• Please review our updated Terms and Rules here

Odd date and time issue

Zap!

Experienced Member
Joined
Jun 20, 2021
Messages
115
Location
Staten Island, New York
Have a 386 that keeps track of the time and date in real time, but only in real time while it's running. For example, if I set the time today (December 7th) and shut it off for three days, it'll say December 7th when I turn it on. That was just an example. I hadn't turned my PC on since September 14th, so when I turned it on and typed "date" it said September 14th. However, if I were to keep it running for three days, the clock would run fine for those three days. When I turn the PC off, it stops tracking time.

What is going on here? The battery (on the MoBo) is remembering things, just not in real time. New battery BTW.
 
Normally that would be a symptom of a battery that's starting to go flat. But you said you've put in a new one. I could only suggest to perhaps try another one. If that makes no difference then a hardware fault.
 
Voltage of the battery too low (CMOS and RTC have different needs) or the clock crystal is dead. But most likely the battery. I have that on many systems on which I replaced the 3.6v NiCd with a 3v coin cell + diode.
 
Normally that would be a symptom of a battery that's starting to go flat. But you said you've put in a new one. I could only suggest to perhaps try another one. If that makes no difference then a hardware fault.

Voltage of the battery too low (CMOS and RTC have different needs) or the clock crystal is dead. But most likely the battery. I have that on many systems on which I replaced the 3.6v NiCd with a 3v coin cell + diode.


It had a soldered in battery, but it's been replaced by a battery holder by the previous owner. The terminals can't be backwards, as I'd imagine it wouldn't save at all if they were. It remembers the date and time to the second that it was shut off. I will try a new one though, and if all else fails, a new battery holder.
 
Diodes will drop the voltage a little as well.
Exactly.

The terminals can't be backwards, as I'd imagine it wouldn't save at all if they were.
Did anyone even assume that? You certainly have no issue due to polarity.

A very simple test would be to connect an external 3.6 - 4v power source (e.g. from a bench PSU) to the battery terminals and see if the clock keeps running then when powered off. If not, replace the clock crystal or check if there is a broken trace around it or the rtc chip.
 
A very simple test would be to connect an external 3.6 - 4v power source (e.g. from a bench PSU) to the battery terminals and see if the clock keeps running then when powered off. If not, replace the clock crystal or check if there is a broken trace around it or the rtc chip.
From my understanding of the OP's posts, the RTC clock (as distinct from the DOS clock) is advancing when the 386 is powered on. If the RTC crystal is dead, the RTC clock would never advance at all.
 
What is going on here?
Reinforcing what others have written:

On many 286/386 computers, the symptom has the well known cause of 'low voltage at the CMOS/RTC chip during the period that the computer is powered off'. Even the IBM AT suffers from the symptom.

When the motherboard is powered on, the motherboard powers the CMOS/RTC chip and oscillator. When the motherboard is powered off, the battery powers the CMOS/RTC chip and oscillator. The oscillator (the thing that 'ticks' the RTC) is more sensitive to low voltage than the CMOS/RTC chip.

Initially, what is noticed is some loss of some time during the period that the computer is off. As the battery voltage drops further, the amount of time loss gets greater. When the voltage drops below a certain point, the RTC does not advance at all when the computer is off.

See the 'Loss of time' section of [here].

(In later motherboards, the functionalities of CMOS/RTC chip and RTC oscillator are in one of the chipset chips, typically an 82C206.)

It had a soldered in battery, but it's been replaced by a battery holder by the previous owner.
If the 'soldered in battery' was a rechargeable battery, and the previous owner intended for a non-rechargeable battery (e.g. lithium) to go into the battery holder, then in addition to the battery holder, there should have been be a diode added to stop the battery charging circuitry on the motherboard from attempting to charge the non-rechargeable battery. This is discussed in the 'TYPE: Motherboard designed for internal battery only' section of [here].

That is the 'diode' that others referred to.
 
We are missing a litle bit of crucial information. What kind of battery is now being used?

I have 2 386 computers, and as usual the batteries went flat and started to leak. The battery was the notorious blue 'barrel' battery, which is soldered in. So I removed this battery, and purchased a new 3.6 volt battery and a holder, which was stuck to the inside of the case with double-sided tape. The new battery should not be charged, so a diode went in series. This reduced the voltage to 2.9 volts, but everything worked.

If instead a button battery was used, its terminal voltage is only 3.0 volts, and a diode will bring it down to 2.3 volts, which could make all the difference.
 
Thanks guys, I finally got around to getting a battery pack (uses three AAA's). I was using a CR2032 inserted into a holder. So far, so good!
 
Back
Top