Barrel Battery Replacement Options

[Chuck(G)]

That's the point with the AA alkalines--you can stick them in a bag off the board, so that if they do leak, it's not a big issue.

I wouldn't try to use a rechargeable lithium coin cell in place of the NiCd one--the charging characteristics are wildly different. You might have some surprises.

The reason for using primary cells in place of the rechargeable is that there's no real point to the rechargeables--the primary will last its shelf life essentially, which may be greater than the lifetime of a rechargeable. A diode (or enhancement-type MOSFET if you want to get fancy) is cheap and durable. If you're stingy, you can usually scavenge a few off some piece of junked gear.

 

[Dwight Elvey]

That's the point with the AA alkalines--you can stick them in a bag off the board, so that if they do leak, it's not a big issue.

I wouldn't try to use a rechargeable lithium coin cell in place of the NiCd one--the charging characteristics are wildly different. You might have some surprises.

The reason for using primary cells in place of the rechargeable is that there's no real point to the rechargeables--the primary will last its shelf life essentially, which may be greater than the lifetime of a rechargeable. A diode (or enhancement-type MOSFET if you want to get fancy) is cheap and durable. If you're stingy, you can usually scavenge a few off some piece of junked gear.

Do use single strand insulated wire to connect them and place them at the bottom of the case. The nasty electrolyte will wick right up stranded wire. Don't ask me how I know. It will eventually go up a insulated solid strand wire but that will take decades.
Dwight

 

[ngtwolf]

That's the point with the AA alkalines--you can stick them in a bag off the board, so that if they do leak, it's not a big issue.

Yeah, so the NiMh phone battery i used cost about $5. Its definitely more than just using some batteries and a diode, but a little more of a drop-in replacement (likely get away without even using a soldering iron)... at the pricepoint of batteries, it probably doesn't matter what option you go with, just a matter of effort and what you're most comfortable with.

Regular lithium coin cell do eventually leak but don't cause as much damage as the alkaline cells do. I know this because one leaked after 17 years in my Suburban's key fob.

Good to know, i've never had one last that long.

 

[Stone]

The other things i've done in this situation is to either not bother replacing... I have a few machines i only use very rarely, so it's no small issue to just put the date in when you boot up. I have to with my 8 bit machines anyway.

If any of these machines are online they can get the time set automatically at boot up or whenever you want them to via NIST or any other time server.

 

[Svenska]

If any of these machines are online they can get the time set automatically at boot up or whenever you want them to via NIST or any other time server.

Keep in mind that a wrongly set system clock will prevent any encrypted network connection from working. That includes encrypted NTP connections - which caused me a major headache a few years ago.

 

[Chuck(G)]

Yabut--the problem is that the RTC chip also stores configuration data.

I never understood that. NVRAM/EEPROM chips are inexpensive and will hold configuration data for eons and need no power to hold their data. We used the little 8 pin Xicor X2444 for our stuff.

On the 5170, I can see the use of a persistent across-boot flags when switching from 286 PM to real mode, but 386 and later chips don't need that feature.

 

[maxtherabbit]

I've never seen a lithium primary leak in my life. I'm sure it is possible but it's exceedingly rare

I actually found a 6V lithium primary designed for RTCs in a junk pile the other week dated mid-90s and it's still reading over 7V no load. I stuck it in a system

 

[tonata]

I am not an electronics guy, but it seems that 1N5817 has a lower voltage drop. But I saw a comment that lower voltage drop goes hand-in-glove with higher reverse leakage. So is 1N5817 actually better than for example 1N4148 diode?

 

[3lectr1c]

I used to use the 3x or 4x AA or AAA packs, but I typically use a 1/2AA 3.6V cell in a holder nowadays. I've found the hermetically sealed Tadiran lithium thionyl chloride cells you can get in 1/2AA basically don't leak unless mechanically damaged or placed in really damp conditions where they can rust from the outside. The really long shelf life is nice, since the current draw is so low anyway.

Just going to add this in, any one of these Lithium Thionyl Chloride barrel batteries can and will leak, and when they do, it's mass destruction. Us vintage mac folks call them battery bombs. Seems they start to blow around 20 years, and will just spew corrosive crap all over the board it's installed on. I've got this outdated page on my website that I need to update about the topic: https://macdat.net/important_issues/remove_your_batteries.html
Updates that I need to add are that I've now heard accounts of every single brand mentioned leaking, though the Varta ones seem to be the least prone to it. Point is that if it's over 15 years old, you need to get it out. They have in general quite varying characteristics. Of the brands mentioned on that page, Maxells leak the most, enough that people coined the term "Maxell Bombed" before the other brands started to blow too. Tadirans are pretty decent but many, many accounts of the 30 year old ones leaking, while others still hold charge. The Varta ones found on early Macs very rarely leak (but can), and many still hold charge, then the Saft batteries have started to blow quite often nowadays, probably the second worst.
So in general, not a bad choice, as long as you buy new ones. To that guy who's using a 1990 Tadiran that still works in their 486, please do yourself a favor and remove it!

Oh, and the block batteries used in some early PCs (like the 5170) can leak too. The Tadiran in my admittedly quite poorly stored 5170 nuked the board

 

[Chuck(G)]

I've got a bucketload of the 2/3 C Lithium batteries (3.0V). Haven't figured out a use for them, or even if should bother. These used to be known as "camera batteries".
I suspect that a rechargeable Li-ion cell is less prone to leakage.

 

[maxtherabbit]

I've never seen a lithium primary leak in my life. I'm sure it is possible but it's exceedingly rare

I actually found a 6V lithium primary designed for RTCs in a junk pile the other week dated mid-90s and it's still reading over 7V no load. I stuck it in a system

Update it didn't even last a year. Still never leaked though

 

[tonata]

I used the external connector with a plastic 2 coin battery holder. I put the 2 coin batteries in serial with average voltage of 5 volts. I have used a long cable so that the batteries are away from the motherboard. No diode is needed as there is no charging. For practical reasons this looks like the best solution to me.
The other one is a 14250 non rechargeable battery with a diode on the internal connector with a battery holder again away from the motherboard. Still I am not sure which diode is the best for the job.

 

[Chuck(G)]

For those applications using a 3.6V rechargeable NiCD cell, I wonder if substituting a cheap 600 maH Li-ion might not be an optimal solution. The charge rate on the old nicads is very low, so it's unlikely to upset an Li-ion cell. Self-discharge is about 1%/month on the average, so you'll go several years without use before fully discharging.
650 maH cells are available as junk--used in vaping devices and discarded when used. I've never seen a leak from one of these little batteries.

Some food for thought.

 

[3lectr1c]

I'd worry quite a lot about substituting a Li-Ion battery in a circuit designed to charge any other type - those guys will not hesitate to explode (with fire) when upset.

 

[Chuck(G)]

Yabut the charging current is miniscule, as is the drain. If you're really uncomfortable, put a resistor in series to limit inbound current--it probably won't affect anything outbound.

 

[GiGaBiTe]

It'd really need to be a case by case basis. Some motherboards properly implemented NiCD charging, while others just slapped the battery directly across the +5v rail. The latter usually used 4.8v (4 x 1.2v button cells in series) which mostly worked, the NiCD batteries just burned the excess off as heat.

A rechargeable lithium battery would work in the first case just fine, because the charge voltage is under the working voltage of the lithium battery. In the latter case, you'd need a charge protection board, which aren't hard to source. There are 1S charge protection boards all over ebay for cheap. The problem then becomes, what is the time from the charge protection board losing power to it feeding power back out its terminals. If it takes too long, the CMOS data will be lost.

 

[Chuck(G)]

Charge protection boards are so cheap that they sell them by the panel of 4.

 

[tonata]

Charge protection boards are so cheap that they sell them by the panel of 4.

Can you give an example?

 

[Chuck(G)]

Sure! How's this for cheap?

 

[tonata]

Sure! How's this for cheap?

So the idea is that you connect this chip between the internal battery (socket) and a 3.7V lithium battery? This is the case where the 386 (for example) does charge the battery.

But then will the battery charge the CMOS memory? Would it work in this direction?