I would check to see if there is a different voltage source on the board going to the RTC. If yes, you could simply remove the resistor (or D3) from circuit. If no, then yeah I guess just do what you said.

There is no connection to second source of voltage for the RTC chip.

Thanks for your opinion, maxtherabbit!

maxtherabbit wrote on 2025-05-05, 13:19:

I would check to see if there is a different voltage source on the board going to the RTC. If yes, you could simply remove the resistor (or D3) from circuit. If no, then yeah I guess just do what you said.

Ok, I was wrong. I checked and measured everything again. It turned out, that one of the two transistors (these look like transistors) near the ATX power connector is actually connected to +5 VDC power line and gives +5 VDC directly to RTC chip when board is alive and running.

So, I tried your suggestion: Desoldered and lifted one leg of D3 diode. And... sh*t! I still get reverse "charging" voltage of +4.3 VDC through D2 diode to the battery positive terminal and the positive pin of the external battery connector:

It seems like I can't modify this board for using 3 V CR2032 non-rechargeable coin cell battery, if I still get +4.3 VDC reverse "charging" voltage when main battery charging part of the circuit is cut-off 🙁

Hi, some thoughts.

a) can a diode be added on the battery side instead?
Some have 0.3v drop (germanium etc), so they don't reduce voltage that much.

b) aren’t there coin cells that can be recharged?
I heard there are both lithium batteries and lithium rechargeables.
If so, then the diodes and the resistor might do enough current-limiting already.

Again, just some thoughts.

Jo22 wrote on 2025-05-05, 19:32:

Hi, some thoughts. […]
Show full quote

Hi, some thoughts.

a) can a diode be added on the battery side instead?
Some have 0.3v drop (germanium etc), so they don't reduce voltage that much.

b) aren’t there coin cells that can be recharged?
I heard there are both lithium batteries and lithium rechargeables.
If so, then the diodes and the resistor might do enough current-limiting already.

Again, just some thoughts.

Hi, Jo22, and thanks for the input!

a) I thought so - see my first post with my first idea for the CR2032 socket modification. Unfortunately after checking the suggestion from maxtherabbit I think this won't work. I can add one SMD diode on the trace to battery, but I don't know if there's a suitable diode with low forward voltage drop, which can fully cut the current flow in reverse direction (after some measurements I found out that diode D2 does not work for this, or maybe there's something wrong with it).

b) Yep, there are some rechargeable alernatives like ML2032, VL2032 (what types are these?) and LIR2032 (this one is Li-Ion and I'm sure it needs a small charging controller board), but they're not widely available as CR2032. I wanted the modification to be a more modern "standardized" one with cheap non-rechargeable CR2032 coin cell battery.

I'm still open for suggestions about modification which includes change of D2 and D3 diodes, as I have no idea what diodes I can use to replace them. I don't know their type or model, I measured their forward voltage drop and it is about 550 mV, but I'm wondering why D2 "leak" so much reverse voltage (4.3 V at 5 V reverse voltage).

I thought this will be an easy modification... Stupid me.

analog_programmer wrote on 2025-05-05, 20:27:

I thought this will be an easy modification... Stupid me.

That wasn't foreseable and it's not your fault, don’t worry.

In retro/vintage computing most of us will encounter such a moment once in a while, I think.

Up 'til yesterday I thought that there are some 386 motherboards with lame rechargeable battery circuits designs, but still most of them can be used with regular non-rechargeable AA or AAA battery packs without any modifications. And what the hell is this sh*t with only "alternative" for external recargeable battery:

D2 & D3: 1N4148 silicon diodes
TR2: C1815 general purpose NPN transistor
TR3: A1015 general purpose PNP Transistor
RTC + CMOS chip: HM6818P

When I lift one of the pins on D3 diode I cut-off the battery direct charging connection to +5 VDC from AT PSU, but this changes nothing since the battery still gets reverse +4.3 VDC from D2's anode, as D2's cathode is connected to TR3's collector and RTC + CMOS chip pin 24 (Vdd/Vcc).

I can draw by hand a proper circuit diagram and attach it here as a picture, if this might help someone to come up with an idea for non-rechargeable CR2032 battery modification.

Test D2 with a multimeter in diode test mode if you're concerned, but I think you're fine. It is normal to see voltage "through" a reverse biased diode in an open circuit, but that does not mean any meaningful amount of current will flow.

https://electronics.stackexchange.com/questio … an-open-circuit

maxtherabbit wrote on 2025-05-06, 14:31:

Test D2 with a multimeter in diode test mode if you're concerned, but I think you're fine. It is normal to see voltage "through" a reverse biased diode in an open circuit, but that does not mean any meaningful amount of current will flow.

https://electronics.stackexchange.com/questio … an-open-circuit

First, thank you very much for the hint with the most simple modification and that you made me checking the whole circuit related to the original NiCd battery.

I already tested D2 diode (and D3 as well) and it's fine. I thought that probably the reverse current will be negligibly low, but still I'm not sure if a 3 V CR2032 battery will be fine in this modified circuit (with cut-off charging voltage through D3 diode). I think I should try to measure the reverse current through D2 by using some small value resistor connected at one of the battery terminals while still there is no battery installed. Just to be sure, I don't want a "BANG" from non-rechargeable battery while still trying to charge it 😁

Ok, If everything with your simple modification solution is fine, there is another question. These 1N4148 Si diodes have relatively high forward voltage drop (I measued 550 mV while tested them if they're still fine) and I can't find any datasheet to check what's the minimum voltage value needed for clock and RAM of HM6818P chip. If 3 V CR2032 in series with 1N4148 diode can't provide enough voltage for HM6818P chip when the board is powered off, I'll have to change D2 for some schottky diode, but then in exchange of lower forward voltage drop the reverse "charging" current will be even higher.

Probably many readers here will say "c'mon don't go into unnecessary details, just try the freakin' mod", but I'd rather prefer to get things sorted before I do something stupid 😉

Stop overthinking it.

Put 3 NiMH AA batteries in a holder and attach it to EXT Battery post.

Those can survive the NiCD charge curves, and are easily replaced.

Suitable holders are COTS on amazon.

wierd_w wrote on 2025-05-06, 15:47:

Stop overthinking it.

Yeah, I know, I know... But this 286 motherboard is fully functional and with just a few small traces of corrosion from the NiCd battery leakage (I've cleaned them and it's visible on the attached pictures). It took me some time to figure out all the jumpers and switches settings without any trace of suitable user manual for this mobo and I don't want to scr*w-up the things right before the "finish line" of its (let's call it) restoration 😁

wierd_w wrote on 2025-05-06, 15:47:

Put 3 NiMH AA batteries in a holder and attach it to EXT Battery post.

Those can survive the NiCD charge curves, and are easily replaced.

Suitable holders are COTS on amazon.

Thank for the suggestion. I'll consider this as "a backup solution" without modifications 😉

IMO you should use 3 AA primaries in a battery holder. If you're using the detachable header anyway, what's the point of a coin cell with terrible capacity and possibly inadequate voltage?

If you use lithium AAs you don't have to worry about leakage. A functional 4148 will not allow enough reverse current leakage to harm any primary battery.

The RTC probably uses as much/little power as a solar pocket calculator or a quartz wrist watch.
The actual current that flows can likely be handled by an OA97 Germanium diode, even, or an BAT48 Schottky diode.
So I wouldn't worry. The voltage level is more important, I assume.

2x AA works, too, but probably only for a while with fresh alkaline batteries.
That's because when fresh they have a higher voltage than 1,5v each, when without a load. Say, 1,7v or so.
2x AA NiCD used to provide 1,2v only, rather than 1,5v. Not sure about NiMH.

Speaking under correction, though.

PS: Coin cells have a lower self-discharge than normal batteries (say AA).
So maybe it makes no difference if AA or AAA batteries are being used.
The RTC isn't depleting the batteries here, they do it all on their own rather.

PS: Another kind of 3.3v or 3.7v power source.. cell phone rechargeables.
If you have an old cell phone battery, it can work as a makeshift clock battery.

Picture attached.

maxtherabbit wrote on 2025-05-06, 16:57:

IMO you should use 3 AA primaries in a battery holder. If you're using the detachable header anyway, what's the point of a coin cell with terrible capacity and possibly inadequate voltage?

If you use lithium AAs you don't have to worry about leakage. A functional 4148 will not allow enough reverse current leakage to harm any primary battery.

The point is not to use an unnecessarily bulky, less accessible and more expensive solution.

We may ask why motherboards for 386 and 486 systems are being modified in a similar way. The reasons are the same.

Jo22 wrote on 2025-05-06, 17:13:

The RTC probably uses as much/little power as a solar pocket calculator or a quartz wrist watch.
The actual current that flows can likely be handled by an OA97 Germanium diode, even, or an BAT48 Schottky diode.
So I wouldn't worry. The voltage level is more important, I assume.

Right now I'm not in mood for tinkering with this mobo, so tomorrow I'll try to measure the reverse current from the 1N4148 D2 diode - the one that gives +4.3 VDC in the modified open battery circuit. I think, it will be fine, but just to be sure.

If one 3 V CR2032 battery (usually gives 3.3 V when new) in series with 1N4148 diode (D2) can't provide enough voltage for RTC+CMOS chip, I already know who to ask for these schottky or Ge diode alternatives 😉

Jo22 wrote on 2025-05-06, 17:13:

2x AA works, too, but probably only for a while with fresh alkaline batteries. That's because when fresh they have a higher volt […]
Show full quote

2x AA works, too, but probably only for a while with fresh alkaline batteries.
That's because when fresh they have a higher voltage than 1,5v each, when without a load. Say, 1,7v or so.
2x AA NiCD used to provide 1,2v only, rather than 1,5v. Not sure about NiMH.

Speaking under correction, though.

PS: Coin cells have a lower self-discharge than normal batteries (say AA).
So maybe it makes no difference if AA or AAA batteries are being used.
The RTC isn't depleting the batteries here, they do it all on their own rather.

PS: Another kind of 3.3v or 3.7v power source.. cell phone rechargeables.
If you have an old cell phone battery, it can work as a makeshift clock battery.

Actually I have a working solution with one rechargeable 3.7 V Li-ion battery (18650) + small charger PCB with TP4056 adjusted for lower charging current. But it's a bulky solution 😁 I had these 18650 battery and TP4056 charger PCB as leftovers from some electronic projects and one day I was wondering what I could use them for, when I suddenly came across this idea for this DIY direct replacement for a 386 motherboard's NiCd barrel battery, which is connected in very stupid "charging circuit" (yeah, the "ingenious" asian engineering thought saving pennies on everything) i.e. with direct connections between the (already leaked) barrel battery and +5 VDC without any resistors, diodes or fuse in between. It works without any problems for almost two years, I think.

maxtherabbit wrote on 2025-05-06, 16:57:

IMO you should use 3 AA primaries in a battery holder. If you're using the detachable header anyway, what's the point of a coin cell with terrible capacity and possibly inadequate voltage?

If you use lithium AAs you don't have to worry about leakage. A functional 4148 will not allow enough reverse current leakage to harm any primary battery.

On a normal board meant for primary lithium (which would take the cr2032 mod directly with very little guff) this is correct.

However, the problem presented here is a persistent, 'cant disable it!' Float charger circuit meant to keep a rechargable NiCD cell full.

That charge circuit will cause alkaline or primary lithium cells to malfunction/leak/burst, which is bad juju.

NiMH is 'able' to be charged in a NiCD charger, but it shortens its life expectency. The curves of a full on NiCD charger will overheat a NiMH cell, but float charging is under 1C, and should be OK(ish). A small resistor on the return voltage line to reduce the current to the recommended 0.05C might do the trick.

https://batteryuniversity.com/article/bu-408- … l-metal-hydride

Failure of the cells is mitigatable by being on a pigtail at a safe distance/location, and by ready replacement with supermarket-available cells.

wierd_w wrote on 2025-05-06, 18:22:

However, the problem presented here is a persistent, 'cant disable it!' Float charger circuit meant to keep a rechargable NiCD cell full.

No it isn't, review the recent posts

analog_programmer wrote on 2025-05-06, 18:13:

The point is not to use an unnecessarily bulky, less accessible and more expensive solution.

Bulky, ok I can't argue with that. But I fail to see what could be more accessible than a battery pack on a string you can put wherever you want. Regarding cost, I don't think a 3xAA box is substantively more expensive than a coin cell holder

maxtherabbit wrote on 2025-05-06, 18:29:

But I fail to see what could be more accessible than a battery pack on a string you can put wherever you want. Regarding cost, I don't think a 3xAA box is substantively more expensive than a coin cell holder

Well, here, where I live, the rechargeable batteries of any type are way more expensive and less available than their non-rechargeable alternatives/substitutes. As for the CR2032 PCB sockets/holders - I have at least a dozen of these taken from already scrapped motherboards. And even if I buy a single piece of these CR2032 battery sockets, it costs only a couple of pennies.

Let's put those rechargeable AA-battery packs aside for now. At least 'til the CR2032 modification attempt fails. If it fails.

No no, I was suggesting PRIMARY AAs in a pack