PDA

View Full Version : RTC CMOS doesn't retain date/time



toolfrolic
June 11th, 2013, 08:44 PM
My Pentium S 120 Mhz clone has a DS12B887 RTC.
I cut away the plastic at pins 16/20 and measured 3.01 V with a digital multimeter.

When power up: enter monitor, update date/time, Save & Exit, switch off, wait, switch on - date/time wrong.

When power up to Windows 95: set date/time, shut down, wait, switch on - date time wrong.

Any suggestions ?

Chuck(G)
June 11th, 2013, 10:44 PM
Well, no good news, I think. The 12887 isn't very complicated--a battery and crystal and a DIP with a bunch of epoxy, mostly.

I think that you can still get 12887s directly from Maxim:

http://www.maximintegrated.com/datasheet/index.mvp/id/2680/t/or#f

High_Treason
June 12th, 2013, 12:13 AM
Also available on eBay between 3-6, at least in the UK. You could probably just order one.

Alternatively, look for an external battery connector on the motherboard, though sometimes these simply don't work.

If you go the route of replacing the package, be prepared to use a soldering iron and order a chip socket too. This way, you only have to desolder the old chip (doesn't matter so much if you ruin it in the process as it is broken already) and sloder the socket in before placing the new one into the socket, this way, no soldering would be required if you had to replace it again in the future.

Edit: If you don't mind it being messy, you could get a diode and two AA batteries then solder them to the appropriate pins on the chip. Be sure to check the pinouts first.

toolfrolic
June 12th, 2013, 02:38 AM
Thanks guys for the comments.
The RTC has a socket, the plastic removal was done on the bench with a high-speed cutter.
Pins 16/20 were identified in several links as being the internal battery connections.
A battery is replaced by opening the package at 16/20, cutting one lead to the internal battery and soldering on a replacement battery.
Commonsense says to measure the existing battery voltage when the pins are exposed before going to the cut/solder stage.
With 3.01 V measured, I didn't go any further.
Tomorrow, I think I will replace the battery anyway as I have one available.

I am thinking, however, that there must be something else wrong.
Can software screw up setting date/time ?

The monitor keys, for example, work backwards - Page Up decreases the value when adjusting the date/time.
Then there is a faulty config.sys file that I simply "REM" out to get Windows 95 up.

If the battery replacement doesn't help, I will get a replacement chip.
I have an IP professional in the wider family that has probably got some 'old' computers I can access.
John.

Chuck(G)
June 12th, 2013, 08:18 AM
Of course, any software can change the clock setting. But the BIOS should be the last word, as no software has a chance to run first. You can always DOS from a floppy to get around the XP boot to see what's happening.

toolfrolic
June 30th, 2013, 03:30 AM
Just an update on this RTC.
I have been through several cycles of re-building this system with W95 involving re-assigning hard drives.
In each case the HDD settings in the CMOS have to be adjusted to get the machine to run.
In every case, setting the date/time in the CMOS is not retained through a BIOS Reset.

Very peculiar.
The BIOS/CMOS connection does work.
The CMOS editing does work.
But something with the Date/Time does not work.
John.