I hate Dallas Semicon engineers: reworking the DS12887A 
Author Message
User avatar

Joined: 2014-09-27 09:29
Posts: 2194
Location: Soviet Venezuela
 I hate Dallas Semicon engineers: reworking the DS12887A
Everybody that owns a device containing one of those dreaded battery-sealed-inside RTC/CMOS bricks eventually has to go through the pain when the battery finally goes flat, your device stops working/booting, and you can't simply go and replace the battery since it comes sealed inside a block of epoxy.

Your options then?
- Buy a new chip: While the DS12887A is still in production, it may be not so easy depending on which chip you have (depending on your specific system, a DS12887 might not work as a replacement for a DS1287, and if you're unlucky, you can't even swap a DS12887 with a DS12887A - sometimes it becomes hit-or-miss) or where you live (easy if you live in 'murrica, impossible if you live in Soviet Venezuela)
- Salvage an used chip from a dead motherboard: finding boards with a working RTC/CMOS brick is getting increasingly hard as such systems are vanishing from existence. Those ICs stopped being used on PCs sometime around the height of Socket 7 CPUs, and were all but dead by the turn of the century. Plus even if you get one, you will end with a battery which is at least 20 years old at this point, so you're effectively solving nothing.
- Rework the fucker: this involves removing enough epoxy to reach the battery pins, sever the connection to the (now dead) battery + pole, and solder wires to a new 3V battery (usually a battery holder for a CR2032 cell). The procedure is well documented around the net, but it can still be scary... and dealing with that goddamned epoxy requires time.

Long short story: I own two DS12887 boxes at home: one of them, my Acer 386SX box had its DS1287 battery die for the second time in 10 years (since I own that box).

When the battery runs flat, CMOS settings will be lost, you won't be able to boot from the HDD, etc. Even worse: trying to access the BIOS setup program to reset things will freeze the machine due to some BIOS bug that doesn't expect to deal with a RTC chip without battery (oddly enough, a fresh chip with a good battery will allow to run the BIOS setup program without freezing, even if the chip was pulled from another system with a completely different memory layout). At this stage, I can only boot from the first floppy drive (boot to DOS -> set the date -> reboot -> naively try to enter BIOS setup thinking that the clock is now valid -> machine freezes because NOPE -> headdesk), but nothing else.

Last time it was in 2009, and while I was able to salvage a (not-so) fresh (and much newer) DS12887A from a dead server motherboard, it was hard to get it replaced: Acer decided to save a couple of cents on a socket and just soldered the bastard to the motherboard. Since I couldn't desolder it (right now I still can't desolder things without destroying them), I had someone at the university computer labs to remove the chip and solder a socket in its place so I can put my replacement. I was really lucky: this specific Acer BIOS will happily initialize the chip with valid data (as long as the battery is good!) - no need for OEM-specific "CMOS initializing programs" (not that you could find those anymore in nearly-2017). Fast-forward to today, where the stupid replacement also ran out of battery (it doesn't help the fact i boot my 386sx only once yearly!). I'm no more a college student, which means no access to a well-stocked junkyard, no way to get a fresh DS12887, and with no option but to perform surgery on my dead chip while cursing the morons at Dallas Semiconductors that decided that it was an AWESOME idea to bury a consumable battery inside a tomb of epoxy (dude, it's a freakin' RTC, not some IP-sensitive component like those encrypted CPUs from Sega arcade games!)

For anyone living on a country with access to luxuries like a Dremel and a soldering station, this would be a piece of cake (but then, if you live there you could simply go and buy a box of RTCs from Aliexpress or eBay). Except that I am on this hellhole, and forced to McGyver. Here are my results. Yes, that's masking tape (not even black electrical tape - that shit sucks, it gets really gooey after stored for too long). And I also managed to break pin 20 (battery +) while trying to kill the internal battery connection, but fortunately enough metal remained at the IC for solder to stick. Yup, that's a paperclip holding the battery wires (too lazy to desolder a battery holder from one of my dead motherboards, I guess) - I ended wrapping it with a small plastic bag so it doesn't short things that shouldn't be shorted.

The end result: http://imgur.com/a/HeAeF

Amazingly enough, it worked at the first try. No sparks, no magic smoke, no dead HDDs. But I really hate those RTC chips. Now I fear for my routerbox since it also uses a DS12887A (originally an Houston Tech something something, which actually was a battery-less VIA 82885 clone chip - that PCCHIPS motherboard happily accepts a genuine Dallas IC, although clock tends to drift a few seconds per week for no good reason at all)

Sidenote: that Acer motherboard had been originally designed for a DS1187 (a DS1287 is nothing but a DS1187 + battery + crystal + epoxy block of doom) - there are mounting spots for a crystal and a battery. I guess that those epoxy bricks were cheaper than just providing a user-friendly solution. Ugh.

_________________
Licensed Pirate® since 2006


2016-12-26 05:40
User avatar

Joined: 2014-09-27 09:56
Posts: 1688
 Re: I hate Dallas Semicon engineers: reworking the DS12887A
Dang, part of me died inside just looking at that photo. I know full well how much difference it makes having access to the right tools and materials, but that just... man :(

_________________
byuu wrote:
Surely I can unite the vi vs emacs crowd like I have the ZSNES vs Snes9X crowd :)



higan WIP builds available here


2016-12-26 06:08
User avatar

Joined: 2014-09-27 09:22
Posts: 5157
Location: A chair.
 Re: I hate Dallas Semicon engineers: reworking the DS12887A
Good job.
I almost did that with my Sun Ultra10, before I managed to track down the right part number for a replacement. Fun fact: On a Sun, that NVRAM stores a bunch of configuration stuff, including the system's serial # and the MAC address of the integrated ethernet port.

_________________
Just in case you thought something could EVER be straightforward, and needed someone to dash your hopes across the rocky shoals of harsh reality.

; write !!!


2016-12-26 06:44
User avatar

Joined: 2014-09-27 09:29
Posts: 2194
Location: Soviet Venezuela
 Re: I hate Dallas Semicon engineers: reworking the DS12887A
qwertymodo wrote:
Dang, part of me died inside just looking at that photo. I know full well how much difference it makes having access to the right tools and materials, but that just... man :(


If that comforts you, on the pages I've linked, there are photos of reworks done in the same terrible way. The design of this bastard IC doesn't really allow for clean reworks, unless if you're willing to remove all of the epoxy (willing to change the crystal too!?)

In my case, the only difference (aside of the ghetto battery holder, something I might address in the future should I manage to desolder a battery holder) is the liberal use of masking tape instead of the Hot Glue Miracle™. I really hate dealing with hot glue (hot, messy, I've had cases of glue too hot that ended desoldering a fragile solder joint, and if you mess it up it means scrapping the entire component away). I know you EE guys are masters of the soldering iron and hot glue gun, but for my first time doing one of these horrible ICs, I'm amazed at the fact it Just Works™ at the first try, rather than ending with a soldering mess* and shorted parts.

Well, some solder ended on some of the IC legs, but fortunately no bridges :)

_________________
Licensed Pirate® since 2006


2016-12-27 12:32
User avatar

Joined: 2014-09-27 09:22
Posts: 5157
Location: A chair.
 Re: I hate Dallas Semicon engineers: reworking the DS12887A
tomman wrote:
qwertymodo wrote:
Dang, part of me died inside just looking at that photo. I know full well how much difference it makes having access to the right tools and materials, but that just... man :(


If that comforts you, on the pages I've linked, there are photos of reworks done in the same terrible way. The design of this bastard IC doesn't really allow for clean reworks, unless if you're willing to remove all of the epoxy (willing to change the crystal too!?)

When I saw people doing it, they were using a fucking pocket knife to whittle away the packaging. It doesn't work very well at all.

I guess what I am saying is that access to the right tools(if such things exist for this job) doesn't mean they will be used.

_________________
Just in case you thought something could EVER be straightforward, and needed someone to dash your hopes across the rocky shoals of harsh reality.

; write !!!


2016-12-27 15:31
User avatar

Joined: 2014-09-27 09:29
Posts: 2194
Location: Soviet Venezuela
 Re: I hate Dallas Semicon engineers: reworking the DS12887A
Doing some cleanup on my room, turns that I actually have a spare CR2032 battery holder (cleanly desoldered, so most likely it wasn't me!), so this story might have a less messy end.

But the masking tape isn't going away, I tell you!

_________________
Licensed Pirate® since 2006


2016-12-29 00:05
User avatar

Joined: 2014-09-27 09:22
Posts: 5157
Location: A chair.
 Re: I hate Dallas Semicon engineers: reworking the DS12887A
MASKIN' TAPE 4 LYFE!

_________________
Just in case you thought something could EVER be straightforward, and needed someone to dash your hopes across the rocky shoals of harsh reality.

; write !!!


2016-12-29 01:15