Conference 7.286::atarist

>    I just upgraded my 520 STFM to 1 Mbyte and would NOT recommend the
>   experience to anyone. You need a light duty soldering iron with

    A word of warning - I noticed (belatedly) that there is a small inductor
filtering the RAM +5, and with the added drain of the expansion RAM, the voltage
drops to about 4.72v. This was on a very early 520; I piggy backed my chips
on top of the existing ones. Does the 520 STFM have room for two banks on the
motherboard?

    My fix to the above problem was to remove the inductor, double the existing
electrolytic capacitor, and add .1 uf caps in the spare holes provided.

							Rob

>>    I just upgraded my 520 STFM to 1 Mbyte and would NOT recommend the
>>   experience to anyone. You need a light duty soldering iron with

I did it this Spring by buying sockets, RAM chips, and caps myself and
installing it.  AT first there was a problem where I had accidentally shorted
a line, but after hours of searching, a friend of mine finally got it working.
If you are an electronics-whiz (I'm not), it ain't bad and darn cheaper than
a commercial expansion board.

>on top of the existing ones. Does the 520 STFM have room for two banks on the
>motherboard?

The 520STFM has an empty bank all ready for you.  Only the holes have to be
emptied of solder.  It is the same motherboard as the 1040, just not as
populated.

CHad

    By the way, I ran into a guy in Berkeley who does RAM upgrades for
    1040's to take them to 2 Mb using 256 kilobit DRAMs. He solders
    them onto the tops of the current RAMs and does it for around $400,
    depending on the current price of RAMs.
    
    According to this guy, the ST's memory controller (contrary to what
    Atari will tell you) can handle 8 rows (x16) of memory chips. In
    theory, one could go to 4 Mb using 256 kilobit chips...
    				- HBM

Oh yea?  I'd like to speak to this guy.

		- Jeff

(I'm desparate for more memory.  With a 200K RAM disk, a few small desk
accessories, and the memory hog Turbodos, I've less than 500K left,
which is inadequete to use the Mark Williams debugger on Whack, or run
the DigiSound demos.)

        
        The local Atari dealer does the upgrade on the 520stfm's,
        complete with chips for $169 for the upgrade to 1Mb.  He also
        will use 1Mb DRAMs (he is currently selling them for $31.95 ea.)
        which should bring it up to 2.5Mb.  He has a complete and proper
        setup ((de)soldering machine).
        
        For those in the Az area, the dealer is Computer works.
        
        Bob
        
                  

    This is intriguing.

    I'm hearing that a 1040ST can easily support 2Mb of RAM, and as much as
    4?  Is there such a thing as a "supported" configuration here? I'd love
    to double or quadruple my memory. 

-b

        The 1040 can support 2Mb of ram (2.5Mb for the 512STfm).  This is
        due strictly to the power supply.
        
        The easiest and safest way to do this on a 1040 or older 520ST is
        with the EZ-Ram memory upgrade.  No soldering or wire cutting,
        etc.  The upgrade with instructions is $149, plus the cost of
        chips.  You can use either 256kb chips of 1Mb chips.  And you can
        install it yourself.  It takes 16 256kb chips or 1Mb chips (can't
        mix them).
        
        E. Author Brown in MN, has just come out with, what looks to be a
        copy of EZ-Ram for about the same price.
        
        No, I'm not an employee nor am I getting kickbacks.  I am
        considering the purchase of this upgrade for many reasons, one is
        the problems we have with foil cracking on our (DEC) modules due
        to the repair process.  The Atari motherboard is less complex,
        however, it is still an additional risk.
        
        Bob

    See 95.0 for Tech Specialties upgrades as well, back when I had
    my 520ST I used their board that would let you go to 4MB.
    
    						-Dave

    
    
    In Germany and Holland there a few companies that will upgrade
    any ST to 4.5 MB. How they do the extra 0.5 MB I don't know.
    Supposedly, the ST's MMU (funny name for a chip selector) only
    supports 4MB. Maybe the extra 0.5 MB is used as a RAM DISK,
    but I wouldn't know.
        
    Or using the Rhotron/Rhobus (or whatever) you can go up to
    11 MB. They'll cut one pin on the GLUE chip, however. You'll
    need the R/TOS OS, yet another multi-tasking operating system.
    It is multi-user, though.
    
    One questions for the Hardy Boys: how dependent is the memory
    on the GLUE Chip. The situation is this: my brother and I both 
    have (very) early 520STs, both with the faulty GLUE chip (display image
    on color monitors is shifted to the right). My brother upgraded
    his ST to 1 MB, just piggy-back, no modifications whatsoever. He
    still has the faulty GLUE installed. I want to upgrade AND replace
    the faulty GLUE for a new one (cost me $80!). Can I upgrade to 1
    MB without modifications or does the new GLUE chip have influence
    on the new memory?
    
    Marc 

    The 1040 can clearly support up to 4 Mb (I've seen the ads for the
    upgrades).
    
    Using a board like EZram, the 1040 goes from 1 Mb to 2.5 Mb (one
    row of 256 kilobit chips replaced with a row of 1 megabit chips).
    Unfortunately, 1 megabit DRAMs are now selling for about $35/ea.
    
    The memory controller itself (and the DMA chip) can support 16 Mb
    (8 rows of 1 megabit chips).
    
    By the way, does anybody know the speed required? (100 nS, 120 nS,
    150 nS) ?
    				- HBM
    

	A french magazine called Micro Systemes has started a series
of at least two articles devoted to modifying the ST to be able to use the
full address range of the 68000.

	Apparently the addressing limit of 4 (or is it 4.5 Mbytes) is
mainly (only?) caused by the glue chip which causes a bus error if
an address outside the allowed space is seen. What the Micro Systemes
article is describing is the addition of a 'super glue' function which
takes the place of the glue chip for addresses outside the Atari defined
address space. This involves intercepting certain signals going to the
glue chip.

	Apparently normal software can be used quite happily, even in
the address space which would normally cause the glue to chip to signal
a bus error. As an additional benefit the add-on board includes a couple
of buffered slots for a standard bus known as G.96 (I have never come
accross this before).

	In principle the adaptor uses:-

		3 x 74LS541
		2 x 74LS645
		1 x 4 switch DIP
	Plus a few resistor arrays and decoupling capacitors.

	The whole thing works because of 3 PALs 2 x 20L8 and 1 x 16L8.

	The board effectively piggy backs onto the 68000 and the link
between the glue pin 12 and 68000 pin 6 is cut. The pin 12 of the glue
chip is then connected to the piggy back board, this is the NOT AS signal
and allows the board to not pass the glue chip signal for addresses outside
its normal limits and generate its own bus error signal if needed.

	Apparently the expansion slots are buffered and can be populated
with memory cards or I/O cards.

	Obviously the modified system will no longer fit in the original
case.

	I'm awaiting this months copy of the magazine to see whether they
publish the necessary information for blowing the PALs.

	The modified memory map then becomes:-

		00 00 00
			 Internal RAM
		3F FF FF

		40 00 00
			 Expansion
		F9 00 00

		FA 00 00
			 ROM and I/O vectors
		FF FF FF

	If I can program the PALs then this appears to be a simple enough
and cheap enough modification that I may have a go at it. However I'd
still like to add memory to what the article calls internal RAM.

	Is there anywhere around Marlborough where I might find the
EZ-Ram board and enough chips to bring my system from 512K to 1 Meg?

	Regards,
	Nigel

    I just received a kit to upgrade my ATARI 520 STF memory to 1M.
    The kit contains : 16 memory chips and 16 capacitors.
    So I plan to solder these memory chips (with chip holders) and their capacitors in the emty
    bank on the board, but I suspect it is not enough.
    
    When looking at the first original bank of memories,
    I feel I need to buy 3 resistors to connect 3 pins of each chips to
    the MMU because I notice 3 empty resistor spaces 
    near the MMU.                   
    
    
    Is it correct ?

    Maybe!!! I received 3 resistors with my kit but did'nt need them.
    Let me have your Email address and I'll try to look them out this
    weekend and let you have them. I remember the vendor asking me
    something about which board I have and when I said I don't know
    he said "I'll include the resistors just in case you need them".
    ...Howard

  I did my from scratch (no kit) and had a few problems that a good friend of
  mine was able to solve. At the time I was at school, not here at DEC, and
  I asked on USENET for help in solving my problems.  One of the questions
  I asked was about the resistors as I was figuring out thewhole thing
  by myself.  The question was (as has been said) a 'maybe'.  Try
  it out without and if you have problems, put in the 220K resistors.
  (I think it was 220K :-) )


 Chad

    If your talking about series damping resistors for RAS and CAS 220K
    is WAY to large.  In memory designs in generel I have seen values
    ranging from 10 ohms to 47 ohms.
    
    	Ray

    The resistors suppled by Ladbroke are 68 ohms. The should be used
    at resistor locations 71, 72 and 73.
     ______   _   ______   _   ______   _   ______   _
    |      | | | |      | | | |      | | | |      | | |
     ------   -   ------   -   ------   -   ------   -
      Existing RAM & Caps.        New RAM & Caps.
    
                    ! ! !   ! ! ! ! ! ! ! ! ! ! ! !
                                              1 2 3 New resistors 71,72,73.
    
    ...Howard

I just went back and noticed that I typed 220K.  Sorry

>  it out without and if you have problems, put in the 220K resistors.
>  (I think it was 220K :-) )


I meant 220 ohms (no K).  That is what I get for typing in auto-pilot mode.
220 was what was suggested to me.  I don't remember if I put 220 or 68
in.  We played with both.  This was a good 6-8 months ago and I haven't
looked since.  Works great though.

Chad

    
    Thanks to every body, my memory upgrade works (it seems).
    Just a few precisions about 68 and 220.
    The value of the resistors is 68 ohms
    The value of the capacitors is 220 nK
    
    The most difficult part is to empty the holes for ground or 
    power supply. 
    I even destroyed two junctions of the printed circuit, but you
    can easily replace them with a wire.
    
    The testing is very easy, you just need an ohmmeter to check 
    the eventual short cicuits.
    
    If you need more precision about testing, I can translate ,in a next
    reply, a short french article about it.
    
    I did not use any anti-static stuff, maybe I'm crazy, I do not know,
    I'm not expert with electronics.
         
    I forgot to mention I had a problem of size with the capacitors
    provided in the kit: they are very big and the heigth is limited
    because of the power supply board of the ATARI. 
    
    
    Eric.

Hi Eric,

	Static electricity is a very strage beast! In my experience it only
strikes those people (and the "parts" they're handling) that believe it exists
and take precautions!

	Whmmmmm, that reminds me of something else.

    Ah yes but by how long have you SHORTENED the life of a component
    by not taking static precautions. Digital have spent a lot of time
    and money researching static and have found that even bipolar
    components can suffer. The moral is "Be safe, wear a strap".
    
        Whmmmmm, are we being reminded of the same thing???
    ...Howard.

Ah, 

	A good point Howard, however, since nothing I ever build
works (not because of dead parts before I power up!) I never get to find
out by how much I shortened component life.......solid state cigarette
lighters anyone (one shot disposable!)? They say ignorance is bliss,
bliss is a state of serene happiness, when it comes to hardware I'm
one of the happiest people you could find.

	Cheers,
	Nigel

    I've been plucking up the courage to upgrade my STFM recently and
    have been told that it can be done by merely 'piggybacking' the
    new chips to the existing ones. No need even for any resistors (on
    the STM these ARE required).
    
    Is what I've been told correct ? Anyone care to give a step by step
    description (for an electronics moron who INSISTS on upgrading without
    the assistance of a dealer) of how to perform the upgrade for the
    STFM ???
    
    
    
    			Thanks,
    
    				Jamie.

    
    	You can piggyback the mem chips. I believe I still have the
    instructions, so send me your mailstop info (to EXPRES::FISTER)
    and I will send them to you (if I still have 'em, that is).
    	I was going to do the same to my OLD 520, but I would have to
    upgrade the MMU chip with it.
    
    					Les
    

    re 200.22
    
    Hello Jamie,
    
    	I thought that STFM's already had spare holes for a second set
    of chips. It was only the ST 's like mine where I have had to piggy
    back chips. 
    
    Cheers
    
    Pat K.
    
    Ps When I did mine I got the upper and lower byte cas lines rouns
    the wrong way. It didnt half make it work funny!!
                                
    PPS Good luck !!
    
                                

    So I just have to solder the chips in ??? Sounds easy enough. What
    has all this talk been of desoldering ??? Could it be that the
    spaces for the chips have been 'filled' in with solder ??? You need
    a special tool for this don't you ??? or would a hammer and a fine
    chisel do the trick ;-)
    
    No resistors or anything needed ???
    
    
    
    				Jamie.

    re .25
    Hi jamie,
    
    		Unfortunately, YES the holes are filled from the solder
    bath. Yes you have to suck them out, Yes you need extra components,
    16 decoupling caps. (and the memories obviously). I have a magazine
    article on how to do it, if you want I'll dig it out of the junk
    heap in the sky, (our loft space)
    
    Cheers
    
    Pat
    

    re my .26 and your .25
    
    Jamie,
    
    	If you are serious in that you dont know how to solder, then
    I would be inclined to say, get a friend who has some ability to
    do the job for you, The St motherboard is NOT the place to learn
    the fundamentals of sticking things together with glue!!
    
    Cheers
    
    Pat.

    
    	Jamie,
    
    	The instructions I have use the piggyback method. If your
    	ST has the holes, I suggest you use Keene's method. If you
    	are not experienced at electronics, the holes should be easy
    	to do, but piggybacking is certainly NOT.
    					Les
     

I "figured it out myself".  This is to upgrade the 520 STFM w/512Kb to
1024Kb.  There are different rev's of the boards that place the memory
in different places on the board.  

Here is what I did.
  
DISCLAIMER:  I am not a hardware hacker, am not that experienced, etc etc etc.
This is just a description of what I did.  It bears no guarantee and I
release myself of any responsibility if anyone attempts to copy what I did.
This is a *use at own risk* info.  No warranty is hereby declared or implied.


Basically,

* open up your machine and familiarize yourself with the motherboard.

* find the memory.  It will either be in a line

* * * * * * * * * * * * * * * *
* * * * * * * * * * * * * * * *

underneath the keyboard

or in a square

* * * * * * * *
* * * * * * * *

* * * * * * * *
* * * * * * * *

on the left somewhat back underneath the power supply somewhat.

I believe most of the "newer" revs have the straight line memory banks. 

(*
 *  = 1 chip )

There will be an identical looking block next to the memory but without chips.
Only the holes (filled with solder).  This is where your sockets go for your
chips.  I do not recommend soldering chips themselves.

* notice the decoupling capacitors on the existing memory.  I forget the
value but get 16 more of the same type.  You will need them.

* gather your materials.

16 256Kbit 150ns or faster chips  ( I used 100 and 120ns)
16 decoupling capacitors (look at existing ones in the existing
memory block)
perhaps some 68 ohms resistors.
16 good quality sockets.  Douple swipe type were recommended to me.

* remember to be careful about *static* electricity.

* desolder the holes for the new chips with a desoldering pump and soldering
  iron.  Be careful not to destroy any etchings or the metal rings on
  the holes.

* also desolder the holes for the caps.

* soldering the sockets and caps in.

* plug in memory chips.  Copy orientation of existing chips.

* try it out.

The repair guy at the store in Utah where I was said that some MMUs need the
resistors and some don't.  It depends on the manufacturer (there are about
three contracters for them he said).  Mine was the straight line of memory

*  *  *  *  *  *  *  *  *  *  *  *  *  *  *  *
*  *  *  *  *  *  *  *  *  *  *  *  *  *  *  *    | | | | |
                                                  | | | | |  resistor bank 68 
*  *  *  *  *  *  *  *  *  *  *  *  *  *  *  *
*  *  *  *  *  *  *  *  *  *  *  *  *  *  *  *


(note: above resistor bank does NOT represent an actual number -- I forget
how many there are.  There are 3 place where resistors could go amongst
the ones that are already there.  They are 68 ohm resistors.  Put three
new 68 ohm resistors in the line with the existing 68 ohm ones.  Try it
without first.


************************************************************************

That is basically what I did.  I had a little trouble desoldering as it was my
first time doing it.  I am not a hardware hacker.  I did end up shorting
something and it took a good friend of mine a lot of time to find it and
fix it but he is a hardware hacker and was mucho experienced at the whole
thing so he took all my memory off, desoldered all the sockets, put them
back in, etc etc etc. It works and saved me a pretty penny.

In my opinion, the hardedt thing is to desolder the holes without destroying
them.  Especially the power bus hole (??? - the big metal encased one).
Be super careful when soldering.  No stray solder is allowed.  
It is a good idea to clean up the board (defluxing ???) afterwards.

Some of my descriptions are vague.  The reason is that I do not remember
exact values for the capacitors or exactly how many reistors are already 
in the bank.  But if you look yourself at your board and the existing
memory and where the holes are these things fall into place.

If you try it, good luck. 

Find a friend or co-worker who knows a bit about hardware to help. ALso,
try a local user's group for help.

**********************************************************************

Chad

    
    	.29 looks good. A few soldering hints...
    
    	1. To remove solder from the boards, use a solder wick. This
    	   should be available at any electronics store. It's a coiled
    	   wick that absorbs solder. Uncoil some of the wick and lay
           it over the solder-filled hole, then place the soldering iron
           on the wick. This method reduces the risk of burning the
    	   board. The wick will absorb the solder and empty the hole.
    	   Cut the used portion of wick off and continue.
    
    	2. Get a GOOD soldering iron with a thin tip. Don't use a monster
    	   cheapo iron. Either buy or borrow a good one (suggest borrow).
    
    	3. I don't know anything about these capacitors, but some types
    	   require you to place them in at the right polarity. The best
    	   way to assure this is using the ones on the board already
    	   as an example.
    
    	4. When filling holes with solder, don't overdo it. Use just
           enough to fill the hole...once again, look at the solder
    	   already on the board, and follow that example
    
    	5. It will be much easier to solder sockets on the board
    	   than the chips. Make sure the chips are plugged into the
    	   sockets right (use the nick at the top of the chip as a 
    	   reference).
    
    	6. Cut the excess leads off the capacitors after you solder
    	   them in. Don't leave the cap sticking up too far.
    	
    	7. Recheck all connections and placement. Recheck again.

       	8.  If you turn the computer on and it doesn't boot, TURN IT
           OFF! Recheck your soldering (look for 'shorts', or where
    	   the solder touches two points and connects them), chip
    	   alignment and capacitor placement. If these look good,
    	   I suggest you find someone who knows electronics and let
    	   them look at it.
         
    	NOTICE: Just by opening the cover you are going to void
    	        any applicable warranty. If you blow the thing up,
    	        YOU will pay to fix it. 
    	   That was to assure you pay attention. Be careful, ask any
    	   questions you may come across, and you should do fine. This
    	   is NOT a hard job.
    
    						Good luck,
    						  Les
    

    Thanks for all the good advice. I'll give it a go in the near future
    and let you know how it goes ...
    
    
    
    				Jamie.

    
    I have performed this op on about 5 machines so far, 4 STFm's and
    my own STM.
    
    You will need 16*IC sockets 16*250 nano farad capacitors a very
    good soldering iron, a good solder sucker and an earth strap.
    
    remove the top cover and the keyboard, take out the floppy disk
    and remove the screening around the mother board, you may have to
    remove the power supply first (can't remember). Unscrew the mother
    board and take off the bottom shield, earth yourself to the board
    when doing all this work.
    
    Locate the 16 IC's and above those are the holes for the additional
    IC's. As was said in a previous note you now have to remove the
    solder from all the holes, it is not that difficult, but the power
    holes, pins 16 and 8 are fiddly as they suck the heat away quickly,
    keep the soldering iron on for awhile, you will find it easier to
    suck it from both sides. Remove the solder from the capacitor holes.
    When you have completed this nasty task check the board for shorts 
    ON BOTH SIDES.
    
    Solder in the 16 caps and the 16 IC sockets, checking your work
    as you go, when you have completed this task partly reassemble the
    unit, ie PSU and floppy and the keyboard, and turn it on.
    
    Hopefully she will work, if not turn off and check all your joints
    for bad solder joints and shorts. If it does boot then use the PD
    memory diagnostic to check it all out. and reassemble.
    
    Voila 1meg of memory.
    
    If you have any questions then please don't hesitate to contact
    me on 844-3356 or by all in one Peter Watt @ESO
    
    best of luck
    
    Peter.

    I want to do this upgrade but have not been able to source any DRAM
    in the UK for the last 2 months. MAPLIN, FARNELLS and many others
    have been out of stock and on back order all this time. Anyone in
    the UK know of a source? My local Atari dealer was offering to do
    it for #300 (he claimed to be one of the few with chips) but his
    supply has now dried up. If I can get the chips I can do it for
    #70.                    

    Have you tried Ladbroke Computing International on 0772-203166.
    That's who supplied mine when they were in "short supply".
    ..Howard

    Has anyone any information with regard the STplus ?
    
    Chris

    I have so far only managed to obtain 8 256K memory chips.  Is
    its possible to upgrade the 520 STFM to 750 K ????
    
    Chris

     re.-1.
    
    (I assume you mean 256Kx1 chips)
    
    'Fraid not. The 68K has a 16-bit memory bus. What you'd actually
    get is an address range of 1024K but with bits missing in the top
    512K.
    
    Like half of them.
    
    For example, if you populated the lower 8 bits of the bus, every
    alternate byte will be missing.

    More likely, the bus wouldn't work at all.
        
    (Where have all the chips gone - I'm interested in upgrading too!)

I am interested in upgrading my 1meg 520stfm to 2.5 or 4 meg

and also a DIY upgrade for an older style 520 to 1 meg.  Anybody have any
info/details/plans for such a do-it-youself upgrade?

What expander boards are available for 520stfms and 1040 st computers
that allow 2.5 and 4 meg upgrades?  

Can I replace one row (16 chips) of the 256kb chips (they are socketed) in
my 520stfm with 1megabit chips and therefore get 2.5 meg?  (that would leave
me the 256kb chips to do the upgrade of the other machine too :-)


Most important now however is a plan for DIY old style 520 memory upgrade.

Thanks!

Chad

RE: .38
(What expander boards are available for 520stfms and 1040 st computers
that allow 2.5 and 4 meg upgrades?) 

See .7 in this topic - EZ-Ram, etc.

From:	DECWRL::"[email protected]" "Randy Hosler  30-Jun-89 0851 PDT" 30-JUN-1989 11:53:01.54
To:	[email protected]
CC:	
Subj:	Re: RAM upgrades

Here's something I got off the net. I haven't tried it.
 
Randy
-----
 
 
                            The 520ST One Meg Upgrade
                            =========================
 
 
by Russell Vail
 
 
WHAT IS REQUIRED
----------------
 
     To do the upgrade, you will need: 16 256K RAM chips, 150 ns access 
time, e.g. NEC 41256C-15 or TMM4256-15 which is what I used.  You can 
order them from MICROPROCESSORS UNLIMITED, INC.,24000 South Peoria Avenue,
Beggs, Oklahoma 74421 (918) 267-4961, or if you know of a place locally.
 
     A good miniature soldering iron, with a small tip.  Approximately 
four foot of wire. If you have it, use wire-wrap wire. But you can use 
ordinary thin single strand wire.  De-soldering equipment. E.g. a sucker.
Normal tools, such as a screwdriver, tweezers, pliers etc.
 
How it's done:
--------------
1. Open up the 520ST case; pull off the keyboard connector; if there is 
shielding round the circuit board, take that off, and remove the board.
 
2. Identify all the RAM chips 
 
            (U45,44,43,42,38,34,33,32,30,29,28,25,24,18,16).
 
Desolder and remove all the capacitors adjacent to them.  This is to give
you more room to do the soldering of the chip pins.  Using a solder 
sucker, open up the holes that used to be occupied by the capacitors.  Do
this step carefully to avoid damaging your ST board.
 
3. Take each of the new ram chips in turn.  Bend pins 4 and 15 so that 
they are horizontal to the package, and cut off the tips of pins 4 and 15
so that they are about half length.  Place the RAM chip on top of the old
one (same way around!)with all the pins except, 4 and 15, touching the 
chip underneath.  Carefully solder each pin, except 4 and 15. The best 
way to do this is to put the tip of the iron on both pins, heat them both
for a second(no more) and add a little solder, which should flow around 
both legs.  After doing each chip, check all pins with a magnifying
glass.
 
4. Replace all the desoldered capacitors and solder them back exactly as 
they were.  Please note that your should not attempt to power up your ST 
until you have completely finished step 6.
 
5. Now for the wires.  The first one connects the RAS on pin 4 of all the 
new chips, to pin 18 of U15.  If you are using wire-wrap wire, then strip 
off five inches of insulation, Solder the end to pin 4 of the new U45. 
Measure the distance to the next pin, and shift over that much insulation
before soldering.  Continue until all the ships are done. Route the wire 
through the hole in the PCB next to U15 and connect to pin 18 of U15.
 
6. The second wire connects all the pin 15's of the new U45-U32 to U15 
pin 22.  The third connects from pin 15 of the new U30-U16 to U15 pin 21.
 
7. Now sit back and think.  Read through these instructions again.  Have 
you done everything? look at the board.  Are all the joints good?  If you
have a meter, check for continuity on all the wires.  Applying power with
errors might destroy a lot of chips.  If you are confident, then put the 
board back in the case (your workbench is probably covered with bits of 
wire and solder), connect it to the monitor, disk and power supply, and 
switch it on.
 
8. If it boots, then you are probably there.  If you have SID (supplied to
the developers only)  or Kissed, then look at the phys-top variable($42E)
which  should read $100000.  If it is only $80000 then you have only 512K.
If you don't have SID, load basic and type in this program.
 
                             10 def seg=0
                             20 loc#=1070
                             30 print peek(loc#)
 
This should print a result of 1048576. If you get half this (524288) then
you have only 512K.
 
 
DISCLAIMER: The author of this doc file or ST Report are not responsible 
for your workmanship, success or failure on this upgrade project.
 
 
 
 
========================================================================
Received: from hp-sde.sde.hp.com by decwrl.dec.com (5.54.5/4.7.34)
	id AA25782; Fri, 30 Jun 89 08:51:30 PDT
Received: from hp-sde.sde.hp.com by decwrl.dec.com (5.54.5/4.7.34)
	for norge::chad; id AA25782; Fri, 30 Jun 89 08:51:30 PDT
Received: from hpsad.HP.COM by hp-sde.sde.hp.com with SMTP ; Fri, 30 Jun 89 08:51:13 pdt
Received: from hpsadpk.HP.COM (hpsadpk9) by hpsad.HP.COM; Fri, 30 Jun 89 08:51:26 pdt
Received: by hpsadpk.HP.COM; Fri, 30 Jun 89 08:51:05 pdt
Message-Id: <[email protected]>
Newsgroups: comp.sys.atari.st
In-Reply-To: article <[email protected]> of Wed, 28 Jun 1989 21:55:40 GMT

    re .40
    
    This procedure is NOT complete. I used a much better one that was in
    the old ATARIST notes file. 
    
    It is essential that the RAS and CAS signals are connectred via low
    value resistors to the MMU chip o/ps. 
    
    Depending on the age of your ST, and the etch rev, there may or maynot
    be physical locations for the resitors. On mine the resistor PTH's were
    present. also output pads to take the wires to the new chips.
    
    Cheers
    
    Pat K

Article 17487 of comp.sys.atari.st
Path: hiatus.dec.com!shlump.nac.dec.com!decuac!haven!purdue!tut.cis.ohio-state.edu!mailrus!csd4.milw.wisc.edu!cs.utexas.edu!uunet!mcvax!ukc!acorn!moncam!emmo
From: [email protected] (Dave Emmerson)
Newsgroups: comp.sys.atari.st
Subject: 1MB piggyback details - how to [LONG]
Keywords: DIY, megabyte, memory, DRAM, piggyback
Message-ID: <[email protected]>
Date: 27 Jul 89 13:39:44 GMT
Organization: Monotype ADG, Cambridge, UK
Lines: 95


Aaagh!
Is it just my imagination, or does 10% of the mail in this group really
consist of requests for details of piggy back expansions?
Perhaps if I reply, they might desist? 

OK peeps, you asked for it..

First off get hold of 16 DRAMs in Dual In Line packages. These look like
the ones already fitted along the front right of your ST, if in doubt,
quote the section of the device marking which includes the number 256,
and ask for an equivalent part. There are far too many to list them all,
but basic specs are : 256K x 1 page mode DRAM, 150 nanoseconds or faster,
preferably in plastic package (cheaper!), 16 pin DIL outline.
They're not cheap, and prices vary enormously, so shop around.

Each chip has a notch at one end, to show which way up it is fitted.
With the notch uppermost, pin 1 is to the left of the notch, pin 8 is
at bottom left, pin 9 bottom right, and pin 16 top right. Armed with
this information, find pins 4 and 15 on each of your new chips in turn,
and bend them CAREFULLY through 180 degrees without breaking them off.
Bend them as close to the body of the chip as you can. These are the
RAS and CAS pins which work as chip select lines. The chips will normally
be supplied in black anti-static foam. Try to keep them there while you 
bend the pins.

Now, one at a time, sit a new chip on top of one of the built-in DRAMs,
with the notch at the same end (!!) and solder each of the other 14 pairs
of pins together. Make sure the chips are accurately lined up, use only 
the minimum of fine multicore solder, and watch out for 'bridges' of
solder between adjacent pairs of pins, and solder splashes on the PCB.
This takes care of all the address, data, power, and read/write_ lines.

Lastly, locate the MMU chip. This is a square, 68 pin device, U15 on
early machines, and probably the one nearest the DRAMs on all versions.
This has a pip in the middle of one side to indicate pins 1 & 68.
From the top of the board, they are numbered anticlockwise. Attach one
end of a short (300mm or so) length of thin INSULATED wire to each of
pins 18 (RAS1), 21 (CAS1LOW), and 22 (CAS1HIGH). Prototyping wire is ideal,
particularly the "solder-wrap" type sold in the UK as "Verowire" (R) for
around 5.5 pounds. Don't connect directly to the MMU, take the wires through
a suitable hole, and trace the pins to the underside of the PCB.
The wire from RAS1 should be soldered to pin 4 of ALL the new DRAMs. If you
are skilled enough to do it, a 33 ohm Resistor should be used in series as
near to the MMU as possible to dampen reflections on all 3 of these wires, 
but its omission is unlikely to cause problems.
The wire from CAS1LOW should be soldered to pin 15 of the LEFTmost 8 new
DRAMs, and that from CAS1HIGH to the RIGHTmost 8.

That's it. Depending on your confidence and proficiency, it should take
about half to 2 hours. DO try to do it all in one session, and avoid 
touching the chips as much as possible, they are static sensitive.
A use ful tip - DISCONNECT the power supply, and attach a lead from the
ground rail around the edge of the board to your metal watch strap. keep the
chips in their foam sitting on a bare area of the ground rail. I'ts not
a static safe workstation, but it's as near as most of you will get.

If you have a hard disk, unhook it before you first try out your handiwork!
If it doesn't work : don't blame me! If your ST runs, but still only with
0.5 MB, check for unsoldered pins, suspect duff chip(s). If it goes crazy,
check for solder shorts, check correct pins used on MMU, take it to your 
nearest hardware wiz with a print of these 'destructions'.

Finally, piggybacking 1Mbit DRAMs instead of 256K for 2.5Mbytes :
This is MESSY, the pinouts don't match up well, and several pins need to 
be bent through impossible angles. You will need to connect another flying
lead for the extra address from MMU pin 64 to A9 on ALL the 1M drams. 
Again the 33 ohm resistor aught to be included as before.
The relevent pinouts follow, I won't go into more detail as I hold the
view that if you can't figure it out from here, you shouldn't attempt to
do it. A useful hint though, apart (perhaps) from A9 (for refresh 
considerations), the address pin numbers can be swapped around anyhow 
you like, ie A0 can be used as A1 etc..

Viewed from above:

256K DRAM pinout		1M DRAM pinout

       notch				notch
1  A8    	16 Vss		1  Data in 	18 Vss
2  Data in	15 CAS_		2  Write_	17 Data out
3  Write_	14 Data out	3  RAS_		16 CAS_
4  RAS_		13 A6		4  UNUSED	15 A9
5  A0		12 A3		5  A0		14 A8
6  A2		11 A4		6  A1		13 A7
7  A1		10 A5		7  A2		12 A6
8  Vcc		9  A7		8  A3		11 A5
				9  Vcc		10 A4


The above is submitted in good faith, but you use it at your own risk.
If you spot any mistakes, please post to this group first, flame me
later!

Dave E.

Some folks in england have designed a ram upgrade board for 1040s and 
similar machines and have hd a batch made.  They are selling the ones
they have extra on a FCFS (fifo) basis.  If you respond to this, do so
at your own risk.  I or DEC am not responsible for your replies and anything
that happens.

These people are not making money from these boards.  Their run was
larger than they needed for themselves.

I have ordered a board.

Good luck

Chad

ps:  they have said they would send a quality photocopy of their pattern
to requestors who wanted to have the board made themselves.

Here is a USENET article talking about it.

------------------------------------------------------

Article 17555 of comp.sys.atari.st
Path: hiatus.dec.com!shlump.nac.dec.com!decwrl!sun-barr!cs.utexas.edu!uunet!mcvax!ukc!acorn!moncam!emmo
From: [email protected] (Dave Emmerson)
Newsgroups: comp.sys.atari.st
Subject: 2 & 4 Meg Expansion PCB
Keywords: DRAM, expansion
Message-ID: <[email protected]>
Date: 31 Jul 89 18:33:51 GMT
Organization: Monotype ADG, Cambridge, UK
Lines: 45

Sorry I'm a little late in posting this, my employer is one of those
traditionalists who expect people to WORK for their salary.
We've built and tested our board, and are happy to report no problems
in either 2.5 Meg or 4.0 Meg configurations.
The result looks reasonably tidy, and by replacing 2 electrolytic
capacitors on the motherboard with ones with longer leads we can *just*
get the metal screen back on. One was the reset timing capacitor
mentioned in 'hard disk autoboot' articles recently, so you might like
to do this anyway. Be sure to leave the leads long enough to lay the 
capacitor down flat against the motherboard.

One last point (before anyone sends any money), I'm not familiar with
the layout of the various releases of the ST, so if you can sketch
out a template to check the board won't foul the keyboard support
pillars etc.. in your machine, you'll maybe save yourself some hassle
later. Cut out a rectangle 85mm high, 108mm wide. Inside it, draw a 
rectangle, 15mm high x 48mm wide, its bottom right corner is 7mm up,
41mm in from bottom right of the template.
Align the template over the shifter chip in your ST, with the rectangle
towards the front of the machine, and check for possible obstructions.
If it won't go in, there's little more we can do but apologise, and 
suggest you use flying leads from a more convenient location, or else
buy a proprietary board.

Now the nitty gritty. The board costs 15.25 pounds each, PLUS carriage
if you can't collect in person. Allow 40 grammes (1.3 ounces), +
airmail rate if required. A further 5 pounds if you want a spare 256K
DRAM, and 4.05 pounds if you want the 16 way cable assembly. Please
pay in pounds sterling if possible, or allow a further 2 pounds for
bank charges. We'll post same day we clear payment, so use POs in the
UK if you're in a hurry. If you include your net address, we'll confirm
receipt & despatch. Make payable to H. Fearnham, & send to :
H. Fearnham
Monotype A.D.G.
Unit 16, Science Park,
Milton Road,
Cambridge	CB4 4FQ
England.

If it weren't for the prattle in some other newsgroups I'd feel guilty
about the amount of bandwidth I've consumed recently. Hopefully this is
my last posting on this subject.
All the best to y'all

Dave E.

    	
    	Here my experience concerning memory upgrades:
    	After comparing do-it-yourself upgrade possibilities
        with commercially upgrade boards, I decided for the
    	cheaper but more error prone do-it-yourself method.
        I upgraded a friend's 520ST from .5 meg to 1. meg
    	using the pick-a-back method. The 16 * 256 K chips
        I took from my 1040ST which I replaced by 16 * 1 M
        chips. For installation of the 1 M chips I modified 
        18-pin sockets to reflect the different pin layout
        by rewiring and cutting pins. Although I'm not very
        much experienced in soldering (I did my last soldering
        exercises some 15 years back), I obviously managed to 
        get everything right, since both machines are now running
        in their new memory configuration without any problems
        (up to now!). However one really needs very good 
        desoldering and soldering tools, an antistatic mat,
        good eyes, patience and quite some time (I needed 
        around 12 work hours altogether). The piece cost for
        this upgrade was essentially the cost for the 1 M chips.
        Now I'm looking for cheap harddisk upgrades - I have 
        already an SH204 - internally there is a 3.5 Tandon.
    
    				Matthias
       
    

Open up your SH204 and verify that there is an Adaptek board in it. 
This is the larger of the two boards, if there is two boards.  If so,
disk upgrade is as easy as hooking up a second drive to this board. 

You need to do the following:

- Make sure you have a complete backup.

- Obtain any ST506-style drive.  An old RD52 works fine.

- Repalce the 34-pin ribbon cable with one that has 3 connectors instead
of 2.  This will connect all three 34-pin connectors in parallel. 
Notice how the Adaptek is connected to the existing drive, and copy
that. 

- Obtain another 20-pin cable identical (but longer) than the one that
runs between the adaptek and the existing Tandon drive.  Notice that
there is an empty 20-pin connector next to the one that's in there. 
Connect the second drive's 20 pin to this connector.  Notice carefully
the orientation of the existing connector.

- Remove the terminator resistor pack from the "center" drive of the three
drives that are bussed together on the 34-pin cable.  Be sure to leave
it in (of find one for) the last drive.

- Configure the new drive to be "unit 2", the first one is already "unit
1".  This is usually a jumper setting.  Often, you can guess which
jumper and get it right.

- Splice into the power to power the new drive, or obtain a second
supply.  I have been able to run two drives from the SH204, but nobody
knows what it's capacity really is.

- Plug it together and turn it on.  You should still be able to acccess
the first drive.

- Get a copy of the SUPRA driver and support software, and use it to
format and repartition both drives.

- You're done.  Restore your backups.

Of course, I make it sound easy.  In reality, it really is.  Obtaining
the parts is the hard part.

     I just received 4 megs of ram through mail order.  Now I need to find
a memory upgrade kit.  I scanned through Start, and I saw about 4 different
kits that allow expansion to 4 meg.  If anyone has had any experience with
these kits could you please fill me in on how easy they were to install,
any soldering/trace cutting/chip pulling anything like that.  Did the board
fit nice and secure?  Any information is appreciated.  The 4 kits I saw were:

1)  Tech Specialties
2)  New World Upgrade
3)  Arco Board
4)  Z-RAM 3D/4Meg

Thanks
Bill Powers

PS - My system is an early 1040st

You might want to ask Robert Fitzpatrick at next Tuesday's NaVAUS
meeting.  He's done a lot of these, and would know what to avoid.

    
    There is a memory board from a german company which holds up to
    4 MB memory and is inserted into the sockets of MMU and video shifter,
    nothing else. In the case of the 4 MB on the board you have a total
    of 5 MB including Atari's own memory. Because the MMU cannot access
    more than 4 MB, one MB is used as a RAM-Disk. The price for an empty
    board without RAMs lists for about 100-150$ as I remember.
    
    The address:     GengTec
                     G. Geng
                     Teichstrasse 20
                     4020 Mettham
                     West-Germany
    
    
    Bernd
    

I have a friend who has upgraded, and has tried a few upgrade boards. He claims
the best is the Tech Specialties board. I think thats the same recommendation
of Bob Fitzpatrick of RF Microtech.

The catch is that while the board itself appears to be technically good, the
company itself (or rather the fellow running it) has a reputation of being
extremely difficult to deal with should something go wrong. No matter what
the problem is, his assumption is that "the customer is always wrong". You
may want to go thru a third party, such as RF Microtech, to avoid this. Or,
always pay via charge card.

I have found out that Bob won't be at tomorrow's meeting.

    	Another thing to consider in new upgrades is the space it occupies:
    does it leave room for any other add-ons like the PC-Ditto II, or the
    16 MHz accelerators?
    	Does anyone offer an expansion box [probably comming off the cart.
    port]  for the 520/1040 with lots of room for all these goodies? 
    
    I can see the ultimate, started-out-as-a-TOS-on-disk 520:
    
    TOS on ROM
    4 Meg memory
    16 MHz accelerator
    PC-Ditto II
    Spectra GCR
    DeskCart
    
    Something tells me that, desirable as such a beast might be, it can't
    be built. Hope the TT is expandible! - Chris
    ps. I'll have the DeskJet at the NaVAUS meeting 1/30 for use with the
       hand scanner. - C 

        I just received the JRI RAM+ memory expansion board. (JRI - John
        Russell Inovations.)  It is the one that allows you to use SIMMs
        and contains a second video shifter socket as well as a gen-lock
        interface connector.
        
        I will be installing it in my 520stfm, as soon as I sort out a
        minor problem with the modem interface - resistor acting as a
        fuse, due to another resistor in the -12V charge pump circuit
        cracking.
        
        The board sits in the area of U43, about the center of the 1st
        bank of rams and leaves enough room for my tweety board.  As to
        whether there is room for a T-16 or PC-DittoII, I don't know, but
        neither is very close to the 68000.
        
        The JRI board plugs into the video shifter socket, the video
        shifter into a socket on the JRI module.  A header attaches to
        U43 - a 256 x 1 DRAM to pick up some of the signals and either 4
        6 wires are soldered - 1 to pin 64 of the mmu, and the others in
        place of the pull-up resistors near the right hand side of the
        memory bank.
        
                . 4-wires for two 256k x 8 simms (1mB total)
                . 4-wires for two 1m x 8 simms (2.5mB total)
                . 6 wires for four 1m x 8 simms (4mB total)
        
        The cost from JRI was $125 plus shipping.  It looks to be nicely
        done.  However, I will be adding some double-sided foam tape to
        both keep the board in place and insulate the bottom.
        
        I'll update this note after I get it working.
        
        Bob

    Re: .46
    
> If anyone has had any experience with
> these kits could you please fill me in on how easy they were to install,
> any soldering/trace cutting/chip pulling anything like that.
    
> 4)  Z-RAM 3D/4Meg
    
    I have a Z-RAM 3D in my early 1040ST (rev 1). I only have 2 Mb in mine,
    however. It was fairly easy to install (took me about 4 hours or so).
    It is completely solderless.
    
    The memory board itself is a sandwich of two PCBs that has the same
    dimensions as the video-shifter box. It plugs into the video-shifter
    socket, and has a ribbon cable connecting it to a "pin-grid board"
    which, umm, plugs "around" the MMU.
    
    Memory ICs are plugged into the top and bottom of the memory board (top
    is first 2 Mb, bottom is other 2 Mb). The first sticking point is too
    see if the board will clear a capacitor in the video-shifter box. Mine
    did not, so I had to move it and re-solder it (so I lied a little). 
    
    The next trick is the pin-grid board. This has pins that wedge into the
    MMU socket. It will take a few tries to get this right (30 min or so). 
    
    Next, the board in my 1040 would not clear the RF shield.
    Recommendation: buy a pair of aviator's snips. Cut the shield.
    
    In addition, when adding 4 Mb, you may have to cut the bus-terminating
    resistors on the motherboard (no other trace cuts are required).
    
    It came with memory diagnostic software, and a RAM-disk. The
    documentation is excellent, including troubleshooting steps for scope,
    multimeter and logic probe. 
    
    Oh, yeah. The company says they'll do the whole upgrade for you for
    $30, and have your machine back to you in 72 hours.
    				- hbm
    
    

    
    I read a new announcement in some magazines which advertises a board
    containing a lot of stuff: 80386SX with optional FPU, 1 to 8 (?) MB
    memory, 16MHz 68000 with optional FPU, on-board cache, one free
    AT-slot for e.g. graphics expansions. I'm not sure if this thing
    is for Mega-STs only and fits into the Mega-bus, anyway the price
    of about 1000$ makes it unreachable for me. It's called "Delta Module".
    
    Bernd
    

        As mentioned in a previous reply, I have finished (?is one ever
        finished?) upgrading my 520stfm to 2.5 mB of memory.
        
        I installed the JRI RAM+ board by following the directions for
        the 1040st under the section for the Rev. C motherboard.  I have
        a Rev. D motherboard, but have found no problems.
        
        I removed r94 and r93 using a desoldering iron (from Radio
        Shack), along with sucking out a couple of +5V pad and a
        ground pad.
        
        I added two wires from +5V to the left pads of R94 and R93.  I
        then ohmed out the connections to insure that I attached the
        wires to +5V.
        
        I ran the green wire from the JRI RAM+ module through a hole in
        the board and attached it to the bottom of the motherboard to pin
        64 of the MMU.  I ohmed this out to insure that I did attach it
        to pin 64.
        
        I tied the red wire from CAS1H to the right pad of R94.
        
        I tied the white wire from CAS1L to the right pad of R93
        
        I tied the blue wire from RAS1 to the right pad between R94 and
        R91.
        
        I tied the ground wires to ground pads.
        
        I then took the video shifter out of its socket and put it into
        the right socket on the JRI board.  I plugged the bus cable into
        the video shifter socket.
        
        I attached the other cable on top of DRAM U43 and tack soldered
        the corner pins.
        
        I rechecked the wireing, put things back together and fired it
        up.  It still worked.
        
        So I turned it off and installed two 1 mB x 8 SIMM modules, 100
        nsec. (I purchased these from Microprocessors Unlimited (they are
        in Oklahoma) for $92 ea., plus shipping - I ordered Tuesday night
        and received them Thursday!!!)
        
        Turned the machine back on and like magic 2.5 mB of memory.
        
        I can increase this to 4 mB by removing R90 and R91 and attaching
        the remaining three wires and adding two more SIMM modules.
        
        I highly recommend this memory upgrade path!  It works and is
        easy to install.
        
        Bob

    
    Do you mean you don't have to take the MMU out of it's socket, just
    solder a cable to one pin?
    I'm asking because I have a 1040 Revision with a surface mounted
    MMU and saw no chance yet to add some memory. All upgrades I found
    had to plug into the 64pin square socket of the MMU in addition
    to the video shifter's socket.
    
    Bernd
    

        That's right - the mmu is not touched.  The only chip that is
        removed is the video shifter.
        
        Bob

On a somewhat different wave length to the previous replies, i have a question
about installed memory.

As some of you may, or may not be aware, i have just installed 2MB of memory.

Now i constantly find that i have the following sucessive error messages :-

Exception Error - Adress Error
                - Illegal Instruction
                - Bus Error
                - Trace Trap

My system then hangs and i need to reboot.

Is this definatively down to my memory upgrade, and if so, what can i do about
it?

Regards,

					Lewis.

- These days, many Macintosh memory upgrades have made 256K SIMMs available for 
  free. (Single In line Memory Module). I'm therefore undertaking the upgrade
  of a 520STf to 1 Meg with such SIMMs, and triying to allow future 2.5 Meg 
  upgrade.
- The signal assignments for these SIMMs are listed below.

  +----------------------------+
  | [] [] [] [] [] [] [] []    |
  |............................|
  1                            30

  (1) +5V	(2) CAS_	(3) data 0	(4) Addr 0	(5) Addr 1
  (6) data 1	(7) addr 2	(8) addr 3	(9) Ground	(10) data 2
  (11) addr 4	(12) addr 5	(13) data 3	(14) addr 6	(15) addr 7
  (16) data 4	(17) addr 8	(18) addr 9	(19) Not conn.	(20) data 5
  (21) WRITE_	(22) Ground	(23) data 6	(24) Not conn.	(25) data 7
  (26) Not conn.(27) RAS_	(28) Not conn.	(29) Not conn.	(30) +5V
  
  (9 signals for +5V, Ground, N-C + 10 Address lines + 8 Data lines + 3 Control
  signals)

- +5V and ground, A0 to A8, WRITE_ can be found on an empty 256K chip seat.
- A9 can be found on the MMU pin 64, via a 33R resistor (1R = 1 Ohm).
- CAS_, RAS_ can be found on the MMU pins 18, 21 and 22 connected to three
  56R resistors. There are 3 empty locations for these resistors on the board.
? Where can one find D0 to D7 ? Some kits seem to find them on the shifter chip.
  Does anyone know either where to find these lines or what are the signals of
  the shifter chip. BTW, I've a Revision B board. The shifter shielded box 
  contains a large IC, the shifter I presume, and the remaining space is crowded
  with discrete components, unlike many pictures I've seen in magazines.
- SIMMs sockets can be found, not too expensive (~8$)
- Should someone be able to provide the missing information, this method appears
  to be nearly comfortable method to make this sort of memory upgrade.

  I'm looking forward to your replies,
  Thanks in advance, Jean-Pierre.

Well, the last time I have been considering extending memory on my own was on
a 8-bit machine, and I did dot figure out that this time, I'm not extending
the memory by 512K bytes, but by 256K 16-bit words.
My guess then, is that the first SIMM will provide D0-D7, and the second 
D8-D15. The RAS_ signal selects the "banks", and therefore is the same for both
SIMM, and the CAS_ selects either odd or even addresses.
I found out that the D0-D15 data line from the cartridge port came to a group
of 4 circuits near the 68000 (pin 1), maybe latches, before going to the 
microprocessor D0-D15 pins. 

Is it possible to connect the memory data lines there ?

I think that it would be easy to extend to 4 Meg, with 4 x 1 Meg SIMMs :
- by disabling the internal memory.
- making 2 new RAS_ signals out of the address lines, in order to address 1 Meg 
  chunks.
- making new CAS_ signals out of A0 line.

Any taker ?
Still looking forward to a wise (and knoledgeable) advice.
Jean-Pierre.

Having no accurate information about ST hardware, I have to test by myself
the statements I do. Having not gotten much help from this note, I carry on
to self-reply and self-correct the information I give.
- The four circuits near the 68000 are connected to the microprocessor data bus.
  I've wasted a few hours trying to connect the SIMMs here : it doesn't hurt,
  it doesn't work either. 
- It seems that the RAM data bus must go first to the shifter, and from there
  to the 68000. I've seen a pins description of the STE shifter, which had both
  a RAM data bus and a microprocessor data bus.
- What I've told about the shielded box containing the shifter is false. 
  What I had seen was a memory extension card, installed inside this shielded 
  box, which "piggy-backed" the shifter, and hiding the discrete components.

  I'm still trying to connect these SIMMs, with less enthousiasm now, 
  but I won't resign!

Why not get most of the signals (including the data bus) off of the existing
ram chips? Thats what the instructions in .40 and .42 have you do.

	Ray

	On the existing chips, the data bus and the address bus seem to be 
	sort of multiplexed, they share the same lines. I don't remember 
	which, but one earlier reply described the pin assignement.
	Too bad !

Since ordinary RAM's don't multiplex address and data, the connections
to the ordinary RAM's won't either.

Rick Saul has done this.  He seems to have disappeared from ELF.  Anyone
heard from Rick lately?  Last I knew he lived in Hudson, MA.

From this notes file :
>>================================================================================
>>Note 200.42                      Memory Upgrade                         42 of 64
>>NORGE::CHAD "Ich glaube Ich t�te Ich h�tte"         108 lines  28-JUL-1989 15:17
>>     -< another procedure  * use at own risk* from USENET -- note 1megdr >-
>>--------------------------------------------------------------------------------
[ .... stuff deleted ]
>>Viewed from above:

>>256K DRAM pinout		1M DRAM pinout

>>       notch				notch
>>1  A8    	16 Vss		1  Data in 	18 Vss
>>2  Data in	15 CAS_		2  Write_	17 Data out
>>3  Write_	14 Data out	3  RAS_		16 CAS_
>>4  RAS_	13 A6		4  UNUSED	15 A9
>>5  A0		12 A3		5  A0		14 A8
>>6  A2		11 A4		6  A1		13 A7
>>7  A1		10 A5		7  A2		12 A6
>>8  Vcc	9  A7		8  A3		11 A5
>>				9  Vcc		10 A4

I'm not a hardware specialist but I can't figure how, unless data and addresses
are multiplexed, both buses can fit into a 16-pin chip ?!?
It's simply a guess. I may be wrong ! What I guess is also that it's the MMU 
that supplies the signal Data In/Out, that makes a difference between data and
addresses. 

>>Viewed from above:

>>256K DRAM pinout		1M DRAM pinout

>>       notch				notch
>>1  A8    	16 Vss		1  Data in 	18 Vss
>>2  Data in	15 CAS_		2  Write_	17 Data out
>>3  Write_	14 Data out	3  RAS_		16 CAS_
>>4  RAS_	13 A6		4  UNUSED	15 A9
>>5  A0		12 A3		5  A0		14 A8
>>6  A2		11 A4		6  A1		13 A7
>>7  A1		10 A5		7  A2		12 A6
>>8  Vcc	9  A7		8  A3		11 A5
>>				9  Vcc		10 A4

>I'm not a hardware specialist but I can't figure how, unless data and addresses
>are multiplexed, both buses can fit into a 16-pin chip ?!?
The list of pin numbers and signal names that you included in your previous
listing show how they fit in a 16 (or 18) pin chip. Data and address' are NOT
multiplexed (just read your pinout in -1). However (assuming you don't know
anything about drams) address' and address' ARE multiplex (commonly known as
row address and column address). The demultiplexing is controled by RAS and
CAS.

>What I guess is also that it's the MMU  that supplies the signal Data In/Out,
>that makes a difference between data and addresses. 
The MMU controls address', without having a pinout and chip description in
front of me, I doubt that it controls the data bus at all.

BTW - It looks like your simm is either 1 megabyte or A8 and A9 are unused.

To connect your SIMM follow the direction in the previous replies that I
mentioned. That is connect your 'addr0' through 'addr8' lines from the simm to
'A0' through 'A8' of the DRAM. Connect 'data 0' from the simm to 'Data in' on
the DRAM which corresponds to data bit 0 (you'll find that 'Data in' and 'Data
out' are connected together on the current DRAMS). Connect 'data 1' from the
simm to 'Data in' on the DRAM which corresponds to data bit 1. etc.. etc..
Keeping in mind that each DRAM (there are 16 of them) only supplies one bit of
data, while each of your simms (you will need two) supply eight bits of data
(for the total required 16 bits).


	Ray

	Thank you very much Ray, I'd just realized that each chip would only
	provide with one data bit (256K x 1 bit) and that 16 chips will provide
	with d0-d15 data lines. Thanks a lot.

	I underline than this way of extending memory is rather safe because
	you only need to solder on empty chips locations. The most difficult
	part is to remove the MMU from its socket and solder a wire from 
	pin #64 to get A9, for 1 Meg SIMMs.

	I'm going summarize the operations :
	You need : 
	- To read all the replies to the basenote. You will learn all you need
	to know and that I won't repeat. (test program, care for static
	electricity, ...)
	- A good friend who has just upgraded a Mac (520 upgraded to 1 Meg),
	or 2 new 1 Meg SIMMs. (upgrade to 2.5 Megs) If you have followed the 
	action, these two SIMMs are R/W in parallel, one for the low byte, 
	one for the high byte. (8-bit or 9-bit SIMMs will work, 100ns)

	- SIMMs sockets, sold by pair. (~8$)
	- Bus cable + wiring cable.	(~6$)
	- 2 x 56R (ohm) resistors.	(green, blue, black)
	- 1 x 33R resistor.		(orange, orange, black)

	This table is not complete, because I need to determine which 
	RAM chip provide which data bit. Not very difficult but, not to
	mention, the label of the chips vary from one revision of the main
	board to another.

	SIMM pin#	Signal		Where to connect it
	---------	-----		----------------------
	1, 30		+5v		empty location pin #	8
	9, 22		Ground		empty location pin #	16

	2		CAS_		via a 56R resistor 
					for the low byte SIMM	MMU pin # 21
					for the high byte SIMM	MMU pin # 22
	27		RAS_		via a 56R resistor	MMU pin # 18
	21		READ/WRITE_	empty location pin #	3

	4		Addr 0		empty location pin #	5	
	5		Addr 1		empty location pin #	7
	7		Addr 2		empty location pin #	6
	8		Addr 3		empty location pin #	12
	11		Addr 4		empty location pin #	11
	12		Addr 5		empty location pin #	10
	14		Addr 6		empty location pin #	13
	15		Addr 7		empty location pin #	9
	17		Addr 8		empty location pin #	1
	18		Addr 9		via a 33R resistor	MMU pin#64

					all to an empty location pin # 2
					Low byte SIMM	High byte SIMM
	3		Data 0		chip # ?	chip # ?
	6		Data 1		chip # ?	chip # ?
	10		Data 2		chip # ?	chip # ?
	13		Data 3		chip # ?	chip # ?
	16		Data 4		chip # ?	chip # ?
	20		Data 5		chip # ?	chip # ?
	23		Data 6		chip # ?	chip # ?
	25		Data 7		chip # ?	chip # ?

	I will try this and let you know.	
	Thanks to all that have tried before, and have contributed to this note.

	This is an update of the report about my experimental memory upgrade 
	using SIMMs. (old 256K from a Mac)
	I'm a bit disappointed because it does not work quite well.
	The best result I got was 640K. I'm suspecting the possibility of
	having blown the SIMMs because of all sorts of bad wiring.

	The value of the resistor to be inserted in the control lines CAS,
	RAS is exactly 68R (blue, gray, black) not 56R.

	About the RAM data bus : The RAM chips are soldered in a matrix
	of 8 rows of 4 columns. The righmost two columns are empty on a
	520ST. The lowest row, the nearest from the MMU, is for data bit 0,
	the next row up is bit 1 and so on. From left to right, column 3
	is for the high byte in a word, and column 4 is for the low byte.

	And this data bus is also connected to the shifter chip :

	shifter			: data bit
	----------------------------------
	pin#2	- pin#9		: d0-d8
	pin#11	- pin#18	: d9-d15
	 
	Can someone interpret the result of 640K ? 
	(No, I've not turned it into a PC :-)

	Can it be some bad order of address lines ? 
	I'm relying on the pin description in note .42.
	I remember that the old 4164 chips showed the same order for A0-A7, so 
	I think it's not bad ... 

    Having just installed a 2Mb SIMM upgrade to my 1040 STF and discovered
    that the new memory will not be seen by the CPU I raise this question:
    
    How does the TOS (1.4) recognize the amount of memory installed at
    power_up time. Trying a read/write at different boundary addresses??
    
    Also I would like to have feedback from anyone having installed the 
    ZYDEC 2/4S upgrade, any problems??
    
    Torbjorn
    

    Torbjorn,
    
    I installed the ZYDEC upgrade in my 520STFM a while back. Since then
    i've been getting lots of problems (i also had the 2x1MB SIMMS).
    Typical problems are that the memory seems to crash certain programs
    like PageStream/FirstWP/LDWorks, and there is sometimes small glitches 
    with NeoDesk. Apart from that there is no real major problem. VidiSt
    digitiser still works, and recognises the memory (which is the main
    reason for getting the upgrade). And the few games i use seem to be
    Ok.
    
    Does your machine boot at all? Did it come up with a screen pattern
    or colour? How firmly did you press down on the MMU adapter? Did you 
    attach the jumper cable supplied? Did you firmly apply the adapter to 
    the Video Shifter chip? I found that when installing, i had to add 
    _loads_ of extra bits to keep chip adapters on the chips. All in all, 
    i found it to be a very messy upgrade - but at least it works! 
    (sometimes ;-)).
    
    Where did you get the upgrade from?
    
    If you need any more answers, you can either reply here or mail me...
    
    Regards,
    
    				Lewis.

    Lewis,
    
    Thanks for the info.
    
    I've followed the installation manual exactly, discovering that I have 
    of the very few "early" motherboards xxx523-001. 
    
    Playing around with the scope last night I discovered that the A0-A7
    addresslines are always zero, while A8-A9 seems to be working. Scope
    probe is here at the SIMM connector. 
    
    I begin to think that one of the octal busdrivers on the board attched
    to the MMU is faulty. The machine boots with 0.5 meg memory OK. 
    0.5 meg is disabled via etch cutting according to the manual.
    
    As one of the above notes suggested I tried to change the PHYSBASE
    address to slightly under 0.5 meg. The result of this is that I see
    the bottom portion of the original picture at the top of the monitor
    and some random patterns when the video addresses are above 0.5 meg.
    The funny thing is that this random pattern is not stable but seems to
    change if you touch the flat cable carrying the address lines. To me
    this indicates that the output of the octal buffer supplying the A0-A7
    lines is floating. 
    
    Scoping the CAS signal at the SIMMS in this case shows that it is
    active (a lot of neg pulses) during the time the video is addressing
    above 0.5 meg.
    
    I'll call the Danish distributor today and talk the problem over and
    it's very likely that I will request a new MMU adaptor board.
    
    BTW, did you have problem to reinstall the top shielding, the plate
    touching the MMU adaptor board before it is settled all the way down??
    
    Thanks to all folks who have contributed with info in this note about
    signal names and pin numbers etc..
    
    Torbjorn
    

    I've now solved the initial problem with this memory upgrade.
    
    A close look at the MMU adaptor board revealed that one of the Address
    buffer chips was not properly soldered. Pin 16 (+5 ??) was loose, only
    touching the solderring pad on the PCB. Over lunch time I soldered it
    in place and for the first time I have indication on 2.5 Mb memory.
    
    Now there are only a couple of minor problems left, reassembling and
    heavy testing with all programs, just to be sure of 100% reliablity.
    
    I must find a way to make sure the MMU adaptor PCB stays in the MMU
    socket, perhaps I can locate some insulated material inside the plate
    shield and simply press the board by the shield itself??
    Any ideas??
    
    Please, keep this note living with your own upgrade experiences. It
    will surely help others.
    
    Torbjorn
     

    Torbjorn,
    
    With this particular upgrade i found i had to fold up some ribbon
    cable, stick  it to  the piggy-back chips and then force the case 
    down on it, to keep it in place. The actually SIMMs themself only
    slightly bulge the top of  the ST case. Mind you, i suppose  that
    for the available room in there, they done pretty well.
    
    The MMU adapter (i had the solderless upgrade) actually had to be 
    forced onto the MMU until it  really sounded like you were trying
    to push it through the board!
    
    I actually had to send the upgrade back, because the first set of
    SIMMs was bad, i forgot to mention  that, but  thankfully, that's
    not where your problem was.
    
    Regards,
    
    				Lewis.

    Over the weekend I put everything back together again and it seems to
    work as expected. I found that grounding the SIMM board ( as described
    in the manual) is very important. Without this I got some mysterious
    glitches and other odd symptoms. 
    
    I had to cut out and slightly bend the top shielding plate and now,
    when properly seated, it applies pressure on the MMU adaptor board.
    
    Everything is now stable and hopefully it remains so .......
    
    Torbjorn
    

	Hi all, and HELP!!!


	Okay, I have my Atari 520 appart in front of me, and I have been
	trying to get this sucker up to 1 meg all week. I have some 
	problems. First, I'm running the memory test (Willie Brown's)
	and it errors out on me with the following...

	Error at address 0008001c hex.
	Bad bit was 00001000 00000000 binary.
	Data bit D11 is probably bad.

	Press any key to continue or <ESC> to (blah,blah, blah)

	What does this mean. Does this mean that the 5th chip is bad?

	Also, when I bring up this test, it tells you how much menory
	you currently have in the system. It shows me I have 640k not
	1 meg. Does this mean that the lower 8 or the upper 8 cannot be
	seen? Lots of questions, no answers. I have called Willie to
	see if he can shed some light on it for me. Please help to 
	troubleshoot it with me.... 


	Thanks

	Harold Swaffield 
	KAOFS::H_SWAFFIELD

There is 1 bit per chip, across the 16-bit word. Swap the chip that you
think might be the offending D11 bit with a known good bit, such a 0 or
15, and see if the problem moves.  If it does, you found the right
chip, and you know the board is wired okay.  Replace it, and all will
be fine.  If You are sure you tried swapping around all the chips that
could possibly be holding D11, and the problem stays in D11, then the
chips are okay and you have to look for a broken connection on the data
bus between the CPU and the expansion board.

>>	And this data bus is also connected to the shifter chip :
>>
>>	shifter			: data bit
>>	----------------------------------
>>	pin#2	- pin#9		: d0-d8
>>	pin#11	- pin#18	: d9-d15
>>	 
>>	Can someone interpret the result of 640K ? 
>>	(No, I've not turned it into a PC :-)
	I see I am not the only one, but we got it by very different
	means.
	
	By the way, I must correct the pin assignement of the shifter :
	shifter			: data bit
	----------------------------------
	pin#3	- pin#10	: d0-d8
	pin#12	- pin#19	: d9-d15

	Same despaired cry : Help !

    
    
    Well now is the time !!!
    
    I have lived with My 1040stf (happily) for about 4 years now. Never had
    any problems, I  have never had to drop it from a height to make it
    work. It has always delivered... Used exclusively for Music/Midi and
    Cubase.
    
    However Cubase tiakes up lots of memory and with the purchase of a Hard
    DIsk this is starting to become something of a limitation..
    
    Lets be clear I will not get involved in soldering the machine myself I
    am not an engineer..
    
    I have seen adds notably from a company who has a product called
    Xtra-Ram Deluxe which they insist is a solderless upgrade (in the
    majority of cases) The system involves the use of SIMMS
    
    Given my lack of dexterity is this going to be as simple as they say
    and how do I find out if I am one of the Minority.
    
    
    					Paul.
    

After reading about the other experiences here, I think even I would not
tackle a memory upgrade on a 1040STf.  (The STe is a different story!). 
I'd call Toad Computers and get a quote, and compare that to the time
and energy required to installand debug a self-installed upgrade.

    	You can get a whole new STe with 4MB for $590 from Toad.  The
    installed upgrades run $350ish.  I bought a whole new STe rather than
    put up with the mechanical tricks it takes to add more memory to
    a regular ST.  I thought I could easily sell the old system for the
    $250 difference in price, but I haven't had any luck yet...
    
    Brian
    

    Yes
     But Toad is about 2500 miles from my place...
    
    These people assure me that any idiot can do this upgrade they say that
    they have thousands of happy customers who never had any problems.
    
    No soldering required is the Boast...
    
    I still don;t quite believe it is going to be taht easy..
    
    
    					Paul.  

	From the experience I have sometimes reported, I should say :
	It depends on your motivation.

	I am interested in knowing (not so) new hardwares, hacking when
	it can significantly save some money.

	My nephew, who owns the computer, is interested in playing with
	it and has been hardly able to stand the starvation my few hacking 
	days forced him to, but he was encouraged by the hope of being 
	able to complete some of his animation projects that hit the memory 
	wall.
	I am very frustrated for not having been able to fulfill his
	expectancies, but as a counterpart I have learned a lot about SIMMS and
	memory addressing, and have updated my knowledge which has remained
	to the state of a few years ago when I built 8bit static memory cards. 

	In all cases, my feeling is that it was already hard with those
	old times hardwares, it is even more difficult nowadays. 
	DO NOT go for it if you have no hardware experience.  
	Even ready-to-mount kits seem to require a little debugging sometimes,
	it seems that the hardware revisions may be another problem
	and finally the documentation is very scarce. 

	It really depends on what you expect.

    Why does the addition of a hard disk make you want to increase memory?
    is it because your usage of the machine has changed or does the disk
    require resident software/drivers that take up memory?
    
    Reason I ask is that I too run a 1mb ST for sequencing, etc. and want
    to get a hard disk. At the moment I find 1Mb enough for my needs
    but will I want to expaand memory after getting the hard disk??
    
    Tony

    Hi, all
    
    
    	I wrote an entry yesterday, but, it seems it never made it here!
    	I am not sure why. Anyways a quick synopsis to keep you up to date.
    	
    	First, I would like to thank Jeff for the wonderfull entries he
    wrote about the troubleshooting. Very neatly organized, but, I did 
    not use them. (sorry Jeff.) What I ended up doing was dialing Willie
    Brown, at supra, and talking directly with him. This was expensive,
    but, it made my memory work, and I still saved money in the process.
    The error message I got was a few replies ago, but, the premis of it
    was an errors at data chip D11. This refers to the actual chip on the
    rev "C" module. Starting at the left hand side of the memory chips and
    counting from zero (0) to 11, I replaced this chip, and voila, 1024
    (1 meg) of memory was the outcome. I still would not recommend this
    upgrade to anyone else. It was very, very angonizing. The piggy back of
    the chips and the cold solder joints were enough to drive you insane.
    As for troubleshooting. It took me aprox 3 days to get it up and
    running, as well as the long distance call. I could have gotten this
    procedure done locally for $200.00 from my Atari Users Group, but, I
    didn't have the cash. 
    
    	Total cost for the upgrade was $40.00 (80 ns 256x1 ram chips)
    				         5.00 (aprox for call to Willie)
    					 5.00 (misc supplies etc...)
    
    	Total savings			$50.00 Canadian.
    
    
    	Not too bad, but, not again. Next time, I'll find the cash and get
    it done professionally. 
    
    
    	Thanks again to all
    
    	Swaff :-) (Who now has 1 meg)

    re -2
    
    The addition of the hard disk has meant that O have a reduced amount of
    memory available due to the addition of I guess some extra drivers. I
    use Cubase and whilst this works OK. I do reach the limits occasionaly.
    More importantly I run Synthworks for my TX802 and this will not run
    ata all once the hard disk is up an running..
    
    I bought the hard disk circuitously from Trevor Worral how built it on
    the roll your own principle. I am not sure what it is exactly that
    takes up the extra memory. Perhaps he could elucidate here ?
    
                                                   Paul.
    					 

    re .84,
    
    I did two problem-free 1M piggyback upgrades to 520STm's just recently,
    so take heart; it needn't be the horrible experience that some people 
    seem to have had!  Not counting extended diagnostic testing it took me
    3-1/2 hours for the first ST and 3 for the second, from go to whoa.
    
    However, I wouldn't recommend the procedure unless you've got good
    tools and you're at least a reasonably proficient solderer.  IMHO a
    top-grade soldering iron is almost essential; i.e. a *temperature
    controlled* 30->60W job with a very fine chisel or conical tip, plus a
    heat-proof dampened sponge to wipe the tip on.
    You also need to use high quality, fine (1mm dia. or less) rosin cored
    solder - spend a few extra pennies here because I've come across a lot
    of "cheap" solders that won't wet properly, or give dry joints no
    matter how careful you are, or both!  I strongly advise using a
    copper-loaded alloy solder, it won't eat into soldering iron tips
    or partially dissolve fine tracks on PCB's.
    
    Just as an example, I use a Weller temp controlled soldering station
    (not cheap, but I've had it since I was in 8th grade and it's still
    going strong!), a 600F conical tip and 0.8mm Ersin Multicore Savbit
    solder.  The conical tip is nice for ST piggyback upgrades because I
    can get to the chip pins without having to remove the bypass caps.
    
    Other things that will make the job quicker and less of a hassle are:
    
    	- A nice BRIGHT work light, preferably an "extended" light source
    	  such as a fluorescent to minimise hot-spot reflections.  The
    	  brighter the better, especially if you're getting a bit
    	  far-sighted with age - the bright light contracts your pupils and
    	  gives you much better depth of focus.  A binocular magnifying
    	  loupe is very good too, but not really necessary.
    
    	- Use fine emery paper to lightly wipe the outside of the pins on
    	  the existing memory chips and the inside of the pins on the new
    	  chips.
    
    	- If you can get it, use solder-through enamel insulated hookup
    	  wire.  You can then daisy-chain a single wire from pin to pin to
    	  pin without cutting or stripping, and solder right through the
    	  insulation.  The "Verowire" that someone mentioned in an earlier
    	  note is a brand that comes in a handy pen-type dispenser.
    
    Above all, you have to be methodical and patient.  Cleaning and bending
    and soldering and wiring up the pins for 16 memory IC's is boring all
    right, but trying to track down and fix solder bridges/unsoldered pins/dry
    joints/etcetera on piggybacked chips with a mess of wires over the top
    of them gets old *really* quickly.  It pays to make haste slowly.
    
    Still, it's really not at all difficult if you do it properly.  And
    it's hard to beat the price - it only cost me A$36.20 (US$28.60) per
    ST!

    re: .-2 (PK)
    
    	The ICD H/D Driver has a cache option. Perhaps this is what's using
    up your memory ?   I would imagine that this caching option is standard
    with other drivers ?
    
    Jim
    
    
    

    Hi,
    	I'm wanting to upgrade my 1040STe to either 2 or 4 Meg (depending
    on price). I've seen lots of adverts for SIMMs upgrades e.t.c. but does
    anyone know the 'format' that the SIMMs chips have to be? Number of
    pins e.t.c.
    	Thanks for all replies.
    
    							rod.
    

    Hi Rod,
    
        As far as I understand it, the STE has slots which take
    up to four 30-pin SIMMS, which can be eight or nine bit.
    If nine bit (parity) SIMMS are used my understanding is
    that the STE will simply ignore the parity bit (whereas
    the parity bit is apparantly a requirement for use in PCs).
        The speed of the SIMM can be as 'slow' as 100nS although
    I believe from what I have seen in various Atari-related
    newsgroups that 120nS should work OK too, but I have no
    experience of this. I have four 100nS 1MB 30-pin SIMMS
    in my STFM (via. a 'Marpet Developments' SIMM upgrade kit)
    and have had no problems with it.
        I don't know if 'three-chip' type PC SIMMS will work
    though.
        FWIW, the Falcon requires 80nS or faster fast-page mode
    SIMMS.
        Perhaps the best plan would be to open up the STE and
    have a look to confirm that 30-pin SIMMS are installed -
    I believe all of the slots need to be populated (i.e. four
    'banks' of memory present) so there may be four off 256k
    SIMMS currently present if the present memory size is 1MB.
    
    I don't actually have an STE but hope this is of some interest,
    
    Brian
     

    Thanks for the information.
    
    					rod.