email:  
For digest mode, email: 
Subscription questions: 
Archive: 
Using a Mac? Free email & more at Applelinks! http://www.applelinks.com

Re: SCSI Manager 4.3 and the Q900/950?

[Jeff Walther]

From: [EMAIL PROTECTED]
Message-ID: <[EMAIL PROTECTED]>
Date: Wed, 1 Oct 2003 23:51:49 EDT
Subject: Re: SCSI Manager 4.3 and the Q900/950?
The idea of the internal bus taking precedence over the external bus
comes from Apple documentation.  I have never bothered to have
identical SCSI IDs on devices residing on both the built-in busses.
I have tested CD-ROMS on the internal and the external buses, both drives
being ID=3, and both are accessible with SCSI Manager 4.3.
So also are hard drives.
That is good to know.  Thank you, Peter.

Jeff Walther

--
Quadlist is sponsored by  and...
Small Dog Electronicshttp://www.smalldog.com   | Enter To Win A |
-- Canon PowerShot Digital Cameras start at $299   |  Free iBook!   |
 Support Low End Mac 

Quadlist info:  
 --> AOL users, remove "mailto:";
Send list messages to:  
To unsubscribe, email:  
For digest mode, email: 
Subscription questions: 
Archive: 
Using a Mac? Free email & more at Applelinks! http://www.applelinks.com

Re: Daystar Cache Speed

[Jeff Walther]

Date: Wed, 1 Oct 2003 20:42:18 -0500
From: Derek Morton <[EMAIL PROTECTED]>

I have found (at work) that if a GAL is overheated in the process of
removal or installation the program can be corrupted.  Whether or not
the device is actually damaged I cannot say as I discard any device
which is overheated (rather than have it act quirky or fail prematurely
in the field).
How do you determine if it was overheated?  Or was that in the case 
where you know what hte contents should be, and if what you read out 
doesn't match what it was programmed with, then you know it was 
overheated?

To program them you will definitely need at least one
PLCC adapter.  As I mentioned, if memory serves, both 16V8 and 20V10
devices are used on that card (although having it in front of you, you
will know for certain) and the 20V10 is a larger device.
I think there's a "universal" PLCC adapter for this programmer which 
does all the PLCC sizes up to 84 pins, but I'll need to double check. 
Another possibility is to wire wrap a PLCC through-hole socket to 
straight pin headers and tell the programmer that it's looking at the 
DIP equivalent to the GAL.  Trade time for money...

As for a GAL getting warm...  It is
possible that if the program became corrupted you could have had
contention on one or more outputs, but realize that GALs consume a LOT
of power (relatively speaking - even the quarter power devices), and
the faster they run the more they use.
So it is normal for them to be warm bordering on hot in operation? 
The chip in question was a Lattice GAL16V8.

A PLCC device is just about the most difficult thing to remove you will
likely find in an older mac.  The most effective way to
remove a PLCC is to use a SMD rework station with the correct size LCC
/ PLCC hot air tool.
I use Chip Quik.   It is an alloy one melts into the existing solder 
which lowers the melting point.   If one forms a bead on each edge of 
the chip which has pins, one can heat all the pins on that edge. 
Then by moving the soldering pencil back and forth between the beads 
(four sides on a PLCC) eventually you get all four sides hot enough 
to come loose at the same time.   With the melting point lowered by 
the Chip Quik, I've found this safe (for the chip) and effective, 
until my experience with the GALs.  But I don't really know for 
certain that they were heat damaged.

 It is always possible to figure out the program in a GAL, but it can
be VERY time consuming and tedious.
How would you do this?  I was thinking, set up on a Proto-board with 
a 5V supply, momentary contact switch for the clock, switches to the 
inputs and LEDs on the outputs.  Then run through combinations (256 
for an 8 input GAL) and do some repeats of the same inputs to see if 
the output changes to test for whether there's a state machine 
implemented.  Then (if it's just combinational) derive the logic 
using a Karnaugh map or software that solves a Karnaugh map, though I 
haven't any such software.

If it's a state machine, I guess I'd test to see how many of hte 
outputs are state machine outputs.  If it's only one or two, then 
once the combinational outputs are solved, the state machines might 
not be too hard.  If all the outputs are state machine outputs, I 
think it might be impossible in a reasonable amount of time.

A faster testing method might be to tell my chip programmer that it's 
looking at a 256 X 8 PROM or EPROM and ask it to read the contents. 
If I wire the address lines to the inputs on the GAL and the data 
lines to the output pins on the GAL, that should give me a file with 
the results I would generate above.  I think.  Does this sound 
feasible?   Is this on the right track?

Whether or not it is worth your effort to play around with this project
of yours is up to you to decide.
The first project I have in line are these old Outbound Laptops. 
These were early unauthorized Mac clones.   Many people remember the 
Outbound Notebooks, but the Laptop predated them and is very cool in 
its way.

Anyway, there seem to be a fair number of the Laptops still around, 
but they came either with an internal hard drive or an internal 
floppy, not both.   There was a connector that could be used for an 
external floppy drive and an external SCSI adapter (not at the same 
time).  However, most of the laptops seem to be missing these 
accessories.   If one's operating system or hard drive gets hosed 
there's no way to reload the laptop if you lack the external floppy 
drive.

The external floppy drive is a PC style floppy mechanism with a 
controller/interface card on the back.   That interface card is not 
too complicated, except that it includes this one GAL (and a somewhat 
hard to find 37C65 and WD92C32).   The SCSI adapter has the same 
issue.   So, if I can read or discover the contents of these GALs I 
can make more of the external floppy drives to support the orphan 
laptops.

This isn't a money making idea, as the cost of the components is more 
than folks are

Re: Apple Displays

[Dean Arthur]

Check with local schools which may have received castoff Mac monitors
and educational Macs when other schools converted to the eMac.  I used
to be able to obtain Mac monitors this way til sources locally dried up.
 Now I use copy of "Upgrading and Repairing Macs" [Que Corp., 1994] to
rewire IBM-type monitors to use with LC Macs when Parkinson's is least bothersome.

Max out the RAM and replace HDs with higher capacity drives.  Then give
away to single-parent families thru valley churches.

-- 
Quadlist is sponsored by  and...

 Small Dog Electronicshttp://www.smalldog.com   | Enter To Win A |
 -- Canon PowerShot Digital Cameras start at $299   |  Free iBook!   |

  Support Low End Mac 

Quadlist info:  
  --> AOL users, remove "mailto:";
Send list messages to:  
To unsubscribe, email:  
For digest mode, email: 
Subscription questions: 
Archive: 

Using a Mac? Free email & more at Applelinks! http://www.applelinks.com

Re: SCSI Manager 4.3 and the Q900/950?

[Jeff Walther]

Date: Thu, 2 Oct 2003 20:49:09 -0500
From: Derek Morton <[EMAIL PROTECTED]>


I really cannot imagine needing more than seven SCSI devices...
CD-ROM, M.O., Printer, Scanner...  This still leaves three for hard
drives (and I doubt many people are running SCSI printers).  What else
is there?  Perhaps I just don't have enough toys!  :-)
Let's see, CD-ROM, CDRW, DVD (unlikely because I don't think there's 
an MPEG2 decoder for the NuBus PM), MO, scanner, two DAT DDS2 drives 
(because I'm cheap, but big hard drives have outgrown DDS2 tapes), 
plus hard drives, and printer if one has one of the old GCC (?) ones 
or that single Apple model.

I do like using a separate CD-ROM and CDRW.  And I already have DAT 
DDS2 but hard drives have outgrown the 8 GB compressed capacity, so I 
pretty much need at least two on the chain, so that Retrospect can 
automatically continue on the second one.   I've never had a SCSI 
printer.

But in practice the hard drives would be on the JackHammer...

In practice you're probably right, but I asked more out of curiosity 
that any practical reason.   It was reading that section of the 
Developer Notes for the 900 that made me think, "What happens 
when...".

Jeff Walther

--
Quadlist is sponsored by  and...
Small Dog Electronicshttp://www.smalldog.com   | Enter To Win A |
-- Canon PowerShot Digital Cameras start at $299   |  Free iBook!   |
 Support Low End Mac 

Quadlist info:  
 --> AOL users, remove "mailto:";
Send list messages to:  
To unsubscribe, email:  
For digest mode, email: 
Subscription questions: 
Archive: 
Using a Mac? Free email & more at Applelinks! http://www.applelinks.com