AST Card Not Ticking When Off

[AgentOrange96]

I have an AST Combo Plus card in my IBM PC XT. While most features of the on-board RTC seem to work fine, it only seems to advance the time while the system is on. So, for example, if I have the time set properly, and I turn the computer off at 4:30 PM, and then I wait half an hour and turn the computer back on, the clock will be set to 4:30 PM, and not to 5:00 PM. This defeats the purpose of the RTC. Otherwise, it works fine. I can use ASTCLOCK.COM /R and then set the time and date and it will save it. It will retrieve the time and date when I start it up (although it'll be the time it thought it was when I shut it down last.) and it will advance time while the system is on. I'm really not sure what would cause this. The battery is only a few months old. I have been away for a while so I don't know if this was a problem when I first put it in or not. If anyone has any ideas, I'd obviously love to hear. Thanks!

 

[3pcedev]

Batteries can sit in a warehouse for a long time; even though you only had it for a few months it may be years old.

I'd start by removing the battery and checking it's voltage with a multimeter. If it's <3V then it's essentially flat.

If the battery checks out OK then I would focus on the crystal oscillator which makes the clock 'tick'. The oscillator circuit may lack continuity with the battery (i.e. the battery is holding up the volatile memory that holds the date/time, however the oscillator circuit is not powered). Check for corrosion around the battery area or obvious signs of damage. Without a board in front of me I can't be more specific; but that should get you started.

 

[Stone]

Sure sounds like a battery issue.

 

[AgentOrange96]

The battery was just above 3V. (3.03V) When in the board though, it was nowhere near 3V. As well, I had previously checked both crystals I found on the board with a scope and didn't see either oscillating. I have three other batteries (the same age) and I tried putting another in. I checked with a multimeter and saw a good voltage well above 3V while in the board. As well I was able to see an oscillation on the smaller of the two crystals. I just booted it, set the time, and then shut it off. A few minutes later and I booted it and lo and behold it had the correct time!

I had figured a low battery would have caused it not to hold time at all, so that's definitely an interesting quark, and one that I'm glad is now documented. (In case anyone else ever happens across a similar issue.) I didn't expect the battery to last that short a time though. I'd replaced it back in August, so that's a bit disappointing. I suppose I should probably pull this one before leaving for school again in mid-January since it will be sitting until May.

 

[modem7]

I had figured a low battery would have caused it not to hold time at all, so that's definitely an interesting quark, and one that I'm glad is now documented. (In case anyone else ever happens across a similar issue.)

The behavior had been seen before, which is why I presume a low battery was suggested in posts #2 and #3.

For example, it is known to occur in the IBM AT (see the 'Loss of time' section at [here]).

As the battery voltage drops:

Phase #1: During computer off time, oscillator (battery powered) feeding RTC chip experiences intermittent operation.
Phase #2: During computer off time, oscillator (battery powered) feeding RTC chip now does not operate at all. [RTC date/time at computer power-on is date/time at last computer power-off]
Phase #3: Battery voltage now no longer good enough for the RTC chip to hold its RAM contents.

 

[Stone]

I had figured a low battery would have caused it not to hold time at all...

No need to figure... if you just pull the battery and set the time with the machine on and note that it's keeping the time correctly you'll then know that the the battery's charge (or lack thereof) doesn't affect the timekeeping with the machine on.

 

[modem7]

BTW.

Many years ago, the place where I worked bought lots of IBM PS/2's, and we used to see the 'date/time is as when the PS/2 was powered off' symptom in some PS/2 model 50's. But, a different cause. The answer was to ensure that the DASDDRVR.SYS patch was installed.

 

[AgentOrange96]

The behavior had been seen before, which is why I presume a low battery was suggested in posts #2 and #3.

For example, it is known to occur in the IBM AT (see the 'Loss of time' section at [here]).

As the battery voltage drops:

Phase #1: During computer off time, oscillator (battery powered) feeding RTC chip experiences intermittent operation.
Phase #2: During computer off time, oscillator (battery powered) feeding RTC chip now does not operate at all. [RTC date/time at computer power-on is date/time at last computer power-off]
Phase #3: Battery voltage now no longer good enough for the RTC chip to hold its RAM contents.

Ah that's interesting. Those were exactly the symptoms!

No need to figure... if you just pull the battery and set the time with the machine on and note that it's keeping the time correctly you'll then know that the the battery's charge (or lack thereof) doesn't affect the timekeeping with the machine on.

I know the machine powers keeps time when on. (It'd do that even without a RTC installed.) I was just surprised that keeping the number in RAM took less power than keeping the clock going at all.

BTW.

Many years ago, the place where I worked bought lots of IBM PS/2's, and we used to see the 'date/time is as when the PS/2 was powered off' symptom in some PS/2 model 50's. But, a different cause. The answer was to ensure that the DASDDRVR.SYS patch was installed.

That's interesting Indeed!

By the way, thank you all for your help. I know in retrospect it seems silly that I didn't just check the battery to begin with, but I really do appreciate the help! :D

 

[NeXT]

Measure the voltage from the solder points for the battery on the PCB.
I have run into the problem as recently as two weeks ago where poor contact between the holder and the battery (from dirt, corrosion or a top clip that was bent out too far) looked physically fitted properly but no power was going to the board.