Is it possible you got a NOS board? The clock doesnt start ticking in those until you first boot it up

I have a P6NP5 obtained in 2002 with original Dallas DS12B887 dated 9622 and use it regularly and it still holds settings and time. I think if the board is in machine that got used daily the battery lasts longer for some reason or just am lucky.

Have a (slightly OT) DS1387 on an EISA board dated 9246 which is still juiced up, tho I seem to remember reading here that EISA boards may be lighter on battery use?

It was not NOS board. I believe the battery is always connected once manufactured.

In datasheet they say the battery life is 10 years.

27 years is remarkable.

There's a register in the nvram that controls whether the internal oscillator runs.
It ships from the factory with it disabled to preserve the battery.
It is possible to disable it via software, which will prevent the unit from keeping time but prolong the battery life during long-term storage.

My lucky Dallas pieces are Shuttle HOT-433 and Asus P6NP5 (anyone? 😁).
The HOT-433 is especially interesting: after long hybernation in the motherboard cupboard it said it forgot the settings yet it could remdmber them after I set up everything. Interesting (and fortunate) behavior on a Dallas' side. I don't know if it means a strange wiring that powers the RTC when the board gets power or it's really on its last legs yet somehow it can retain settings for an hour or so.

All in all, I consider both boards ticking time bombs. I'll have to learn how to patch Dallas chips while they're soldered in. Crazy. Or remove these bastards, solder in a socket then patch the RTC chips. Even crazier. Bad that I like both boards.

Yes. I know about control registers. Arguably I'd expect that in longer term self-discharge of the battery is bigger problem than oscillator drain if you are talking 20+ years.

They say the in the active state with the oscillator ticking the device consumes 500 nA (100 nA when inactive).

I normally prefer to unsolder them and replace with DIP24 socket and a newer unit. Don't like how modded chips look.

I much prefer sealed Dallas over leaky battery. That's a time bomb. Dallas is fine IMHO.

a dallas is the literal opposite of a time bomb, when they fail they fail gracefully and do no harm

mpe wrote on 2021-01-09, 14:02:

I normally prefer to unsolder them and replace with DIP24 socket and a newer unit. Don't like how modded chips look.

this is the way - I don't even care about the "look" as much as the loss of the encapsulation protecting surrounding circuits from battery failure

lithium primary self-discharge is minuscule btw, the 10 year shelf life that gets bandied about is routinely exceeded geometrically

maxtherabbit wrote on 2021-01-09, 15:32:

lithium primary self-discharge is minuscule btw, the 10 year shelf life that gets bandied about is routinely exceeded geometrically

Yes ! They can can last multiple times that "10 year" average shelf life.
Here are two NOS Dallas bought about 6 months ago, date code 9612 and 9920. Used a dremel and ground just enough to measure the volts when I received them.
Still shows 3.3v tested again today. Yes the one pic is fuzzy, hard to hold two probes and a camera at same time 😀

maxtherabbit wrote on 2021-01-09, 15:30:

a dallas is the literal opposite of a time bomb, when they fail they fail gracefully and do no harm

I beg to differ. They really do no physical harm, that's fine. This is why they're still better than typical barrel batteties.

But if I get it cofrectly they're designed in an utterly stupid way that they can not be powered from outside - ie through a powered on motherboard - so when the built-in battery fails they render the motherboard unworkable, while a removed (I mean in time) barrel battery leaves it in a completely working state apart from forgetting the BIOS settings (and the date and time) when turned off which is but a minor nuisance.

Also these RTC chips won't be produced forever. Okay, these boards can die in the upcoming decades of many reasons but it's better not to add another one.

alvaro84 wrote on 2021-01-09, 18:50:

But if I get it cofrectly they're designed in an utterly stupid way that they can not be powered from outside - ie through a powered on motherboard - so when the built-in battery fails they render the motherboard unworkable

Generally, no - they just forget their settings. What usually happens is someone discovers their Dallas is dead and removes it, then can't figure out why the board won't boot (it has no RTC/CMOS).

jmarsh wrote on 2021-01-09, 19:01:

Generally, no - they just forget their settings. What usually happens is someone discovers their Dallas is dead and removes it, then can't figure out why the board won't boot (it has no RTC/CMOS).

Then I really hope that this is what happened to my Shuttle 486. I routinely go through BIOS settings on that (and many other) board without problem.

If so, I'll be more tolerant towards Dallas chips.

27 years is remarkable.

I have similar experience with my Iwill 430FX board, if you remember the topic about acync cache.

Batman is one of these board where it doesn't even boot without working cmos.

Really? All I had to do is to reset everything via jumper and the board was functional, although refused to load into BIOS setup.

Generally, no - they just forget their settings

RTC motherboards more often than not will refuse to hold any settings even while the system is working. But humorously enough can keep date and current time for some period.

jmarsh wrote on 2021-01-09, 12:59:

There's a register in the nvram that controls whether the internal oscillator runs.
It ships from the factory with it disabled to preserve the battery.
It is possible to disable it via software, which will prevent the unit from keeping time but prolong the battery life during long-term storage.

Yes, and you can turn it off easily.
It's like parking a hard disk.
Conserving Dallas RTC chips