SERVICING THE CMD RAMLINK... my first experience
                latest updates and/or corrections 3-4-2020

     Until I was asked by a friend to repair his CMD RAMLink, I had no
experience with those devices... never even saw one up close. Turns out he
actually had five of them, three older and two newer versions, and they all
ended up on my repair bench for evaluation. He mentioned that one of them (a
later version) had a problem of losing data stored on the RAM after it was
on awhile. Rather than trying to dig into that one right away, I thought I
should first get as familiar with a working one as I could.
     Starting from scratch on any electronic device, I try to get all the
information I can before I even open the case. Bad information is difficult
to un-learn. Fortunately, the owner had the original user manual and a
utilities disk, so I found out how to properly connect the device to the
computer and how to use it. A user needs both the book and the disk. I
believe there are copies on the Internet. One source I found for a pdf of
the later spiral bound manual is from Daniel Mackey. As of this writing:
http://ftp://n2dvm.com/Commodore/Programs/Misc/Ramlink-manual.pdf It's a 50 Meg
file and took over an hour to download on my 1.5M DSL.
     My next step as a servicer, like a doctor, is to examine the "patient".
Taken completely apart, I found two of the early models had poor solder
connections on the 4 pin DIN PS input jack (main board). They were bad from
the factory, and I had to scrape the pins a bit to get solder to properly
flow around those connections. All early models should be checked for that
shortcoming as it could cause intermittent operation or an outright failure.

POWER REQUIREMENTS AND EXTERNAL CONNECTORS:

     The early version RAMLink uses a 1581/1541-II power supply. It has a
four pin DIN power connector and puts out two voltages: +5VDC at 1 Amp and
+12VDC at 0.5A. NOTE: A smaller version of that same PS was sold by Commodore
but which probably should not be used with the RAMLink because its 5V output
is only 0.75A (750mA). Those specs are printed on the PS. A coaxial or
"barrel" type of connector is there for the backup battery, if used. The
battery is rated at 6 volts, 6.5AH and is used to keep the RAM alive in the
event of a power failure or if the RAMLink is moved to another location and
the data on it must be maintained.

     The later version RAMLink uses a single 9 volt DC "wall wart" AC to DC
adapter rated at 1 Amp. The DC input connector on that later RAMLink is
therefore different: a single coaxial or "barrel" type. The plug on the PS
is about 10mm long, 5mm outside diameter, and it mates with a socket center
pin diameter of approximately 2.5mm. The backup battery connector on the
later version RL is a mini-phone rather than a coaxial (barrel) type. The
connector is essentially the same as the older mono earphone type. The
polarity of that connector is tip=positive, ring=negative. I didn't need a
backup battery for my tests but I understand a fully charged battery should
hold the memory for 6 to 8 hours, depending on how much RAM is installed
and the state of charge of the battery. It's slowly "trickle" charged by
the RL and so it would take a lot of time (several days) to fully charge a
low battery.

NOTE: If you didn't get an original Commodore power supply for the later
version RAMLink, most important is the polarity of the PS output: the
center pin is positive and the barrel is negative. Using a supply with a
reversed voltage output could damage the RAMLink and/or the PS!
     I should mention one other thing here. I never like using a mini-phone
plug as a battery connector. The exposed ends of the plug could short out
if that plug touches metal. Also, there are many manufacturing variations of
that plug and its jack. The wrong combination could potentially short out
the battery when the plug is inserted or removed, or it just will not fit. 
It's best to plug or unplug that connector only when the battery itself is 
disconnected from its wires. A shorted battery could explode or at the very 
least burn its wiring.

     One connector common to both versions of the RAMLink is the large
circular 14 pin DIN parallel port. It's used to connect a CMD hard drive to 
the RL with the appropriate cable and HD DOS version in the drive.

RAMLINK INTERNALS

     The main (bottom) board in the two RAMLink versions is, of course,
different. They use some different IC's although both have a date stamped
on that board of 1990. The plug-in RAM board situated on top of that bottom
board appears to be the same (can be marked either 1990 or 1992) except
for a very early one that doesn't have a real time clock. RAM boards should
be interchangeable between versions of the RAMLink. If the RAM is to be
upgraded, a jumper must be cut on that sub-board (see appropriate users
manual for details). In the units I examined, there were four RAM module
sockets filled with DRAM strips for a total of 4 Meg in the early units and
16 Megs in the later ones. Total BYTES FREE in the early units was 16064,
and 65216 in the later ones.
     The one RL (later version) that was defective was found to have had a
liquid spill onto the internal boards. I disassembled that unit, scrubbed
the boards with Windex and an old toothbrush, rinsed with water and dried
with a hair dryer. I replaced a corroded IC socket and a burned resistor
(R49, a 5.1 ohm 1/4 watt flameproof) but that RL still didn't work reliably.
Swapping chips back and forth between the good and bad units, I finally
found the DOS 27512 EPROM had part of its code corrupted. As stated, that
unit would initialize but couldn't reliably hold its partition and program
data. I was able to "clone" a copy of the DOS from the good RL onto a fresh
EPROM and replace the bad one. All the other socketed chips in those
RAMLinks are GALs, programmable chips I don't presently have the hardware
with which to "burn" replacements. I clearly got lucky with this one. One
last thing... I noticed that some of these RL's would not always show up
with the JiffyDOS opening screen when the computer was booted. I cleaned
the two toggle bypass switches on the case top by spraying control cleaner
in the openings at the base of the paddles and working the switches back
and forth a few times. That took care of it.
     The paper stickers with the factory part numbers were glued to the
chips in these RAMLinks but they were all loose and falling off, so I
glued the ones I found back on. All the labels are on programmed socketed
chips only, namely the DOS EPROM and GAL's.

Ray Carlsen