Club Amiga de Montreal - CAM/CAM_CD_1.iso/files/233.lha/LucasProject/Building

Text File | 1989-04-30 | 16KB | 345 lines

BUILDING IT:

Notes about Parts

I strongly suggest that you socket all the parts on the board. Since
the 020 and the 881 make up most of the price for the board, its a
good idea not to skimp on the sockets especially the 020, 881, xtal
and dual pin header. Gold Augat machine sockets are great. The Amiga
1000 is a very noisy machine, good quality sockets will help reduce
the noise.

You must buy at least 16 meg. 020 & 881 to guarantee proper
performance. I have tested the lucas board at 16 Meg. and it works
well, however some of the people who have built the board are running
just fine at 20 Meg. using 16 Meg. parts. We have tried installing 25
Meg. parts and have had some limited success, but as of this time I've
done little testing. If you have the parts please try it, and let me
know how it goes. Although these higher speeds are possible the board
was designed for operation at 16 Meg. so be adventureous at your own
risk and expence.

If you didn't buy the pals from me make sure you use -B2 pals. I have
used both National and MMI and they work fine. The pal equations in
the pals directory will complile with CUPL. If you are having any
problems let me know and I will modem you the JEDEC files. One of the
things I would like to try and haven't yet for operation above 16 Meg.
is using -D2 parts for the pals.

The four discrete TTL parts, with the exception of U9 must be F parts.
(More about U9 later)

The Caps should be of the type specified. You could use tants for the
.47 parts if you cannot find them in monolithic.

It is important to use the 330 ohm and 220 ohm resistors as specified.
They are standard for TTL clock termination.

The 30 ohm resistors are used to help control the over and under shoot
on the control lines *UDS, *LDS, R/*W, and *AS.

The 16 Meg clock oscillator should be inserted into a gold socket.
When you buy your parts you may want to pick up a 20 meg. xtal. Half
the people so far are getting away with it even though they are using
16 Meg. 020's and 881's.

The part you will need for U9 varies from one Amiga to another. Buy
one each of 7474, 74LS74, 74S74, and 74ALS74 (they're cheap). I will
explain later how to determine which one of these you should use and
why.

Soldering

If you haven't soldered up a board before, find someone who has. It is
a simple job bit requires some degree of manual dexterity and a good
fine tip iron. Everyone has their own method. I start with the
sockets, then do the caps and resistors and then the two connectors.
If everything is on hand you can do an excellent job in about an hour.
This is an important stage, don't rush it.

Dual Pin Header

Unless your in the BIZ the connector which connects the Lucas board to
the 68000 socket can be a bitch to find. However it is pretty easy to
make one up yourself from readily available parts. The easiest method
is to buy two 64 pin 68000-like sockets (again gold is best) and
connect them together with 64 bits of wire. The cutoffs from
capacitors and resistors are ideal. The distance between the two
sockets should be about 0.6 inches. Basically you want to raise the
Lucas board to the same height as the daughter board. I suggest you
cut four bits of wire, attach them to the four corners of the socket
and see if the board is of the correct height. Once this is done you
can cut 60 more bits of wire and install them between the sockets.
Test the height again, if its OK I suggest you solder in the four
corner posts (wires) then solder the assembly to the bottom (solder
side) of the LUCAS board. DO NOT use .025 sqaure posts this will
permanently destroy your 68000 socket.

When you put in the tant capacitors make sure you observe the correct
polarity. Positive always goes to the square pad. Note C19 has no
square hole, the positive side of the cap goes to the terminal closest
half inch hole in the board. If you still confused get out an ohm
meter an make sure the plus side of the Tants is connected to the 5
volt rail (pin 20 of U4).



INSTALLATION:

Installation is quite straight forward, rip out the 68000 and replace
it with the Lucas Board. Q.E.D.

To elaborate, remove the plastic cover and the EMI shield. 5 screws
for the plastic cover, 14 for the EMI shield. Gently and with
reverence, pry the 68000 out of its socket. I use a small bent
screwdriver. Now push the 64 pin header on the bottom of the Lucas
board into the 68000 socket. Do this carefully to insure that each pin
goes in properly and that you didn't bend any.

Things to be careful about.

The ribbon cable from the disk drive can be a pain. Make sure it is
lying as flat as possible, and that the bend in the cable takes up as
little space as possible.  Make sure the board is the same height as
the daughter board. If it is too low the Lucas board will not clear
the cable and header for the disk drive. This is also important if you
want to add some 32-bit wide memory to the LUCAS memory port (96 Pin
DIN connector) as this board must be high enough to clear the metal
oscillator can in the middle back of the Amiga motherboard.  Once you
have installed the Lucas board we are ready for check out. When it
gets working you can replace the plastic cover. The EMI shield will
have to have a small section cut out of the back, if you wish to
replace it too. So far no one has done this.  Be careful, don't force
anything, and don't PANIC !



CHECK IT OUT!

In all the years I've been hacking around with hardware stuff, I have
only discovered two golden rules for hardware design. Inside every
device is some magic smoke which makes the device function. As long as
you don't let this magic smoke out everything will work fine, however
in all the cases where I have inadvertanly let this magic smoke
escape, the device ceased to function. Please remember the following
two rules: 1) Don't let the magic smoke out.  2) Don't let rule one
scare you.

Okay, were going to turn the Amiga on now, if it doesn't work first
time don't panic. No one so far has blown an 020 or a 881, and no one
so far has done any damage to their Amiga. The only damage that has
been done so far is someone used .025 square posts to mate the LUCAS
board to the AMIGA. This damaged the socket. Once he replaced the
socket and replaced those tree trucks with the proper pins his board
worked just fine.

If you are really paranoid, use an ohm meter to make sure that there
is no short between the 5 volt rail and ground. You should get about
270 ohms.

At this time we are going to put a 74ALS74 in the socket for U9.

So, turn on the Amiga, and watch the power light on the front of the
machine. It should flash five or so times and then stay on. This is
the basic test, if it doesn't do this it will never do anything. If
you haven't done so already check to make sure all the chips are in
the correctly oriented ( Pin #1 in the top left corner as the board is
now) (( NOTE!!! IT IS POSSIBLE TO PUT THE 68881 IN INCORRECTLY. THERE
IS NO KEY PIN AS THERE IS ON THE 68020. MAKE SURE THE ORIENTATION OF
THE 68881 IS THE SAME AS ON THE LUCAS BORAD SILK SCREEN)) NOTE IT DOES
NOT HAVE THE SAME ORIENTATION AS THE 020. Also make sure that all the
pins are correctly seated. If you still have problems see below
or the article as to how to get hold of me.

Now back to the 99% of you who have had no problem so far. The next
indication that everything is OK is you will hear the Amiga's little
boot song. Don't let the fact that the song is slower disturb you. The
cashe hasn't been enabled yet and alot of useless fetches are going
on. Next you should see the Kickstart screen. We are now about 70%
sure everything is going OK.

Insert the KickStart disk and away we go again. The hardest test that
the Lucas board has to pass is putting up the Workbench screen. Just
prior to doing this, interrupts are enabled, the 020 cashe is enabled,
the 68881 is talked to to see if its there, and the autoconfig stuff
happens. If you get to the workbench screen your probably not going to
have any futher problems. So insert the Workbench of your choice. Try
running the bench marks to see if you are operating at the same level.
Run the Mandfast program in the Mandelbrot directory this will verify
that the 68881 is working properly. Next connect your peripherals and
insure that they work as well.

Something your going to have to do is tune your particular AMIGA to
the LUCAS board. This is done with U9. That is why I asked you to buy
several different types of 7474. If your operation isn't what it
should be try different types of 7474 in the U9 socket. The first to
try is a 74LS74. My buddy Eric Haberfellner is convinced that a Texas
Instruments 74LS74 works with all configurations he has run into. I
contend that the LUCAS board insists on having one piece of
unexplained magic. I've tried to fix it but it seems what works for
one AMIGA doesn't work for another. The important thing is that so far
no one who has bought a LUCAS Board has been unable to get it going.
Just play around and have faith. I really do know how the rest of
board works and once you get the right chip it is solid as a rock.

Historical Note:

I got in to this U9 mess when I did the board layout. PCAD allows you
to switch identical gates or sections of a chip in order to improve
the layout. When I gave the command that it was OK to switch gates I
forgot that I had 2 74F74 and one 74LS74 on the board and PCAD wasn't
going to complain so it switched an F part for an LS part and now we
all have to go through this nonsense. I think it sounds better though
to say you have to tune your AMIGA to the LUCAS board. It does show
how critical the timing for DSACK is, read the article if you want to
bored to death about DSACK.

Seriously it isn't a big problem.



EVEN FASTER..

So far, 20 meg. seems to be the limit. I've had it working at 25 meg.
but it was a little flakey. I'm trying -D2 PALS and the inevitable U9
tricks. I'll post it if I'm successful.

We have found that 12 meg. parts work just fine at 20 meg. The reason
for this, I think, is that when used in an AMIGA, none of the more
exotic 020 modes are used.  I designed the board for operation at 16
meg. and you should use 16 meg parts, but if you want a cheapest
possible way to get a LUCAS board going try 12 meg. parts, with a 12
meg. xtal. If it works you can try xtals all the way up to 20 meg. I
can't tell you definately it will work but no one so far with 12 meg.
parts has found any problem operating at 20. "Ya play de game, Ya
takes yer chances"



EXPANSION:

I am currently designing a 32-bit wide memory board using 100ns 1 meg
DRAM's. Unless your doing alot of floating point stuff, the Lucas
board won't give you miracles. If your LUCAS board is running at 20
Meg. (this is the configuration I know best) than you can expect a 35%
- 40% speed increase across the board. The whole point of putting the
020 in your Amiga is to one, give you a 32 bit wide upgrade path, and
two, the 256 byte instruction cashe sure helps. If you connect 32 bit
wide memory to the LUCAS board you will get about 200% speed increase.
Ofcourse floating point stuff, if it uses the 68881 will indeed appear
miraculous.

If anyone also wants to have a go at a memory board the expansion
connector is documented in the drawings and in the PCAD documents. If
you have access to fast SRAM an SRAM board with no wait states would
sure fly.

Once I get a good memory board layout I'll make the bare board
available as cheaply as I can, but don't let that stop anyone from
doing one also.



COMPATABILITY:

The LUCAS Board is works with most of the Peripherals I've tried.
Comspec Memory Board, Comspec Hard Disk, Microbotics Memory Board,
EASYL, and Supra Hard Drive. The only peripheral that has failed is
the Allegra 512K memory board. I can't guaratee anything beyond this
at this time. I will however post compatible and incompatible hardware
and software as reports come in.

So far LUCAS is very compatible with the software we've tried. The
only software that seems to fail are those games which use heavy copy
protection, like Carrier Command. Jet will work if you turn the cache
off with SetCPU. DeciGel will allow those programs which do move SR's
to behave. If you find programs which don't work please let me know
and I will post them for those who are still considering getting a
LUCAS board.

By the way FA-18 is gorgeous with the LUCAS board installed.

I welcome any information or bugs which you can provide.



A500 and A2000 OPERATION:

For the 500 there is good news and bad news. The good news is the
LUCAS board works in the A500. The bad news is that you will never get
the case back on. No one so far has tried it in a 2000 but I'd bet the
same is true. I have no plans to do a board for the 500 or the 2000. I
will gladly give anyone the PCAD files so they can layout the board to
a different form factor.

I built the board for the 1000 because I have a 1000 and I still like
it best (No 1 meg chip ram eh!) and because I'd like to save the 1000
from becoming a door stop.



PROBLEMS?

I've spent considerable time debugging the board to make it work. Here
are some of the problems I have run into and perhaps they might help.

Please remember that the whole purpose of making this board PD was so
that the Amiga community at large can help those with problems, but
more importantly, with a full disclosure of the design we can benefit
from all those hackers out there who will make improvements to the
board. Already we have boards that work at 20 Meg. and 25 isn't far
off. I will post on USENET and BIX all the traffic good and bad that I
feel should be generally known. I will also post the bugs, software
that doesn't work, peripherals which don't work, etc.

If you insert the Kickstart disk and it runs for a couple of seconds
then pops the kickstart screen back at you, and you are sure that this
is a valid kickstart disk, try changing U9 from an ALS part to a LS
part, or a 7474 part. this should solve this timing problem. This is a
generic fix for all problems, changing the type of 74 for U9 will make
some peripherals which don't work, work like a charm. It is a quick
test so try it if you getting strange behaviour.

If Kickstart is taken in OK but it gets to the point that it is about
to pop the workbench screen then goes bonkers try changing U9 again.
If this doesn't help you might have a gounding problem.

Gounding problems.

We have found that the Amiga 1000 has a very noisy ground plane. Some
Amiga's will only take one peripheral, two and they cough.

Dominic at COMSPEC came up with a grounding mod that I now do as a
matter of course to all Amiga's I open up. The problem he solved is
that there is a slight voltage differential between the Amiga daughter
board and the expansion connector ground planes. To remedy this
connect a piece of wire between the four pals ground pins on the
daughter board and the ground which surrounds the expansion connector.
These pals are J6, K6, L6, and N6 at the far end of the daughter
board. The ground pin is pin 10 which is the pin at the upper left of
the pal as you look at it with the Amiga facing you. Bus these four
ground pins together then run a wire to the gound which surrounds
expansion connector (Use an ohmmeter to be sure). Scrape away a small
section of the solder mask which covers the ground and solder the pal
ground wire there. This can make your Amiga much more robust with 3
and 4 peripherals attached. There is also a place on the LUCAS board
to attach a secondary ground wire. If you look at the board there is a
dedication which says "Special Thanks to Dr. Brain Gaines and ANAKIN".
Look right on top of the N in ANAKIN and you will see a small GND
marked. Solder a wire through this hole on the bottom of the baord
(scrap solder mask first) and attach it to the same place as the pal
ground wire.

It is always a good idea to if your having problems to check the
seating of all the chips, especially the seating of the 64 pin socket
which mates the LUCAS Board to the AMIGA 68000 socket. It is also a
good idea to check that none of the disrete chips or pals have a pin
bent underneath them instead of into the socket.

If you still are having problems E-Mail me on BIX or USENET (see the
article) and I'll try to help.