History of the CMOS battery?

[CuriousChord]

Just wondering how the CMOS battery has evolved over the years. What computer first used a CMOS battery connected to a motherboard? What types of CMOS battery have been "industry standard" during what time periods? What's the standard CMOS battery today, and where is it headed?

For example, and please forgive my ignorance, did a TRS-80 and an Apple ][ employ CMOS batteries?

I'm only familiar with the barrel-type soldered-onto-the-MB battery, the B-220, and the CR2032. Before and after those examples, I have no idea.

Thanks for any insights.

Cheers! cc

 

[Vlad]

As far as I know its still the CR2032 going in modern times. I do remember around the G3 era macs had the PRAM battery, isn't that the same concept in a way? Some of the PS/2's had that battery pack thingy and so did generic mainboards.

I always kinda liked the CR-2032 ones the best. The battery just died and that was that. The battery pack ones could sometimes be hard to find and now a days are and the soldered on ones tended to corrode and damage the PCB. The other two were the lesser of the evils in my opinion.

 

[RichCini]

Here are some that I remember: the Apple /// used 3-AA batteries in a crappy metal holder. The original Mac used an AA-sized 3.6v lithium battery that was in a battery holder embedded in the rear shell. Later Macs of the II era used a 1/2-AA 3.6v lithium battery on the motherboard. The PC-AT used a 4.5v square or rectangle block battery (which was probably 3-AA batteries anyway). The Model 100 used a 3.6v stacked coin cell NiCd battery soldered to the board. I'd have to pull out my original PC, but I think that used an AA battery holder setup like the Apple ///.

Add-on clock boards used varying batteries. Most that I remember used a variation of the coin cell. There were some for the S-100 buss that used these GE "DataSentry" batteries which were 3.6v NiCd blocks but based on the physical size, don't seem to fit a particular standard battery size (3-AAA in a triangle configuration at the most). I replaced one with a cordless phone battery of similar mAh rating and it seems to work fine.

From the progression, you can see the movement from readily available (and replacable) standard batteries like the AA, to more customized power solutions to reduce board space and increase the service interval. Also, I don't think that the RTC chips back they were as power-thrifty as they are now, thus requiring a battery with a larger capacity so you could achieve the desired service interval.

In the early-80's I know that "hearing aid" batteries existed but I don't think that 3v lithium coin cells did. No one wants to open their PC every three months to replace a stack of A76 low-mAh hearing aid batteries, so I guess the designers resorted to what was available that could handle the power draw from the RTC or PRAM chips.

 

[Chuck(G)]

I think that it's odd that the battery is called as a "CMOS" battery. The battery has nothing to do with the complementary metal-oxide fabrication technology used, but rather serves as a clock or watch battery.

That the MC146818 memory-with-clock is implemented in CMOS is irrelevant--so are other chips on the AT motherboard.

If the battery were needed only to keep configuration RAM contents intact, a simple NVRAM chip would have been much simpler--and smaller, with no battery involved.

On one of my 188 boards, there's a DS1215 "phantom time chip" that uses a 1F cap to keep clock time across power cycles. No batteries at all--and no external address decoding.

 

[Dwight Elvey]

I think that it's odd that the battery is called as a "CMOS" battery. The battery has nothing to do with the complementary metal-oxide fabrication technology used, but rather serves as a clock or watch battery.

That the MC146818 memory-with-clock is implemented in CMOS is irrelevant--so are other chips on the AT motherboard.

If the battery were needed only to keep configuration RAM contents intact, a simple NVRAM chip would have been much simpler--and smaller, with no battery involved.

On one of my 188 boards, there's a DS1215 "phantom time chip" that uses a 1F cap to keep clock time across power cycles. No batteries at all--and no external address decoding.

The RAM was called CMOS RAM. There was no NVRAM other than Dallas chips
that were actually CMOS RAMs with a battery. Early FLASH parts required special
programming voltages.
The battery was usually chosen was the NiCad.
This was intended to last of ever. Of course, they all seem to leak after
some time.
I didn't know that the DS1215 used the 1F caps. That is news to me but
then I've not kept track.
Dwight

 

[Chuck(G)]

The RAM was called CMOS RAM. There was no NVRAM other than Dallas chips
that were actually CMOS RAMs with a battery. Early FLASH parts required special
programming voltages.

I was using the Xicor X2444P chips in 1985--I have some with early 1984 date codes in my hellbox. I certainly wasn't designing on the bleeding edge then. 8 pins, 256 bits, 5V power--more than enough for holding configuration data.

The battery was usually chosen was the NiCad.
This was intended to last of ever. Of course, they all seem to leak after some time

NiCads eventually fail, even without leaking. Eventually they refuse to hold a charge, even absent "memory" effects. They were probably the best available technology at the time, but I doubt that manufacturers deluded themselves that NiCads had an unlimited lifetime.

 

[CuriousChord]

NiCads eventually fail, even without leaking. Eventually they refuse to hold a charge, even absent "memory" effects. They were probably the best available technology at the time, but I doubt that manufacturers deluded themselves that NiCads had an unlimited lifetime.

The whole "lifetime" approach to design is interesting as a part of CMOS battery history. Assuming that CR2032 is a reasonable current standard, and assuming there's not a viable replacement candidate on the horizon, what made NiCad more attractive than watch-cell batteries in the 70s and 80s? Can't recall the correct battery model number, but during the 70s I had an LED watch that ran on a cell much like the CR2032 in appearance. Perhaps the output was not suffucient compared to today's CR2032 (as RichCini noted about sufficient power for the RTC chip and "desired service interval")?

As for NiCad, was there a component of planned obsolescence in the design thinking? In other words, what user will need to replace the computer/MB in 10 or even 5 years, given that newer, faster, better machines are in the production pipeline or on the drawing boards? But then there's the external battery connector, which would address the longevity issue, assuming it was a standard feature on motherboards. Even so, why were there B-220 batteries with a "4-pin" connector that really had 3 holes--thus requiring that a user bend or clip a pin on the board to connect the battery?

Just curious and learning here. I have no real insight into the world of computer design considerations.

 

[Chuck(G)]

The amount of current required to keep the clock going and the RAM contents on an MC146818 is minute--just about any stable source would work.

But most common primary cells have a problem with shelf life--they will self-discharge even if not used. So a NiCd secondary battery was a logical choice. Its life is longer than most common primary cell types (carbon-zinc, alkaline) and the cost isn't very high. It is a mismatch in that NiCd cells do tend to function best as high-output current devices, but it was a decent, if imperfect, solution.

Lithium (Li-Mn) cells do have an advantage in this situation--they have a very long life (10 years is not unusual) particularly when used in a very low-current application. Those blocky Dallas clock chips are nothing more or less than a standard clock chip with a lithium cell and resonator sealed under an epoxy potting compound.

Long life is also why you can often see smoke detectors with lithium batteries installed. Apartment landlords love them (as do meth cooks, but that's another story). But eventually even they discharge.

My own experience with on-board PC NiCads was that every one eventually failed. The same for the alkaline batteries. Those got replaced with a simple 2 or 4 cell set of AA alkaline batteries, located off the board and preferably outside the PC case. I'll have to change the batteries every few years, but if they leak, they won't damage any electronics. And AA cells aren't exotic.

 

[CuriousChord]

...a simple 2 or 4 cell set of AA alkaline batteries, located off the board and preferably outside the PC case.

Interesting...how have you located batteries outside the case?

 

[Chuck(G)]

Interesting...how have you located batteries outside the case?

Wire with connectors and battery holder. If you want to be neat, use some double-stick foam tape to attach the holder to the case. Many older 286 and 386 mobos have a 4-pin header for off-board batteries. If you don't care about looks, put the batteries in a plastic bag to catch the goo if it should escape the cells.

 

[Dwight Elvey]

Hi
If you want to use NiCads like the original, use solid strand wire.
if it leaks, the stuff won't run through the wire to the MOBO.
( a trick I learned from pinballs ).
Also, if the original had a NiCad, and you intend to use
regular non-rechargeable cells, you may need to disable the
charging circuit.
This is located as follows:
Trace the positive lead of the NiCad. You should find a diode and
a resistor. Lift a lead of the resistor but leave the diode.
Dwight

 

[andy]

The first IBM to use a CMOS battery was the AT. All the ones I've seen had 3 AAs in a holder, but I don't know if that was original. Later clones had a Ni-Cd soldered to the motherboard.

I've always wondered why Macs run down the PRAM battery so quickly. They rarely last more than a few year, but it's not unusual to see 15 year old PCs with the original CMOS battery still working.

 

[CuriousChord]

Wire with connectors and battery holder. If you want to be neat, use some double-stick foam tape to attach the holder to the case. Many older 286 and 386 mobos have a 4-pin header for off-board batteries. If you don't care about looks, put the batteries in a plastic bag to catch the goo if it should escape the cells.

How do you feed the wire through the case? Remove the cover on an adapter slot, or use any small opening at the back of the case? I was thinking to attach the battery holder to the base of the power supply with a velcro strip.

The 4-pin header sounds like the same external battery connector for which the B-220 was used. Or maybe battery type has nothing to do with the 4 pins, as long as the battery connects. Funny, I thought "external" meant "off the MB"... "outside the case" never occurred to me.

The first IBM to use a CMOS battery was the AT. All the ones I've seen had 3 AAs in a holder, but I don't know if that was original. Later clones had a Ni-Cd soldered to the motherboard.

I've always wondered why Macs run down the PRAM battery so quickly. They rarely last more than a few year, but it's not unusual to see 15 year old PCs with the original CMOS battery still working.

Does the PRAM battery on a Mac do more than an AT-style CMOS battery? Trying to keep all the RAMs and ROMs straight--PRAM, SRAM, DRAM, NVRAM, etc.--and wondering what a Mac and an AT had in common that was powered by a watch cell or AA's or a NiCad barrel soldered onto the board. At the very least is the RTC plus some kind of RAM, as a guess.

A Wikipedia excursion indicates that prior to a CMOS/PRAM battery being used, there was magnetic core media (the 60s), then solid state (the 70s), and into the 80s with PROM before the AT and Mac (mid-80s) introduced the PRAM/CMOS battery concept. Although this could be incorrect if the various components stored different things. The common denominator seems to be...what memory would the computer need after the power was turned off, and how was this retained?

Cheers! CC