Keatah wrote:

I recently fixed my old machine and got to thinking about why it would have 2 forms of battery backup. […]
Show full quote

I recently fixed my old machine and got to thinking about why it would have 2 forms of battery backup.

It has (had) a 3.6v NiCd on the soldered to the motherboard, and the mobo will trickle-charge it. By calculations it should last 3 or so months with the power off. I thought it would be longer.

It also has provision for an external battery that is not charged by the mobo, there's a diode there. And it checks out. The system came equipped with 4.5v alkaline pack connected by wires and a connector. It can be replaced in a minute. Calculations show it's good for 4 or 5 years.

The batteries are wired such that the alkaline pack will charge the on-board NiCd cell.

---

So why have two batteries going? Why not just go with the external pack and be done with it? What is the reasoning behind doing things this way?

Actually the "second" battery is normally not there, there would be only the connection for it. It is meant to be an extra measure, to be used after the soldered battery died, because, well, it's soldered and not easy to change by an end user.

So, if you already have an external battery connected, then most probably your soldered 3.6 volt battery is already dead. I STRONGLY suggest you to remove the soldered battery, since they have a "reliable" tendency to leak and destroy the motherboard with acidic goo. Actually, I suggest you to remove it even if it is NOT dead.

tayyare wrote:

Keatah wrote:

I recently fixed my old machine and got to thinking about why it would have 2 forms of battery backup. […]
Show full quote

I recently fixed my old machine and got to thinking about why it would have 2 forms of battery backup.

It has (had) a 3.6v NiCd on the soldered to the motherboard, and the mobo will trickle-charge it. By calculations it should last 3 or so months with the power off. I thought it would be longer.

It also has provision for an external battery that is not charged by the mobo, there's a diode there. And it checks out. The system came equipped with 4.5v alkaline pack connected by wires and a connector. It can be replaced in a minute. Calculations show it's good for 4 or 5 years.

The batteries are wired such that the alkaline pack will charge the on-board NiCd cell.

---

So why have two batteries going? Why not just go with the external pack and be done with it? What is the reasoning behind doing things this way?

Actually the "second" battery is normally not there, there would be only the connection for it. It is meant to be an extra measure, to be used after the soldered battery died, because, well, it's soldered and not easy to change by an end user.

So, if you already have an external battery connected, then most probably your soldered 3.6 volt battery is already dead. I STRONGLY suggest you to remove the soldered battery, since they have a "reliable" tendency to leak and destroy the motherboard with acidic goo. Actually, I suggest you to remove it even if it is NOT dead.

But what leaks the NiMh and NiCd isn't Nickel Oxide hydroxyde? if it is what the battery leak, then it's not acidic, its basic,but corrosive.

Edit: Correction: idk if Nickel Oxide Hydroxide is one base, but seems that this compound can be attacked by acid and bases.

Yeah... and you also got the main idea. It corrodes 🤣

Also don't forget that back in the day if you lost your CMOS settings and hadn't written them down, you could likely kiss all your data goodbye on your hard drive. So having the two batteries mean you would be able to switch out the external battery for a new one without losing your settings. Even with a mostly dead onboard battery you'd still likely have a couple of hours of charge from it in which to spend with the machine off changing the external battery.

Yes, you could do it with the power on, but I imagine the manufacturer didn't want inexperienced people poking around inside with the power still on, especially in the days of AT PSUs which generally didn't have short-circuit protection, so if you accidentally shorted something, the PSU would just deliver max current until something else blew up.

Malvineous wrote:

Also don't forget that back in the day if you lost your CMOS settings and hadn't written them down, you could likely kiss all your data goodbye on your hard drive...

Sorry but why is that? For me "back in the day" means early 90s and all you had to do is either taking notes of your hard drive parameters (either the "type" number or just three numbers denoting heads, cylinders and sectors) to somewhere you can find again, or you need to go to any service tech who most probably have the documents on hand to tell them to you (for a fee probably)

I know during that time internet was far from what we know today, devices was not coming all the parameters printed on their labels, and "autodetect HDD" is not a feature that is exisiting yet, but it was not that bad. I never ever had a situation like this according to my personal experiences (helping friends and family) or my experiences as a part time service tech, back in the day.

I had the issue personally in 1997 with a 3.2GB IDE hard drive that my Pentium 100's BIOS couldn't autodetect and required manual entry of all the parameters - head, cylinders, sectors, etc. The docs that came with the drive basically said it didn't matter what you picked and gave some tips, so I picked values that came out with the correct size and everything worked fine. Then one day a year or two later I lost the CMOS settings and I hadn't written them down, and I couldn't remember exactly what I had put in. I remembered the manual saying it didn't matter so I picked some values to get another approximate value, but unbeknown to me at the time, the new values I had chosen resulted in a different CHS -> LBA translation layout, so when I booted the machine all the sectors appeared out of order.

Realising my mistake I somehow managed to fix it (maybe I did write the numbers down but didn't realise how important it was to put them back exactly the same, my memory is a bit hazy now), and the end result was that the second partition on the drive was completely corrupted (beyond repair by Scandisk) but the first partition survived.

I ended up writing a program that would search the disk sector by sector looking for signatures of important files and picking them out, which worked for those files that weren't fragmented so in the end I did manage to recover a large number of files, but I still lost quite a bit of stuff.

But that was my lesson that even in the late 90s with autodetecting hard drives, the CMOS values were still very important if there was any sort of manual data entry at all!