11 Dec
2015
11 Dec
'15
5:19 a.m.
From: Jacob Ritorto
Would you happen to have notes or references about how
to do it?
It's not too hard; basically, one has to wire pins BC1, BD1, BE1 and BF1
(BDAL 18-BDAL21, respectively) on all _QBUS_ slots together into a bus. So
wire BC1 on slot 1 to BC1 on slot 2, slot 3, etc, etc.
A couple of notes: First, I said '_QBUS_' because if you have a Q/CD
backplane, clearly one doesn't run the extra BDAL lines to the CD slots, only
the QBUS slots (which run down the left-hand side, when facing the backplane).
Second, for optimal analog behaviour, the 'out' slot on the backplane should
be the last slot you wire to, so that there are no branches in the
transmission line for BDAL18-BDAL21 (which can produce reflections - aka
noise - on the transmission lines). How to do this efficiently (in terms of
the wiring) can be a bit tricky, depending on the backplane configuration.
E.g. if one has the standard 'serpentine' backplane, i.e. one with the slots
in the following kind of order (facing the backplane from the board side):
1-2
4-3
5-6
8-7
9-10
etc., one might naively think one has to run the extra bus lines back and
forth to match. However, only the _grant_ lines have to follow this pattern
(and they are already there); the added lines don't have to follow the same
pattern, as long as there are no branches.
So, for the example 5-slot backplane above, one could/would wire:
1-4-5-8-9--2-3-6-7-10
i.e. a single vertical run on the left hand side, a single diagonal from 9
back to 2 (shown with "--"), and then another vertical run on the right hand
side. Much simpler than wiring back and forth in slot order; there are no
branches; and the last slot is the 'out' slot.
For backplane with an _even_ number of layers, e.g.:
1-2
4-3
5-6
8-7
it's a little more complicated: a single vertical run on each side
cannot be connected in such a way as to have the 'out' slot (8) be the
last slot. One has to do something a little more complex:
1-4-5--2-3-6-7--8
with a vertical run on the left side, stopping short of the last slot; then
a vertical run on the right side, then a lateral back across on the last
layer.
Obviously one _could_ run the wires back and forth, in slot order, but that
will take a lot more wire, which at the very least is more work (especially
on backplanes which don't have full wire-wrap pins, just the little stubby
pins that have to have the wires soldered to); whether it also increases the
delay down those transmission lines enough to be noticeable is something I
don't know the answer to.
All the obvious caveats apply: make sure not to get confused by the mirror
pin and slot numbers on the front and back sides (you'll be wiring on the
back, whereas the diagrams above are on the front), etc.
Noel
8 Feb
8 Feb
3:55 p.m.
This is extremely good info (thank you to Robert, Joseph and Noel!) and I
plan to do something like this someday when I get more q-bus stuff, but I
must apoligise for my inaccuracy because what I was originally trying to
ask was:
Can I tear apart my little BA23 (which currently has a power supply problem
and not enough space for my high capacity 8" SMD disks) and put the
Micro/PDP-11 backplane (with all its nice 11/73 cards and SMD disk
controllers, etc.) into the spot that my 11/03 backplane currently occupies
and run it via the (working) stock 11/03 power supply? This "11/03
chassis" is bolted into what appears to be a common, official Digital 19"
rack surrounded by some RL02s, mid-height style (don't know the name of
this racking option).
Specifically, would I have to butcher power and clock lines to do this,
or is it all plug compatible? I'm considering tearing the systems down and
assessing the situation, but wanted to ask in advance in case somebody's
already gone there and can save me the heartache. The recent mention of
the dual 11/73 in a 19" rack prompted me to reconsider my original intent
instead of rewiring old backplanes to be 22-bit compliant.
thx
jake
On Fri, Dec 11, 2015 at 8:19 AM, Noel Chiappa <jnc at mercury.lcs.mit.edu>
wrote:
From: Jacob
Ritorto
Would you happen to have notes or references
about how to do it?
It's not too hard; basically, one has to wire pins BC1, BD1, BE1 and BF1
(BDAL 18-BDAL21, respectively) on all _QBUS_ slots together into a bus. So
wire BC1 on slot 1 to BC1 on slot 2, slot 3, etc, etc.
A couple of notes: First, I said '_QBUS_' because if you have a Q/CD
backplane, clearly one doesn't run the extra BDAL lines to the CD slots,
only
the QBUS slots (which run down the left-hand side, when facing the
backplane).
Second, for optimal analog behaviour, the 'out' slot on the backplane
should
be the last slot you wire to, so that there are no branches in the
transmission line for BDAL18-BDAL21 (which can produce reflections - aka
noise - on the transmission lines). How to do this efficiently (in terms of
the wiring) can be a bit tricky, depending on the backplane configuration.
E.g. if one has the standard 'serpentine' backplane, i.e. one with the
slots
in the following kind of order (facing the backplane from the board side):
1-2
4-3
5-6
8-7
9-10
etc., one might naively think one has to run the extra bus lines back and
forth to match. However, only the _grant_ lines have to follow this pattern
(and they are already there); the added lines don't have to follow the same
pattern, as long as there are no branches.
So, for the example 5-slot backplane above, one could/would wire:
1-4-5-8-9--2-3-6-7-10
i.e. a single vertical run on the left hand side, a single diagonal from 9
back to 2 (shown with "--"), and then another vertical run on the right
hand
side. Much simpler than wiring back and forth in slot order; there are no
branches; and the last slot is the 'out' slot.
For backplane with an _even_ number of layers, e.g.:
1-2
4-3
5-6
8-7
it's a little more complicated: a single vertical run on each side
cannot be connected in such a way as to have the 'out' slot (8) be the
last slot. One has to do something a little more complex:
1-4-5--2-3-6-7--8
with a vertical run on the left side, stopping short of the last slot; then
a vertical run on the right side, then a lateral back across on the last
layer.
Obviously one _could_ run the wires back and forth, in slot order, but that
will take a lot more wire, which at the very least is more work (especially
on backplanes which don't have full wire-wrap pins, just the little stubby
pins that have to have the wires soldered to); whether it also increases
the
delay down those transmission lines enough to be noticeable is something I
don't know the answer to.
All the obvious caveats apply: make sure not to get confused by the mirror
pin and slot numbers on the front and back sides (you'll be wiring on the
back, whereas the diagrams above are on the front), etc.
Noel
5:45 p.m.
I top post only once or twice a year. This is the exception since my reply
does not explicitly address the question.
An internal built-in power supply is just the ordinary standard power
supply which provides power to the board on which the CPU is
located.
I have managed to avoid many of the problems associated with power
supplies by using the built-in power supply only for the BA23 boards,
tape drives and, if any, floppy drives.
Note that when I use a TK70 tape drive in a BA23 box, it functions
correctly ONLY when connected to the built-in power supply.
Specifically, I always use a SEPARATE PC power supply for the hard
disk drives. This is probably essential when I use THREE Hitachi ESDI
hard disk drives since their power requirements, when added to the
rest of the load, really challenge the BA23 box built-in power supply.
But I believe it is probably best even when just an RD51 is used
although the load from a single RD51 is probably OK to run using the
built-in power supply of the BA23.
In addition, the Hitachi ESDI hard drives also require an external fan
to keep them at a reasonable temperature. The fans are also powered
by the separate PC power supply.
This procedure also works well with other systems, so I recommend
it to reduce the load and keep internal built-in power supplies from
being overloaded on any system, especially when there is more than
one hard drive.
Jerome Fine
Jacob Ritorto wrote:
This is extremely good info (thank you to Robert,
Joseph and Noel!) and I
plan to do something like this someday when I get more q-bus stuff, but I
must apoligise for my inaccuracy because what I was originally trying to
ask was:
Can I tear apart my little BA23 (which currently has a power supply problem
and not enough space for my high capacity 8" SMD disks) and put the
Micro/PDP-11 backplane (with all its nice 11/73 cards and SMD disk
controllers, etc.) into the spot that my 11/03 backplane currently occupies
and run it via the (working) stock 11/03 power supply? This "11/03
chassis" is bolted into what appears to be a common, official Digital 19"
rack surrounded by some RL02s, mid-height style (don't know the name of
this racking option).
Specifically, would I have to butcher power and clock lines to do this,
or is it all plug compatible? I'm considering tearing the systems down and
assessing the situation, but wanted to ask in advance in case somebody's
already gone there and can save me the heartache. The recent mention of
the dual 11/73 in a 19" rack prompted me to reconsider my original intent
instead of rewiring old backplanes to be 22-bit compliant.
thx
jake
On Fri, Dec 11, 2015 at 8:19 AM, Noel Chiappa
<jnc at mercury.lcs.mit.edu> wrote:
> > From: Jacob Ritorto
>
> > Would you happen to have notes or references about how to do it?
>
>It's not too hard; basically, one has to wire pins BC1, BD1, BE1 and BF1
>(BDAL 18-BDAL21, respectively) on all _QBUS_ slots together into a bus. So
>wire BC1 on slot 1 to BC1 on slot 2, slot 3, etc, etc.
>
>A couple of notes: First, I said '_QBUS_' because if you have a Q/CD
>backplane, clearly one doesn't run the extra BDAL lines to the CD slots,
>only
>the QBUS slots (which run down the left-hand side, when facing the
>backplane).
>
>Second, for optimal analog behaviour, the 'out' slot on the backplane
>should
>be the last slot you wire to, so that there are no branches in the
>transmission line for BDAL18-BDAL21 (which can produce reflections - aka
>noise - on the transmission lines). How to do this efficiently (in terms of
>the wiring) can be a bit tricky, depending on the backplane configuration.
>
>E.g. if one has the standard 'serpentine' backplane, i.e. one with the
>slots
>in the following kind of order (facing the backplane from the board side):
>
>1-2
>4-3
>5-6
>8-7
>9-10
>
>etc., one might naively think one has to run the extra bus lines back and
>forth to match. However, only the _grant_ lines have to follow this pattern
>(and they are already there); the added lines don't have to follow the same
>pattern, as long as there are no branches.
>
>So, for the example 5-slot backplane above, one could/would wire:
>
>1-4-5-8-9--2-3-6-7-10
>
>i.e. a single vertical run on the left hand side, a single diagonal from 9
>back to 2 (shown with "--"), and then another vertical run on the right
>hand
>side. Much simpler than wiring back and forth in slot order; there are no
>branches; and the last slot is the 'out' slot.
>
>For backplane with an _even_ number of layers, e.g.:
>
>1-2
>4-3
>5-6
>8-7
>
>it's a little more complicated: a single vertical run on each side
>cannot be connected in such a way as to have the 'out' slot (8) be the
>last slot. One has to do something a little more complex:
>
>1-4-5--2-3-6-7--8
>
>with a vertical run on the left side, stopping short of the last slot; then
>a vertical run on the right side, then a lateral back across on the last
>layer.
>
>Obviously one _could_ run the wires back and forth, in slot order, but that
>will take a lot more wire, which at the very least is more work (especially
>on backplanes which don't have full wire-wrap pins, just the little stubby
>pins that have to have the wires soldered to); whether it also increases
>the
>delay down those transmission lines enough to be noticeable is something I
>don't know the answer to.
>
>
>All the obvious caveats apply: make sure not to get confused by the mirror
>pin and slot numbers on the front and back sides (you'll be wiring on the
>back, whereas the diagrams above are on the front), etc.
>
> Noel
>
9 Feb
9 Feb
7:18 a.m.
On Mon, Feb 8, 2016 at 8:45 PM, Jerome H. Fine <jhfinedp3k at compsys.to>
wrote:
[...]
Specifically, I always use a SEPARATE PC power supply for the hard
disk drives.
Yep, this sounds like exactly where I went wrong. Two RD32s running
24/7. Took it only a couple months to burn out. In the next incarnation
of this machine (i.e. if I ever learn to repair power supplies), I'll set
it up as Jerome describes.
11 Feb
11 Feb
8:57 a.m.
Jacob Ritorto wrote:
On Mon, Feb 8,
2016 at 8:45 PM, Jerome H. Fine <jhfinedp3k at compsys.to> wrote:
Specifically, I always use a SEPARATE PC power
supply for the hard disk drives.
Yep, this sounds like exactly where I went wrong. Two RD32s running
24/7. Took it only a couple months to burn out. In the next incarnation
of this machine (i.e. if I ever learn to repair power supplies), I'll set
it up as Jerome describes.
3189
days inactive
3251
days old
4 comments
3 participants
participants (3)
-
jacob.ritorto＠gmail.com -
jhfinedp3k＠compsys.to -
jnc＠mercury.lcs.mit.edu