Coin battery in computer drains overnight

[Averagesupernova]

TL;DR Summary

The battery that holds up the memory for the bios drains overnight.

I have an older computer running Windows 98 in a mobile application. I know, Windows 98, just replace it. Not a pleasant option. This system runs an application that dispenses dry granular material out of a hopper that holds about 9 tons of material. It is spread over an 80 foot swath. So basically, the computer functions as to control the rate that the material is dispensed.
-
The issue is as the title states that the internal battery drains overnight. Replacing the computer is not like ordering a laptop for a couple hundred dollars. The environment this computer runs in is dirty and corossive. The original unit is enclosed in a die-cast case that is very well sealed and I don't believe anything has migrated into the case to cause leakage on the motherboard. The replacement cost is around $2K. I find it difficult to believe that a silicon device on the motherboard is causing this. Anyone have experience with this? I am suspecting a leaky cap. This is territory I have never ventured into.

 

[Wrichik Basu]

If the CMOS battery drains overnight, there is some serious trouble in the motherboard.

Try this: Before leaving the computer overnight, shut down Windows as usual, but do not turn off the PC from the mains. If you leave the mains power on, then the PSU will keep on supplying +5VSB (5V standby), and the CMOS will be powered from that rather than the battery. Let us know what you find next day.

 

[Averagesupernova]

When I get it shipped back to me (I don't have it at the moment) I'll do a little poking around. But, like I said, it's mobile equipment. No mains power, just 12 volts. Not sure what they would have done there. I understand what you're saying, just not sure it applies to this situation. Thanks a lot for the reply.

 

[Wrichik Basu]

But, like I said, it's mobile equipment.

Oh, sorry. I missed that part in the OP. But there is some sort of charging mechanism, right? Or does it run on a battery that has to be replaced and cannot be recharged?

 

[anorlunda]

How about an external USB battery backup, plus a voltage divider to drop 5V down to 1.5V? The power draw should be tiny, so the voltage divider resistance could be relatively high. The power drain can't be a short circuit, or it would not take overnight to empty the battery.

Some of them come with their own solar panel.

You could also cannibalize a solar garden light. The cheap ones have a solar panel, a charger circuit, and a 1.5V AA battery.

I know this is not amusing to you, but I see here a variant of the Y2K problem. I'm thinking of computer hardware/software solutions that far outlive their expected lifetimes. I'm sure there are PDP-8 machines somewhere still running vital functions.

 

[Averagesupernova]

Oh, sorry. I missed that part in the OP. But there is some sort of charging mechanism, right? Or does it run on a battery that has to be replaced and cannot be recharged?

I'm not sure about recharging. I suspect it is not meant to have any voltage to the motherboard when turned off. The reason I say this is because there is a master switch, and when this switch is flipped, it provides power to the unit as well as the out board units that take in data from sensors and control various servos. As soon as I turned it on, the monitor comes up and goes through boot-up that would be similar to an old Windows 98 machine. When I shut down from windows, it shuts down on it's own, does not give me the old 'it is now safe...' but there is still power to the unit as long as the master switch is on. To power the computer again the master switch needs to be cycled. Pretty sure the master is not meant to just be left on since the outboard units will remain powered up and run down the 12 volt battery. I will have to wait until I get the unit back.

 

[Averagesupernova]

@anorlunda I'm considering something like this. To even find these machines anymore is all but impossible. Replacement costs of the whole system to compete with the functionality of the old one is above $10K.

 

[jedishrfu]

Here's another idea:
- ditch the old hw
- get new homework that meets what you need
- run the app inside a docker container running WIndows 98

Here's a reference to running Windows 98 as a docker container:

https://github.com/liudonghua123/windows98

It extends your code lifetime for as long as docker is around and you might be able to run it other homework hosting linux or macos too. This also provides you with some measure of security from threats both foreign and domestic. :-)

 

[Baluncore]

I know, Windows 98, just replace it. Not a pleasant option.

How many times have you replaced the battery since new?
When did you last replace the battery with a known good battery?

 

[DaveE]

There's a good chance that it has a relatively standard MB inside. If you can id the form factor, you may be able to swap from another old computer. Old legacy PCs aren't that unusual, you may find people on the web that sell new or used replacement HW, although I don't know of any.

Also, if you can get your hands on a good one you could insert an ammeter and compare the battery currents. Then maybe trace the leakage current. It's a PIA though. Reminds me of my summer job at Atari repairing the Space Invaders MBs. We only had a cheap scope and a DVM. Cutting and resoldering traces was a normal troubleshooting technique in that cheap electronics world. Fortunately the PCBs for an 8080 PC weren't dense or complicated.

If you're desperate and adventurous, you could try replacing the battery with a 1.5V isolated DC-DC converter powered by it's own larger rechargeable battery. Of course it has to run all the time, so you wouldn't want to use the vehicle battery which may be disconnected sometimes.

 

[Averagesupernova]

I've never replaced the internal battery because I've never used this hardware. I've only had the machine a short time. I sent the computer, it's mobile display and a mobile keyboard off to a place that repairs these. This was last year. The display was damaged and they repaired it as well as what the keyboard uses as a mouse. It was shipped back to me and I powered it up last week. Everything is completely new to me. The rest you've all been told. So, with over $1K in repairs already and the motherboard supposedly unrepairable, I'm a bit ticked off. I honestly suspect that the first time I sent it off to be gone through, only parts of it were evaluated. I'll try to post some pix for those interested.

 

[Averagesupernova]

Here it is. Sorry I don't have pix of the internals of the case. I'll have to wait until I get it back. Being new to delving into a computer motherboard, wondering just how many places the power from the battery goes. I've always assumed pretty much just one device. And maybe a capacitor. Someone can correct me if I'm wrong. There's a bit more to the keyboard than it looks. On the back side there is a 3.5 floppy drive as well as a USB port. Plus what they use as a mouse. A little flaky to use the mouse but the short time I've played with it I think I could get used to using it. The outboard units that this computer talks to are not pictured.
-
Edit: BTW, the battery is soldered to the board. This is why I sent it back. They advised I not do it myself even though I used to do rework for a living. Plus, I had something else to ship to them in the same box and I wanted them to know that they dropped the ball so-to-speak.

 

[Vanadium 50]

But there is some sort of charging mechanism, right?

No. He says it's a coin battery.

I'm not sure what we can do. You don't have the computer, and they are going to replace the battery. We can't do anything or learn anything until it comes back.

Windows 98 is 23 years old now. A CR2032 battery, the most common, has a shelf life of 10 years, and Panasonic says "10 year life under normal use." So after two decades, it is not surprising that this needs replacing. That may not be the only problem, of course.

Honestly, $2K over 20 years is not something I think is exorbitant. It's $2/week. You're probably paying twice that in electricity. I'm not so surprised that the battery has failed. I'm surprised nothing else has failed, particularly fans or hard drives.

 

[Averagesupernova]

I'm not surprised the battery had failed either. The issue is that supposedly a new one won't last overnight.

 

[Wrichik Basu]

No. He says it's a coin battery.

Actually I thought there were two batteries — the CMOS battery, and another separate battery for powering the system while it's working.

 

[Keith_McClary]

There are many questions about this:
https://www.google.com/search?q=CMOS+battery+drains

 

[Averagesupernova]

There are many questions about this:
https://www.google.com/search?q=CMOS+battery+drains

Yep. In the last couple of hours I've read through some things there that give me hope.

 

[Averagesupernova]

Actually I thought there were two batteries — the CMOS battery, and another separate battery for powering the system while it's working.

You are correct. There is a CMOS battery that motherboards typically have and the vehicle battery. Large truck chassis so actually 3 large truck batteries. I have reservations about attempting to power the CMOS as I suspect a leaky diode. This could potentially try to power up the 5 volt power rail. I think I'll try to track down where the leak is.

 

[Tom.G]

Here are a few possibilities I've run across with extremely short coin cell life; the first and third ones are the most likely, otherwise no particular order.

1) If the soldered-in battery has been replaced, the board around the soldering MUST be well cleaned. Leaving even a little bit of flux on the board will drain a coin cell in a couple weeks. The solder flux is slightly hygroscopic and absorbs moisture from the air. This creates a leakage path to drain the coin cell.

2) After many years in a mixed industrial & city environment, there was enough pollution deposited on the motherboard to act as the above situation. Actually it was a bit worse; there would be occassional program crashes. These cleared up after removing and washing the motherboard, not for the faint-of-heart or in-experienced.

3) Your conjecture about a bad electrolytic is a very real possibility. The one (typical?) circuit I can find at the moment has the clock kept alive during power-off by the battery but powered from the DC supply when operating.
The switching is accomplished with a "diode or" circuit, that is there are two diodes in series with the battery and another diode from the DC supply, both feeding the clock chip. There is often (always?) an electrolytic connected to the clock chip Power pin. The one example I can find at the moment has a 22uF, 16V cap. That should be relatively easy to locate once you get access to the motherboard, especially if you can identify the clock by the chip Part No. Or just trace the circuit from the coin battery "+".

4) The replacement battery was not 'fresh' when installed. For instance some of the Big Box department stores seem to buy in huge quantities and then store the product in non-climate-controlled warehouses. This is a problem here in the South West of the USA, the warehouses are in Desert areas where the land is cheap. This affects just about any non-food with an expiration date, batteries, over the counter drugstore products, cosmetics, small tubes of paint, etc.

Cheers,
Tom

p.s. A few schematics for older machines are findable on the WWW with a concerted search. Might be worth it.

 

[Rive]

...wondering just how many places the power from the battery goes.

Usually it's some diode&capacitor thing, then it goes to the south bridge (at the win98 era it's south bridge already).

BTW, the battery is soldered to the board.

Did you see it, or this is what 'they' told you? It gives some chance that the battery is not the standard coin cell.
Do you have a type for the unit? Or some pictures?

If the CMOS battery is some old style 3.6V NiMh or NiCd thing (for the Win98 era it's rare, but still happens) that may be a bit worrisome

This kind of thing requires quite some work to clean up (and it'll never be like the new ever again )

 

[Averagesupernova]

It is in fact soldered to the board. I have seen it. It is a lithium type thing with leads. Not the leaky thing in the above pic.

 

[Rive]

From those I had only a very few similar cases but those were about the south bridge (apart from the occasional piece of faulty battery, of course). I would risk it and do some rework to measure a current draw from a new battery.

A south bridge still can be replaced, but usually just don't worth it.

Could you please tell us the type of the computer?

 

[Vanadium 50]

The issue is that supposedly a new one won't last overnight.

How do you know, since you say that you need to solder it to replace it and have not?

Maybe it's time to describe the actual symptoms and not your theory of what is causing them.

 

[Averagesupernova]

How do you know, since you say that you need to solder it to replace it and have not?

Maybe it's time to describe the actual symptoms and not your theory of what is causing them.

Maybe it's time you read the whole thread to learn what has actually been done.

 

[Vanadium 50]

Maybe it's time you read the whole thread to learn what has actually been done.

That's not very nice. You do not explain why you think the battery is dead. Did you get a message that says "your battery is dead"?

But hey, you want to act this way to people who are trying to help you, your choice. In the words of Barbara Billingsley, " Chump don' want no help, chump don't GET da help! "

 

[MikeeMiracle]

This computer system sounds like a great case for being virtualised, removing the operating system from being tied into the underlying hardware. Any copy if windows server can run Hyper-V virtualisation software for free. Worth looking into if it's a critical machine that you can't afford to lose.

 

[Averagesupernova]

@Vanadium 50 read the whole thread already. If you had you would know it has been sent in for service the second time now for the reason of having a dead CMOS battery after having set unused for about a year after (supposedly) being checked out thoroughly. Do you really think I just have a 'theory' of a CMOS battery going dead overnight? After all I've posted here, you assume there hasn't been any communication with the company that services these machines? Btw, it's not $2K over twenty years. It's more like$12K to15K original system cost. It's getting very difficult to find replacements for anything involved.
-
In all likelihood what will happen is I will either find the leak myself since the company that I sent it to won't go any farther with it or the whole system will be replaced. I'm not dumping any more money into this system. I was Leary about dropping $1K into it last year, but did because I was told it's ready to go. I can likely recoup some losses by selling the remaining pieces and I will simply start over.

 

[AlexCaledin]

- could some infrared thermal imaging equipment help to detect such a leak??

 

[Averagesupernova]

- could some infrared thermal imaging equipment help to detect such a leak??

If only. Lol. I'm afraid not. Certainly nothing I can afford. Most likely the drain is in the low milliamp range. 5 microamp is probably an acceptable draw indefinitely for running the real time clock and holding memory up. The unit is on its way back. I plan on keeping folks here informed.

 

[256bits]

having set unused for about a year

Kind of says it all I think.

 

[Averagesupernova]

Kind of says it all I think.

Lol. These machines can likely go for ten months without being powered up. A twenty year old lithium coin battery is expected to soon be at the end of its life. No doubt. But the service tech claims a fresh one gets drained overnight. What I really suspect happened is they put in a new battery the first time I shipped it off when they likely found a dead one. I would not have been surprised either to find a dead one that is around twenty years old. The rest of the unit was checked out, powered down and returned to me. The unit probably had already drained the battery by the time I received it back. I just didn't check it out immediately after I received it. Bigger things to worry about at the time.
-
I did manage to find a factory reman of this unit from the OEM. The price tag is $6500 with exchange. Dealer price. Not an e-bay seller. Without the exchange the price goes to $8500. I won't be going this route.

 

[Averagesupernova]

The latest:
-
The unit arrived back to me yesterday. I put it on the bench tonight and here is what I found: The lithium battery is in fact dead. I removed it and hooked a three volt supply in place of it with a 100K resistor in series. The voltage drop across the 100K is 95 mV. The thing is drawing less than a uA. I question whether there was an attempt to even repair it. I have to say it was not the most fun thing to disassemble. For whatever reason the tech decided it would be a good idea to blob the little battery with black RTV silicone. Worried about vibration or wanted to make sure the next guy to deal with it would have fun? Here are some pix.
-

 

[Averagesupernova]

Here are a few more:

Not sure what the little nicad bank in there does. I assume provides enough power to make sure nothing is lost in a power loss. Powers unit long enough to save everything I would guess. I never really want to find out. Lol.

 

[AlexCaledin]

The current may be 1 microampere at some logical state of the CMOS chip and more than 10 milliamperes at another state, if the CMOS is damaged.

 

[Baluncore]

For whatever reason the tech decided it would be a good idea to blob the little battery with black RTV silicone.

Was that a neutral cure silicone, specified to be an excellent insulator?
It only takes a finger print on a coin battery insulator to discharge that battery if there is moisture condensation later. Don't handle coin cells with your fingers.

Check for a ceramic capacitor across the Real Time Clock chip. If it is multilayer ceramic then it may become mechanically or thermally intermittent, but with insufficient power available to fuse that layer, which is expected to happen in most power supplies.