13 Aug
2006
13 Aug
'06
5:01 p.m.
Looking for information on a QUAD width omnibus RAM card.
Not interested in hex wide as it doesn't fit the older
Omnibus 8E or F boxes. I used to have a 4kx12 card that
used 2102 type rams but had no data for it. That went to
and earlier 8E I had.
Even schematics would help. In the end I may do my own as
a pair of 61256s will certainly fit the bill and barely fill
a corner of a board.
Allison
14 Aug
14 Aug
3:04 a.m.
From: "Allison" <ajp166 at bellatlantic.net>
Looking for information on a QUAD width omnibus RAM
card.
Not interested in hex wide as it doesn't fit the older
Omnibus 8E or F boxes. I used to have a 4kx12 card that
used 2102 type rams but had no data for it. That went to
and earlier 8E I had.
Even schematics would help. In the end I may do my own as
a pair of 61256s will certainly fit the bill and barely fill
a corner of a board.
I looked at doing one of these, but got kind of lost about how
the timing signals are used. I looked at the MOS RAM for the 8/A,
but there's all this cruft for refresh in there, some stuff about
suppressing access during ROM access, etc. And then it looks like
the memory timing signals are derived somehow from the memory
refresh stuff??
Is there a nice place where the Omnibus memory interface control
signals are explained somewhere?
Vince
12:54 p.m.
vrs wrote:
I looked at doing one of these, but got kind of lost
about how
the timing signals are used. I looked at the MOS RAM for the 8/A,
but there's all this cruft for refresh in there, some stuff about
suppressing access during ROM access, etc. And then it looks like
the memory timing signals are derived somehow from the memory
refresh stuff??
Is there a nice place where the Omnibus memory interface control
signals are explained somewhere?
I did a semiconductor memory design for the OMNIBUS several years ago.
It was for an -8/e and 32kW using two SRAMs. I never got to documenting
it outside of my notebook, but if there is enough interest, I could do
some schematics. It is battery backed, so its pretty much equivalent to
the core I was replacing.
The Small Computer Handbook, the maintenance manuals and timing diagrams
from the processor print sets had enough information
for me.
-chuck
1:11 p.m.
If you are able to read german text, then the documentation of a 32KW
dynamic memory modul
from mico for the pdp8/e/f/m and /a may give some
hints.
I've made a scan available under
http://pdp8.de/download/mico-mem.pdf
- Gerold
C H Dickman wrote:
vrs wrote:
I looked at doing one of these, but got kind of
lost about how the
timing signals are used. I looked at the MOS RAM for the 8/A, but
there's all this cruft for refresh in there, some stuff about
suppressing access during ROM access, etc. And then it looks like
the memory timing signals are derived somehow from the memory refresh
stuff??
Is there a nice place where the Omnibus memory interface control
signals are explained somewhere?
I did a semiconductor memory design for the OMNIBUS several years ago.
It was for an -8/e and 32kW using two SRAMs. I never got to
documenting it outside of my notebook, but if there is enough
interest, I could do some schematics. It is battery backed, so its
pretty much equivalent to the core I was replacing.
The Small Computer Handbook, the maintenance manuals and timing
diagrams from the processor print sets had enough information for me.
-chuck
7:07 p.m.
If you are able to read german text, then the
documentation of a 32KW
dynamic memory modul
from mico for the pdp8/e/f/m and /a may give some hints.
I've made a scan available under
http://pdp8.de/download/mico-mem.pdf
Thanks -- unfortunately that mostly documents the parts I already understood
:-).
The timing diagram is somewhat helpful, though.
Vince
7:29 p.m.
vrs wrote:
If you use 2167's ( 16k x 1 ) ram's
you can have 16k of memory ( 12 chips ) for
about $25 not counting glue logic and other stuff.
Does anybody make a wire-wrap proto-type board for the
PDP-8/e slots?
Thanks -- unfortunately that mostly documents the
parts I already understood
:-).
7:37 p.m.
woodelf wrote:
vrs wrote:
If you use 2167's ( 16k x 1 ) ram's
you can have 16k of memory ( 12 chips ) for
about $25 not counting glue logic and other stuff.
Does anybody make a wire-wrap proto-type board for the
PDP-8/e slots?
Yes, but 2 61256s reduce the memory chip count to two (2)
and they are available in CMOS cutting the heatload.
I have a few quad width protoboards that are PDP-11 (not sure
if unibus or Qbus) but with strategic cuts for power and
ground they work for omnibus.
Allison
Thanks -- unfortunately that mostly documents the
parts I already
understood
:-).
15 Aug
15 Aug
12:58 p.m.
From: "Allison" <ajp166 at bellatlantic.net>
I have a few quad width protoboards that are PDP-11
(not sure
if unibus or Qbus) but with strategic cuts for power and
ground they work for omnibus.
I have a couple of these, too, but mine are all 16-pin sockets. I'd
have to mount a daughter board or something to be able to use them
for this.
Vince
14 Aug
14 Aug
7:43 p.m.
On 8/15/06, woodelf <bfranchuk at jetnet.ab.ca> wrote:
If you use 2167's ( 16k x 1 ) ram's
you can have 16k of memory ( 12 chips ) for
about $25 not counting glue logic and other stuff.
Why not use a pair of 62256s? 32Kx16 and ignore 4 bits. It'd even be
easy to use a Dallas DS1210 or something like it to provide battery
backup.
Does anybody make a wire-wrap proto-type board for
the
PDP-8/e slots?
Douglas used to. Don't know what they have in stock.
-ethan
8:11 p.m.
On Monday 14 August 2006 19:43, Ethan Dicks wrote:
On 8/15/06, woodelf <bfranchuk at jetnet.ab.ca>
wrote:
I'd probably do something like what the SBC-6120 does... use some typical 486
motherboard cache SRAM. Cheap, and relatively easy to find. :)
You'd end up with one or two chips to do 32k x 12.
Pat
--
Purdue University ITAP/RCAC --- http://www.rcac.purdue.edu/
The Computer Refuge --- http://computer-refuge.org
If you use 2167's ( 16k x 1 ) ram's
you can have 16k of memory ( 12 chips ) for
about $25 not counting glue logic and other stuff.
Why not use a pair of 62256s? 32Kx16 and ignore 4 bits. It'd even be
easy to use a Dallas DS1210 or something like it to provide battery
backup.
8:54 p.m.
On 8/15/06, Patrick Finnegan <pat at computer-refuge.org> wrote:
On Monday 14 August 2006 19:43, Ethan Dicks wrote:
32Kx8 are easy to find as old cache RAM. They are not, however, low
power, and would not be the best solution if you wanted to hang a
battery off of them. OTOH, unless you are using paper tape software
only, chances are you have a mass storage device and are booting into
OS/8 and don't really need battery backup.
-ethan
Why not use a pair of 62256s? 32Kx16 and ignore
4 bits. It'd even be
easy to use a Dallas DS1210 or something like it to provide battery
backup.
I'd probably do something like what the SBC-6120 does... use some typical 486
motherboard cache SRAM. Cheap, and relatively easy to find. :)
You'd end up with one or two chips to do 32k x 12. 
8:59 p.m.
On Aug 14, 2006, at 8:54 PM, Ethan Dicks wrote:
32Kx8 are easy to find as old cache RAM. They are not, however, low
power, and would not be the best solution if you wanted to hang a
battery off of them. OTOH, unless you are using paper tape software
only, chances are you have a mass storage device and are booting into
OS/8 and don't really need battery backup.
True, but I think the idea there was to duplicate the functionality
of core, i.e. the nonvolatility in this case.
That said, though, I've not found 486 motherboard cache memory to be
particularly easy to find. I did, however, buy a box of about 80 6264
chips on eBay a few months ago for fifteen bucks. And 62256s are still
manufactured, as far as I'm aware.
-Dave
--
Dave McGuire
Cape Coral, FL
Why not use a pair of 62256s? 32Kx16 and ignore 4
bits. It'd even
be
easy to use a Dallas DS1210 or something like it
to provide battery
backup.
I'd probably do something like what the SBC-6120 does... use some
typical 486
motherboard cache SRAM. Cheap, and relatively easy to find. :)
You'd end up with one or two chips to do 32k x 12.
9:17 p.m.
On Monday 14 August 2006 20:59, Dave McGuire wrote:
On Aug 14, 2006, at 8:54 PM, Ethan Dicks wrote:
>> > Why not use a pair of 62256s? 32Kx16 and ignore 4 bits. It'd even
>>
>> be
>>
>> > easy to use a Dallas DS1210 or something like it to provide battery
>> > backup.
>>
>> I'd probably do something like what the SBC-6120 does... use some
>> typical 486
>> motherboard cache SRAM. Cheap, and relatively easy to find. :)
>>
>> You'd end up with one or two chips to do 32k x 12.
>
> 32Kx8 are easy to find as old cache RAM. They are not, however, low
> power, and would not be the best solution if you wanted to hang a
> battery off of them. OTOH, unless you are using paper tape software
> only, chances are you have a mass storage device and are booting into
> OS/8 and don't really need battery backup.
I thought the point was to use them in a real PDP-8 e/f/m, in which case the
extra power would probably be a non-issue..
True, but I think the idea there was to duplicate
the functionality
of core, i.e. the nonvolatility in this case.
That said, though, I've not found 486 motherboard cache memory to be
particularly easy to find. I did, however, buy a box of about 80 6264
chips on eBay a few months ago for fifteen bucks. And 62256s are still
manufactured, as far as I'm aware.
I guess I have a better supply of old 486 motherboards than you do, then. :)
Pat
--
Purdue University ITAP/RCAC --- http://www.rcac.purdue.edu/
The Computer Refuge --- http://computer-refuge.org
9:46 p.m.
On 8/14/2006 at 9:17 PM Patrick Finnegan wrote:
I guess I have a better supply of old 486 motherboards
than you do, then.
Dig into an old 33/56K external modem. Lots of good SRAM there. One model
of the Hayes has a nice 32Kx16 0.3" DIP in it. There's SRAM in lots of
places.
Cheers,
Chuck
15 Aug
15 Aug
12:38 a.m.
On 8/15/06, Patrick Finnegan <pat at computer-refuge.org> wrote:
The extra power has to do with battery life, not power-on consumption.
The amount of power that two cache SRAMs and some buffer chips takes
up is utterly invisible compared to eight set of 4K core stacks (but
you need two OMNIBUS chassis to hold that many cards; a more typical
installation is four 8K stacks or a couple of third-party 16K stacks).
-ethan
On Aug 14,
2006, at 8:54 PM, Ethan Dicks wrote:
>> > Why not use a pair of 62256s? 32Kx16 and ignore 4 bits. It'd even
>> > be easy to use a Dallas DS1210 or something like it to provide battery
>> > backup.
>
> 32Kx8 are easy to find as old cache RAM. They are not, however, low
> power, and would not be the best solution if you wanted to hang a
> battery off of them.
I thought the point was to use them in a real PDP-8 e/f/m, in which case the
extra power would probably be a non-issue..
1:17 a.m.
On Tuesday 15 August 2006 00:38, Ethan Dicks wrote:
On 8/15/06, Patrick Finnegan <pat at
computer-refuge.org> wrote:
The extra power has to do with battery life, not power-on consumption.
The amount of power that two cache SRAMs and some buffer chips takes
up is utterly invisible compared to eight set of 4K core stacks (but
you need two OMNIBUS chassis to hold that many cards; a more typical
installation is four 8K stacks or a couple of third-party 16K stacks).
Yeah, I realized that after re-reading your message. :)
Still, if you're just looking for memory, and don't care about battery backup
(like MOS memory replacement), I think that cache SRAM is a reasonable
choice.
Pat
--
Purdue University ITAP/RCAC --- http://www.rcac.purdue.edu/
The Computer Refuge --- http://computer-refuge.org
On Aug
14, 2006, at 8:54 PM, Ethan Dicks wrote:
>> > Why not use a pair of 62256s? 32Kx16 and ignore 4 bits. It'd
>> > even be easy to use a Dallas DS1210 or something like it to
>> > provide battery backup.
>
> 32Kx8 are easy to find as old cache RAM. They are not, however, low
> power, and would not be the best solution if you wanted to hang a
> battery off of them.
I thought the point was to use them in a real PDP-8 e/f/m, in which case
the extra power would probably be a non-issue.. 
12:08 a.m.
On Mon, 14 Aug 2006, Dave McGuire wrote:
True, but I think the idea there was to duplicate
the functionality
of core, i.e. the nonvolatility in this case.
That said, though, I've not found 486 motherboard cache memory to be
particularly easy to find. I did, however, buy a box of about 80 6264
chips on eBay a few months ago for fifteen bucks. And 62256s are still
manufactured, as far as I'm aware.
I have a couple tubes of 62256s left over.
--
David Griffith
dgriffi at cs.csubak.edu
A: Because it fouls the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?
14 Aug
14 Aug
11:44 p.m.
Ethan Dicks wrote:
On 8/15/06, Patrick Finnegan <pat at
computer-refuge.org> wrote:
Depends on what grade device you use. I often use 32Kx8's as
BBSRAM.
On Monday 14 August 2006 19:43, Ethan Dicks
wrote:
32Kx8 are easy to find as old cache RAM. They are not, however, low
power, and would not be the best solution if you wanted to hang a Why not use a pair of 62256s? 32Kx16 and ignore
4 bits. It'd even be
easy to use a Dallas DS1210 or something like it to provide battery
backup.
I'd probably do something like what the SBC-6120 does... use some
typical 486
motherboard cache SRAM. Cheap, and relatively easy to find. :)
You'd end up with one or two chips to do 32k x 12. battery off of them. OTOH, unless you are using paper
tape software
only, chances are you have a mass storage device and are booting into
OS/8 and don't really need battery backup.
9:01 p.m.
I checked with my friend, and his coworker, who died a few years ago,did
make some 32k mos boards for the 8E. He is going to check to see if he has
any complete units, unpopulated boards, parts, and documentation for
them.This will probably take a few days.
Thanks, Paul
On 8/14/06, Patrick Finnegan <pat at computer-refuge.org> wrote:
On Monday 14 August 2006 19:43, Ethan Dicks wrote:
On 8/15/06, woodelf <bfranchuk at
jetnet.ab.ca> wrote:
I'd probably do something like what the SBC-6120 does... use some typical
486
motherboard cache SRAM. Cheap, and relatively easy to find. :)
You'd end up with one or two chips to do 32k x 12.
Pat
--
Purdue University ITAP/RCAC --- http://www.rcac.purdue.edu/
The Computer Refuge --- http://computer-refuge.org
If you use 2167's ( 16k x 1 ) ram's
you can have 16k of memory ( 12 chips ) for
about $25 not counting glue logic and other stuff.
Why not use a pair of 62256s? 32Kx16 and ignore 4 bits. It'd even be
easy to use a Dallas DS1210 or something like it to provide battery
backup.
11:42 p.m.
Patrick Finnegan wrote:
On Monday 14 August 2006 19:43, Ethan Dicks wrote:
Outrageous overkill. What's that, 10ns memory? What's the 8's
memory cycle time?? ;-)
On 8/15/06, woodelf <bfranchuk at
jetnet.ab.ca> wrote:
I'd probably do something like what the SBC-6120 does... use some typical 486
motherboard cache SRAM. Cheap, and relatively easy to find. :) If you use 2167's ( 16k x 1 ) ram's
you can have 16k of memory ( 12 chips ) for
about $25 not counting glue logic and other stuff.
Why not use a pair of 62256s?
32Kx16 and ignore 4 bits. It'd even be
easy to use a Dallas DS1210 or something like it to provide battery
backup. You'd end up with one or two chips to do 32k x
12.
Pat
15 Aug
15 Aug
12:32 a.m.
On 8/15/06, Don <THX1138 at dakotacom.net> wrote:
Outrageous overkill. What's that, 10ns memory?
What's the 8's
memory cycle time?? ;-)
The machine cycle time for an -8/e is 1.2uS. ISTR that a core read
cycle on an MM8E (Omnibus core stack), including replacing the old
value after the read destroys it, is on the order of 950ns, but that's
from memory, not from looking it up. In any case,
*any* SRAM, even
stuff from the mid-1970s is fast enough. As for overkill, it's
not
about memory access time, it's about modern availability and package
count - if you want to use a 32Kx8 SRAM, your choices are,
essentially, uber-fast cache RAM in a skinny DIP or still-too-fast
CMOS SRAM (62256) in a wide DIP form factor.
One could even wire up a socket with 3 rows of pins so the user could
choose skinny or wide - the RAMs have, essentially, the same pinout.
-ethan
12:45 a.m.
Ethan Dicks wrote:
On 8/15/06, Don <THX1138 at dakotacom.net>
wrote:
But cache RAM is power hungry and has little practical chance
of being converted to BBSRAM
Outrageous overkill. What's that, 10ns
memory? What's the 8's
memory cycle time?? ;-)
The machine cycle time for an -8/e is 1.2uS. ISTR that a core read
cycle on an MM8E (Omnibus core stack), including replacing the old
value after the read destroys it, is on the order of 950ns, but that's
from memory, not from looking it up. In any case, *any* SRAM, even
stuff from the mid-1970s is fast enough. As for overkill, it's not
about memory access time, it's about modern availability and package
count - if you want to use a 32Kx8 SRAM, your choices are,
essentially, uber-fast cache RAM in a skinny DIP or still-too-fast
CMOS SRAM (62256) in a wide DIP form factor. One could even wire up a socket with 3 rows of pins so
the user could
choose skinny or wide - the RAMs have, essentially, the same pinout.
14 Aug
14 Aug
9:14 p.m.
Ethan Dicks wrote:
Why not use a pair of 62256s? 32Kx16 and ignore 4
bits. It'd even be
easy to use a Dallas DS1210 or something like it to provide battery
backup.
I had the same idea back in December of '98.
I was just looking over my notes, which are not as clear as they should
be, but I'll give a try at converting them to a schematic. Here is a
summary though:
Two 62256 low power SRAMs with 4 bits ignored. The MA bus is buffered
and inverted with 74HCT14s. The MD bus transceiver consists of more
74HCT14s, 74LS240 tri-state buffers and 74LS01s OC drivers, all wired
together as sort of a DEC005.
In retrospect, the 74LS01s are probably a bit weak for this purpose and
the 74HCT14s might not have the ideal threshold.
The control logic is trivial. The SRAM /CS is effectively grounded. /OE
is controlled by MD DIR L. /WE is INHIBIT H NANDed with /OE. Note that
just like core, the memory is written after every read.
I complicated things a bit because I wanted to have the 3 core fields
installed and have the rest semiconductor memory so there are switches
that de-assert /CS when accessing fields 0, 1, or 2. I also have a write
protect for field 7 so there is a switch that prevents asserting /WE
when accessing field 7. I used this for storing a boot loader.
The battery backup circuit is bit complicated with lots of 1N914 diodes.
I was worried about standby leakage. I even used a reed relay to
interrupt /CS on loss of POWER OK H. It seems to have worked though
because I am still using the same 3.6V lithium battery after almost 9 years.
-chuck
10:25 p.m.
C. H. Dickman wrote:
Ethan Dicks wrote:
How about two Ramtron FM18L08's? Nonvolatile like core with no battery
to die or leak. Would need level shifters to 3.3 volts
and slightly different timing. Low power, and in DIP or SMT. I'm just
tired of cleaning boards that had batteries that leaked all over them.
Bob
Why not use a pair of 62256s? 32Kx16 and ignore
4 bits. It'd even be
easy to use a Dallas DS1210 or something like it to provide battery
backup.
10:58 p.m.
On 8/14/2006 at 7:25 PM Bob Rosenbloom wrote:
How about two Ramtron FM18L08's? Nonvolatile like
core with no battery
to die or leak. Would need level shifters to 3.3 volts
and slightly different timing. Low power, and in DIP or SMT. I'm just
tired of cleaning boards that had batteries that leaked all over them.
How long will a supercap keep that much SRAM alive?
Cheers,
Chuck
11:30 p.m.
A supercap can keep low power SRAM alive for many months. The FRAM will
stay unchanged for >10 years.
Bob
Chuck Guzis wrote:
On 8/14/2006 at 7:25 PM Bob Rosenbloom wrote:
How about two Ramtron FM18L08's? Nonvolatile
like core with no battery
to die or leak. Would need level shifters to 3.3 volts
and slightly different timing. Low power, and in DIP or SMT. I'm just
tired of cleaning boards that had batteries that leaked all over them.
How long will a supercap keep that much SRAM alive?
Cheers,
Chuck
11:52 p.m.
Bob Rosenbloom wrote:
A supercap can keep low power SRAM alive for many
months. The FRAM will
stay unchanged for >10 years.
Cycle time? And, is it byte writeable?? And, does it have
a durability specification? (e.g., even an '8 can write
an address a few million times in a few seconds :> )
Bob
Chuck Guzis wrote:
> On 8/14/2006 at 7:25 PM Bob Rosenbloom wrote:
>
>
>
>> How about two Ramtron FM18L08's? Nonvolatile like core with no
>> battery to die or leak. Would need level shifters to 3.3 volts
>> and slightly different timing. Low power, and in DIP or SMT. I'm just
>> tired of cleaning boards that had batteries that leaked all over them.
15 Aug
15 Aug
12:37 a.m.
Don wrote:
Bob Rosenbloom wrote:
It looks just like an SRAM but has slightly different timing. It's not
FLASH.
Check it out at:
http://www.ramtron.com/doc/Products/Nonvolatile/Detail.asp?ID=13&gr=6
Bob
A supercap can keep low power SRAM alive for many
months. The FRAM
will stay unchanged for >10 years.
Cycle time? And, is it byte writeable?? And, does it have
a durability specification? (e.g., even an '8 can write
an address a few million times in a few seconds :> )
14 Aug
14 Aug
11:50 p.m.
Chuck Guzis wrote:
On 8/14/2006 at 7:25 PM Bob Rosenbloom wrote:
Why not use a *big* battery/cell... aren't we talking
about a design that would end up WAY undersized? Even
scaled would leave lots of real estate that you could
install a conventional AA, 9V, etc. battery and forget
about it "forever"
How about two Ramtron FM18L08's? Nonvolatile
like core with no battery
to die or leak. Would need level shifters to 3.3 volts
and slightly different timing. Low power, and in DIP or SMT. I'm just
tired of cleaning boards that had batteries that leaked all over them.
How long will a supercap keep that much SRAM alive?
11:59 p.m.
Don wrote:
Why not use a *big* battery/cell... aren't we
talking
about a design that would end up WAY undersized? Even
scaled would leave lots of real estate that you could
install a conventional AA, 9V, etc. battery and forget
about it "forever"
Batteries DIE...
Core memory lives Forever ... They even used that in
Dr Who as part of a plot line ... OK so Dr Who is not
the most reliable Si-Fi show around :)
Other than boot strap loaders I can't think where
non - volatile memory on the 8 is needed.
15 Aug
15 Aug
12:25 a.m.
On 8/15/06, woodelf <bfranchuk at jetnet.ab.ca> wrote:
Don wrote:
> Why not use a *big* battery/cell... aren't we talking
> about a design that would end up WAY undersized? Even
> scaled would leave lots of real estate that you could
> install a conventional AA, 9V, etc. battery and forget
> about it "forever"
One could dangle an external battery pack off the board, but given
board-to-board clearances, only a lithium coin cell is likely to fit
_on_ the board.
Other than boot strap loaders I can't think where
non - volatile memory on the 8 is needed.
It's a nice demonstration of my -8/L to be running a FOCAL program,
say, and walk up and turn the key, wait, then turn the key back and
the machine is doing exactly what it was doing before the power was
cut off. In practical usage, though, I agree - core on an -8/L saves
having to toggle in the RIM loader every time. For any of my OMNIBUS
machines, except for quickie diagnostics, I don't do much toggling -
they have ROM bootstraps.
-ethan
1:14 a.m.
From: "woodelf" <bfranchuk at jetnet.ab.ca>
Other than boot strap loaders I can't think where
non - volatile memory on the 8 is needed.
It is pretty much an essential feature if you don't have
block-replaceable media of some kind to boot from. Maybe
everyone is hiding block-replaceable media in their terminal
(PC) these days?
I you have a working block device and boot card, then it is
just a luxury :-).
Vince
1:40 a.m.
vrs wrote:
It is pretty much an essential feature if you
don't have
block-replaceable media of some kind to boot from. Maybe
everyone is hiding block-replaceable media in their terminal
(PC) these days?
Yep!!! I will be for the moment.If I can get a IDE drive
working when I get my SBC6120 (PDP 8/e Clone )built I plan to use
a real terminal and dump the Pee-Cee. I just my PC for email,
Adult Games,and some surfing.
I you have a working block device and boot card, then
it is
just a luxury :-).
It still amazes me what they did with 4k of memory on the
8's
and 8Kb on 16 bit machines.
Vince
12:24 a.m.
On 8/14/2006 at 8:50 PM Don wrote:
Why not use a *big* battery/cell... aren't we
talking
about a design that would end up WAY undersized? Even
scaled would leave lots of real estate that you could
install a conventional AA, 9V, etc. battery and forget
about it "forever"
I have always HATED the idea of a battery where one wasn't absolutely
needed. They have a nasty tendency to go south when you can least afford
it--and they can leak or explode.
Or, in the case of Dell, catch fire:
http://news.yahoo.com/s/nm/20060815/tc_nm/dell_batteries_dc_4
Cheers,
Chuck
12:35 a.m.
Chuck Guzis wrote:
On 8/14/2006 at 8:50 PM Don wrote:
That's why using a *regular* battery/cell instead of something
"designed for battery backup" (e.g., Li coin cells, etc.).
I try to design with batteries tht users can purchase at
their own grocery store, etc. (having replaced my share of
overpriced batteries in products that *I* have purchased).
Unless you can make a battery application last A VERY LONG TIME
(not just a LONG time since those are the cases where you get
screwed because you have forgotten about the battery that
you replaced 1 or 2 years ago... *but*, if you replaced it
*10* years ago you don't mind -- as much -- when it dies
after that long of a service life) it will frustrate users.
Why not use a *big* battery/cell... aren't we
talking
about a design that would end up WAY undersized? Even
scaled would leave lots of real estate that you could
install a conventional AA, 9V, etc. battery and forget
about it "forever"
I have always HATED the idea of a battery where one wasn't absolutely
needed. They have a nasty tendency to go south when you can least afford
it--and they can leak or explode.
1:19 a.m.
On 8/14/2006 at 9:35 PM Don wrote:
Unless you can make a battery application last A VERY LONG TIME
(not just a LONG time since those are the cases where you get
screwed because you have forgotten about the battery that
you replaced 1 or 2 years ago... *but*, if you replaced it
*10* years ago you don't mind -- as much -- when it dies
after that long of a service life) it will frustrate users.
Well, consider the Dallas battery-inside-the-chip clock-calendar circuits.
7 years sounded like forever--and it may well have been to the original
owners of the equipment. But now, they're starting to fade and finding
replacements can be a bit of a hassle. In 20 years, it'll probably be
nearly impossible.
For this particular application, any keep-alive power source needs only to
last as long the longest power-off interval. If, as Don, says a supercap
will last several months, that should be good enough--and it's a permanent
solution. Heck, I wonder if you could trickle-charge the battery with
ambient light via a solar cell. That should extend the keep-alive period
some.
I just don't like chemical reactions in my equipment--it's sort of like
bird's nest soup. It might taste good, but I don't like the idea of eating
bird spit..
Cheers,
Chuck.
12:15 p.m.
Chuck Guzis wrote:
On 8/14/2006 at 9:35 PM Don wrote:
Exactly! That's why you don't use a proprietary battery!
(and, a battery *inside* a chip is the most proprietary
form I know :> ).
I deliberately redesigned the clock/bbsram's in my SPARCs to
accept an external battery pack -- so I can just swap the
batteries when the time comes (instead of having to hunt down
replacement parts with stale batteries!)
Unless you can make a battery application last A
VERY LONG TIME
(not just a LONG time since those are the cases where you get
screwed because you have forgotten about the battery that
you replaced 1 or 2 years ago... *but*, if you replaced it
*10* years ago you don't mind -- as much -- when it dies
after that long of a service life) it will frustrate users.
Well, consider the Dallas battery-inside-the-chip clock-calendar circuits.
7 years sounded like forever--and it may well have been to the original
owners of the equipment. But now, they're starting to fade and finding
replacements can be a bit of a hassle. In 20 years, it'll probably be
nearly impossible. For this particular application, any keep-alive power
source needs only to
last as long the longest power-off interval. If, as Don, says a supercap
But how often do you power up your '8? I'm sure many folks
have them sit in a corner and only use them intermittently.
E.g., my Voyager hasn't seen power applied in 6 mos or more :-(
OTOH, if you *know* that those AA cells will keep it
running for years AND you just discipline yourself to
put a new set in as a regular maintenance issue (just
like servicing your smoke detector, PDA, etc.), then
you never find yourself worrying if you've not powered
it up often enough to recharge a supercap!
will last several months, that should be good
enough--and it's a permanent
solution. Heck, I wonder if you could trickle-charge the battery with
ambient light via a solar cell. That should extend the keep-alive period
some.
I just don't like chemical reactions in my equipment--it's sort of like
bird's nest soup. It might taste good, but I don't like the idea of eating
bird spit..
I haven't found a "chemical mess" in a piece of equipment
since some paperclad eveready D cells leaked in a flashlight
a few decades ago. I last replaced the batteries in the
SPARCs 3 years ago and my Compaq portable 3 was probably even
longer than that (it is a more "involved" effort to disassemble
than the SPARCs so it doesn't happen as often -- despite the
fact that the consequences of a BBSRAM data loss in the compaq
are potentially more severe than in the SPARCs).
1:44 a.m.
I have always
HATED the idea of a battery where one wasn't absolutely
needed. They have a nasty tendency to go south when you can least afford
it--and they can leak or explode.
That's why using a *regular* battery/cell instead of something
"designed for battery backup" (e.g., Li coin cells, etc.). Unless you can make a battery application last A VERY
LONG TIME
(not just a LONG time since those are the cases where you get
screwed because you have forgotten about the battery that
you replaced 1 or 2 years ago... *but*, if you replaced it
*10* years ago you don't mind -- as much -- when it dies
after that long of a service life) it will frustrate users.
Er, I don't think it's the longevity of the battery _power_ which is at
issue here; it's the physical longevity of the battery itself.
I have mini-Maglites in my toolkit - three of 'em, and they all get use; I
work in the dark a great deal, but that's another story * - and recently
had the unpleasant surprise of having one of the "regular" batteries
corrode and die (and the batteries had been replaced less than a month
earlier). This did an _astonishingly_ effective job of destroying the
interior of the Maglite as well, the acids eating away the Maglite's
shell in a way I literally didn't think was possible. Defective battery?
Some other problem? We may never know.
Whatever the cause, I'd hate to even _contemplate_ the idea of that
possibly happening right next to a DEC backplane.
-O.-
* - http://flyingmoose.org/stage/
12:18 p.m.
O. Sharp wrote:
[...]
What brand of battery? I haven't seen a battery fail like this
in decades. Was the light stored in a harsh environment?
I have always HATED the idea of a battery where one
wasn't absolutely
needed. They have a nasty tendency to go south when you can least afford
it--and they can leak or explode.
That's why using a *regular* battery/cell
instead of something
"designed for battery backup" (e.g., Li coin cells, etc.). Unless you can make a battery application last A
VERY LONG TIME
(not just a LONG time since those are the cases where you get
screwed because you have forgotten about the battery that
you replaced 1 or 2 years ago... *but*, if you replaced it
*10* years ago you don't mind -- as much -- when it dies
after that long of a service life) it will frustrate users.
Er, I don't think it's the longevity of the battery _power_ which is at
issue here; it's the physical longevity of the battery itself.
I have mini-Maglites in my toolkit - three of 'em, and they all get use; I
work in the dark a great deal, but that's another story * - and recently
had the unpleasant surprise of having one of the "regular" batteries
corrode and die (and the batteries had been replaced less than a month
earlier). This did an _astonishingly_ effective job of destroying the
interior of the Maglite as well, the acids eating away the Maglite's
shell in a way I literally didn't think was possible. Defective battery?
Some other problem? We may never know. Whatever the cause, I'd hate to even _contemplate_
the idea of that
possibly happening right next to a DEC backplane.
6:26 p.m.
> I have mini-Maglites in my toolkit - three of
'em, and they all get use; I
> work in the dark a great deal, but that's another story * - and recently
> had the unpleasant surprise of having one of the "regular" batteries
> corrode and die (and the batteries had been replaced less than a month
> earlier). This did an _astonishingly_ effective job of destroying the
> interior of the Maglite as well, the acids eating away the Maglite's
Actually, I thought the electrolyte was alkaline....
shell in a way
I literally didn't think was possible. Defective battery?
Some other problem? We may never know.
What brand of battery? I haven't seen a battery fail like this
in decades. Was the light stored in a harsh environment?
7:08 p.m.
On 8/15/2006 at 11:26 PM ard at p850ug1.demon.co.uk wrote:
Indeed. I was simply commenting on the OP's statement that the 'acid'
made a right mess of the Maglite.
Incidentally, if the corrosion isn't too bad, a good thing to use to
clean it up is a chemically weak acid. I use citric acid, disolved in
water and applied with a cotton bud. It's good on PCBs that have suffered
Actually, I thought the electrolyte was
alkaline....
It is--and stong alkaline solutions are bad news for aluminum, magnesium
and zinc. Combines with the metal, liberating hydrogen. from NiCd leakage, for example.
-tony
8:27 p.m.
Quoth Tony Duell on my recent battery failure (and for those who have
asked, they were typical off-the-shelf Eveready AAs):
Actually, I thought the electrolyte was alkaline....
Ah, okay.
> > shell in a way I literally didn't think
was possible. Defective battery?
> > Some other problem? We may never know.
>
> What brand of battery? I haven't seen a battery fail like this
> in decades. Was the light stored in a harsh environment?
Not at all. They've been in and out of my toolkit in the same manner for
many years. (I treat them kindly. They're tools, so deserve good treatment
for that alone; and if they fail, it's _ugly_ being stuck in the dark
when scenery is coming at you very fast. <g>)
With some (most?) alkaline primary cells (the
'Duracell' type of thing),
the outer casing of the cell is the +ve connection. There's an insulating
sleeve around it, of course. And the outer casing of a Maglite is
connected to the -ve side of the battery via the spring at the bottom end.
So if the insualting sleeve on a cell fails, then that cell, or the whole
battery, is shorted out. That will cause it to discharge darn fast, and
probably leak.
Ahhh, that's likely what happened then. _Why_ it happened - how the
insulating sleeve came to be damaged to begin with - may well remain a
mystery, but a sleeve failure leading to a short->leakage->alkaline
corrosion->mangling the interior of the Maglite certainly matches the
damage I found.
...Anyway, rare an event as it may be, it would give me serious pause
about using anything battery-backed in a classic mini. They're a _lot_
harder to come by than Maglites. :) :)
-O.-
4:21 p.m.
O. Sharp wrote:
Er, I don't think it's the longevity of the
battery _power_ which is at
issue here; it's the physical longevity of the battery itself.
I have mini-Maglites in my toolkit - three of 'em, and they all get use; I
work in the dark a great deal, but that's another story * - and recently
had the unpleasant surprise of having one of the "regular" batteries
corrode and die (and the batteries had been replaced less than a month
earlier). This did an _astonishingly_ effective job of destroying the
interior of the Maglite as well, the acids eating away the Maglite's
shell in a way I literally didn't think was possible. Defective battery?
Some other problem? We may never know.
Could you have left the flashlight on the last time you put it back into
the toolbox? I have had batteries left on for a long time do rather nasty
things. Probably nasty things happen when they go past "discharged".
Whatever the cause, I'd hate to even _contemplate_
the idea of that
possibly happening right next to a DEC backplane.
12:10 p.m.
On Monday 14 August 2006 11:50 pm, Don wrote:
you could install a conventional AA, 9V, etc. battery
and forget
about it "forever"
Until the damn thing leaked, which has proven to be an issue with several
items of equipment here... :-(
--
Member of the toughest, meanest, deadliest, most unrelenting -- and
ablest -- form of life in this section of space, a critter that can
be killed but can't be tamed. --Robert A. Heinlein, "The Puppet Masters"
-
Information is more dangerous than cannon to a society ruled by lies. --James
M Dakin
12:56 p.m.
Roy J. Tellason wrote:
On Monday 14 August 2006 11:50 pm, Don wrote:
I must have extraordinary luck (?) I just don't see this problem
here. :-( Perhaps you *literally* forgot about it "forever"?
you could install a conventional AA, 9V, etc.
battery and forget
about it "forever"
Until the damn thing leaked, which has proven to be an issue with several
items of equipment here... :-(
1:14 p.m.
On Tuesday 15 August 2006 12:56 pm, Don wrote:
Roy J. Tellason wrote:
Darn near. I had to smack my old simpson 260 to get the highest ohms range to
work, and it turned out that one of the AA batteries in there had leaked,
which are only used on that range.
The bit of masking tape on the back showed them having been changed out in
1987. I'm not sure if they'd been changed in the interim or not...
--
Member of the toughest, meanest, deadliest, most unrelenting -- and
ablest -- form of life in this section of space, a critter that can
be killed but can't be tamed. --Robert A. Heinlein, "The Puppet Masters"
-
Information is more dangerous than cannon to a society ruled by lies. --James
M Dakin
On Monday 14 August 2006 11:50 pm, Don wrote:
I must have extraordinary luck (?) I just don't see this problem
here. :-( Perhaps you *literally* forgot about it "forever"? you could install a conventional AA, 9V, etc.
battery and forget
about it "forever"
Until the damn thing leaked, which has proven to be an issue with
several items of equipment here... :-( 
1:39 p.m.
And thusly were the wise words spake by Roy J. Tellason
On Tuesday 15 August 2006 12:56 pm, Don wrote:
If these batteries have the Eveready 9-lives cat logo on them, then
they probably have not been changed... ;)
Cheers,
Bryan
Roy J. Tellason wrote:
Darn near. I had to smack my old simpson 260 to get the highest ohms range to
work, and it turned out that one of the AA batteries in there had leaked,
which are only used on that range.
The bit of masking tape on the back showed them having been changed out in
1987. I'm not sure if they'd been changed in the interim or not...
On Monday 14 August 2006 11:50 pm, Don wrote:
I must have extraordinary luck (?) I just don't see this problem
here. :-( Perhaps you *literally* forgot about it "forever"? you could install a conventional AA, 9V, etc.
battery and forget
about it "forever"
Until the damn thing leaked, which has proven to be an issue with
several items of equipment here... :-(
12:27 a.m.
On 8/15/06, Bob Rosenbloom <bobalan at sbcglobal.net> wrote:
Don't FRAMs have a maximum write count? Other than that, they are an
excellent idea (and pin-compatible with 62256s, IIRC).
The level shifters, etc., are a bit of a hassle, though.
Old-fashioned SRAM is cheap and plentiful and TTL-friendly.
Ethan Dicks
wrote:
> Why not use a pair of 62256s? 32Kx16 and ignore 4 bits. It'd even be
> easy to use a Dallas DS1210 or something like it to provide battery
> backup.
How about two Ramtron FM18L08's? Nonvolatile like core with no battery
to die or leak. Would need level shifters to 3.3 volts
and slightly different timing. Low power, and in DIP or SMT. I'm just tired of cleaning boards that had
batteries that leaked all over them.
Understandable. I have at least one dead Amiga from a leaky NiCd.
:-( I've never seen Li Coin Cells leak, though.
-ethan
9:03 a.m.
Don't FRAMs have a maximum write count? Other
than that, they are an
excellent idea (and pin-compatible with 62256s, IIRC).
They did, but apparently they have worked that out. From the website:
"You can now access each address a million times per second for
hundreds of years with no wear out"
I remember getting sample of their first chip years ago, and tought it
was pretty neat.
--
Will
11:14 a.m.
On Aug 15, 2006, at 12:27 AM, Ethan Dicks wrote:
Don't FRAMs have a maximum write count?
Well yes, 10^14 writes! ;)
-Dave
--
Dave McGuire
Cape Coral, FL
> Why not use a pair of 62256s? 32Kx16 and ignore 4
bits. It'd
even be
> easy to use a Dallas DS1210 or something like
it to provide battery
> backup.
How about two Ramtron FM18L08's? Nonvolatile like core with no battery
to die or leak. Would need level shifters to 3.3 volts
and slightly different timing. Low power, and in DIP or SMT.
11:52 a.m.
On 8/15/2006 at 11:14 AM Dave McGuire wrote:
So, let's see--for a system powered on and writing once per microsecond
would hae 10^8 seconds worth of writing; given that there are about 3*10^7
seconds in a year, that would be roughly 3.3 years of continuous operation,
no?
But the Ramtron squib for the FM18L08 says:
"FRAM is known for high write-endurance. Until now, some applications
needed more cycles than FRAM could offer. No longer! The FM18L08 offers
virtually unlimited read/write cycles. You can now access each address a
million times per second for hundreds of years with no wear out. "
So, is 10^14 an accurate figure?
Cheers,
Chuck
Don't
FRAMs have a maximum write count?
Well yes, 10^14 writes! ;)
16 Aug
16 Aug
9:07 a.m.
On Aug 15, 2006, at 11:52 AM, Chuck Guzis wrote:
So, let's see--for a system powered on and writing once per microsecond
would hae 10^8 seconds worth of writing; given that there are about
3*10^7
seconds in a year, that would be roughly 3.3 years of continuous
operation,
no?
But the Ramtron squib for the FM18L08 says:
"FRAM is known for high write-endurance. Until now, some applications
needed more cycles than FRAM could offer. No longer! The FM18L08 offers
virtually unlimited read/write cycles. You can now access each address
a
million times per second for hundreds of years with no wear out. "
So, is 10^14 an accurate figure?
Well I got it from this month's Circuit Cellar article on FRAMs,
which I happened to have open on my desk when this came up. I've not
checked any datasheets.
-Dave
--
Dave McGuire
Cape Coral, FL
Don't FRAMs have a maximum write count?
Well yes, 10^14 writes! ;)
12:32 p.m.
On Aug 15, 2006, at 12:33 PM, woodelf wrote:
20 years * 365 * 24 * 60 * 60 * 1/1.24us is well over 10^14
for your typical PDP-8. :(
Sure, but you can buy a spare FRAM when you build the board, and get
another twenty years.
-Dave
--
Dave McGuire
Cape Coral, FL
Don't FRAMs have a maximum write count?
Well yes, 10^14 writes! ;) 
12:50 p.m.
woodelf wrote:
Dave McGuire wrote:
20 years * 365 * 24 * 60 * 60 * 1/1.24us is well over 10^14
for your typical PDP-8. :(
My calculator says 5 x 10^14.
You should also derate by the fact that no program will spend 100% of
its cycles writing to a single location.
20 years is safe.
Don't
FRAMs have a maximum write count?
Well yes, 10^14 writes! ;)
1:18 p.m.
From: "Jim Battle" <frustum at pacbell.net>
woodelf wrote:
Mine also gets a little under 4 years for 10^14, not 20 years.
20 years * 365 * 24 * 60 * 60 * 1/1.24us is well
over 10^14
for your typical PDP-8. :(
My calculator says 5 x 10^14. You should also derate by the fact that no program
will spend 100% of
its cycles writing to a single location.
Actually, I think the program
JMP .
does just that. IIRC, JMP doesn't have and EXEC cycle, and the next cycle
is the FETCH from the new location. Since all cycles, are RMW, the current
location would be then written during the last half of every cycle.
But, I think we can be sure that no-one wrote an operating system where that
was the idle loop, and is running it 24/7 :-).
20 years is safe.
I'd go with 12 (maybe call it 10, to be overcautious :-)). Then you can run
the more common
ISZ FOO / or KSF or whatever
JMP .-1
loop :-). The way I run mine, another 12 years of operating time is well
outside
my expected lifespan!
I think the more compelling problems with FRAM have to do with availability
and
3.3V buffering and such.
Vince
1:50 p.m.
Rumor has it that vrs may have mentioned these words:
I think the more compelling problems with FRAM have to
do with
availability and 3.3V buffering and such.
Gwarsh - then why not just order the 5V version? Unforch, the 1Mbit parts
are 3.3V only, but the 256 & 64Kbit parts (the FM 1808 & FM1608
respectively) have 5V versions in both 120ns & 70ns speeds.
http://www.mouser.com/search/refine.aspx?Mpc=Ramtron%2BMemory%2B%2D%2BParal…
or here, if line-wrapping sucks...
http://tinyurl.com/nweux
Not too expensive, if you ask me. (But you didn't ask me! ;-)
Laterz,
Roger "Merch" Merchberger
--
Roger "Merch" Merchberger | "Profile, don't speculate."
SysAdmin, Iceberg Computers | Daniel J. Bernstein
zmerch at 30below.com |
2:12 p.m.
--- Roger MercMerchbergererzmerchacctalkelow.com>
wrote:
nsspeeds.
The 5 volt parts have a write cycle limitation that's
not there with the 3.3 volt parts.
Bob
I think the more compelling problems with
FRAMFRAMe
to do with
availability and 3.3V buffering and such.
GwarGwarshhen why not just order the 5V version?
UnfoUnforche 1MbitMbitts
are 3.3V only, but the 256 & 64KbitKbitts (the FM
1808 & FM1608
respectively) have 5V versions in both 120ns &ns0ns
3:51 p.m.
Rumor has it that Bob Rosenbloom may have mentioned these words:
--- Roger MercMerchbergererzmerchacctalkelow.com>
wrote:
nsspeeds.
The 5 volt parts have a write cycle limitation that's
not there with the 3.3 volt parts.
Ah, yes - 10 *billion* writes. Dang. ;-)
I do see where replacing core with this stuff could be affected, as core
*needs* to be rewritten every read (reads are destructive, aren't they?) ...
... but you know what? This looks like some *schweet* stuff to use to
rebuild a few of my old Tandy Model 100/200/102 computers with. -0- battery
drain to keep the memory. It looks to be about the same amount of power
consumption whilst active as the existing low-power static RAM installed,
but you wouldn't have to worry as much about keeping the internal nicad
charged (for BBSRAM).
Heck, if one hacked in a switchable /WE line to the ROMs & switched this
thru, one could possibly patch/rewrite the ROMS on the fly, too! ;-)
Things that make you go "Hummmmm"...
Laterz,
Roger "Merch" Merchberger
--
Roger "Merch" Merchberger -- SysAdmin, Iceberg Computers
zmerch at 30below.com
Hi! I am a .signature virus. Copy me into your .signature to join in!
I think the more compelling problems with
FRAMFRAMe
to do with
availability and 3.3V buffering and such.
GwarGwarshhen why not just order the 5V version?
UnfoUnforche 1MbitMbitts
are 3.3V only, but the 256 & 64KbitKbitts (the FM
1808 & FM1608
respectively) have 5V versions in both 120ns &ns0ns
8:23 p.m.
On 8/16/06, Bob Rosenbloom <bobalan at sbcglobal.net> wrote:
The 5 volt parts have a write cycle limitation
that's
not there with the 3.3 volt parts.
Ah... I did not know that 3.3V FRAMs were substantially longer-lived
than the original 5V versions. I'm still in favor of a pair of 62256s
(since they are so common and very cheap), but if I were to do an FRAM
board in the future, it's probably worth the extra buffers and
regulator to get chips that will last.
-ethan
3:26 p.m.
Roger Merchberger wrote:
Rumor has it that vrs may have mentioned these words:
The 5 volt stuff has read/write cycle limits if I remember right.
I think the more compelling problems with FRAM
have to do with
availability and 3.3V buffering and such.
Gwarsh - then why not just order the 5V version? Unforch, the 1Mbit
parts are 3.3V only, but the 256 & 64Kbit parts (the FM 1808 & FM1608
respectively) have 5V versions in both 120ns & 70ns speeds. 
4:47 p.m.
20 years * 365 * 24 * 60 * 60 * 1/1.24us is well over
10^14
for your typical PDP-8. :(
My calculator says 5 x 10^14. Mine also gets a little under 4 years for 10^14, not
20 years.
THAT is PC product life (replace all hardware with every OS "upgrade"),
not PDP product life.
12:45 a.m.
On 8/15/06, C. H. Dickman <chd_1 at nktelco.net> wrote:
Ethan Dicks wrote:
OK. Is a DEC005 OMNIBUS-friendly? We used them on our Qbus COMBOARD,
so that's the context I think of them in.
Why not use a pair of 62256s? 32Kx16 and ignore
4 bits. It'd even be
easy to use a Dallas DS1210 or something like it to provide battery
backup.
I had the same idea back in December of '98.
I was just looking over my notes, which are not as clear as they should
be, but I'll give a try at converting them to a schematic. Here is a
summary though:
Two 62256 low power SRAMs with 4 bits ignored. The MA bus is buffered
and inverted with 74HCT14s. The MD bus transceiver consists of more
74HCT14s, 74LS240 tri-state buffers and 74LS01s OC drivers, all wired
together as sort of a DEC005. In retrospect, the 74LS01s are probably a bit weak for
this purpose and
the 74HCT14s might not have the ideal threshold.
Perhaps. Sorted and rebadged TTL 7401s _are_ OMNIBUS-friendly, aren't they?
The control logic is trivial. The SRAM /CS is
effectively grounded. /OE
is controlled by MD DIR L. /WE is INHIBIT H NANDed with /OE. Note that
just like core, the memory is written after every read.
OK... I thought the core stack handled re-writes, not the CPU...
perhaps I don't understand that part of the machine (never having had
to fiddle with it in depth).
I complicated things a bit because I wanted to have
the 3 core fields
installed and have the rest semiconductor memory so there are switches
that de-assert /CS when accessing fields 0, 1, or 2. I also have a write
protect for field 7 so there is a switch that prevents asserting /WE
when accessing field 7. I used this for storing a boot loader.
That's a reasonable feature. No reason to throw out real core if it
can be used in concert with a MOS board.
The battery backup circuit is bit complicated with
lots of 1N914 diodes.
I was worried about standby leakage. I even used a reed relay to
interrupt /CS on loss of POWER OK H. It seems to have worked though
because I am still using the same 3.6V lithium battery after almost 9 years.
The Dallas DS1210 handles all of that in an 8-pin DIP, including
blocking /CS on power-down. It's a trivial circuit to implement...
just wedge the DS1210 between Vcc coming onto the board and the SRAM,
attach battery power to the DS1210. The only other attachment is the
/CS pass-through, IIRC. One can even design in the DS1210 and install
power and /CS jumpers across the socket if you don't happen to have
the chip handy (or don't care about BBU).
-ethan
1:29 a.m.
From: "Ethan Dicks" <ethan.dicks at gmail.com>
OK. Is a DEC005 OMNIBUS-friendly? We used them on
our Qbus COMBOARD,
so that's the context I think of them in.
Interesting question, even though I'd never design in a DEC005. They are
made of even more unobtainium than the regular tranceivers!
Vince
8:13 p.m.
On 8/15/06, vrs <vrs at msn.com> wrote:
From: "Ethan Dicks" <ethan.dicks at
gmail.com>
Agreed. I have a few, but I used to make Qbus cards. For most folks,
the only source is other cards.
-ethan
OK. Is a DEC005 OMNIBUS-friendly? We used them
on our Qbus COMBOARD,
so that's the context I think of them in.
Interesting question, even though I'd never design in a DEC005. They are
made of even more unobtainium than the regular tranceivers!
12:14 a.m.
From: "Ethan Dicks" <ethan.dicks at gmail.com>
On 8/15/06, woodelf <bfranchuk at jetnet.ab.ca>
wrote:
That's just what I had in mind :-). (I was thinking to design for the
SRAM and let people substitute the NVRAM if they wanted.)
Vince
If you use 2167's ( 16k x 1 ) ram's
you can have 16k of memory ( 12 chips ) for
about $25 not counting glue logic and other stuff.
Why not use a pair of 62256s? 32Kx16 and ignore 4 bits. It'd even be
easy to use a Dallas DS1210 or something like it to provide battery
backup. 
14 Aug
14 Aug
1:41 p.m.
Chuck wrote...
I did a semiconductor memory design for the OMNIBUS
several years ago. It
was for an -8/e and 32kW using two SRAMs. I never got to documenting it
outside of my notebook, but if there is enough interest, I could do some
schematics. It is battery backed, so its pretty much equivalent to the
core I was replacing.
Maybe it was your diagram I got years ago. Someone did a nice 32kw board for
the 8e using two 2114's I think it was. I have the schematics somewhere
still. The only reason I didn't build it right away was because at the time
I got them, the person said "these schematics aren't quite final, there may
have been some tweaking afterwards that didn't make it to the schematic" so
I just sat on them.
A known working design - I'd want to build that in a heartbeat :)
Jay West
6:10 p.m.
Dave wrote....
> That'd be a pretty good trick!
32kW (122 bit words) in 1k bytes of RAM, I'd like to see that done too...
I'm going from foggy memory, but as I recall the
schematic was for less than
32kw, but had ----- lines showing "add additional chips & lines here" for
the 32kw.
If you're going to use 2114s (and I'd recoemnd against it), then you want
to use them in 3s. Each chip is 4 bits wide, so 3 of them will store 1k
12 bit words.
-tony
6556
days inactive
6559
days old
72 comments
23 participants
participants (23)
-
ajp166＠bellatlantic.net -
ard＠p850ug1.demon.co.uk -
bfranchuk＠jetnet.ab.ca -
bobalan＠sbcglobal.net -
bpope＠wordstock.com -
cclist＠sydex.com -
chd_1＠nktelco.net -
cisin＠xenosoft.com -
dgriffi＠cs.csubak.edu -
ethan.dicks＠gmail.com -
frustum＠pacbell.net -
gerold.pauler＠gmx.net -
jwest＠classiccmp.org -
kth＠srv.net -
mcguire＠neurotica.com -
ohh＠drizzle.com -
pat＠computer-refuge.org -
rtellason＠verizon.net -
THX1138＠dakotacom.net -
useddec＠gmail.com -
vrs＠msn.com -
wdonzelli＠gmail.com -
zmerch-cctalk＠30below.com