Re: Z-RAM ultra 2... replacing the battery?

[mjmahon@aol.com (Michael J. Mahon)]

Gary replied:

>Michael J. Mahon wrote:
>
>> That's right.  The circuit operates from the battery _only_ when
>> the computer power is off.
>> 
>> That means that the software is choking because the clock
>> currently contains an invalid date/time.  Try running the clock
>> setting software and see if that fixes it for a warm boot (since
>> the battery is still not known to be working).
>
>I had booted up the clock setting disk before, and the setting software
>crashed into the monitor, too. Just now, I realized that there was the off
>chance that the setup disk would actually run the clock driver before it
>ran the setup program... (which, it turns out, it does... sorta stupid
>design there) so I booted up from another disk, then ran the clock setting
>software directly. And it worked! The clock had a date somewhere in the
>year 52. I reset it to the current date and time, then did a warm boot. 

Having the clock reading driver on the disk with the setting program
is not a bad design decision, but having a clock read routine that is
not robust in the situation of a corrupted clock state is bad design.
The driver should always complete, possibly returning a "garbage"
date, regardless of the state of the clock.

>The clock reading software was able to run without crashing, but it read
>back the exact date and time I had set with the clock setting program. 
>After about 20 minutes, rebooting into any of the disks that has the clock
>driver still resulted in the exact time I had set being read. So it looks
>like the clock accepted the data, but won't actually update it.

Agreed--it sounds like either the "hold" signal is not being released
by the setting software (improbable if you told it to start) or the
oscillator is not running.

>Next I tried a quick powercycle to see what would happen. Loading the clock
>driver results in the crash to the monitor again. Running the clock setup
>software also ends up in a crash to the monitor, with the time on the
>screen showing a date in 04, but a different month and random time. Just
>before the crash, the time flips around madly, as if the software is
>reading from a spinning clock.
>
>So... I'd guess I'm looking at both a dead battery and a screwy clock chip?
>A bit of googling turned up a PDF version of the Z-RAM's manual
>(http://tinyurl.com/6xn6l), which identifies the clock chip as an OKI
>6242B. The spec sheet I found:
>
>http://www.okisemi.com/jp/datadocs/doc-eng/msm6242b.pdf
>
>says it has an onboard crystal... so I wouldn't be anle to swap out a
>crystal to see if that's the culprit.

Actually, it says that it has an onboard crystal oscillator--but _not_
the crystal or its loading capacitors.  They are located on the PC board
adjacent to the clock chip.  If you have a delicate soldering hand, you
could try replacing the 32kHz crystal (with heat sinking!).  Better yet,
if you have an oscilloscope with a high-impedance (10x) probe, you
can use it with a 5pf capacitor in series (or just a "gimmick" of insulated
wire wrapped around a crystal lead a couple of times) to each of the leads
of the crystal and see if it is running.  (This is a _very_ high impedance
circuit, so any extra loading may stop the oscillator.)

That actually suggests another possibility worth trying.  Use a Q-tip
with some non-oily alcohol to clean the board around the chip and the
crystal and its loading capacitors.  (One of the loading capacitors is
probably a small trimmer.)  Try not to get alcohol into the trimmer, since
it would take a while to dry.  Follow close behind with a dry Q-tip to
clean up whatever the alcohol may have dissolved.  The purpose of
this exercise is to ensure that the PC board is not contaminated with
a dirt-oil-flux film that might be enough to stop the oscillator.

> I'm not able to discern many other
>relevant details from the sheet (it's been almost 20 years since I flunked
>my last EE class ;-). The datasheet is dated 2002, so I assume the part is
>still made... perhaps I could just buy another one and swap out the
>existing chip (I think everything on the board was socketed, so I wouldn't
>have to tax my limited soldering skills). 

That is a very practical approach, though you may still need to replace
the crystal.

>Thanks for all the help and suggestions so far!

You are to be congratulated on your persistence and the diagnostic
successes you have had!

-michael

Check out parallel computing for 8-bit Apples on my
Home page:  http://members.aol.com/MJMahon/