A bit of feedback, nearly 2 years on. I replaced the bupbat on four of my N900s and it worked fine for a while, but I noticed a few months ago that the RTC time is lost again once the main battery is taken out for more than a few seconds. That would suggest that the bupbat has ran out and is not being recharged. Wasn't there supposed to be a fix in the power kernel for that?

hmm, the charging *should* be enabled per default, afaik. You might want to check the registers, and/or probe voltage on bupbat during and after inserting main battery and attaching wallcharger
maybe useful: http://www.ti.com/lit/an/swca025/swca025.pdf
actually seems it's not on per default: http://www.datasheetarchive.com/SWCU050G-datasheet.html

[ SWCU050G – April 2008 – Revised December 2010 p438 "7.4.2.1 Backup Battery Charger"]
I would have expected this getting done by NOLO

OK, I made a few tests. Please note that I have the package, whatever it is called, that puts up the "enter date" dialog on boot up removed. My phone boots up straight to desktop every time.

Test 1: Remove main battery for 5 minutes. Put it back, power up. System tme is preserved.
Test 2: Remove main battery for 2 hours. Put it back, power up. System time is reset.

So it looks like the bupbat keeps some charge but not for very long.

which bupbat you are using? The polyacene capacitor type is not supposed to provide nearly as long a backup as the original LiMg cell. Anyway I'll try to find the data and do the math how long actually

to start with:


PAS414 CAP PAS COIN-TYPE 3.3V 0.07F
0.07 As/V * 3 V * 3 V = 0.63 VAs; aka Ws aka J (actually * 0.5 since voltage drops linear with discharge)
0.63 Ws / 7.2E-6 W = 87.5E3 s = 24 h (* 0.5; assuming the capacitor could power the VBRTC domain until voltage = 0, so actually probably another time half of that, ~6h in the end)

Anyway BB_CFG register reset default is 0b00000000, for polyacene cap PAS414 (!!!) I suggest setting it to 0b00011111 as soon as possible during boot

I am using the ones from you. If this is as expected then fine. I think my test proves that the bupbat does get recharged, otherwise the time would be lost right away and not after two hours. I was under a misconception that an RTC backup battery should last for days, if not years. We used some RTC chips back in mid-1990's, can't remember the type number, that had the battery built in. So clearly not vety big. IIRC, the catalogue page claimed the battery should last 10 years.

Infamous DALLAS chip, had a Lithium battery which had quite a few more Ws than the polyacene. Probably something like 100mAh * 3V = 18Ws. And the clock in DALLAS was less "greedy" than the TPS65950

Looking for new bup, Tayo Yuiden part is out of production. Could be ELNA DSK-414-HL a suitable replacement?

yes, looks like a perfect match, though I didn't check the exact specs (millimeters etc), anyway generally it should just work as long as it fits.
Thanks for the pointer
/j

The N900 has a clock backup battery? All these years of ignorance I assumed it was just s lost technology and reseting date/time when changing battery wss a feature. Thanks to this recent activity in this thread I'll see if I can find that potential replacement.

The things one learns every day here!

And digikey stocks the Elna part. Makes it easy to get just one.

It's probably realted to broken firmware in mine case, but does Nokia N9 got similar RTC battery?