22 Jan
2017
22 Jan
'17
10:20 a.m.
Jon,
Then why not use a dip compatible version of the DS chip? I mean yes this allows for
switchable battery which is very nice but SMT soldering is not for everyone.
I wonder if there is a way to determine if there will be BIOS issues by switching the
DS12887?
-------- Original message --------
From: Jon Elson <elson at pico-systems.com>
Date: 1/22/17 9:16 AM (GMT-08:00)
To: General at classiccmp.org, "Discussion at classiccmp.org:On-Topic and Off-Topic
Posts" <cctalk at classiccmp.org>
Subject: Re: DS12887 pcb substitute with battery
On 01/22/2017 10:07 AM, Ali wrote:
Al,
I thought the problem with switching these chips was that part of the ROM code was
embedded in them? I.e. it isn't just an issue of battery? Am I wrong? If I am then why
not use one of the replacement chips that are available?
These don't have a lot of memory on them.? many early PCs
stored some config info there, but generally the BIOS can
reconstruct it if it isn't there.? I suppose there is a
possibility that random data in the CMOS memory could cause
the BIOS to try to use unavailable features and hang.? I
don't think anybody put actual executable code in there.
Jon
22 Jan
22 Jan
10:57 a.m.
I don't know about most people but this solution has been around for
decades.
I locate the battery on the failed part with a small magnet, then grind the
epoxy down to it then pick it out with a sharp pointed tool. Once I expose
the connection point I older two wires then epoxy a small coin-cell holder
in that spot and it s done. I've done this more times than I care to count
and its effective and the replacement battery some over 10 years old
now have not failed. But just in case I have a bag of NOS replacements
(and pulls from socketed boards) all with dead batteries from age. There
is no magic to this.
Allison
On 01/22/2017 01:20 PM, Ali wrote:
Jon,
Then why not use a dip compatible version of the DS chip? I mean yes this allows for
switchable battery which is very nice but SMT soldering is not for everyone.
I wonder if there is a way to determine if there will be BIOS issues by switching the
DS12887?
-------- Original message --------
From: Jon Elson <elson at pico-systems.com>
Date: 1/22/17 9:16 AM (GMT-08:00)
To: General at classiccmp.org, "Discussion at classiccmp.org:On-Topic and Off-Topic
Posts" <cctalk at classiccmp.org>
Subject: Re: DS12887 pcb substitute with battery
On 01/22/2017 10:07 AM, Ali wrote:
Al,
I thought the problem with switching these chips was that part of the ROM code was
embedded in them? I.e. it isn't just an issue of battery? Am I wrong? If I am then why
not use one of the replacement chips that are available?
These don't have a lot of memory on them. many early PCs
stored some config info there, but generally the BIOS can
reconstruct it if it isn't there. I suppose there is a
possibility that random data in the CMOS memory could cause
the BIOS to try to use unavailable features and hang. I
don't think anybody put actual executable code in there.
Jon 
11:46 a.m.
On 01/22/2017 10:57 AM, allison wrote:
I don't know about most people but this solution
has been around for
decades.
I locate the battery on the failed part with a small magnet, then
grind the epoxy down to it then pick it out with a sharp pointed
tool. Once I expose the connection point I older two wires then
epoxy a small coin-cell holder in that spot and it s done. I've done
this more times than I care to count and its effective and the
replacement battery some over 10 years old now have not failed. But
just in case I have a bag of NOS replacements (and pulls from
socketed boards) all with dead batteries from age. There is no magic
to this.
It's worth noting that the original post was about fabricating a
replacement using the DS12885A RTC chip. However, most old PCs used the
DS1285 RTC (inside of a DS1287 module). The DS12885A is supposed to be
drop-in compatible with the DS1285, but apparently, in some cases is not.
So there's logic in reworking the old DS1287 modules, as the DS1285 chip
is long out of production--you'll most likely have to be content with
pulls or the occasional NOS lot.
--Chuck
1:49 p.m.
On Sun, 22 Jan 2017, Chuck Guzis wrote:
It's worth noting that the original post was about
fabricating a
replacement using the DS12885A RTC chip. However, most old PCs used the
DS1285 RTC (inside of a DS1287 module). The DS12885A is supposed to be
drop-in compatible with the DS1285, but apparently, in some cases is not.
Of course it's not, because the DS12885 has additional 128 bytes of
internal RAM addressable with the use of the AD6 line, which used to be a
don't-care for the address register in the DS1285. So if a system or a
piece of software uses bit #6 of the address for some other purposes while
letting it through down to the chip, expected to ignore it, then the
effect will be different with the expanded chip compared to the original.
Likewise with all the module variations made around these cores.
So there's logic in reworking the old DS1287
modules, as the DS1285 chip
is long out of production--you'll most likely have to be content with
pulls or the occasional NOS lot.
A problem with reworking is there is sometimes very little clearance
available, so any modification has to be made in a clever way or you risk
a short-circuit. For example in DECstation 5000 systems their DS1287 chip
is located in the TURBOchannel option slot area with the top of the chip
almost reaching the solder side of the PCB of any option plugged in there.
Maciej
2:12 p.m.
On 01/22/2017 01:49 PM, Maciej W. Rozycki wrote:
A problem with reworking is there is sometimes very
little clearance
available, so any modification has to be made in a clever way or you
risk a short-circuit. For example in DECstation 5000 systems their
DS1287 chip is located in the TURBOchannel option slot area with the
top of the chip almost reaching the solder side of the PCB of any
option plugged in there.
Although there are several illustrations of the DS1287 rework where a
coin-cell holder was placed atop the chip, there's absolutely no reason
why that has to be so. A couple of wire leads away to an off-board
battery holder, holding, say, an ER14250 half-AA.will work just fine and
keep the battery away from the PCB--or a couple of alkaline AA cells,
which should last a few years. No reason to stick to lithium.
You know--"use your wetware".
--Chuck
12 Nov
12 Nov
4:44 p.m.
Hi Chuck,
It's taken me a little while to respond to this one, however certain
matters do need time to develop.
On Sun, 22 Jan 2017, Chuck Guzis wrote:
On 01/22/2017 01:49 PM, Maciej W. Rozycki wrote:
Yeah, putting the aesthetic of the solution aside, in the DECstation
there's the practical issue of making sure the wires do not obstruct
shuffling TURBOchannel option cards and also finding a place where the
substitute battery does not disturb the airflow such as to cause a
component somewhere to overheat -- the whole side of the enclosure is a
grille.
A problem with reworking is there is sometimes
very little clearance
available, so any modification has to be made in a clever way or you
risk a short-circuit. For example in DECstation 5000 systems their
DS1287 chip is located in the TURBOchannel option slot area with the
top of the chip almost reaching the solder side of the PCB of any
option plugged in there.
Although there are several illustrations of the DS1287 rework where a
coin-cell holder was placed atop the chip, there's absolutely no reason
why that has to be so. A couple of wire leads away to an off-board
battery holder, holding, say, an ER14250 half-AA.will work just fine and
keep the battery away from the PCB--or a couple of alkaline AA cells,
which should last a few years. No reason to stick to lithium. You know--"use your wetware".
That always helps, as long as it's got enough input to process, in
particular as to what resources are available.
Mine has got that now, thanks to this thread of discussion, where I
learnt of the existence of this small coin cell holder, and also the other
thread mentioning wire-wrap wire and its excellent properties for small
circuit repair/modification works. Naturally I picked up the green
insulation color to match what DEC used to use, at least in their
TURBOchannel equipment. :)
So here: <ftp://ftp.linux-mips.org/pub/linux/mips/people/macro/ds1287/>
are a couple photos of the result, meeting both the clearance requirements
and my subjective aesthetic criteria, and avoiding the issues I identified
above.
Taking into account the rated DS1285's standby power consumption of 0.5?A
and the 35mAh 2.5V discharge level capacity of the CR1220 cell it will
guarantee keeping the chip powered for 8 years. Which I think is enough
given the ease to replace the cell now.
For comparison the 48mAh capacity of the BR1225 cell originally embedded
in the chip allowed for 11 years of operation and the CR2032 cell people
seem to commonly choose allows for 50 years at its 220mAh capacity (which
I find a bit of an overkill).
Note that I have calculated approximate cell capacities based on
discharge characteristic curves published by cell manufacturers, as the
rated capacities published are calculated for the 2.0V discharge level
whereas DS1285 is specified for 2.5V minimum battery voltage required for
correct operation.
For the record, I have measured the clearance available and it's 0.1875"
(3/16") or 4.76mm from the top of the RTC chip to the bottom of the PCB of
a TURBOchannel option right above it. The TURBOchannel mechanical spec
allows for 0.075" (3/40") or 1.91mm underneath component or lead height
(e.g. 24-plane HX+ options have memory chips there). Which leaves 0.1125"
(9/80") or 2.86mm of space only.
I may yet stick a piece of insulating tape on top of the RTC chip just to
make sure no contact happens by chance, though I think it might be an
overkill after all. It's not like the cell is going to pop out by itself.
Maciej
13 Nov
13 Nov
8:01 a.m.
Good stuff! I recently designed a module to build new DS1287 and DS12887
modules from the bare DS1285 and DS12885 ICs:
https://imgur.com/a/cgKm5
Just did a small run of 100 boards with the GW-12887-1 part number (they of
course work with the DS1285, though the number would be misleading). It
uses a 1225 cell as did the original, though the life of the original was
supposedly boosted by the potting not allowing the cell to evaporate as
much electrolyte through its (presumably imperfect) seal. I've see another
replacement that uses the surface mount version of the DS12885, but that
module requires surface mount bits and solder-down IC legs, whereas my
module just gets glued on top of the IC, and the appropriate legs bent up
and soldered into the castellated holes.
The motivation for the board was, as with yours, height restrictions. I did
a similar repair board for MK48T02 and MK48T08 timekeeper NVRAMs as used on
Sun systems -- there's an SBus slot directly over the NVRAM on some Sun
machines, which prevents the "glue a battery on top" method from working.
Thanks,
Jonathan
On Sun, Nov 12, 2017 at 7:44 PM, Maciej W. Rozycki via cctech <
cctech at classiccmp.org> wrote:
Hi Chuck,
It's taken me a little while to respond to this one, however certain
matters do need time to develop.
On Sun, 22 Jan 2017, Chuck Guzis wrote:
On 01/22/2017 01:49 PM, Maciej W. Rozycki wrote:
Yeah, putting the aesthetic of the solution aside, in the DECstation
there's the practical issue of making sure the wires do not obstruct
shuffling TURBOchannel option cards and also finding a place where the
substitute battery does not disturb the airflow such as to cause a
component somewhere to overheat -- the whole side of the enclosure is a
grille.
A problem with reworking is there is sometimes
very little clearance
available, so any modification has to be made in a clever way or you
risk a short-circuit. For example in DECstation 5000 systems their
DS1287 chip is located in the TURBOchannel option slot area with the
top of the chip almost reaching the solder side of the PCB of any
option plugged in there.
Although there are several illustrations of the DS1287 rework where a
coin-cell holder was placed atop the chip, there's absolutely no reason
why that has to be so. A couple of wire leads away to an off-board
battery holder, holding, say, an ER14250 half-AA.will work just fine and
keep the battery away from the PCB--or a couple of alkaline AA cells,
which should last a few years. No reason to stick to lithium. You know--"use your wetware".
That always helps, as long as it's got enough input to process, in
particular as to what resources are available.
Mine has got that now, thanks to this thread of discussion, where I
learnt of the existence of this small coin cell holder, and also the other
thread mentioning wire-wrap wire and its excellent properties for small
circuit repair/modification works. Naturally I picked up the green
insulation color to match what DEC used to use, at least in their
TURBOchannel equipment. :)
So here: <ftp://ftp.linux-mips.org/pub/linux/mips/people/macro/ds1287/>
are a couple photos of the result, meeting both the clearance requirements
and my subjective aesthetic criteria, and avoiding the issues I identified
above.
Taking into account the rated DS1285's standby power consumption of 0.5?A
and the 35mAh 2.5V discharge level capacity of the CR1220 cell it will
guarantee keeping the chip powered for 8 years. Which I think is enough
given the ease to replace the cell now.
For comparison the 48mAh capacity of the BR1225 cell originally embedded
in the chip allowed for 11 years of operation and the CR2032 cell people
seem to commonly choose allows for 50 years at its 220mAh capacity (which
I find a bit of an overkill).
Note that I have calculated approximate cell capacities based on
discharge characteristic curves published by cell manufacturers, as the
rated capacities published are calculated for the 2.0V discharge level
whereas DS1285 is specified for 2.5V minimum battery voltage required for
correct operation.
For the record, I have measured the clearance available and it's 0.1875"
(3/16") or 4.76mm from the top of the RTC chip to the bottom of the PCB of
a TURBOchannel option right above it. The TURBOchannel mechanical spec
allows for 0.075" (3/40") or 1.91mm underneath component or lead height
(e.g. 24-plane HX+ options have memory chips there). Which leaves 0.1125"
(9/80") or 2.86mm of space only.
I may yet stick a piece of insulating tape on top of the RTC chip just to
make sure no contact happens by chance, though I think it might be an
overkill after all. It's not like the cell is going to pop out by itself.
Maciej
21 Nov
21 Nov
11:16 a.m.
On Mon, 13 Nov 2017, systems_glitch wrote:
Good stuff! I recently designed a module to build new
DS1287 and DS12887
modules from the bare DS1285 and DS12885 ICs:
https://imgur.com/a/cgKm5
Nice!
That wouldn't solve my problem though, given the apparent unavailability
of DS1285 chips (in any packaging, whether surplus or used), compared with
the ubiquity of used DS1287 and DS1287A ones, possibly because they were
often socketed. Using my systems as a reference for the Linux port I want
to avoid any deviation from their original specification so that software
does not make use of it by chance.
As to the DS12887 and DS12887A chips -- these are still manufactured and
readily available, as someone mentioned a while ago, although a bit highly
priced, so either reworking an old one or using your alternative does help
cutting cost, which may especially matter if you need more than just a
few.
Maciej
6:54 p.m.
What are you talking about?
https://www.digikey.com/product-detail/en/maxim-integrated/DS12885/DS12885-…
-Alan
On 2017-11-21 14:16, Maciej W. Rozycki via cctech wrote:
On Mon, 13 Nov 2017, systems_glitch wrote:
Good stuff! I recently designed a module to build
new DS1287 and
DS12887
modules from the bare DS1285 and DS12885 ICs:
https://imgur.com/a/cgKm5
Nice!
That wouldn't solve my problem though, given the apparent
unavailability
of DS1285 chips (in any packaging, whether surplus or used), compared
with
the ubiquity of used DS1287 and DS1287A ones, possibly because they
were
often socketed. Using my systems as a reference for the Linux port I
want
to avoid any deviation from their original specification so that
software
does not make use of it by chance.
As to the DS12887 and DS12887A chips -- these are still manufactured
and
readily available, as someone mentioned a while ago, although a bit
highly
priced, so either reworking an old one or using your alternative does
help
cutting cost, which may especially matter if you need more than just a
few.
Maciej
23 Nov
23 Nov
3 p.m.
Hi Alan
What are you talking about?
https://www.digikey.com/product-detail/en/maxim-integrated/DS12885/DS12885-…
Thank you for your input, however a DS12885 is not a DS1285 (and neither
a DS12887/A is a DS1287/A), and I did explain why I only consider an exact
replacement suitable for my use.
Maciej
16 Nov
16 Nov
1:07 a.m.
New subject: Forth for PDP-8?
Kyle Owen wrote:
Just a little bit of cursory looking turned up
nothing. Has someone
written a Forth interpreter for PDP-8? Preferably using 24-bit
integers, but I'll take what I can get. OS/8 support would be nice.
I recently wrote a Forth cross compiler with PDP-8 support.
https://github.com/larsbrinkhoff/xForth/tree/master/target/pdp8
23 Jan
23 Jan
2:57 a.m.
On 01/22/2017 02:46 PM, Chuck Guzis wrote:
On 01/22/2017 10:57 AM, allison wrote:
I haven't seen that version for a while. The later are fully versions
are epoxy filled.
Same for the MT48T part, same fix.
The problem with NOS parts is manufacture date. Some are really old.
Allison
I don't know about most people but this
solution has been around for
decades.
I locate the battery on the failed part with a small magnet, then
grind the epoxy down to it then pick it out with a sharp pointed
tool. Once I expose the connection point I older two wires then
epoxy a small coin-cell holder in that spot and it s done. I've done
this more times than I care to count and its effective and the
replacement battery some over 10 years old now have not failed. But
just in case I have a bag of NOS replacements (and pulls from
socketed boards) all with dead batteries from age. There is no magic
to this.
It's worth noting that the original post was about fabricating a
replacement using the DS12885A RTC chip. However, most old PCs used the
DS1285 RTC (inside of a DS1287 module). The DS12885A is supposed to be
drop-in compatible with the DS1285, but apparently, in some cases is not.
So there's logic in reworking the old DS1287 modules, as the DS1285 chip
is long out of production--you'll most likely have to be content with
pulls or the occasional NOS lot.
--Chuck 
9:04 a.m.
On 01/23/2017 04:57 AM, allison wrote:
On 01/22/2017 02:46 PM, Chuck Guzis wrote:
Well, it doesn't really matter. If you can find one of the
"really old" ones, the battery can be replaced and you are
good for 5 years or so. There must be TONS of these old
clock/RAM chips out there, and somebody must have saved a few.
Jon
On 01/22/2017 10:57 AM, allison wrote:
I haven't seen that version for a while. The later are fully
versions
are epoxy filled.
Same for the MT48T part, same fix.
The problem with NOS parts is manufacture date. Some are really old.
I don't know about most people but this
solution has been around for
decades.
I locate the battery on the failed part with a small magnet, then
grind the epoxy down to it then pick it out with a sharp pointed
tool. Once I expose the connection point I older two wires then
epoxy a small coin-cell holder in that spot and it s done. I've done
this more times than I care to count and its effective and the
replacement battery some over 10 years old now have not failed. But
just in case I have a bag of NOS replacements (and pulls from
socketed boards) all with dead batteries from age. There is no magic
to this.
It's worth noting that the original post was about fabricating a
replacement using the DS12885A RTC chip. However, most old PCs used the
DS1285 RTC (inside of a DS1287 module). The DS12885A is supposed to be
drop-in compatible with the DS1285, but apparently, in some cases is not.
So there's logic in reworking the old DS1287 modules, as the DS1285 chip
is long out of production--you'll most likely have to be content with
pulls or the occasional NOS lot.
--Chuck
11:39 a.m.
On 01/23/2017 09:04 AM, Jon Elson wrote:
Well, it doesn't really matter. If you can find
one of the "really
old" ones, the battery can be replaced and you are good for 5 years
or so. There must be TONS of these old clock/RAM chips out there, and
somebody must have saved a few.
Probably being sold on eBay as NOS. ;)
Reminiscent of the time when I ran across a seller offering a deal on
some bipolar PROMs. When I asked about the history of the parts, he
said that they were pulls, but he'd checked them out and they worked fine.
Oooookay...
--Chuck
22 Jan
22 Jan
6:34 p.m.
On 01/22/2017 12:57 PM, allison wrote:
I don't know about most people but this solution
has been around for
decades.
I locate the battery on the failed part with a small magnet, then grind the
epoxy down to it then pick it out with a sharp pointed tool. Once I expose
the connection point I older two wires then epoxy a small coin-cell holder
in that spot and it s done. I've done this more times than I care to count
and its effective and the replacement battery some over 10 years old
now have not failed. But just in case I have a bag of NOS replacements
(and pulls from socketed boards) all with dead batteries from age. There
is no magic to this.
Most of these Dallas clock/RAM chips can be popped open VERY
easily! Take an Xacto knife and pick around the edges on
the pin side, and usually the plastic cover will pop free
from the epoxy fill. Then, the cover can be lifted,
and you
will find a standard 28-pin (I think) DIP chip with a button
cell wired onto the top of the chip. The ones I've opened
were not totally filled with epoxy, the IC package acted as
a sort of dam to prevent the cover from completely filling
with the stuff.
Jon
2351
days inactive
2656
days old
14 comments
8 participants
participants (8)
-
ajp166＠verizon.net -
alan＠alanlee.org -
cclist＠sydex.com -
cctalk＠ibm51xx.net -
elson＠pico-systems.com -
lars＠nocrew.org -
macro＠linux-mips.org -
systems.glitch＠gmail.com