11 Jun
2006
11 Jun
'06
11:18 a.m.
Found a problem in my IRIS 3k - didn't have a terminator on the IM1 memory bus. I
doubt that this is still available through
retail channels
Gerhard helped quite a lot by taking pictures of the board and giving a description of
where it went, now I need to fab one.
The board looks to be 4-layer (front&back traces with power and ground planes), which
is a bit difficult to do at home. This runs
in a 16MHz 68k system. Would it be possible to fab it using a barely-etched dual sided
board for power/ground planes and running
point-to-point wires for the signals, or is this asking for trouble as far as timing
issues go?
Termination is standard 331/331 resistors.
Scott Quinn
11 Jun
11 Jun
1:14 p.m.
On 6/11/2006 at 10:18 AM Scott Quinn wrote:
The board looks to be 4-layer (front&back traces
with power and ground
planes), which is a bit difficult to do at home. This runs
in a 16MHz 68k system. Would it be possible to fab it using a
barely-etched dual sided board for power/ground planes and running
point-to-point wires for the signals, or is this asking for trouble as far
as timing issues go?
Golly, at that bus speed, I'd think that even wirewrapped terminator boards
would do the job. Your method would probably be fine--and may even be
superior to PCB (reduced crosstalk if you keep the wires short).
Cheers,
Chuck
1:20 p.m.
Scott Quinn wrote:
Found a problem in my IRIS 3k - didn't have a
terminator on the IM1 memory bus. I doubt that this is still available through
retail channels
Gerhard helped quite a lot by taking pictures of the board and giving a description of
where it went, now I need to fab one.
The board looks to be 4-layer (front&back traces with power and ground planes), which
is a bit difficult to do at home. This runs
in a 16MHz 68k system. Would it be possible to fab it using a barely-etched dual sided
board for power/ground planes and running
point-to-point wires for the signals, or is this asking for trouble as far as timing
issues go?
Termination is standard 331/331 resistors.
How large is it? You can probably get it done "for free" using one of
the "$500 promotions" that PCB manufacturers often have. Of course,
they are expecting you to SAVE $500 on your > $500 order... but, if
you are only doing one or two and can live without things like
silkscreen, you can probably send them a layout tweeked to come
in UNDER that $500 number (ie. FREE)
Otherwise...
What sort of connector does it have on it (sorry, I don't have an
IRIS)? I suspect you can probably hand-wire the whole thing
(depends on what they used to drive the bus and how sharp the
edges are). That's not a particularly fast bus (I've wirewarpped
68K designs that fast, easily).
--don
2:06 p.m.
Sunday, June 11, 2006, 7:20:11 PM, you wrote:
What sort of connector does it have on it (sorry, I
don't have an
IRIS)?
I'll just post the image link - maybe that helps in finding good
ideas:
http://sgistuff.g-lenerz.de/pictures/iris.php#imterm
This board is plugged into the two connectors of the last memory board
in the chain. The cables are then connected to the terminator instead
of the board itself.
--
Best regards,
Gerhard mailto:mail at g-lenerz.de
Old SGI Stuff http://sgistuff.g-lenerz.de/
3:27 p.m.
Gerhard Lenerz wrote:
Sunday, June 11, 2006, 7:20:11 PM, you wrote:
Huh? Perhaps you mean that the way the board connects to
the rest of the machine is *through* the cables (i.e. there
is not some other set of connectors on the board that plug
into a "backplane")? So, *normally* the cables plug into the
memory board but, on the last board, the terminator sits
"between" the cables and the board?
In other words, any solution *must* have 4 connectors on it
(you can't just take two INDIVIDUAL connectors and solder
the RPAK's onto the back of those connectors (i.e. skip the
PCB entirely!)
What sort of connector does it have on it (sorry,
I don't have an
IRIS)?
I'll just post the image link - maybe that helps in finding good
ideas:
http://sgistuff.g-lenerz.de/pictures/iris.php#imterm
This board is plugged into the two connectors of the last memory board
in the chain. The cables are then connected to the terminator instead
of the board itself.
3:54 p.m.
Sunday, June 11, 2006, 9:27:08 PM, you wrote:
Huh? Perhaps you mean that the way the board connects
to
the rest of the machine is *through* the cables (i.e. there
is not some other set of connectors on the board that plug
into a "backplane")? So, *normally* the cables plug into the
memory board but, on the last board, the terminator sits
"between" the cables and the board?
It actually does both. The IRIS has a Multibus backplane into which
all boards plugged. In addition to that for some busses ribbon cable
connections are used. So a look into the cardcage of an IRIS shows
some boards hidden by quite a lot of ribbon cables:
http://www.g-lenerz.de/tmp/IMG_1869.JPG
As the leftmost board in that picture you see the memory board along
with the terminator installed. The following boards also connected
by the two ribbon cables at the upper left corner are FPU and CPU. The
reason for the termination not being on the memory board itself is
that additional memory is installed by using further boards. The last
one (leftmost) is then terminated as shown.
Just for another view I also upload a picture of a pulled memory board
(with attached terminator):
http://www.g-lenerz.de/tmp/IMG_1876.JPG
--
Best regards,
Gerhard mailto:mail at g-lenerz.de
Old SGI Stuff http://sgistuff.g-lenerz.de/
4:12 p.m.
Gerhard Lenerz wrote:
Sunday, June 11, 2006, 9:27:08 PM, you wrote:
But the question still stands: Could the OP just take a pair
of the *female* connectors (that plug into the memory board)
and solder some RPAKs onto them, then plug onto the board
(*whichever* board).
Or, would the absence of the *male* connectors (blue in your
pix) render the board useless?
Or, do it the other way around... keep the blue connectors
to mate to the ribbons, solder the RPAKs to those connectors
and omit the "black" connectors?
Huh? Perhaps you mean that the way the board
connects to
the rest of the machine is *through* the cables (i.e. there
is not some other set of connectors on the board that plug
into a "backplane")? So, *normally* the cables plug into the
memory board but, on the last board, the terminator sits
"between" the cables and the board?
It actually does both. The IRIS has a Multibus backplane into which
all boards plugged. In addition to that for some busses ribbon cable
connections are used. So a look into the cardcage of an IRIS shows
some boards hidden by quite a lot of ribbon cables:
http://www.g-lenerz.de/tmp/IMG_1869.JPG
As the leftmost board in that picture you see the memory board along
with the terminator installed. The following boards also connected
by the two ribbon cables at the upper left corner are FPU and CPU. The
reason for the termination not being on the memory board itself is
that additional memory is installed by using further boards. The last
one (leftmost) is then terminated as shown.
Just for another view I also upload a picture of a pulled memory board
(with attached terminator):
http://www.g-lenerz.de/tmp/IMG_1876.JPG
3:57 p.m.
On 6/11/2006 at 12:27 PM Don Y wrote:
In other words, any solution *must* have 4 connectors
on it
(you can't just take two INDIVIDUAL connectors and solder
the RPAK's onto the back of those connectors (i.e. skip the
PCB entirely!)
Maybe--didn't/doesn't someone like AP miake a cute little "T" adapter
for
probing the signals on a ribbon cable? The details are fuzzy, but with a
couple of those, that might solve the problem of 4 connectors. I suppose
one could just solder the resistors right to the backplane, too.
Cheers,
Chuck
4:14 p.m.
Chuck Guzis wrote:
On 6/11/2006 at 12:27 PM Don Y wrote:
I know someone makes an assembly that fits in series and
lets you *open* (via small slide switches) individual conductors.
But, I can't see an easy way to exploit that.
I'd have to look at pix closer to see if you couldn't make two
(identical) "half terminators"... one for each M-F pair
of connectors...
In other words, any solution *must* have 4
connectors on it
(you can't just take two INDIVIDUAL connectors and solder
the RPAK's onto the back of those connectors (i.e. skip the
PCB entirely!)
Maybe--didn't/doesn't someone like AP miake a cute little "T" adapter
for
probing the signals on a ribbon cable? The details are fuzzy, but with a
couple of those, that might solve the problem of 4 connectors. I suppose
one could just solder the resistors right to the backplane, too. 
12 Jun
12 Jun
4:09 a.m.
On Sun, 11 Jun 2006 10:18:13 -0500
"Scott Quinn" <compoobah at valleyimplants.com> wrote:
Found a problem in my IRIS 3k - didn't have a
terminator on the IM1
memory bus. I doubt that this is still available through retail
channels Gerhard helped quite a lot by taking pictures of the board
and giving a description of where it went, now I need to fab one.
Three ideas:
If you have unused connectors on the ribbon cables (e.g. last memory
board not instaled) just use a suitable male connector for the ribbon
cable, solder the resistors to it and connect it to the last, unused
connector of the cable.
If you need all connectors on the ribbon cables make a new cables with
one additional connector, then see above.
Use a male and female connector, solder the pins at the tips together,
so that you get a 1:1 male-female connector. Solder the resistors to the
sides of this.
All of this is not hard if you know how to handle a soldering iron. I
build SCSI terminators that way.
--
tsch??,
Jochen
Homepage: http://www.unixag-kl.fh-kl.de/~jkunz/
6610
days inactive
6611
days old
9 comments
5 participants
participants (5)
-
cclist＠sydex.com -
compoobah＠valleyimplants.com -
dgy＠DakotaCom.Net -
jkunz＠unixag-kl.fh-kl.de -
mail＠g-lenerz.de