Are you sure the old barrel battery was 3.6v ?
I ran across an odd one that was 9v once. (Yes, on a motherboard.)
.

The motherboard manual states 3.6 V for the rechargeable. I was not aware that the battery was even needed for the seconds to count up correctly when the system is powered. The manual states 6 V if using an external battery (on different pins, and requires a jumper to be moved), however, I am using the rechargeable pins. The leads from the 6 V run thorugh some diodes which drop the voltage down, probably to 4.6 V if going through 2 diodes. Perhaps the 3.6 V is a typo? But still, I would not expect the date counter to be off by a factor of 5 when the power is up.

I'd like to know if you make progress with this. My Opti-based 386 board does weird things with the RTC where it ticks intermittently. It actaully keeps time fine, but it ticks at irregular intervals like;
1,2...345.6.7,89. and onward. Never figured out why and never worried about it, but it does sit in the back of my mind as a potential problem. It is a very strange behavior for sure. This board was modified before I got it and now uses a coin battery, there is still evidence of the original barrel battery having leaked and started eating things, though it was all cleaned up and won't ruin anything anymore.

I have a theory as to what might be going on with my system. I could not find any evidence that the battery is connected to any pin on the 85C206, neither via direct contact (a PCB trace), nor through a resistor or diode. It may have been an inner-layer trace which got severed by the battery acid leak. But why does the system keep its CMOS settings. It is possible that the 85C206 uses some type of FLASH ROM in its chipset or some other type of low leakage internal capacitor to store the CMOS settings. I know my AMI Mark V Baby Screamer board (VLSI-based) does not need any battery on the board to maintain clock time or CMOS settings. The Chaintech (SiS Rabbit) motherboard manual does mention that the 85C206 contains 14 bytes of clock and control registers and 50 bytes of general purpose RAM and that it uses CMOS technology.

feipoa wrote:

The motherboard manual states 3.6 V for the rechargeable. I was not aware that the battery was even needed for the seconds to count up correctly when the system is powered. The manual states 6 V if using an external battery (on different pins, and requires a jumper to be moved), however, I am using the rechargeable pins. The leads from the 6 V run thorugh some diodes which drop the voltage down, probably to 4.6 V if going through 2 diodes. Perhaps the 3.6 V is a typo? But still, I would not expect the date counter to be off by a factor of 5 when the power is up.

The battery powers an oscillator which is the timer circuit for the clock.
If the supplied voltage is too low the oscillator can run slow because the cap(s) in the circuit don't charge as fast as they should.
(In that circuit the caps would be very small. Probably SMD and maybe even inside an IC package.)
.
The oscillators on motherboards usually use a quartz crystal and, although rare, they sometimes go bad by drifting off their intended frequency.
That will also slow your clock.
.
The CMOS Memory is most likely powered through a different path from the battery than the oscillator.
.
Most boards that appear to have no battery actually do but it's located inside the RTC package where you can't see it.
Boards that have that -and- battery provisions are intended to give you a way to work around a dead RTC battery.
.
.
"Baby Screamer" - What a freakin' COOL name. I like it!
.

And after I wrote all that I realized you probably know already.

The AMI Mark V Baby Screamer board does not contain a dedicated RTC module, like the common ODIN/Dallas DS12887, nor does it contain a battery anywhere visible on the motherboard. Nor do I need to use an external battery for the system to keep its time or CMOS settings. All I can find in the manual concerning this is "The Mark V has 128 bytes of battery-backed CMOS RAM to store system configuration information." There is no mention of where that battery is, but I suspect it is inside the VLSI chipset. Anyway, that motherboard is a different story.

I also thought the name was pretty neat, however I was informed by another member that the origin of the name is not what I thought it to be. I originally thought that some motherboard designer with a newborn screaming baby named the motherboard, however I was informed that the larger brother to this board is the Mark V Screamer, and the "Baby" refers to the motherboard's footprint.

Back to the Chaintech 340SCD, aka "SiS Rabbit". There is some small crystal oscillator outside the 85C206 chipset, which houses the real-time clock. There are some ceramic caps in this vicinity as well. Are you telling me that the motherboard's battery should connect to this crystal for it to function with the date/clock correctly? Is this crystal just for the date clock? Let me type out what the manual says in this regard.

TIMERS
The 85C206 chip provides three programmable timers, each with the same timing frequency of 1.19 MHz.

Timer 0 - The output of this timer is tied to the interrupt request 0 (IRQ0).
Timer 1 - This timer is used to trigger memory refresh cycles.
Timer 2 - This timer provides the speaker tone. Application programs can load different counts into this timer to generate various sound frequencies.

REAL-TIME CLOCK AND NONVOLATILE RAM
The 85C206 chip contains a real-time clock omponent that maintains date and time information, in addition to storing configuration information about the computer system. It contains 14 bytes of clock and control registers and 50 bytes of general prupose RAM. Because of the use of CMOS technology, it consumes very little power and can be maintained for long periods of time with an inexpensive battery (one rechargeable battery and one 6V external battery connector are supported on the main board).

I am attaching a photo of the SiS 85C206 and surrounding circuitry. So that little can oscillator should be connected to the battery? Click image for larger view.

Also attached are images of the past damage done by the leaking barrel battery. A nasty trace repair job. I could not determine where the positive lead from the battery is supposed to go. The can oscillator?

feipoa wrote:

Are you telling me that the motherboard's battery should connect to this crystal for it to function with the date/clock correctly? Is this crystal just for the date clock?

Maybe and probably not.
Since I can't see the thing to trace circuits and don't have any datasheets or schematics for those chips I can't say anything specific to that board.
What I'm telling you is general in nature. Methods I've seen used.

They use quartz crystals for their high accuracy and long life.
They are persistent (resilient) enough to keep the frequency in spec even when related oscillator components drift a little with age.

One crystal can be used for multiple circuits and by way of additional circuitry one crystal can provide more than one frequency. (Although it's better to use more than one crystal.)
It's not usual to control BUS and CPU frequencies with crystals.
The next part you wrote is a good example. The 85C206 times for three different circuits.

The date-time clock has to keep time even with the board powered down so -if- one of the functions of that crystal is the date-time clock then there should be some kind of voltage on it -BUT- the crystal is not where Vcc enters the oscillator circuit so it won't be battery voltage or even DC voltage. The crystal should be oscillating. That isn't DC. Via related oscillator components the end output is probably a square wave.

I could be wrong but I don't think they would have put a quartz crystal inside the 85C206 chip.
An oscillator right next to memory just seems like a bad idea and quartz crystals are rather big to be inside an IC package.
I suspect the crystal you see supports the 85C206 by supplying it with the square wave for the purpose of timing other things.

I hope that made sense.

feipoa wrote:

Let me type out what the manual says in this regard. […]
Show full quote

Let me type out what the manual says in this regard.

TIMERS
The 85C206 chip provides three programmable timers, each with the same timing frequency of 1.19 MHz.

Timer 0 - The output of this timer is tied to the interrupt request 0 (IRQ0).
Timer 1 - This timer is used to trigger memory refresh cycles.
Timer 2 - This timer provides the speaker tone. Application programs can load different counts into this timer to generate various sound frequencies.

REAL-TIME CLOCK AND NONVOLATILE RAM
The 85C206 chip contains a real-time clock omponent that maintains date and time information, in addition to storing configuration information about the computer system. It contains 14 bytes of clock and control registers and 50 bytes of general prupose RAM. Because of the use of CMOS technology, it consumes very little power and can be maintained for long periods of time with an inexpensive battery (one rechargeable battery and one 6V external battery connector are supported on the main board).

Do you have the actual datasheet for the 85C206?

Yes, pictures help bunches. Did not see yet before my last post.
Yes, in the first pic that looks like an oscillator presumably there to support the 85C206.

feipoa wrote:

I am attaching a photo of the SiS 85C206 and surrounding circuitry. So that little can oscillator should be connected to the battery?

As I said in the previous post that canned device is just the crystal, not the whole oscillator.
Some of those nearby caps, inductors and resisters are also parts of the oscillator..
You would have to trace out the circuit to find where the battery voltage inputs to the oscillator.
Is not likely at the crystal itself.
.
.
I went looking for pics of the board between posts and it appears there are two other crystals on the board in the 'sardine can' style case but those are more likely for something other than CMOS.
Your pics are better than those I found insofar as the part of the board I needed to see.
.

Real Time Clock crystals are typically 32.768 KHz if that is of any help to you. Chips like those Odin ones (Basically like the Dallas things but without an internal battery) require one of these across two of their terminals.

I'm no good with crystal oscillators as I've only ever worked with modern clock generators outside of radio, generally single chip or a chip with a simple oscillator I set up by following diagrams in the datasheet - and I never got my head around those either. Think I might investigate my own tomorrow though, if I have time, supposed to start packing given I move on Monday.

I have never been able to locate the datasheet for the 206. I did some retracing and the battery goes to a standard rectifier diode, which then splits off to a few parallel destinations - pins on the 85C206, as Vdd to the HEX inverter (MC14069UBCP), and to that resistor which is adjacent to the crystal can and parallel with it. Said resistor then goes to GND. Based on the colour code, the resistor is 1 K-ohm. Instead of this 3 V coin cell battery (which is brand new and measures 3.3 V), I connected 4.5 V and 6.1 V of battery up, but the date seconds still counts very slow. Even with no battery connected, the seconds counting is overly slow.

As I have now traced the battery leads to the 206 chipset, it seems more plausible that either the crystal can is defective or the RTC module inside the 206 is defective.

Did you try to disconnect mains, then disconnect battery, and wait for the remaining charge in capacitors to decay (or use a clear CMOS jumper if there is one), and power the board up without battery, only with PSU?

What I see that could go wrong here is that the acid has corroded something important, or the extra wiring has some problems, or the circuitry on motherboard is just not meant to work with coin cell battery.

Because the coin cell battery has relatively high internal resistance and lower voltage compared to the original battery, it may cause the onboard backup voltage to be too low or rise too slowly when battery is connected, so the oscillator circuit does not start or work as intended. (I've seen a RTC crystal resonating much faster at some integer multiple overtone mode, until you touched it with a finger so it stopped, and removing the finger made it start oscillating at the rated frequency).

The MC14069 is used as oscillator for the RTC. For the schematic check some of these: https://www.google.de/search?q=4069+crystal+o … o=u&source=univ

Its output goes to the 85C206. From the battery leak some of the parts may have internal damage. The crystal, the MC14069, the resistors or the capacitor C9. Or there may be a broken trace in the oscillator cirquit, or some conductive 'juice' is still on the board (i.e. between the capacitors pins) causing current flow where it shouldn't. To get rid of the latter you should bath the board in some sort of acid.

Thanks for your additional comments. Some replies below.

Jepael - yes I have tried to clear the CMOS, I also tried using a different BIOS chip, reflashing the BIOS, powering up without battery. I have also tried using the EXT BATT connector with three 1.5 V AA batteries (4.5 V) and four 1.5 V AA batteries (6V). All circumstances still indicate a very slow seconds counter. Unfortunately, I do not possess a rechargeable battery to test in its place. Come to think of it, I think I have a rechargeable battery in my Mac SE/30, but I'd need to canabolise that system.

h-a-l-9000 - I believe what you suggest is in most likely culprit. That whole area around the crystal can and the MC14069 was blue with battery acid. The board has already been washed with acetic acid and ISO a few times. So you think that some of the battery acid made its way inside either the discrete components or the MC14069 inverter IC? If that is the case, then there could be quite a few inner-layer traces which have been severed as well. I guess I will need to start probing the board with my oscilliscope, although it might just be easier to replace all those components. I wish I knew the 206 pin assignments. My worst fear is a defective 206 chipset.

I have gone over my trace fixes about a dozen times over the past 2 years. I know the repair work looks ugly, but I do not believe there are any errors with it. If there are disconnects, it is inner-layer.

If it has enough time the stuff creeps along the pins, under the plastic seals, into the components.

With some luck the 82C206 has a similar pinout:
http://www.elenota.pl/datasheet-pdf/60708/others/82C206

Pin 72 would be the clock input for the RTC. This should go to some output pin on the 4069. A resistor might be in between.

I'd start with the easier to replace parts, crystal, resistor, capacitor. And always make sure there are no invisible breaks in the traces. The battery fluid also eats its way under the solder finish of the PCB, causing breaks in the traces directly at the pads.

I'm not sure what the changes were between the 82C and 85C, but at least the number of pins are the same. On page 40 of that 82C206 PDF, there is this "Power Conversion and Reset Circuitry" diagram. I believe this, or part thereof, is implemented on this motherboard. From what I measured yesturday, I have that 1K resistor going through a small cap to GND. Where you see the 6 V battery node, the "rechargeable" battery's positive terminal gets connected between those two series diodes, meaning the coin cell battery has about 0.7 V less of voltage drop compared to the EXT BATT. So I'm not sure why the manual calls for a 3.6 V BATT for the rechargeable and 6 V for the external battery. I will try to probe around this motherboard this evening if my kids/wife let me.

Thanks to everyone for their comments. I solved the problem. I ultimately traced out the oscillator circuit and probed it with an oscilliscope. The square wave was at about 5 KHz instead of 32.768 KHz. I replaced the crystal and now the seconds count up correctly.

By the way, that power conversion and reset circuity diagram is implemted on this motherboard exactly as shown. The oscillator circuit uses a 27 pF capacitor in series with 2 inverters in a loop. It then branches off to a 3rd inverter before heading to pin 72 (OSC1) of the 85C206.

The power-up amplitude of the oscillator circuit is 5 V, while the power-down amplitude is 3.3 V, which is the current emf of the coin cell battery. Hopefully a single coin cell is sufficient, otherwise I'll need to stack two in series.

You mentioned 1.19 MHz earlier. - And you have an O'scope.
Presumably the 1.19 MHz comes from the oscillator.
Check and see if the 1.19 MHz there. If it is then none of the oscillator parts are the problem.
.