If you set the time using the dos time command, and then leave it powered off for a while, when you power it back on, does it keep time accurately, freeze the time, or lose the time entirely?

I wouldn't trust an original 12887 to still have a good battery. I bought several more recent 12887+ for when I need replacements. No idea really what happens when a 12887 is not connected to power for over a decade. I don't know how to decode the date on them, so no idea how long that had sat in a warehouse, but I would hazard a guess, based on similar codes on my 12887+ that they are 2002 week 25, so they had no external power for 18 years. Not sure what happens when the battery in them dies, but it may be destroyed internally.

jakethompson1 wrote on 2020-09-05, 22:51:

If you set the time using the dos time command, and then leave it powered off for a while, when you power it back on, does it keep time accurately, freeze the time, or lose the time entirely?

I set the time and date. Unplugged the machine for 20 minutes. And the time was reset back to 1980. This is with the glitchworks version.

debs3759 wrote on 2020-09-05, 23:52:

I wouldn't trust an original 12887 to still have a good battery. I bought several more recent 12887+ for when I need replacements. No idea really what happens when a 12887 is not connected to power for over a decade. I don't know how to decode the date on them, so no idea how long that had sat in a warehouse, but I would hazard a guess, based on similar codes on my 12887+ that they are 2002 week 25, so they had no external power for 18 years. Not sure what happens when the battery in them dies, but it may be destroyed internally.

Did you read the whole post? I tried a modern glitchworks solution, too.

kingcake wrote on 2020-09-05, 23:59:

I set the time and date. Unplugged the machine for 20 minutes. And the time was reset back to 1980. This is with the glitchworks […]
Show full quote

I set the time and date. Unplugged the machine for 20 minutes. And the time was reset back to 1980. This is with the glitchworks version.

If you read the Oscillator Control Bits section of the DS12887 datasheet https://datasheets.maximintegrated.com/en/ds/ … 5-DS12C887A.pdf you can see that the Dallas chips supposedly shipped with the oscillator turned off so the battery doesn't drain.

I've seen others claim they had to turn those bits on. But, plenty of people left reviews on the Glitchworks one saying nothing about it. Maybe it depends on whether the BIOS automatically enables it or not?

It might be worth reaching out to Glitchworks and asking if there is anything you may have to do to enable it re: oscillator control bits, or if the chip is expected to work immediately when plugged in.

debs3759 wrote on 2020-09-05, 23:52:

I wouldn't trust an original 12887 to still have a good battery. I bought several more recent 12887+ for when I need replacements. No idea really what happens when a 12887 is not connected to power for over a decade. I don't know how to decode the date on them, so no idea how long that had sat in a warehouse, but I would hazard a guess, based on similar codes on my 12887+ that they are 2002 week 25, so they had no external power for 18 years. Not sure what happens when the battery in them dies, but it may be destroyed internally.

Yep the 0225XX is 2002 25th week, the xx is some type of batch code afaik. I just bought some DS12C887A dated 2016 week 20, and have some 1999 DS12887A that still have 3.2v (ground the side to measure). One thing is that some boards have to have the exact model DS and (a 12C887 will quirk when original was 12887 plain) due to this: DS12885, DS12887, DS12887A: 114 Bytes of General-Purpose Battery-BackedRAM, 113 Bytes in the DS12C887 and DS12C887A
If the BIOS expects 114Bytes CMOS RAM and sees only 113Bytes OR visa-versa: it can cause an issue, it is rare but does exist.
I do not know about Glitchworks but assume they make a good product and the issue is the battery on it....

jakethompson1 wrote on 2020-09-06, 00:13:

If you read the Oscillator Control Bits section of the DS12887 datasheet https://datasheets.maximintegrated.com/en/ds/ … 5-DS12C […]
Show full quote

kingcake wrote on 2020-09-05, 23:59:

I set the time and date. Unplugged the machine for 20 minutes. And the time was reset back to 1980. This is with the glitchworks […]
Show full quote

I set the time and date. Unplugged the machine for 20 minutes. And the time was reset back to 1980. This is with the glitchworks version.

If you read the Oscillator Control Bits section of the DS12887 datasheet https://datasheets.maximintegrated.com/en/ds/ … 5-DS12C887A.pdf you can see that the Dallas chips supposedly shipped with the oscillator turned off so the battery doesn't drain.

I've seen others claim they had to turn those bits on. But, plenty of people left reviews on the Glitchworks one saying nothing about it. Maybe it depends on whether the BIOS automatically enables it or not?

It might be worth reaching out to Glitchworks and asking if there is anything you may have to do to enable it re: oscillator control bits, or if the chip is expected to work immediately when plugged in.

I found code examples for how to turn off the oscillator (Conserving Dallas RTC chips) but none for how to enable it

kingcake wrote on 2020-09-06, 00:55:

jakethompson1 wrote on 2020-09-06, 00:13:

If you read the Oscillator Control Bits section of the DS12887 datasheet https://datasheets.maximintegrated.com/en/ds/ … 5-DS12C […]
Show full quote

kingcake wrote on 2020-09-05, 23:59:

I set the time and date. Unplugged the machine for 20 minutes. And the time was reset back to 1980. This is with the glitchworks […]
Show full quote

I set the time and date. Unplugged the machine for 20 minutes. And the time was reset back to 1980. This is with the glitchworks version.

If you read the Oscillator Control Bits section of the DS12887 datasheet https://datasheets.maximintegrated.com/en/ds/ … 5-DS12C887A.pdf you can see that the Dallas chips supposedly shipped with the oscillator turned off so the battery doesn't drain.

I've seen others claim they had to turn those bits on. But, plenty of people left reviews on the Glitchworks one saying nothing about it. Maybe it depends on whether the BIOS automatically enables it or not?

It might be worth reaching out to Glitchworks and asking if there is anything you may have to do to enable it re: oscillator control bits, or if the chip is expected to work immediately when plugged in.

I found code examples for how to turn off the oscillator (Conserving Dallas RTC chips) but none for how to enable it

I tried helping someone enable it another time, but in their case it was already enabled. YMMV:

Run debug. At the - prompt, type A and hit enter (assemble)

You will get a xxxx:0100 prompt. Enter the following instructions pressing enter after each one. Be careful with lowercase l vs. number 1.

cli
mov al,8a
out 70,al
in al,71
mov bl,al
mov al,8a
out 70,al
mov al,bl
and al,8f
or al,20
out 71,al
sti
int3

After the last instruction int3, leave a blank line at :0117, and you should get the - prompt back.

Type G (for go) and enter. Could you please report back when AX= and BX= values you get back as that will tell us if it's indeed the problem.

After that, type Q (for quit) and enter. For good measure, power off your machine entirely. Then try setting the date and bios settings. Save, power off entirely again, and see if it works.

jakethompson1 wrote on 2020-09-06, 00:58:

I tried helping someone enable it another time, but in their case it was already enabled. YMMV: […]
Show full quote

kingcake wrote on 2020-09-06, 00:55:

jakethompson1 wrote on 2020-09-06, 00:13:

If you read the Oscillator Control Bits section of the DS12887 datasheet https://datasheets.maximintegrated.com/en/ds/ … 5-DS12C887A.pdf you can see that the Dallas chips supposedly shipped with the oscillator turned off so the battery doesn't drain.

I've seen others claim they had to turn those bits on. But, plenty of people left reviews on the Glitchworks one saying nothing about it. Maybe it depends on whether the BIOS automatically enables it or not?

It might be worth reaching out to Glitchworks and asking if there is anything you may have to do to enable it re: oscillator control bits, or if the chip is expected to work immediately when plugged in.

I found code examples for how to turn off the oscillator (Conserving Dallas RTC chips) but none for how to enable it

I tried helping someone enable it another time, but in their case it was already enabled. YMMV:

Run debug. At the - prompt, type A and hit enter (assemble)

You will get a xxxx:0100 prompt. Enter the following instructions pressing enter after each one. Be careful with lowercase l vs. number 1.

cli
mov al,8a
out 70,al
in al,71
mov bl,al
mov al,8a
out 70,al
mov al,bl
and al,8f
or al,20
out 71,al
sti
int3

After the last instruction int3, leave a blank line at :0117, and you should get the - prompt back.

Type G (for go) and enter. Could you please report back when AX= and BX= values you get back as that will tell us if it's indeed the problem.

After that, type Q (for quit) and enter. For good measure, power off your machine entirely. Then try setting the date and bios settings. Save, power off entirely again, and see if it works.

I appreciate all the help regardless. I did what you requested. Here's the output:

symptoms remain unchanged

The fact that it read FF makes me think it isn't talking to the new chip at all. Everything goes back to how it was before when you put the original one back in? Got a picture of it? Perhaps you could just do the battery modification to it, if there is some special reason that it works and the others do not.

One interesting tidbit. I put in a diagnostic POST card for the heck of it. And when you press F1 to enter the Bios the card goes nuts.

https://www.youtube.com/watch?v=UfqQI_Gxi98

I wouldn't read too much into that, because writing to the POST card (80h) is one way to introduce a delay. In fact, I didn't put a delay in the asm code I gave you and perhaps I should have. If you want to for the heck of it, you could try with a delay:

cli
mov al,8a
out 70,al
out 80,al
in al,71
out 80,al
mov bl,al
mov al,8a
out 70,al
out 80,al
mov al,bl
and al,8f
or al,20
out 71,al
sti
int3

As a side effect that code should make the number on the post card change to 8a.

kingcake wrote on 2020-09-06, 01:17:

I found code examples for how to turn off the oscillator (Conserving Dallas RTC chips) but none for how to enable it

Do you have the original RTC chip ? If so can you take a picture ? I want to see the original markings.
All the Dallas RTC with built in batteries can be hacked to provide external power if you know how, not sure why you did not try that route.
Sure may not look pretty but: working ugly is better than non-working pretty....

Horun wrote on 2020-09-06, 02:34:

Do you have the original RTC chip ? If so can you take a picture ? I want to see the original markings. All the Dallas RTC with […]
Show full quote

kingcake wrote on 2020-09-06, 01:17:

I found code examples for how to turn off the oscillator (Conserving Dallas RTC chips) but none for how to enable it

Do you have the original RTC chip ? If so can you take a picture ? I want to see the original markings.
All the Dallas RTC with built in batteries can be hacked to provide external power if you know how, not sure why you did not try that route.
Sure may not look pretty but: working ugly is better than non-working pretty....

here's the original

Have you tried setting a year below 2000 in DOS?

Had something resembling this happen on an Intel Hendrix board after replacing the Dallas module, it would lock (with garbage on screen, afaicr) when trying to enter BIOS. Booted DOS normally, TIME and DATE would return some non-sense values (year was 99999 or something like that). After setting the correct date/time, no more problems. Yours doesn't seem to default to abnormal values, but maybe it doesn't accept the year being past 1999.

quicknick wrote on 2020-09-06, 13:03:

Have you tried setting a year below 2000 in DOS?

Had something resembling this happen on an Intel Hendrix board after replacing the Dallas module, it would lock (with garbage on screen, afaicr) when trying to enter BIOS. Booted DOS normally, TIME and DATE would return some non-sense values (year was 99999 or something like that). After setting the correct date/time, no more problems. Yours doesn't seem to default to abnormal values, but maybe it doesn't accept the year being past 1999.

that's a good suggestion, unfortunately it still defaults back to 1980 on reset.

Was this socketed or did you desolder the old chip?

vetz wrote on 2020-09-06, 19:37:

Was this socketed or did you desolder the old chip?

That's a good point. In addition, I wonder what happens if you try to boot with no chip at all?

Well it's fixed.

I measured Vcc and Ground on the RTC and it was reading 3 volts instead of 4.5-5. Power supply rails all measured good. Noticed ground pin of socket seemed crushed/corroded.

So I popped out the socket with my hot air gun and put in a new socket. Boom. It works now.

What's weird is this machine will not post without a RTC installed. Yet it somehow worked good enough to post just not good enough to set date/time. Weird one.