Dallas Clock chips

[pentium]

We all hate them.
Those stupid dallas chips that were put in everything from SGI's to IBM computers.
My Model 30 greeted me the other day to a 161 and 163 error that could not be corrected and indicating that the dallas chip had gone south. Then today I managed to get hold of a handful of regular (but dead) dallas chips.
I then thought about the battery hack for sun computers and got an idea. Please refer to the photos below for even more detail.
I grabbed one unlucky subject and with my dremel set out to see what was under the hood. It took a while but I first found the crystal followed by the battery in the middle and then finally the battery terminals themselves.
Going with my volt meter I determined that the battery was indeed dead.
now knowing where the battery terminals were, I grabbed another chip (of a different model) and with the dremel dug out a cavity to the two battery terminals. A quick check with the meter indeed proved it was the battery terminals.

My theory is that if I were to figure out how to cut the battery leads (not soldered on like in a sun) and install a regular CMOS battery the chip would be as good as new.
Do you think it's possible? I can see at least four computers I own that would need this.

On another note. The powder made during the drilling closely resembles moon dust.

[linuxguru]

Great work - you're kinda persistent, aren't you?

Anyway, Lithium batteriess die into a low-impedance (almost short state), so you can't parallel another battery with a dead battery - you have to cut one of the leads (either +ve or -ve) very carefully, to isolate the dead battery before soldering another battery into the circuit. Perhaps a Dremel drill or cutting wheel could do the trick. Are both battery terminals passed through to the pins of the IC? That simplifies the new battery installation a bit - it can be anywhere in the system and connected by a pair of wires to the corresponding pads on the mobo.

[Maxxarcade]

I have a ton of machines that use the DS1225Y and DS1230Y chips. They are all dying, and they are not cheap.

Nothing worse than something that depends on a battery to work, and the battery is sealed. Wait until you can no longer get these chips at all, think of all the wasted electronics. Though I think someone would make a replacement circuit to plug in the socket, using discrete componenets and a standard battery.

[pentium]

I am incredibly persistent and I was also bored.

By the looks of it, They just bent two pins the opposite direction and...
Wait, that means they used solder.
Well in that case I just desolder the leads and wire up a new coin battery.
My SS 1+ is happy with a 3V cell. I wonder if it's the same here?

I think someone would make a replacement circuit to plug in the socket, using discrete componenets and a standard battery.

It won't work, the battery pins are not visible from the outside, you have to dig to find them.

[gonzo0815]

All my older 486 and 386 systems all are equipped with an external battery connector (at the jumper area).
So there are those really big external lithium battery to put in, after the build in DS battery had been failed.
So check this first, before going trough the hassle of replacing the chip or battery.

[Maxxarcade]

Originally Posted by pentium

It won't work, the battery pins are not visible from the outside, you have to dig to find them.

I was aware of that... I meant someone making a functional replacement for the whole chip, except not sealed in epoxy.

[Maxxarcade]

Some links on dallas chips:

Interesting info on SGI dallas chips
http://www.pimpworks.org/sgi/dallas.html

[pentium]

Yeah, I was also thinking of my Indy and Indigo when I started working on this project. I presume their batteries will be going south soon too.

All my older 486 and 386 systems all are equipped with an external battery connector

I doubt you will see one though in a Sun or SGI system or we wouldn't have this problem. My three IBM systems don't have these connectors.

making a functional replacement for the whole chip, except not sealed in epoxy.

First off it would be expensive (some of the Dallas chips are no longer made) and don't forget about that crystal. I could just see my way being a little easier.
Second, The battery seems to only be connected to the chip and lacks the pins needed to plug into the socket.

[linuxguru]

> The battery seems to only be connected to the chip and lacks the pins needed to plug into the socket.

This sucks - it means that you need to access both leads, cut one carefully and still have enough metal surface to be able to solder in two new leads. All of this without damaging the chip and the plastic too much.

Maybe the approach is to drill two holes - one to cut the battery lead, and the other just reaching the surface of the lead on the chip side. Fill a drop of solder into this hole to make contact with the lead as well as bring it to the surface of the package - the rest is just a matter of technique.

[pentium]

All it looks like is that they bent the pin upwards and not downwards like all the rest of the pins. It should be easy to do and I'll try later this afternoon.

[999999999]

I suspect you might be making this harder than it needs to be. Well, (LOL) after reading what I had written below it doesn't seem so easy either but once you get the general concept it is just a matter of determining which solution is best.

That some boards have an external connector for a battery pack, is an indication that at least some Dallas RTC can run from an external battery regardless of whether you have that connector or have to be more creative (soldering an external battery pack to a VBAT pin on the board and/or soldering it to VCC (and Gnd)adding isolating diodes and/or minimal charging circuit possiblity as mentioned below).

You did not mention the specific Dallas RTC part number. That is always the place to start so the datasheet can be examined. Taking one example, Dallas DS1285/12885/1287/12887/12C887

Datasheet

It clearly indicates on page 1, "A lithium coin-cell battery can be connected to the VBAT input pin on the DS12885 to maintain time and date operation when primary power is absent."

Some people used a 4 x AA battery holder and alkalines instead of a lithium. Suppose though that you have a Dallas RTC module without a VBAT pin, it does still have a VCC pin for 5V system rail power when system is on, and would switch from power by that pin to the internal battery when that VCC level falls below a certain threshold. I saw mention of the values 4.25V threshold, below which the data I/O is shut down, and 3V at which point an internal switch, switches to the integral battery. This leads to the presumption that if you just keep the VCC above 3V (that might be a typical value, it might vary slightly from one module to the next?) then it continues to run from the VCC input. The remaining question is the battery life running like this.

Suppose it consumes about 7mA, you won't get acceptible life from a 4 X AA alkaline pack BUT if you used an OR-ing diode arrangement (one diode serially connected from each supply to the VCC pin) to the Dallas chip's VCC from both the system 5V rail (having to cut the trace if such a diode is not yet present on the board and adding it) and from the battery pack, AND a diode from the 5V rail to the battery pack along with a current limiting resistor for a low mA trickle charge rate, THEN you can use 4 x rechargable ~ 1.2V NiMH or NiCad batteries and they will recharge while system is on, and discharge powering the Dallas RTC when system is off... the key being that if you left the system off long enough to completely drain the rechargable battery pack, you have to set the clock again.

On the other hand, the standby current consumption is bound to be lower, so if your battery pack voltage is below the threshold for the RTC to drop into standby mode, it "might" get an acceptible life from 4 x AA alkaline (non-rechargable) then of course you would not need or want the diode plus resistor trickle charging the pack. You could try it and see, I can't speculate much more about this, going off on a tangent before knowing which Dallas RTC(s) you have.

[999999999]

If I didn't make it clear in my last post, that the internal battery contacts are not directly wired to an externally accessible pin soldered to the motherboard, is not necessarily an indication the module can't take an external power supply (battery pack) and switch between which battery input it uses. This would also remove the problem of whether the original battery had died shorted since switching instead of a common power connection won't be shorting the input power.

[pentium]

I still think it sounds easier to just disconnect the old battery and solder in a new one. I have yet to see a PS/2 or RISC system that uses a dallas chip but laso has a connector for an external battery.

[999999999]

Well... now that you have

1) Desoldered the Dallas RTC module.
2) Dremeled it away, accepting that a certain % of them may be damaged doing so.
3) Probed and now will seek a specific new battery of the right size.

Then yes, but do not solder in a battery. The heat will be damaging, it should be spot-welded to the contacts.

Here is the short and simple condensed version of my prior post:

1) Get the datasheet for your specific Dallas RTC
2) See if it has the VBat pin. If it does, all you have to do is solder a wire to a 4 x AA battery pack, or a Lithium cell, and negative lead to any ground.
3) If it does not have VBAT, it only takes a few minutes to solder a couple diodes to create an ORing connection to VCC.

This is how is is done on a board with rechargable battery on it (was NiCad back in those days), or how it is done with a board having the battery add-on pack pins. Any and all methods will work, but desoldering the whole Dallas RTC module then dremeling it out seems pretty extreme and unnecessary, IMO.

[pentium]

I don't think you understood me.
After dinner I'll mock-up what I am talking about. I see it's hard to understand.

[999999999]

How could it seem easier to do all this work and higher risk desoldering and dremeling instead of just looking on the datasheet for the VBAT pin (which it probably has) and soldering the battery lead to that?

Also, some of the Dallas/Maxim RTC modules aren't very expensive, here's a popular one that only costs $9.40, but you still haven't mentioned which model yours is, so we can't be sure this is even the same layout as any other Dallas module, I mean the internal positioning of the parts (relating to where to dremel it), not just the pinout.

[pentium]

Also, some of the Dallas/Maxim RTC modules aren't very expensive, here's a popular one that only costs $9.40, but you still haven't mentioned which model yours is, so we can't be sure this is even the same layout as any other Dallas module, I mean the internal positioning of the parts (relating to where to dremel it), not just the pinout.

This page will sum it up for you. They also use the DS1287 which is the same chip I use.

Yes, Dallas chips can be cheap but that is for newer electronics (in most cases). Older equipment that used dallas chips are different. Either they no longer produce a replacement or the replacement is expensive ($20+)
You do however say that this might not work with other chips and you might be right. In the case of my Onyx, the chip is larger. The only way to find the battery terminals there would be to either:

a)look up a datasheet and hope you find pins for external power
b)expose the pins that are bent upwards and into the epoxy-filled case and look for two pins that have a voltage and THEN perform the battery hack

Also, have you ever tried to easily get an older datasheet? (I have a feeling you will say yes.)
If you are lucky you will find it in a sea of wrong links (they mention the chip but no datasheet) and if you are even luckier you will be able to read it for no cost (many sites have it but you need to pay to view/download it).

By the way, I finished my hack (which it seems has already been done. Rats.)
I didn't find a way to desolder the battery so I used a disc attachment for my dremel to cut the lead in half and carry on from there.

[999999999]

The DS12887 I linked for $9.40 is "usually" a pin compatible replacement. I would see dremeling into a RTC as a last resort.

[gonzo0815]

I think the suggestion with a diode decoupled battery feed via power pins isn`t that complicated, if the chip have an external power pin (not external battery pin).
This would be the far less messiest solution compared with dremeling etc.
Any way, if it is working, don`t bother.

[pentium]

You know what, let's just forget about this.
We are fighting too much over what's the best fix.