From - Fri Nov 15 21:37:37 1996
Path: 
nntp.ulg.ac.be!news.belnet.be!swsbe6.switch.ch!swidir.switch.ch!news.grnet.gr!bt
net-feed2!btnet!newsfeed.internetmci.com!news.sprintlink.net!news-
peer.sprintlink.net!cs.utexas.edu!ennfs.eas.asu.edu!noao!not-for-mail
From: chavez@noao.edu (Enrique Chavez/)
Newsgroups: comp.sys.hp48
Subject: Re: SC 32K Memory Cards
Date: 14 Nov 1996 18:08:55 GMT
Organization: National Optical Astronomy Observatories, Tucson, AZ, USA
Lines: 251
Message-ID: <56fn7n$5at@noao.edu>
References: <328aaefe.12755265@nntp.ix.netcom.com>
NNTP-Posting-Host: sunset.tuc.noao.edu
X-Newsreader: TIN [UNIX 1.3 BETA-950824-color PL0]

Scott Vanzandt (j-vzandt@ix.netcom.com) wrote:
>Has anyone tried using the Smith Corona memory cards for their word
>processors in the HP?

Sure have!  Card is over 3 years old and works like a charm.  
Wish I had another card to modify.

A long time ago there was an article on how to modify the SC memory 
cards so that they can properly work in the hp48sx.  


Here is the original article:
--------------------------------------------------------------------
Date: 2 Jul 93 13:39:27 GMT
From: Allan Petersen <ap@dde.dk>
Newsgroups: comp.sys.hp48
Subject: How to use Smith Corona RAM cards in HP48


Here  it  (finally  :-))  is,  the  article on modifying Smith Corona
DataStore 32K RAM cards for use in the HP48 calculator....

For a couple of years many  people  have  talked  about  using  Smith
Corona  RAM cards in their HP48.  I have for a long time tried to get
one of them to see if it behaved like the Epson  cards  that  I  have
modified for use in the 48. Until recently I had no succes in getting
a SC card, but then suddenly I got the opportunity (thanks, Garth).

I have now succesfully modified a Smith Corona 32k RAM card  for  use
in the HP48SX.  It involves opening the card,  cutting a trace on the
PCB, and adding one wire from a test pad to Vdd.

It requires the use of a solder iron to solder a  thin  wire  between
two testpoints, but is otherwise straight forward to do...

I will describe here basically how to do it and what I found.

 However, first this:
*------------------------------------------------------------------*
*                ** DISCLAIMER **                                  *
*   This posting does not encourage anyone to do any physical      *
*   manipulating with their own products.  This posting may        *
*   contain errors and any damage done to your own machine or      *
*   to your cards is your own responsibility.                      *
*   I have no affiliation with HP other than owning some of        *
*   their products.                                                *
*   If the chance of a dead card (or calculator) after this        *
*   modification worries you I suggest that you leave it alone.... *
*                                                                  *
*------------------------------------------------------------------*

Well,  I  first  tested the original clamping voltage level and found
the clamping to be released at 4.21 volts and  to  be  enabled  again
when  the voltage was lowered to 4.16 volts.  That's too high for the
card to be used safely unmodified. When I last measured the Vcc on my
48,  I  got  4.22  to  4.24 volts and that's not much higher than the
clamp release voltage of the SC card.  Any kind of noise  or  voltage
tolerance  may  then  let your calc go dead....  and that's just like
what people report. If you're lucky, it may work.... otherwise not.

I then opened it and tried the "one wire modification".

The modification effectively disables the clamping of the  inputs  of
the  card when the supply voltage is low and makes it possible to use
any of the SC cards in the 48 (you loose the ESD protection  provided
by the clamping, though).

This  modification is much easier than the modification (changing the
voltage detector) that I described long time ago and seems  to  work.
Something  similar was proposed earlier from someone else on the net,
but until a few weeks ago I had not been able to get  a  SC  card  to
verify wether it really works..... but it does :-)

Here we go...

Step one:

Remove the battery and using a thin, sharp blade (scalpel or similar)
gently pry off the top cover from the card (the metal shutter  is  on
the  bottom  side)  starting  at  the  end where the battery was.  Be
careful not to bend the thin aluminum cover. It has to be put back in
place later. There is a small spring near one side of the card.

You'll  most  likely  damage  this,  but you can put a small piece of
conductive rubber in there when  you  assemble  the  card,  so  don't
worry. It's used to connect the two metal cover plates to avoid built
up of static charge on the top cover.

**BE CAREFUL** at this point - don't damage any of the components  on
the card.

When opened, SC card looks almost identical to the Epson card...

It looks like this (drawing provided by Garth Sweeth):

+-----------------------+       Legend:
|  /------\     ==B==   |       ------------------------------------
|  |  A   |     +-----+ |       A   - Battery Area
|  |      |     | @@@ | |       B   - Write protect switch
|  \------/     +-----+ |       C   - Voltage Detector
| +-------------------+ |       D   - Gate Array
| |  @@@@@@   7:::::: | |       1-8 - Gold test points (1-3) are the
| |  @@@@@@   8:::::: | |             ones mentioned in article.
| +-------------------+ |       @@  - Epoxy coated chips
| +----++-------------+ |       ::  - Blank chip mounting points        
| |  ::||::   5:::::: | |
| |  ::||::   6:::::: | |       * Cutout areas represent sections
| +----++-------------+ |         viewable through the original 
| +-----+  +-+          |         plastic frame.
| |     |  | |4  +----+ |
| +--+  |  +++-+ |@D@@| |
|    +-+|    |C| +----+ |
|      ++    +-+  3 2 1 |
+-----------------------+

The  test  points  1-3  (and  4)  are  exactly what I refer to in the
original Epson article.

The clamp enable input is on testpoint 4.
Ground is on testpoint 1 and Vdd is on testpoint 3.

Step two:

Again using a scalpel or a sharp, thin knife, cut the PCB trace going
from the voltage detector (marked "C" on the drawing) to testpoint 4.
It  must  be cut between the detector and the testpoint just where it
comes out under the epoxy covering the detector.  The best way to cut
the  trace  is  to make two cuts spaced about half a millimeter apart
and then remove the the thin strip of copper  between  the  two  cuts
with the tip of a pointed solder iron.

Now step three:

Solder  a  thin  wire (wirewrap wire) from testpoint 4 to testpoint 3
i.e.  to Vdd.  The wire should be isolated between the solderings  to
avoid  short circuits to the metal cover later.  Testpoint 4 connects
to the clamping disable input of the gate array (marked  "D"  on  the
drawing).

Use  the solder iron to melt a small grove in the plastic frame where
the wire has to pass above the frame. This will prevent the wire from
being  pressed  directly  against  the  top  cover  when  the card is
reassembled.

Step four:

Using some kind of glue (I used a two component epoxy cement)  fasten
the cover to the frame in its original position.

Remember  to  reinsert  a  spring  or a piece of conductive rubber to
interconnect the two metal covers....  hmm, well, I usually leave  it
out, but it *does* improve the resistance to statics :-)

You mau also want to glue on some kind of handle on the outer side of
the cover to make it easier to remove the card from your HP48  again.
Place it over the greenish bar just above the Smith Corona logo.

Step five:

Now  you  are  ready  to test the card in your HP48,  so put back the
battery and insert the card in the 48.

When you switch the calc ON, you get the message "Invalid Data Card".
That's ok,  the card hasn't been inilitialized yet.  Just proceede to
entering the port number,  where you have inserted the card and press
MERGE. Now check the amount of memory,  which should be 32k more than
before. The card is now ready for use.

One warning regarding step three:
The  gate  array input should never be left floating.  It's very high
impedant and an electric charge accumulated there may bring the input
to  any unpredictable state,  so it's not good enough just to cut the
trace.  It has to be high to disable the clamping,  so  it  can't  be
wired  directly  to  ground  (that  would  have been a nice soultion,
though).

Assuming it's a normal CMOS input means that it will  be  in  a  high
state when the input voltage is higher than typically half the supply
voltage.  It could have a small hysteresis, but wiring it directly to
the supply then implies that it could probably be high as soon as the
supply voltage goes higher than,  say,  0.8  Volt.  This  could  then
remove the clamping and everything should be ok.... but it might take
a higher  supply  voltage  to  bring  the  clamping  FET's  to  their
non-conducting state,  so the effective treshold could be much higher
after all.  It might also differ markedly from card to card and could
in some cases be as high as 3-4-? volts. At least it's unpredictable,
and that's why I first recommended changing the voltage detector.

However,  contrary to my expectations, after the modification I found
that  there  was  no  clamping action at all,  i.e.  wiring the clamp
control input of the gate array directly to Vdd entirely disables the
clamping.

Well, that's good news and that's bad news:

   It's  good news in the sense that you can then safely use the card
   in the 48 as long as you keep it in there.

   It's bad news in the sense that the  static  discharge  protection
   provided by the clamping is then gone...  so be careful not to zap
   the input lines when handling the card.

The metal shutter mechanism will greatly improve resistance to static
discharge though, but don't entirely rely on it.....

Well,  I'll  certainly use it in my calc.  I'll just avoid taking the
card out and carry it around in my shirt pocket.

As long as people keep this in mind I think this simpler mod will  be
very useful to lots of people. Also, those people who got a card just
to find out that their 48 didn't like it  will now have a way to make
use of it (or regret having thrown it out :-)).

As the price for a 128k card from EduCalc is about $US 90, i.e. about
$US 22 per 32k of RAM, I would consider using  the modified  SC cards
if available for less than, say, $US 12. At $US 1.00 (or 5.00 for that
matter) they're just a bargain.

I didn't try to wire the clamp directly to Vcc on the Epson card back
then,  but as the SC card and the Epson card seems identical, I would
consider this modification to be valid for Epson cards (32k and 128k)
as well.

Well,  I have also modified a  standard  512k  bytes  Epson  card  to
include  a  manual bank switch (a Hall sensor built into the card and
operated by passing a magnet under the calc :-)).
But that's a totally different story :-)
It was lots of fun to make it work, though.... and lots of work. It's
real crowded in there. There are 16 memory chips from the start.

Anyway - that's what I found - hope it proves helpful (or if not, then
just informative) to someone.

Have fun,  and remember that you get to use this information AT YOUR
OWN RISK.

   ---  Allan

*------------------------------------------------------------*
! Allan Petersen   Dansk Data Elektronik A/S   +45 4284 5011 !
! ap@dde.dk        Copenhagen, DENMARK                       !
!"They call it common sense. Then why does so few have it ?" !
*------------------------------------------------------------*

-- 
-----------------------------------------------------------------------------
                                Enrique Z Chavez 
Global Oscillation Network Group                   National Solar Observatory
chavez@noao.edu                              http://www.gong.noao.edu/~chavez
-----------------------------------------------------------------------------

