Just saw the yt video, these are very nice 😀 kudos!

Thank you very much! Hope this will be helpful for the one or another.

Nice work, but it's worth mentioning the DS12887A is still being made in great numbers. I recently bought one and it came with a 2119 date code. I realize that it may not be compatible in all situations and the replaceable battery is nice, though.

Ok, thanks, good to know. I just bought in China three times, they were all dead. Then I was tired of it and got five in Germany from a big official distributor, sold as new for 10€ each. Turned out, that it was old stock as well. One was dead on arrival, one died after 6 months, the others after a year. May be there are some new indeed out there, which you can buy, but I just had no luck to find one. However, now I have enough of them in different variants 😁

tag! going to build one and try in a system that did NOT work with a DS12887+ but did work with an Odin OEC12C887 and see what it does

Yes, even if those modules are actually identical and many boards do work with both, there are some mainboards, which are picky about it. I think, it's something about timings. I even had once two Odin OEC12C887, where one worked in a mainboard and another one not. Both modules were ok and did work in another mainboard though. I have a feeling, that Odin used different RTC ICs in the same module version internally. Probably to keep the costs low, but the side effect was, that the timings were not quite inline. And some mainboards probably relied too much on particular timings.

Hello. I am sorry to resurrect this old thread to ask for support but it seemed to be the best way to do it? I have built one of your nwX287 adapters for my DataExpert EXP-8046 motherboard's dead DS12887. I have built to module as directed by the Github page, using a 32.768KHZ 6pF Crystal and a BQ3285S.

When I plug it into the motherboard, the board reports correctly that the CPU is a 486SX 25Mhz unlike the ??? 486 66MHz it incorrectly reported with the dead DS12887, so something is working. Furthermore, I no longer get a any post errors complaining about a flat CMOS battery, so that is also working. Unfortunately, the board still reports a "Real Time Clock Error" and when I go into the BIOS settings, time does not advance. What could be wrong here? Bad crystal?

Hi. The CPU (speed) detection routine in BIOS is probably relying on RTC clock on this board, which seems to not work. The reason could be a defective crystal, but also the chip could be bad. Unfortunately meanwhile a reliable sellers are hard to find. First of all pay attention to have the same pins as on the original, remove or bend outwards the pins, which shouldn't be used. Then you can check the crystal with an oscillator, but if that doesn't swing, you still will not know, if it's the crystal or the IC. As an experiment you can try to use a crystal from another board, may be you have something in scrap. And if nothing helps, you can still go with this solution:

https://youtu.be/ecTZtZhE9bI

Hello necroware, thank you for responding. I really appreciate it. I built a second adapter last night, and the behaviour was exactly the same..

Dallas chip dead battery: CPU not detected, CMOS settings not retained, clock works
nwX287: CPU detected, CMOS settings retained, clock is stuck

I am inclined to think that the crystals I got may be wrong or dead? The markings on them is very difficult to read but IIRC it was something like JC32Q3R or something so I believe they are 32Khz.. Should work, normally?

Well, may be faulty parts, you never know. As I said, you can try a crystal from scrap if you have something at hand. They are usually around dedicated RTC chip on older boards or near the chipset in the newer ones until today. Also you didn't mention the pins. Are they cut in the same way as on the original Dallas module? Some unused pins should remain floating, but could be pulled up or down by the mainboard if you leave them in place.

Hey scorp, popping by here to ask one last question.. I noticed an error in my diagnoses, previously I had reported that with your adapter the system clock does not advance. I was wrong. It advances. It advances VERY SLOWLY. LI go into BIOS and set the time 01:00:00 (and at this point it should start ticking along like 01:00:01, 01:00:02, etc. but stays as is) but if I do a cold reboot and to back into BIOS the time is now 01:00:03 even though it has been, probably, around 30 seconds since I saved and exited. I have checked to make sure that I have the right pins clipped, and that all the SMD component legs are connected to the right components. Does that ring any alarm bells for you?

I have built yet one more of the adapters, and it behaves exactly the same. To make sure that my issue is not board related, I modded a DS12887+ with a CR2032 and it works flawlessly: CMOS retains settings and the clock ticks on in BIOS.

I have checked my adapters and tBQ3285s gets power from the battery and retain CMOS settings. That leaves only the crystals as suspect in my mind, unless my board is somehow incompatible with your adapter in a strange way.. I think maybe the supplier may have sent me the wrong crystals? The ones I am using are marked J32CQ3R (3R is for the 3mm radius I believe, and J32 leads me to believe it SHOULD be a 32KHz crystal..) The one thing I can think of that may have gone wrong is, while soldering the crystal down to the Ground plane I may have used a bit too much heat (it did not solder down easily) which may have damaged it? I will order more 32.768KHz 6pf crystals and replace the ones I have on regardless and see if that helps..

Hi there. Yes, that could be bad or wrong crystal indeed. Overheating crystals can end up in them been damaged. They are quite sensitive. I also once got a bag of crystals from Aliexpress where every second one was defective. Since then I always buy those from LCSC. You don't need to solder the body if you want just try another one. As I mentioned previously you might want to salvage a crystal from scrap. Good luck.

Nice project! Do you know if this could work for Houston Tech HT12888A ?

Didn't test it, but those modules seem to use VIA chips internally, which are pin compatible. So most probably yes.

I used one of these in a PC Chips M530 board. I updated this thread with results:

www.vogons.org/viewtopic.php?t=73635

I believe the Dallas DS1287 is not fully compatible in this board. The other pictures I've found show a Houston Tech HT12888A being used. I saw a video from vswitchzero about a similar problem on a Gigabyte GA-586ATV Socket 7 motherboard, where the machine could boot with a Dallas RTC module but not save any settings. So I installed a necroware BQ3285 module and that fixed it. I can now save BIOS settings, date/time, etc.

It's because DS1287 is not the same as DS12887. The first is older and was used on 286, 386 and early 486 boards. For later boards you need DS12887 which has more memory and the BQ3285 is a clone of that. That's why it fits. You can usually use DS12887 in systems which need DS1287, but not the other way around.