
Accelerating your Amiga 500 or How to Void your Warrantee with Class

by John Flanagan and Brad Fowles


At this time the management of AmigoTimes would like to warn its readers
that the following article contains information and situations which some
members of our audience may consider too frightening to contemplate.  Those
readers with faint hearts and delicate constitutions are asked at this time
to kindly proceed to the next article

In August, 1988 I released a public domain hardware project for the Amiga
1000, called the LUCAS project.  It consisted of a bare board, 4 pals, and
documentation, which would allow you to put a 68020 and 68881 into your
1000 for as cheaply as you could get the parts.  It was one the first
designs which would allow the '020 to run asyncronously to the Amiga clock
at speeds up to 20 megahertz.  This provided a substantial increase in
performance for applications like ray tracing and the calculation of
Mandelbrot sets, but only about a 1.4 times increase in general
performance.  In July of 1989 it was followed by a 4 megabyte 32-bit wide
memory board called FRANCES which provided a generic 2.3 times speed
increase by extending the 32-bit wide data path to the memory and remapping
the KickStart code into 32-bit wide memory space.  There were two major
reasons I did the project.  One, the A1000 had just been orphaned by the
A500 and the A2000 and I wanted to provide A1000 owners like myself with an
inexpensive upgrade to more performance.  Secondly, I wanted to help revive
the art of hardware hacking so folks could save some cash and hopefully
learn a little along the way.  The project was fairly well received, and my
faith in the ingenuity of other hackers out there was demonstated by a
number of hacks which solved some of the design flaws, fixed some interface
problems with other peripherals and a neat hack to allow the 68000 to
remain resident so it rather than the '020 could be selected at boot time.
As of the end of October I have shipped slightly over 1000 LUCAS boards.
In the last few months the orders for the LUCAS board have increased
substantially due to the fact that 68020's and 68881's have become
available on the surplus market for between $35.00 and $50.00 each.  Some
of the original builders of the LUCAS board paid as much as $300.00 for the
68020 alone.  There is information on the documentation disk for the
project on where these cheaper parts may be obtained.  I might as well go
on record at this time and tell you that in response to many requests I am
designing a 68030 daughter board for the LUCAS project.  The running joke
around here is that if I keep this up by 1994 my Amiga 1000 will be 4 feet
tall.  It will be a pretty plain vanilla design, but I intend to do a
little work latching the data bus so we can get speeds over 20 megahertz.
We still won't be as fast as those other boards because the FRANCES memory
doesn't support DMA and there is no special hard disk interface so the hard
disk's performance will be far below theirs.  Still for raw compute power
we should be their equal.  68030's have also started to become available on
the surplus market.

LUCAS and the A500

I get a great deal of mail asking me if the LUCAS board will work in the
500 or the 2000.  The simple reply used to be, no, I don't recommend it.
The power supply for the 500 can't handle the load and there are much
better though more expensive solutions for the 2000, besides, the form
factor is all wrong.  The time and money required to redo the layout the
board just wasn't available.  It wasn't long however before I started
getting mail telling me of successful installations into A500's and a few
into the A2000.  The last thing I want to do is to discourage hackers, so I
relented a bit to say yes, its possible for the A500, but say goodbye to
the case.  I still think it defeats the economics of it to try it in a
A2000.  So if folks were going to try hacking it into an A500 anyway, maybe
I could get one of the people who had successfully done it to document how
they went about it, so I could include it on the docs disk.  At this point
two things happened.  One, AmigoTimes asked me to do an article, and two,
John Flanagan sent me a UseNet message describing how he had put the LUCAS
and FRANCES boards into his 500, and kindly offering to write it up for me.
There is one very positive advantage of mounting LUCAS and FRANCES in an
A500 versus an A1000, and that, ofcourse, is that with the A500 you can
have access to 1 megabyte of chip ram.  Greg Tibbs will hopefully be soon
releasing his Rejuvenator board for the A1000 so that those of us with the
original AMY can have access to 1 meg.  of chip ram as well.  Over the next
couple of weeks John and I will work up a more detailed account of how to
do this hack so we can include it on the Documentation disk for the
project, but for those of you who would like to know what your getting
yourself into, here is an overview written by John on how he did it.
Personally I love the concept, and I intend on buying one of the used 500's
I see in the newspaper and trying it myself, (I can always use another
rendering engine) but I warn you, this is a REAL hack.
						Brad Fowles

Mounting LUCAS and FRANCES in the A500

When I first heard about the LUCAS board, I knew that I wanted to build
one.  I was working on an animation program from which I just couldn't get
enough speed to suit me.  Code optimization was only getting me modest
improvements in speed, so I figured it was time to try brute force.  The
LUCAS board sounded like a good design, which would provide an upgrade path
to 32-bit wide memory at an affordable cost, and which would also be fun to
put together.  The only catch was that Brad had designed LUCAS for the
A1000, while I had an A500.  Nevertheless, there did not seem to be any
fundamental reason why I couldn't use LUCAS, so I decided to give it a
shot.  The principal difficulty, of course, was the form factor; there was
no way to fit LUCAS inside the A500 case.  This did not bother me, since I
had never liked the A500 case style anyway.  To tell the truth, I much
prefer the A1000 case style to that of either the A500 or the A2000.  I
especially did not like the attached keyboard on the A500, but I wanted
compatibility with future chip-sets, and I couldn't afford the A2000, so I
bought the A500.  I decided to remount the A500 motherboard in a PC XT
clone case, which would give me plenty of room to hold LUCAS, and would
also provide the opportunity to detach the keyboard.  I already had an XT
clone case which held some PC-compatible floppy drives, which had been
hooked up to my Amiga with a home-made interface circuit.  The case I used
is available from Jameco for about $US 35.00, and has a flip top for easy
access while working inside it.  I would recommend purchasing a fairly
large case for this sort of project.  The motherboard is fairly good-sized,
and since the expansion bus ends up inside the case, any expansion devices
need to be able to fit inside the case as well.  Now, by remounting the
system in another case I had undone all the work put in by Commodore's
mechanical design crew, and a couple of packaging issues arose.  For one
thing, the keyboard needed to be detached and an extender cable made for
it.  The keyboard is ordinarily attached to the motherboard through a short
8-wire cable, which I extended by about 3 feet.  I also added some rubber
feet to the bottom of the keyboard in order to prop it up at a comfortable
angle.  Since I did not build a case for my keyboard, its exposed circuitry
lends a bit of Max Headroom atmosphere to my desk top.  It would be
possible to attach an A2000 keyboard to the motherboard if a suitable
adaptor were made for its cord, but the A2000 keyboard is not quite
compatible with the A500 keyboard:  the Control-Amiga-Amiga reset signal is
handled differently, and the power and disk-drive L.E.D.'s are missing from
the A2000 keyboard.  Another issue to be dealt with was the internal floppy
drive -- it had to come out.  I made a pair of two-foot extension cables
for the drive.  One cable was a 34-lead ribbon cable with female headers on
the end of it, identical to the one which comes with the machine except for
the length.  The other cable was a 2-wire power cable with headers to match
the original ones.  I remounted the drive in a spare drive case I happened
to have, outside the main system box.  I could have mounted it in one of
the internal drive bays of the XT case, but they were already filled by
other floppy drives.  The most difficult aspect of extending DF0:  was
finding a suitable face-plate for it.  I eventually made my own by cutting
slots in a blank disk-drive face-plate, but I would still like to find a
"real" face-plate.  Finally, with my A500 reduced to a bare motherboard
connected to its peripherals by extender cables, I mounted the motherboard
backwards in the bottom of the XT case, with the rear connectors facing
towards the front.  One of the support legs for the disk-drive rack at the
front of the case had to be bent up to allow the motherboard to fit
underneath.  Also, extender cables for the mouse, monitor and external disk
drives proved convenient.  My Amiga was now repackaged and ready for
turbocharging.  I put the LUCAS board together according to Brad's
instructions, and plugged it into the 68000 socket.  In order to accomadate
LUCAS, part of the RF shielding had to be removed.  There is a panel of the
shielding over the 68000 socket which is removable, and with a long enough
socket adaptor on the LUCAS board it should be possible to sit it on top of
the shielding.  I actually just cut some of the shielding away from that
end of the motherboard, but this is not good for the resale value of the
machine, so I would not recommend doing this.  There is one problem I found
with LUCAS on the A500.  When powering up the machine, there is about a
50/50 chance that the LUCAS clock will start up in the wrong phase to sync
with the rest of the Amiga.  When this happens, the machine fails to start,
and sits there with a green or grey screen.  The only cure is to turn the
power back off and then on again.  There is a fix for this problem on
A1000's, but I have not tried it on the A500, so I can't say whether it
will work.  I've just put up with it so far.  (Note from Brad; The best fix
for this is to use a 74F32 OR gate and 74F04 Inverter and invert the *C1
clock (pin 16 on the expanision connector) and run it and the *C3 clock
(pin 14 exp.  conn.) into the OR gate, and run the output to PAL U6, pin 7
(7MB2) on the LUCAS board disconnecting the signal which goes there now.
This solves the problem.) Adding FRANCES really made my machine a joy to
use.  Again, it went together according to the directions, and installed
without much trouble.  There are a couple of hacks described in the FRANCES
documentation which I had to perform.  One was the CAS* hack, which is
needed to reduce bus noise, and the other was the construction of an
expansion bus terminator, which also reduces bus noise.  The bus terminator
had not been necessary when I just had a LUCAS board.  I also added an
"A-Max switch," which is need in order to be able to use the FRANCES memory
with the Mac emulator from ReadySoft.  One last change needed to be made to
acommodate FRANCES:  the power supply.  The original power supply had been
sufficient to drive the LUCAS board (I had one of the 65 Watt supplies),
but for FRANCES I made an adaptor to a PC power supply.  FRANCES only needs
+5V and Ground, so the adaptor is fairly simple to make.  I also cut the
power connector off of the Commodore power supply, and attached it to the
PC supply, so that everything in my machine was running off the same
supply.  Thankfully, Commodore provides pinouts in the back of their A500
manual, so it is not too difficult to figure out how to make the power
cable.  The drawback to using the PC supply is that there is no room inside
the case to put the thing once the Amiga is in there, so it has to sit
behind the case.  Fortunately, A500 owners are already conditioned to put
up with external power supplies, so this is not too jarring.  It might be
possible to find a smaller supply which would fit inside the case -- an
A1000 supply, for instance, or perhaps a Baby AT supply.  In the end, I
have a machine which looks nothing like an Amiga on the outside, but
behaves on the inside like an A2500.  And it certainly is not just a
personal computer, but a personalized computer.  Not too bad for less than
the cost of a stock A2000 and some elbow grease.

					John Flanagan

The original article describing the LUCAS board appeared in Amiga
TRANSACTOR Volume 1 issue 3, December 1988 and the FRANCES board in Volume
2 issue 5, August,1989.

For more information or to order contact:

The Lucas Project
c/o Brad Fowles
RR #5
Caledon East
Ontario, Canada
L0N1E0


LUCAS 68020/68881 Accelerator board

LUCAS bare board ............. $40.00
4 pals ....................... $30.00
Disk, Docs, and mailing ...... $ 5.00

FRANCES 4 megabyte 32-bit memory board for LUCAS

FRANCES bare board ........... $60.00
2 pals ....................... $10.00
Disk, Docs, and mailing ...... $ 5.00
