.5M to 2M UNIX-PC RAM expansion instructions

[James E. Prior]

With all the curiosity about expanding the .5M on DDD's el-cheapo UNIX-PCs,
I figured that it was about time to rummage though my archives for this old
gem that was passed along to me by Gary Sanders (gws at n8emr).  Of course,
ordering another UNIX-PC from DDS has kept needling me to find this stuff.

I cannot vouch for the '>'ed instructions, as my current UNIX-PC is a 3B1
that already had 2M on it.  I don't expect to expand the RAM on the 
stripped system I just ordered until chip prices fall.

<Hi Folks!
<
<Well, after the having my mailbox stuffed with requests (yea!) I decided
<I'd better get this out, so here's my story on "How I Changed My Memory
<Board From 0.5MB to 2.0MB".
<
<Before I start, let me say that Mark Hillard (gizzmo!mark), who was at
<one time, a NASA solder technician, requests that I emphasize
<the fact that attempting to de-solder components can cause damage to the
<board. Excess heat causes the foil to weaken which makes it very susceptible
<to damage. I guess what he's trying to say (and I agree) is DON'T TRY THIS
<WITHOUT HAVING DE-SOLDERING EXPERIENCE FIRST!! Even if you have experience,
<go to Radio Shack (like I did) and buy an experimenters board and practice
<soldering and un-soldering a garbage chip (or socket) until you get the
<knack back! Remember; the less time the iron spends on the board, the better!
<
<Ok, now for the good stuff...
<
<Equipment:
<
<	PCA D-60-00226-00 REV D Copyright 1984 0.5 Megabyte Memory Board
<		The only weird thing about the board is that there is a factory
<		cut trace near C32 and a 33 Ohm resistor across it. It looks
<		like R8 on the schematics, and since I can't find R8 on the
<		board, that's what I think it is. It's off of a clock
<		(20MHz ??) from the motherboard.
<
<	Weller TC202 60w Soldering Station
<		My girlfriend got this for me for about $100.00 for xmas.
<		They are GREAT! If you don't want to get this, there are
<		others at lower prices. These have a thermostat in the tip
<		to regulate the temp., that's the important part. Try to get
<		the pencil tip, all I had was the blunt end tip.
<
<	Silverstat AS196 Soldapult - LS197 tip
<		For those of you that have never seen one, it's a long (5")
<		tube with a plunger in it like a big syringe. You cock it by
<		pushing the syringe all the way in against a powerful spring
<		and it locks in place. When you hit the button, the plunger
<		jumps back sucking air through the hole in the tip. If you
<		have the tip over liquid solder, it sucks it right up!
<		I borrowed this (thanks Larry!) but they run about $20.00 I
<		think. Get extra tips too, that will help.

I bought two more of these used tools last Saturday at the Dayton Hamvention.
If you have trouble finding some yourself, I can stand to part with them for
ten bucks each.  

<	72 Sockets
<		I don't remember who made them, but mine had round holes
<		instead of pressure clamps. Find a hardware guru and have
<		him suggest some to you, that's what I did.

Get only "machine screw, high reliability" sockets.  Yes, they do cost a 
bundle, but are worth every penny.  Notice that the few sockets on a UNIX-PC
motherboard are this good style.  Make sure that the sphincter is gold plated.
It is nice to have gold plated leads, but not necessary.

<	72 Samsung KM41256-15 6222 "Made in Korea" Memory Chips
<		I think the "-15" means 150nS. These work fine, and are among
<		the cheapest around.
<
<	1 74F258 Memory Decoder Chip
<	1 10 Ohm resistor pack
<		You only need these if they were not initially put on your
<		board. Check to see if there is a 74F258 chip at location
<		2-B and resistor pack at RP4. If not, put them there.
<		
<	Solder
<		Make sure you have enough, I ran out about 1/2 way through.
<		Mine was Radio Shack rosin core solder, I like the rosin core
<		because it cleans the connections before the solder sets. The
<		only bad thing is all the rosin left (see next item).
<
<	Denatured Alcohol
<		I have a "squeeze" bottle of this that I use to remove excess
<		rosin from my solder connections. It works (fairly) well on
<		the rosin while not effecting the green "paint" on the back
<		of the circuit board. Any better suggestions out there??

Aside from the commercial flux stripper solutions, I've found the best thing
to be a two step process: First use acetone.  It strips the actual rosin 
damn fast, but leaves a power residue where the flux actually have to do
work on some oxide.  Second, use a heavy Freon solvent such as TF.  It does
an excellent job of dissolving the powder.

The acetone is nasty powerful stuff.  Don't come in contact with it.  If
you can smell it at *all*, that's too much.  I always set up a window fan
do suck away the vapors.

Many people use Freon alone.  It *does* work, but Freon alone is very slow
at stripping the bulk of the rosin.  

<	Old Toothbrush
<		What you do is douse the back of the board with the alcohol,
<		then brush it off with the toothbrush - it really works!

I often use Q-tips to both apply the acetone and remove the crud.  Roll it
over the board.  This creates much less vapor than spraying or pouring 
acetone over the board.  It also eliminates the need for having something
to catch the excess solvent.

<	Ohm Meter
<		To check electrical continuity between points
<
<	Orange Sticks
<		These were great! They're a stick of wood about pencil size
<		with tapered ends. TV repair guys use them for poking around
<		in High Voltage areas. The nicest thing is that if you try to
<		force something too hard, the orange stick breaks instead of
<		the part! How I used them is described below.
<
<	The Computer Shopper Catalog
<		This made a GREAT workspace, it's thick enough to be "cushy",
<		but doesn't "give" too much to bend the board; and the size is
<		just perfect! I didn't ground myself to anything in particular,
<		but I was in my bare feet most of the time on my kitchen 
<		linoleum.
<
<	Jewelers Loupe
<		I had a 5x one to look at traces when I needed to.
<
<	Light
<		LOTS of it!
<
<Following will be a description of how I did my board. Many people say that
<it's easier to clip the chips off with small, sharp clippers, then unsolder
<each pin individually. I may use this method for the motherboard, but for
<the expansion board I didn't, so I don't know which way is easier. My method
<was to free each pin on the entire chip and extract it intact, this worked
<well, and I still have the chips ... maybe I'll put 'em in my mom's apple!
<
<Things to remember EACH time the iron touches the board:
<----------
<If you're not sure of something ... ANYTHING (is the iron touching the board
<correctly, etc), I mean ANYTHING... STOP and start all over on that pin!
<Remember if you screw up once, that could be it!
<
<If the pin is not cleaned out properly, re-solder the pin completely back again,
<then start over (I know this sounds dumb, and hard to fight temptation on, but
<it's really better that trying to get the last bit out 5 or 6 times!).
<----------
<
<I'd also suggest that you do 1 or 2 memory chips by themselves first (taking
<out the old memory, putting in the socket, then the memory, then running the
<board through diagnostics) just to get into the swing. Once your ready to go
<into FULL production, follow these steps:
<
<Step 1: De-solder a row (9) chips. Personally, I prefer holding the iron at a
<	45 degree angle on the right and the soldapult at a 45 degree angle to
<	the left. I used a count of 4 making sure I saw the solder turn liquid
<	during the count, then hitting it with the soldapult. A successfully
<	unsoldered pin allows you to see the round foil and the pin sticking
<	thru the hole. If it looks clean from the foil side, you're ok.
<
<Step 2: Push each pin back and forth at least once from the foil side with
<	the orange stick to loosen it from that side (sometimes leftover rosin
<	from the initial solder makes them stick)
<
<Step 3: Rub off any excess solder splashes around each pin with
<	the tip of the orange stick. While doing this check for lifted foil,
<	missed pins, and any other "bad stuff" you can think of.
<
<Step 4: Turn board over to component side and push each pin of each chip with
<	the end of the orange stick to loosen it, you may hear a slight
<	"tick" as the pin breaks free.
<
<Step 5: When you're convinced the chip is loose, use the end of the orange
<	stick to pry up on end of the chip a bit. Then go to the other side,
<	and pry it up a bit. Keep going back and forth until it comes out.
<
<Step 6: ALLWAYS check the leads of the removed chip to see if any foil was
<	extracted with it. If so, you're already screwed, but at least you'll
<	know where the problem is.
<
<Step 7: Carefully check both sides of where the chip was and make sure
<	everything is ok.
<
<Step 8: Install the sockets and solder them down. You all know the "be
<	carefull's" of this step.
<
<Step 9: Check both sides again and remove the excess rosin with the alcohol
<	and toothbrush.
<
<Step 10: Install the memory chips (make SURE they're seated properly) and run
<	 diagnostics.
<
<Once you've replaced all of the memory chips and made sure you have the
<address decoder chip (2-B) and resistor pack (RP11), all you need to do
<is the jumpering and you're set! If you haven't run diagnostics after each row,
<now is the time to do it, just to be sure that 0.5MB of the memory you've
<installed works.
<
<Jumpering:
<
<On the board is a set of 3 holes in a box labeled "w4" (lower left about at
<chip location 1-K). The two lower holes are grouped together, separated by
<a vertical line from the third (upper) hole. If you look at the foil side,
<the 2 grouped holes (lower) have foil running between them. Remove this foil
<and check it with an ohm meter to be sure there is no connection. Now solder
<in a small bit of wire coming out of the middle hole and looped around, into
<the isolated (upper) hole. Check your work with an ohm meter just to be sure.
<This jumper will disconnect address line 11 from the memory address decoders,
<and connect up address line 19 (This part is similar to the mother board
<upgrade). At this point you can run diagnostics again. If you do this, you
<will be checking a different 0.5MB chunk within the 2MB on the board.
<
<There are 2 more similar sets of holes labeled "w1" and "w2" (upper left near
<A-1). Use the same procedure to disconnect the existing holes (foil is on the
<component side this time) and jump the middle hole with the isolated hole.
<These jumpers ground pins 13 and 18 of the 74F373 chip at 1-A (check it with
<your ohm meter if you want, ground foil runs around the board edge)
<and, in effect, hardwire the board for slot 1 using up ALL available
<expansion memory (even though the machine can take 4MB, only 2MB of it
<can be on the expansion bus).
<
<W3 simply pulls up +5 volts on pin 73 of the expansion plug. Convergent
<claims it is "unused". I've found no reason to disbelieve them. You don't
<have to touch this one.
<
<Testing:
<
<THE BOARD MUST BE PLACED IN EXPANSION SLOT 1 OR IT WILL NOT WORK! Expansion
<slot 1 is directly under the power supply, or, the slot on the same side as
<the volume control for the speaker, or, the right side if you look at the
<machine from the front, or, the left side if you look at the machine from the
<back. When you boot up from the diagnostic floppy to run diagnostics, you
<should see 2M of memory in the expansion slot. The total memory as reported
<by the Administration function from your "office" should be 2560k. Be sure
<to run diagnostics on it to make sure everything is correct, and you're all
<set!
<
<By this point you will have de-soldered and re-soldered 1152 pins!
<
<The only other cautions that come to mind is always handle the board with
<care, I have heard of printed circuit boards cracking and rendering them
<useless. If you constantly remind yourself that if you screw up once, you're
<done for, and never continue when you have a slight doubt in your head, you
<should be fine.
<
<I am willing to help anybody who wants, but if you decide to do this, nobody
<can take responsibility for what happens but yourself, so "Lets be careful
<out there..."
<
<Good Luck!
<
<Craig Votava
<AT&T Bell Labs, Naperville IL
<IH 4G-324 312/979-2542
-- 
Jim Prior    {ihnp4|osu-cis}!n8emr!oink!jep    jep at oink.UUCP

Pointers are my friend.