Replacing battery in Dallas DS1230Y-150?

[GearTechWolf]

It's a nonvolatile SRAM module off this oddball Taiwanese ATI/AMD ISA VGA card I unearthed from my shed today.
Doing a search with the GPU model netted a few similar cards, including the top card-edge connection, but no exact matches.
It has three AM27C256 EPROMS, three GAL chips, an AMD N80535, an an INMOS9127-F/IMSG176P-66G.
Model(?) text on one end reads: "S LCD BD 91406 48.51112.000-SB"
Sticker on back reads: 55.51109.0012030062A M
Here's some pics:
Any information on the card would be appreciated too!

 

[sorphin]

You can still buy (and sample) Dallas DS1230Ys...

​

 

[GearTechWolf]

You can still buy (and sample) Dallas DS1230Ys...

​

Sure, but there are a few reasons I'm not interested in doing that.
First, where's the fun in that? Second, buying costs money and I don't feel like supplying the data required to sample.
Third, and perhaps most important, odds that I'll get a module with a good battery are not great.
After all, how do I tell if these are recently made units or NOS ones with pre-failed batteries?
I'll attempt to de-cap and repair the one I have eventually, if nobody chimes in with their own attempts at it before then.
Still, thanks for the heads-up! I hadn't bothered to check, aside from incidental results when hunting a data-sheet.

 

[Plasma]

It looks like a custom ATI VGA Wonder that can drive an LCD panel (34-pin connector at the top). The SRAM is probably for saving the LCD panel configuration, and not necessary for VGA output.

 

[GearTechWolf]

Oh! So it'd be for an integrated display with optional VGA output? Cool!
Finding a driver would probably be tricky, but I should test the card before I worry about that too much.
Unfortunately I don't have an ISA-capable machine to plug it into, as my Win98se machine has developed power-rail issues.

 

[Chuck(G)]

I assume that you've seen this.

 

[Plasma]

Oh! So it'd be for an integrated display with optional VGA output? Cool!
Finding a driver would probably be tricky, but I should test the card before I worry about that too much.
Unfortunately I don't have an ISA-capable machine to plug it into, as my Win98se machine has developed power-rail issues.

Yes, it's from ye olde days when LCD panels sucked, so they weren't used unless absolutely necessary. There were no standalone LCD "monitors" with VGA connectors back then.

ATI VGA Wonder+ drivers should work, assuming the BIOS hasn't been changed too much.

 

[GearTechWolf]

I assume that you've seen this.

Oh wow, thank you! No, I hadn't. That's a great option! I'll still probably try fixing the old one, especially now that I know that exists.

Yes, it's from ye olde days when LCD panels sucked, so they weren't used unless absolutely necessary. There were no standalone LCD "monitors" with VGA connectors back then.

ATI VGA Wonder+ drivers should work, assuming the BIOS hasn't been changed too much.

Ah, that makes sense. Whenever I get it set up to where I can test it, I'll give those a try, thanks!

 

[sorphin]

Sure, but there are a few reasons I'm not interested in doing that.
First, where's the fun in that? Second, buying costs money and I don't feel like supplying the data required to sample.
Third, and perhaps most important, odds that I'll get a module with a good battery are not great.
After all, how do I tell if these are recently made units or NOS ones with pre-failed batteries?
I'll attempt to de-cap and repair the one I have eventually, if nobody chimes in with their own attempts at it before then.
Still, thanks for the heads-up! I hadn't bothered to check, aside from incidental results when hunting a data-sheet.

whatever floats your boat. Mine came directly from the mfg. Made the same year. re: fun.. i don't consider dremeling chips fun, but to each their own. You can tell by the date code fyi.

 

[GearTechWolf]

Module after decapping/skinning the outer shell off, for reference. (tried to keep the top section with the labeling intact but it was just too stuck to the potting within, despite the potting having been formed in a separate vessel and not poured into the casing. How can I tell? Markings cast into the potting that are not present on the inside of the casing!)

Okay, managed to cut into the module by hammering a razor-blade into it about a battery-thickness from the top, as helpfully indicated by one battery showing through the potting at one end.

It cracked apart where the batteries lay atop the SRAM chip* and I managed to disconnect three of the battery leads in such a way as to leave a leg to solder holders to, the fourth I'll have to dig out** the pin/leg a little.
*(which I think may be surface-mount, as you can see the edges of the circuit-board showing through the potting)
**(might do some very careful dremel-work for this, as I haven't needed it for anything else on this chip thus far)
Can actually see a little bit of the markings on the chip now! Kinda want to expose more, but don't want to damage the chip.

 

[GearTechWolf]

No dremel needed for the above chip after all, thankfully. Was able to chip out, expose, and lift the fourth batter pin with the razor-blade.
But I did use my dremel to disassemble this Dallas Smartsocket (with its surface-mounted DIP DS1213 chip) and extract its two batteries.
I'd already previously removed the socket from the top and the pins from the bottom in previous attempts, which didn't actually help.
So no pictures of mine intact, but those can be easily found online, I think.
I first hammered a razor-blade in at each end to start the potting separating from the circuit-board, then I cut the embedded pins along the sides with my dremel and used small screwdrivers to lever the potting away from the board.

In prying the batteries away from the board I did pull out a couple plated vias, but those points only attached to one side of the board anyhow.
The non-battery connections the eagle-eyed among you might have spotted perplexed me at first.
I had to carefully dissect the epoxy around these tiny pins to trace their paths and eventually figured out that they are redirected pins!
While most of the pins of whatever ram chip you couple with the Smartsocket simply pass straight through, a few of them are routed to the DS1213 and the batteries before returning to the pin exiting the bottom of the Smartsocket into the board you plug it into.
I'll have to replicate these redirections when I attempt to rebuild the board with new pins and a new socket.
But hey, managed to save the board and chip! (helped that the chip was on the top inside the hollow of the socket instead of being potted)
While I think I can rebuild it with what I've documented during disassembly*, I wouldn't say no to a schematic if anyone has one!
*(the pins didn't divert far and went in a straight line to the nearest point on the board, I'm almost certain)
Anyhow, that's all the Dallas modules I have to break down for refurb, so that's it for now!
Until next time, keep on fixing, bodging, hacking, and bringing these old things back to life!

 

[GearTechWolf]

Oh, small note, both units had Japanese Panasonic batteries in them.
Two little BR1225's in the Smartsocket and a pair of CR1632 for the DS1230.
Don't know what the DS1287A's usually have in them, haven't cracked one of mine open yet.

 

[Hugo Holden]

I assume that you've seen this.

I started the "craze" of replacing Dallas BBNVRAMs (battery backed up non volatile RAM) with Ramtron FRAM (ferroelectric RAM).

It had been suggested of course, before I did it, but oddly some theoretical problems related to the fast read modes, prevented anyone from actually trying it. And, because of that, it was said on a couple of Forums, that it could not work.

Not many know but the Ramtron company and FRAM technology started out here in Australia. They moved to Colorado Springs (Nikola Tesla's old stomping ground).

The one I did was to replaced the Dallas DS1225 with the FM16w08 (or the FM18w08 using a page select switch).

I published the article in 2013:

https://www.worldphaco.com/uploads/TEKTRONIX_2465b_OSCILLOSCOPE_CALIBRATION___REPOWERING_THE_DS1225.pdf

It became obvious that FRAM was an ideal substitute for the Dallas battery NV RAMs. So the idea caught on.

The SOIC package versions of FRAM, such as the FM16w08 and FM18w08 were ideal for fitting to SOIC to DIL adapter to replace Dallas BBNVRAMs.

It turned out after my investigations that common programmers like the GQ-4x, could program the Ramtron FRAMs if they were set for the Dallas parts. This was very helpful. MCUmall, the makers of the GQ-4x, added FRAM to their supported device list.

Also in the original article I wrote back in 2013, it explained how, even when the Dallas rams appeared to have failed, or stopped working, as the batteries went flat, this happened at a point where the data in them was still intact. And that data could be retrieved by adding a support battery. This occurs because the management IC inside the module, as the battery goes flat, disables the function of the IC, before the data is lost. That is explained in the article.

One other oddity is that, if a battery in the Dallas RAM is on its last legs, it is a bad idea to attempt to read the IC's contents in a reader. Unless you support that with an external added battery. If not , with each read the data becomes progressively corrupted.

Another option as mentioned in the article is Auto-store or Shadow RAM. It does work, but requires some support components, specifically a capacitor to store some energy. At power down the data in the main Sram memory is transferred to the shadow non volatile memory (called a Quantum Trap with a 10^6 cycle endurance) enabled by the charge stored in the 100uF capacitor. At “turn on” the shadow memory writes back to the main Sram memory & restores it. When I experimented with these though, I found that they were not resilient to rapid power cycling, so for this reason I preferred the FRAM.

 

[Chuck(G)]

Way back in the 200x years, I suggested FRAM as a replacement for various UVEPROM chips. I think TI owns the technology now and incorporates it into their MSP430 MCU line.

Another such technology is MRAM, still offered by Everspin.

 

[Hugo Holden]

Way back in the 200x years, I suggested FRAM as a replacement for various UVEPROM chips. I think TI owns the technology now and incorporates it into their MSP430 MCU line.

Another such technology is MRAM, still offered by Everspin.

Yes I mentioned the MRAM by Everspin in my 2013 article with these remarks:

C) THE MAGNETORESISTIVE RAM: Another type of non volatile ram is Magneto-Resistive, or Mrams. Cypress were developing this technology around 2004 to 2005. They produced a preliminary data sheet on a CY9C6264 and the pdf datasheet for this item circulated the internet, often taken up by the search engines “Stocking Parts Supply Companies” some of whom claim to actually have it in stock. It was a byte-wide 8k x 8 non volatile memory IC in a SOIC 28 package. However it appears they were never mass produced or commercially shipped to suppliers at all because Cypress sold off their Mram technology. It seems that this IC would have been just as good as the FM16w08, if not even better if that were possible. Everspin took up manufacturing of Mrams, but unfortunately they do not appear to make an 8k x 8 product currently and their products are generally for 3.3V applications too.

I have come to call the type number CY9C6264 "The Probe". It has a very interesting use. If you are looking at a parts supplier that claims to have this in their stock, they are not a genuine stocking supplier. The are a company that attempts to source an IC from elsewhere after a customer requests it. And their IC stock lists are fake, acquired only from internet data searches. Because the physical part does not exist, despite its data sheet being on the internet.

 

[Chuck(G)]

In fact, MRAM and FRAM are probably best considered to be niche products, the function being taken over by flash. In fact, there have been several technologies that have briefly appeared and been scrapped. I'm thinking of Intel's phase-change Optane. The "next big thing" appears to be STT MRAM, but it hasn't made much of a splash. HP gave up on memristor technology.

I wonder what the fate of Cypress MRAM devices will be now that they're part of Infineon.

Back in the 1980s, I recall working with those little Xicor X2444 8-pin DIP chips with what, 256 bits of NOVRAM. Basically SRAM overlaying flash. Very handy for storing limited bits of information and present as writable configuration ROM on many NICs.

Here's an interesting tidbit about a change replacing the Dallas NVRAM module with a plain old EEPROM: http://www.repeater-builder.com/motorola/maxtrac/replacing-dallas-nvram.html

Depending on the application, an EEPROM may be good enough.

 

[GearTechWolf]

Well, I'll have to try reading my chips once I attach batteries to them then! Be neat if I could salvage data from them.
Any idea what settings/reader I should use? I have an old parallel-port BPM EP-1132 EPROM programmer and the WinXP program for it.