Correction on the PCChips (ILON, Hedaka, probably Amptron, and 'Triple D Technology' going by the BIOS that came with my unmarked M-205), the CMOS circuit doesn't get powered at all if the charging jumper is left open (and the residual capacitive charge doesn't power it either), so to use a non-rechargeable cell, you need to connect a diode in series where the jumper normally goes (on the center two pins of the ext battery header).

You could probably make a handy plug-in diode jumper for such boards, but I haven't done that yet.

I'm yet more confused by the behavior of some BIOS settings, though. The date/time gets wiped with CMOS cleared, and a CMOS checksum error shows at POST, but not saved hard disk parameters, foppy disk configurations, and some RAM settings. (shadow RAM settings get reset, but base/EMS/XMS memory configuration stays)

The AMI BIOS used does explicitly mention EEPROM functionality, so maybe it saves some parameters in ROM instead, but that's unusual, moreso since the later 386SX PCChips board doesn't do that.

Hi,

I am opting for this solution for my 386 Commodore motherboard:
- 3.7 V lithium battery rechargeable 2500 (it is bigger than 2AA)
- mountable battery holder for 18650 (it has holes for mounting somewhere in the case) from Ebay or Aliexpress

The battery will be mounted far from the motherboard and connected where the old 3.6 barrel battery was. The motherboard does charge the battery when the computer is on.

There are a few small details though:
- as it is a new battery the output is actually 3.9 volts, but it should not be a problem I think ...
- it will be charged by 5 V when the computer is ON. I am not sure if there are some specific requirements for charging such a battery ?

What do you think?

Is it just me, or do some motherboards drain those CR2032 batteries substantially faster than other boards? It seems like I have a few boards in which the battery only lasts a year or two, whereas others last 10 years. Was there that much variance in the circuit (current draw) to hold the CMOS/clock settings?

feipoa wrote on 2021-06-09, 20:17:

Is it just me, or do some motherboards drain those CR2032 batteries substantially faster than other boards? It seems like I have a few boards in which the battery only lasts a year or two, whereas others last 10 years. Was there that much variance in the circuit (current draw) to hold the CMOS/clock settings?

Yep ! Not just you. Some of the cmos draw 20uA or more where others only need about 5uA to stay alive from what little research done, the older ones required more. Is very chipset dependent and Dallas seemed to get it right with very low draw in their old RTC which is why they would last 10+ years compared to some early 486/586 (with embedded CMOS) using coin cell that some barely lasted a few years. Just my observation.

feipoa wrote on 2021-06-09, 20:17:

Is it just me, or do some motherboards drain those CR2032 batteries substantially faster than other boards? It seems like I have a few boards in which the battery only lasts a year or two, whereas others last 10 years. Was there that much variance in the circuit (current draw) to hold the CMOS/clock settings?

Most of them are higher power draw components than modern counterparts that realistically require 3.6V instead of 3V. Also, the capacity in CR2032s is quite lower than the barrel batteries they replace. Add it all up, you get this result. I find that external 3xAA or 3xAAA power packs tend to last a lot longer.

For me the issue is that I have numerous boxes of motherboards and if I go to pull out a motherboard for test and it has been sitting in the bin for 3 years, I want the CMOS settings to be preserved. I don't want to figure out all the most optimal settings that may have taken me a few hours to figure out, especially on these pesky 386/486 boards. I also don't want to create a settings sheet for each motherboard because those tend to get lost. Since I didn't a barrel battery leaking, I replaced them all with CR2032 holders and cut the charging lead. It takes up quite a bit more storage space in the bins and would have parts dangling around if I were to use those AA battery holders.

It would be more ideal to have a battery replacement option for these boards which was still small, but lasted a lot longer. Has anyone looked into different coin holder/battery combinations with more amp-hours, yet was still low profile enough not to interfere with expansion cards? Perhaps if there was a battery that was 4x as tall as the CR2032 and had 6x the capacity. Or even a CR2032 holder which held 4 CR2032's in parallel, not series.

Can you please comment on my solution on using a 18650 lithium ion rechargeable battery. Can these batteries be safely recharged by a 386 motherboard while in use?

Here is a 18650 https://fr.aliexpress.com/item/1005002420574151.html charger if a comparison can be made. Or maybe I should put a diode to be on the safe side reducing to it 3.3 V ?

I am not expert, but I am affraid, that NiCd batteries have a different charging strategy than the newer Li-XXX batteries. These require regulatory and control circuits, that the motherboard does not have.

tonata wrote on 2021-06-10, 08:02:

Can you please comment on my solution on using a 18650 lithium ion rechargeable battery. Can these batteries be safely recharged by a 386 motherboard while in use?

Here is a 18650 https://fr.aliexpress.com/item/1005002420574151.html charger if a comparison can be made. Or maybe I should put a diode to be on the safe side reducing to it 3.3 V ?

No, a rechargeable Li Ion battery's charging characteristics are completely different from the NiCd it replaces, this is a recipe for disaster. If you want to play safe use NiMH.

feipoa wrote on 2021-06-09, 20:17:

Is it just me, or do some motherboards drain those CR2032 batteries substantially faster than other boards? It seems like I have a few boards in which the battery only lasts a year or two, whereas others last 10 years. Was there that much variance in the circuit (current draw) to hold the CMOS/clock settings?

ATX machines, if left plugged in while "off", have the capability of being able to maintain date and time and settings off the 5V standby supply, so in general, batteries in ATX should last longer than batteries in AT systems. That's not to say that they all take advantage of that.

Then you've got boards designed for the rechargables, where the board designer might have reasonably thought the machine would be used somewhat regularly, so he didn't take very much care on how many milliamps the clock/settings ate. Design from the beginning to use lithium cells may have taken more care.

The Dallas clock chips are meant to consume in the neighbourhood of a microwatt... if all the capacitances on the board store something like 100microwatt/hours, then you get 100 hours for free, this could go by a factor of 10 in either direction, according to how much capacitance there is, and whether it leaks through other sources, or remains while board is off. This would be why flipping the battery out and back in briefly without shorting the contacts or jumper provided, may not clear settings. Such a board you might only have to turn on twice a week to avoid it using the battery at all.

Then of course there's heat, it accelerates chemical reactions... including the self discharge rate of the battery... if it's sitting up cozy with the CPU at a constant 40-50C it might not last all that long even if the machine is seldom off... whereas if it's out by the edge of the board near some fresh air, it might be near ambient.

EDIT: Oh yah and some boards will stop the clock first to maintain settings as long as possible beyond a certain point of discharge, or that's just how it happens, the settings cling on at say 2V but the clock won't run.

appiah4 wrote on 2021-06-10, 12:09:

tonata wrote on 2021-06-10, 08:02:

Can you please comment on my solution on using a 18650 lithium ion rechargeable battery. Can these batteries be safely recharged by a 386 motherboard while in use?

Here is a 18650 https://fr.aliexpress.com/item/1005002420574151.html charger if a comparison can be made. Or maybe I should put a diode to be on the safe side reducing to it 3.3 V ?

No, a rechargeable Li Ion battery's charging characteristics are completely different from the NiCd it replaces, this is a recipe for disaster. If you want to play safe use NiMH.

Can you please give more details? What is different? A link maybe? It is just different current and voltage or first it starts at certain voltage and then it changes to another during the charging time?

I do not not want to use NiMH because it will leak at some point, that is for sure. Even if it takes 3-4 years.

tonata wrote on 2021-06-10, 08:02:

Can you please comment on my solution on using a 18650 lithium ion rechargeable battery. Can these batteries be safely recharged by a 386 motherboard while in use?

Here is a 18650 https://fr.aliexpress.com/item/1005002420574151.html charger if a comparison can be made. Or maybe I should put a diode to be on the safe side reducing to it 3.3 V ?

That's a terrible idea. You can learn about the details of different batteries here:

https://batteryuniversity.com/learn/article/c … based_batteries

https://batteryuniversity.com/learn/article/c … m_ion_batteries

tonata wrote on 2021-06-10, 13:22:

I do not not want to use NiMH because it will leak at some point, that is for sure. Even if it takes 3-4 years.

then use a non-rechargable primary and connect it through a diode, simple as

Thanks. I am not much into electronics. I think this one 1N5817 diode will do the job.

It says:
Vf = 450 mV
Ir = 500 uA

So I will loose 0.45 V and the battery will be charged but negligibly (500 uA is very low I think) during the powered state of the computer.

Is this OK?

I'd risk leaking every 10 years vs exploding incendiary device every 6 months, but that's just me.

tonata wrote on 2021-06-10, 15:56:

Thanks. I am not much into electronics. I think this one 1N5817 diode will do the job. […]
Show full quote

Thanks. I am not much into electronics. I think this one 1N5817 diode will do the job.

It says:
Vf = 450 mV
Ir = 500 uA

So I will loose 0.45 V and the battery will be charged but negligibly (500 uA is very low I think) during the powered state of the computer.

Is this OK?

If you like living dangerously, sure why not.

tonata wrote on 2021-06-10, 15:56:

Thanks. I am not much into electronics. I think this one 1N5817 diode will do the job. […]
Show full quote

Thanks. I am not much into electronics. I think this one 1N5817 diode will do the job.

It says:
Vf = 450 mV
Ir = 500 uA

So I will loose 0.45 V and the battery will be charged but negligibly (500 uA is very low I think) during the powered state of the computer.

Is this OK?

As others have pointed out, no, it's not OK. There is potential fire hazard involved, so be careful!
I myself prefer to remove the barrel battery and connect a coin cell to the external battery header (which is present on many boards), using a cheap battery holder. One coin cell battery is usually enough for many motherboards, while for others you might need multiple AA batteries in series. Either way, the external battery connector is definitely the easiest and safest solution (if your motherboard has it).

If you are not heating or piercing a lithium-ion battery why it should be a problem? And also I do not intend to charge it.

tonata wrote on 2021-06-10, 20:49:

If you are not heating or piercing a lithium-ion battery why it should be a problem? And also I do not intend to charge it.

Dude, did you even read the links above?

appiah4 wrote on 2021-06-11, 08:56:

tonata wrote on 2021-06-10, 20:49:

If you are not heating or piercing a lithium-ion battery why it should be a problem? And also I do not intend to charge it.

Dude, did you even read the links above?

Don't bother. This is the point where the thread needs to die. 😉