Good day everyone,
A friend gave this game to me, said "It won't keep save files". I tested it, it seemed to save with no issue. Borrowed it, played it for a little over a week when it eventually, lost my save.
Replaced the battery*, started playing again. got 2 weeks in, again; save file gone.
I have basic knowledge of using a DVM and can solder, But have no idea where to go now.
Maybe i'm not using the right keywords, but I can't find any help with google.
Can anyone give some insight to this issue?
I can't replicate the problem. Tried taking the cart out and leaving it out. leaving the cart in, unplugging the system and draining the power & taken the cart out while i drained power on the system. I have even tried flicking the power on and off in rapid succession to attempt a save loss and still nothing.
I thought the battery was just used for making the save, not holding it (it kept a save file when i replaced the battery, which i assume supported this)
It would help if anyone knew how the save system actually works on the hardware level.
My 1st post in a forum, im a long time collector but fairly new to development (have made a few basic nes repros)
Hope I got this topic in the right section and thanks in advance!
Many Thanks
Ehks
ehks wrote:
I thought the battery was just used for making the save, not holding it (it kept a save file when i replaced the battery, which i assume supported this)
The battery is responsible for holding it, and has almost nothing to do with making it. That the save remained when you replaced the battery is fairly surprising.
It's worth finding out whether the RAM is entirely losing its contents, or if it's instead a few random bytes lost (causing the game to reject the save)
Is this an original unmodified game? Or some repro conversion of an original board for a translation or something? What mapper is it? Perhaps just a picture of the PCB would answer my questions.
Use the official super nintendo cleaning kit on all the consoles that this game is going into, and clean the edge contacts of all game cartridges that are going to be plugged into these consoles.
If you get quality cotton swabs, you can use them with isopropyl alcohol to clean the metal contacts on the cartridge edges. Keep spinning the swab as needed to keep exposing the surfaces to new cotton, then just toss. don't dare put a swab directly into the connector of the snes though as you'll be picking out shredded cotton. Even on the cart, you might need to pull off the occasional cotton thread, but if you keep it damp (not drenched) and move mostly parallel to the contacts in a kind of zig-zag manner (don't just drag across horizontally) it's unlikely to do that.
Isopropyl alcohol dries out pretty quick, so just be a little patient to let it dry out before inserting the cartridge and powering on again. isopropyl alcohol is mostly gentle on ABS plastic, although I've seen it draw out that fire retardant in the especially browned areas to make a kind of sticky film. Isopropyl alcohol is also compatible with common fiberglass circuit boards, and is actually really great for removing the rosin flux residue that migrates all over the place on these aging boards. (I didn't believe the flux migration issue until I've actually started taking these old systems apart... wow that stuff travels far).
Anyways, before you try and come up with crazy mechanisms, just rule out the obvious (dirty contacts).
Also, if you taped the battery in, and didn't solder it down, or didn't solder in a proper battery holder resistant to vibration, then obviously it's going to fail.
Yes, the battery is needed to hold the data.
If the SNES is powered-on, then the battery isn't needed at all.
Even without battery (or empty battery), SRAM can hold the data for a while (around 30 seconds, or maybe even a few minutes).
The battery should have 3 Volts, if it's only having 1-2 Volts then something is wrong... quite possible that this could result in slow data loss after some days.
Best use a multimeter to verify that the battery has 3Volts, best at a time when the cartridge wasn't powered by the SNES for some days (the battery isn't rechargeable - but I think it might still get "refreshed" a little after it was assisted by the SNES power supply).
And be sure that you got the polarity (+ and -) the right way :-)
nocash wrote:
Even without battery (or empty battery), SRAM can hold the data for a while (around 30 seconds, or maybe even a few minutes).
Small technicality, it's the capacitor that's maintaining voltage on the SRAM long enough to maintain data for ~30sec. The length of time is a function of the capacitor size, and standby current of the SRAM. If the supply voltage drops well below the Vdr (Vcc data retention) the SRAM will loose it's data instantly.
nocash wrote:
Best use a multimeter to verify that the battery has 3Volts, best at a time when the cartridge wasn't powered by the SNES for some days (the battery isn't rechargeable - but I think it might still get "refreshed" a little after it was assisted by the SNES power supply).
From what I understand, testing a battery by measuring directly across its terminals with a multimeter doesn't give an accurate depiction of the battery's charge level. You need to add a load to get an accurate measurement. Basically, that just means placing a resistor (fairly large, in the 10K-1Mohm range, small resistance = high current = heat = bad) between the terminals and measuring across the resistor.
I hope this post looks right, this is my 1st reply
Quote:
It's worth finding out whether the RAM is entirely losing its contents, or if it's instead a few random bytes lost (causing the game to reject the save)
The game plays fine other wise, i would think a few bad bytes would be more likely then the chip going bad, its been this way for a while and im still unsure as to why its throwing its file
Quote:
Is this an original unmodified game? Or some repro conversion of an original board for a translation or something? What mapper is it? Perhaps just a picture of the PCB would answer my questions.
It is an original game, no idea what mapper it is (unless they are marked by the mapper chips?) im attaching a pic

Quote:
Even without battery (or empty battery), SRAM can hold the data for a while (around 30 seconds, or maybe even a few minutes).
The battery should have 3 Volts, if it's only having 1-2 Volts then something is wrong... quite possible that this could result in slow data loss after some days.
the battery i took off actually registered 3.1, so its not the battery (it was stupid not to test it on the game, im aware of my folly lol)
Quote:
Small technicality, it's the capacitor that's maintaining voltage on the SRAM long enough to maintain data for ~30sec. The length of time is a function of the capacitor size, and standby current of the SRAM. If the supply voltage drops well below the Vdr (Vcc data retention) the SRAM will loose it's data instantly.
this is what i assumed it was after the battery was replaced, but unsure where to start to even find out what is misfiring/dieing
forgot to add the pic

ehks wrote:
Quote:
It's worth finding out whether the RAM is entirely losing its contents, or if it's instead a few random bytes lost (causing the game to reject the save)
The game plays fine other wise, i would think a few bad bytes would be more likely then the chip going bad, its been this way for a while and im still unsure as to why its throwing its file
SRAMs don't "go bad"; a few bad bytes are a sign of the bad-writes memory protection circuit failing. Whereas the entire SRAM losing its contents would implicate the battery and switchover. Hence why I asked which. If the game provides multiple saves, and ever only one of them disappears, it's probably the former. If they only ever all disappear at the same time, it's probably the latter.
Quote:
SRAMs don't "go bad"; a few bad bytes are a sign of the bad-writes memory protection circuit failing. Whereas the entire SRAM losing its contents would implicate the battery and switchover. Hence why I asked which. If the game provides multiple saves, and ever only one of them disappears, it's probably the former. If they only ever all disappear at the same time, it's probably the latter.
I have some 6264's here, thats what i see on the board, is that not the SRAM?
The MS6264CLL chip is indeed SRAM (specifically 8Kx8 CMOS 28-pin DIP made by Mosel Vitelic Corp).
FYI -- not sure if it matters, but one of your cartridge edge connectors is damaged/scratched quite badly -- it looks like pin 49, while 48 also has some damage (less so).
A SNES pinout diagram indicates these are probably BA7 (pin 48) and /CART (pin 49), but I might have those inverted (front side of board vs. back side of board):
http://www.caitsith2.net/snes/flashcart ... nouts.html
Quote:
FYI -- not sure if it matters (don't have a cart pinout diagram handy), but one of your cartridge edge connectors is damaged/scratched quite badly -- it looks like pin 49, while 48 also has some damage (less so).
I have tried as many things i can think aside from pulling the plug while the game is on, in an attempt to throw a save file (it is only 1 save that disappears) I cant honestly say, but i dont think that scratch is the issue. I could replace the 6264 (really dont want to) but i dont want to replace the 6264 if its actually a diode or capacitor.
Well, what you're describing is usually what happens when the game code itself attempts to verify the checksum of the memory-mapped SRAM contents and they don't validate. Some games throw a on-screen error telling you of this fact, others just treat it as corrupt and consider the entry invalid and say nothing to you.
You really have no other choice other than to start replacing chips, or just buying another cart somewhere (eBay, etc.). I see one on eBay that's going for US$10 right now. *shrug*
You're not using this cart with a copier (ex. SWC, MGH, etc.) in front of it, are you? Have to ask, because the SRAM backup mechanism there is done differently (from a user/administrative perspective).
Quote:
You're not using this cart with a copier (ex. SWC, MGH, etc.) in front of it, are you? Have to ask, because the SRAM backup mechanism there is done differently (from a user/administrative perspective).
Not entirely sure about those, i have a game doctor. but i have only been using it directly on an offical snes with licensed rf and adapter.
Thanks for the info, guess i will have to talk to my friend and see what he would like to do lol (Thanks for the ebay tip too

)
Thanks to everyone who helped. If anyone has anything they could add, please feel free. Its been a warm welcome to the community and I look forward to learning much from your posts

Ehks
Well, what I'm asking is, is the cart plugged into the Game Doctor, or is the cart plugged into the SNES directly?
If you've been doing it through the GD -- remove the GD and plug the cart into the console directly and see if the issue persists.
sorry about that. definitely not using any peripherals.
koitsu wrote:
one of your cartridge edge connectors is damaged/scratched quite badly -- it looks like pin 49, while 48 also has some damage (less so).
A SNES pinout diagram indicates these are probably BA7 (pin 48) and /CART (pin 49)
Pin 48 (BA7 aka A23) still looks intact enough, and isn't connected to anything anyway.
Pin 49 (/CART) looks as if it got cut completely? Or is it only some cosmetic scratch? Do you still measure 0 ohms between the lower/upper halves of that pin? The /CARTridge select signal is important, if it's really cut - that would explain all sorts of problems (would be only a little bit surprising that the game is still playable at all).
For testing the battery it may be best to measure it directly on the SRAM chip (SRAM pin 14 and 28) (or the two pins on C8 capactitor), to see if the voltage does really arrive there. Before that, best shortcut C8 for a moment to discharge it, to be sure that the volts are really coming from the battery.
ehks wrote:
Quote:
If the game provides multiple saves, and ever only one of them disappears, it's probably the former. If they only ever all disappear at the same time, it's probably the latter.
I have some 6264's here, thats what i see on the board, is that not the SRAM?
Aye, that's the SRAM. I think lidnariq meant that the data could disappear (not the whole 6264 chip).
Quote:
Pin 48 (BA7 aka A23) still looks intact enough, and isn't connected to anything anyway.
Pin 49 (/CART) looks as if it got cut completely? Or is it only some cosmetic scratch? Do you still measure 0 ohms between the lower/upper halves of that pin? The /CARTridge select signal is important, if it's really cut - that would explain all sorts of problems (would be only a little bit surprising that the game is still playable at all).
I tested the contact on the top and bottom, it is just cosmetic.
Quote:
For testing the battery it may be best to measure it directly on the SRAM chip (SRAM pin 14 and 28) (or the two pins on C8 capactitor), to see if the voltage does really arrive there. Before that, best shortcut C8 for a moment to discharge it, to be sure that the volts are really coming from the battery.
Tested the SRAM chip and the c8 (actually says cb). they both read about 4.1 (i dont understand why its 4 instead of the 3 that it reads when not on the board) I have no idea what shortcutting is (im guessing touch different points with the volt meter, but have no clue which pins to shortcut with
Quote:
ehks wrote:
Quote:
If the game provides multiple saves, and ever only one of them disappears, it's probably the former. If they only ever all disappear at the same time, it's probably the latter.
I have some 6264's here, thats what i see on the board, is that not the SRAM?
Aye, that's the SRAM. I think lidnariq meant that the data could disappear (not the whole 6264 chip).
My bad, i didn't word what I meant right. I knew the whole chip wouldn't disappear. I was asking if in fact perhaps the 6264 has a few bad bytes on it and needed to be replaced? sorry if im asking stupid questions, im a huge novice to all of this.
The SRAM (and the CB capacitor) receive the power from the battery (3V), and - if the SNES is switched on - additionally from the SNES power supply (5V). The capacitor works like a tiny rechargeable battery, if it has more that 3V, then it must be still charged up from the SNES power supply.
In that situation the SRAM works even without the battery, which is fine... unless you want to test if the battery is working, then it's rather disturbing (you won't see battery problems until the capacitor discharges, which may take some minutes.. or days... whatever).
To get rid of that effect: Remove the CB capacitor (desolder it, it's rather useless anyways). Or shortcut the two CB pins with each other, that should discharge the capacitor after less than a second (for example: bridge the two CB contacts with a screwdriver, that should be totally harmless on that PCB) (assuming that the SNES power-supply is switched off, else you might see some smoke & sparks).
And be sure to count the chips after each test... just in case that some of should happen to disappear (no, not seriously).
Quote:
To get rid of that effect: Remove the CB capacitor (desolder it, it's rather useless anyways). Or shortcut the two CB pins with each other, that should discharge the capacitor after less than a second (for example: bridge the two CB contacts with a screwdriver, that should be totally harmless on that PCB) (assuming that the SNES power-supply is switched off, else you might see some smoke & sparks).
Thanks for the info. I shortcut the cb cap, didn't change what it read on the dvm. the battery, cap and pins 14&28 of the SRAM all read out about the same except the battery which is about 4 (the others are about 3.7)
To be clear. removing the cb cap is going to do what? the cart doesn't need it to keep its saves?
Removing the cap will make the cart more susceptible to power fluctuations, its purpose is to "recharge" when you power on the console, and any time the power fluctuates, it helps smooth out the transients. It would be easier to just short the two leads together (TEMPORARILY, just touch the leads with a wire or something then remove it, and with the power DISCONNECTED) in order to fully discharge it when you test. What you're seeing with the 4v on the SRAM chip is that the console is charging the capacitor up to 5v, and it's slowly dropping after the console is turned off, acting as a very small battery. This may be why the saves are being retained for short periods of time. Removing the capacitor, or shorting the leads, removes this factor from your debugging, so you can actually see what the battery is doing on its own.
What's the longest the game can be off for while retaining your save data? You mentioned earlier a period of like 2 weeks, which suggests some other issue than the battery not working.
Quote:
qwertymodo wrote:
Removing the capacitor, or shorting the leads, removes this factor from your debugging, so you can actually see what the battery is doing on its own.
Thanks for the explanation, that really helps. I shortcut the cap, but im not sure what im looking for after that (assume this has something to do with the save file because my next reply answers this)
Would shortcutting the cap wipe the saves?
Quote:
MottZilla wrote:
What's the longest the game can be off for while retaining your save data? You mentioned earlier a period of like 2 weeks, which suggests some other issue than the battery not working.
I just tested it and this time BOTH the save files are gone (playing it threw the file i was playing on both file 1 and file 2 separately, while at the same time keeping another file) I don't know if this is from the shortcut on cap cb or not, im testing that now
Quote:
MottZilla wrote:
What's the longest the game can be off for while retaining your save data? You mentioned earlier a period of like 2 weeks, which suggests some other issue than the battery not working.
At least a few days had gone between me playing with working and such (probably 3 or 4, maybe even up to 5 days without play. still left in the snes, but most of the time i unplug the snes)
UPDATE*
Just saved a file, took it out and shortcut the cb capacitor. that made the save file disappear. Does that mean the cb cap is what needs to be replaced? (fingers crossed)
Shorting out the capacitor should lose your saves. (Sorry.) Shorting it out is only useful for testing whether the battery works.
lidnariq wrote:
Shorting out the capacitor should lose your saves. (Sorry.) Shorting it out is only useful for testing whether the battery works.
Yeah, forgot to mention this... sorry. When you short the capacitor, you're also shorting the entire supply rail to ground, which would have essentially the same result as removing the battery, as far as the SRAM is concerned. The point of shorting the capacitor would be to do it, then test the voltage on the various Vcc pins, the battery terminals, etc. to see whether or not you still had a reasonable supply voltage (3.4ish volts). If you short the capacitor and then read much lower voltages than you did before, then you know it was the capacitor maintaining the voltage before. If you want to test data retention, you'll want to remove the capacitor (just the big electrolytic one, the smaller ones shouldn't hold enough charge to matter). If the battery is dead, then with the capacitor gone, the saves should be lost almost instantly when you power down. Otherwise, if the saves are still retained for several days, then the issue probably lies elsewhere, like maybe a bad trace or solder joint.
I made 2 save files, and verified they would stay (at least for when i pull the cart). gonna leave it out for about a week or so (unless someone could give a guideline to how long i should wait) to see if the file throws again.
In the meantime, thanks again everyone; you have been a great help.
I'll report back when i try again

Is the capacitor still on the board? Replacing the capacitor won't fix the save problem (unless it's due to supply transients during power-on/off... but that's rather unlikely), it's the battery that is responsible for maintaining power to the SRAM, not the capacitor. The only reason it was suggested to remove it was because the capacitor might be masking a problem with the battery.
since you have the board out of the cartridge shell, try just cleaning off those contacts with a pink pencil eraser. There is a major color difference between the top and bottom of each of the contact pads. I still stand behind just cleaning it with isopropyl alcohol, but finding a pencil eraser takes almost no effort.
The pencil eraser will clean by purely a friction action (don't go overboard and rip up the traces), whereas alcohol will dissolve most of the crud, and then the remainder removed by friction/exposure from the cotton swab.
If you have a bad contact to one of the middle Address lines (A11-ish) on powerup, but the game code can get far enough to where it checks the SRAM, then it might think the save data integrity is bad and then attempt to initialize it.
sorry for the late response, got sick

Quote:
Is the capacitor still on the board?
Yes. i left it on in hope of not needing to replace it. Its been out of the system for 5 whole days now, any idea how long I should wait? could i unplug the snes, put the cart in and turn it on? would this drain the battery/cap?
Quote:
since you have the board out of the cartridge shell, try just cleaning off those contacts with a pink pencil eraser. There is a major color difference between the top and bottom of each of the contact pads. I still stand behind just cleaning it with isopropyl alcohol, but finding a pencil eraser takes almost no effort.
The pencil eraser will clean by purely a friction action (don't go overboard and rip up the traces), whereas alcohol will dissolve most of the crud, and then the remainder removed by friction/exposure from the cotton swab.
If you have a bad contact to one of the middle Address lines (A11-ish) on powerup, but the game code can get far enough to where it checks the SRAM, then it might think the save data integrity is bad and then attempt to initialize it.
I clean every game i play with swabs until the swab no longer becomes dirty (usually go through a few each game lol)
As for the address lines, you would need to provide and explanation along with a suggestion to test as i don't know what ya mean (im not against reading. just dont know what ya mean)
Thanks
Well, we could keep trying to debug the board, or you could just give up and consider other options. Are you able to desolder the ROM cleanly? I'd be willing to send you a compatible donor board if you like. I have a bunch of crappy sports games I picked up for research, and now I'm pretty much at a point where I don't have any real use for them. I'm sure I have a few 1A3B-13's lying around. Or, if you're not able to do the transplant yourself, shoot me a PM, I'd be willing to do it for you.
Edit: Whaddya know, I have one with the ROM already pulled in my toolbox right now. Want it?
Tomorrow will be 7 days its been out of a system. im gonna try it tomorrow and post back if it still has its files or not.
I have made repros out of donor carts, but i have always cut the mask roms out. If i needed to save the rom from shadow run, I dont know how comfortable I would feel about doing that myself. Are ya within Canada?
ehks wrote:
Tomorrow will be 7 days its been out of a system. im gonna try it tomorrow and post back if it still has its files or not.
I have made repros out of donor carts, but i have always cut the mask roms out. If i needed to save the rom from shadow run, I dont know how comfortable I would feel about doing that myself. Are ya within Canada?
I'm in the US. If you've done desoldering before, it's a piece of cake. If not (as it sounds), you probably don't want to start with something you actually care about...
just popped the game in, both save files still in-tact. I have never saved a chip from a board (only removed by way of cutting) i have an old mobo around here im going to practice on.
sending it to the US might be worth it, but its not my game. i will ask my buddy what he thinks tomorrow.
Thanks again XD
Apropos solder: How good are your batteries new soldering points? On the photo it looks as if you solder was a bit 'sticky'. In case it was your first soldering project: Heat up both contacts (the battery leg, and the PCB), and add some fresh solder (that'll make it more fluid, old solder can get sticky, and isn't shiny). After that you should have a nice shiny soldering point.
If the soldering point is bad, it might happen that the battery looses contact (like when you shake/move the cartridge). A bad contact (on the battery, or elsewhere) would effect the cartridge "works only sometimes" effect.
Removing the CB capactitor could really help on testing such problems (at the moment the SRAM is double-backed, by the battery and by the capacitor - so you won't notice the problem even if the battery looses the contact for some seconds - if you remove CB then you'll see such problems more instantly).
I've been having an identical problem with a Secret of Mana 2 cart that I made out of a Seiken Densetsu 3 donor cart. It works fine for a week, or sometimes only 15 minutes, and then the save files disappear. Sometimes its only one of the 3 save slots, other times its all of them. At first I thought this was a bad battery, so I replaced it with a brand new one....Same issue happened within a day. I copied the save file onto the cart using a Retrode from an .srm file I have for this game.
Next I figured it was something on the board, so I transplanted the flash chip onto another compatible donor cart, added a new battery and tested it. Save file was there for a while, and then again within ~2 days it erased itself. Strange...
Now, I have tested with 3 different SNES systems and i've seen it happen on all of them. All my other rpgs have kept their saves since the day I bought them and they have held their saves with no issues for 10-20 years with plenty of use.
So if the board, the systems and the battery have been ruled out, what else can it be? Is it possible there could be a checksum or whatever conflict between my save file and the version of the rom on the chip causing this to happen?
Are your cartridge contacts clean? Not dirty or oxidized? Is your SNES connector pretty clean? I have a Chrono Trigger game which we believed had a dead battery. Turns out the original battery still works great, but dirty contacts were causing the saves to be lost. After a cleaning it works fine.
I had thought of that, so I cleaned both the carts with isopropyl alcohol during my tests. All squeaky clean, still happened.
There are two ways battery-backed memory loses its contents:
* Temporary voltage sag below the minimum necessary to retain cell voltage, causing partial or total loss
* Failure to disable the RAM when it shouldn't be enabled (powerup and powerdown, or incorrect decoding from bad contacts), causing random writes and thus partial loss
Do you ever lose saves with the game pak left in the SNES? What if you leave the SNES on but only ever use the reset button? How confident are you there's no mechanical issues?
Quote:
How confident are you there's no mechanical issues?
Very confident.
Again, i've tested these carts on 3 different systems and it happened on all of them, unless all three systems have the same fault
The only thing that I thought maybe had something to do with it was that I am writing the save file to the sram using the Retrode. But I figured if the save file shows up even once when the game is booted up then the sram took the file properly, and that any loss thereafter must be with the cart...but 2 donor carts doing the same thing? And again, all my other rpgs (ToP, FF3, Super Metroid, etc...) all work perfectly. FF3 for example has over 200 hours total on the cart played on this same system, and the save files are intact on the original battery.
Quote:
Do you ever lose saves with the game pak left in the SNES?
Yes, I left it in the system for 2 days and when I went to check it the saves were gone.
Quote:
What if you leave the SNES on but only ever use the reset button?
If I reset the game and check the save file is there. The fastest I've seen a save disappear was around 5 min after power down, haven't held the reset button for that long so I can't say for sure.
Holding the reset button doesn't affect the Vcc line, it just pulls the reset line on the CPU, holding the CPU in a reset state, then jumping to the code's reset vector once the button is released. So basically, this means that the issue is with power loss to the SRAM chip, not with spurious garbage writes, although that was already kind of apparent by the fact that time removed from the console seemed to be a factor in the save retention. My only suggestion at this point, since you've already replaced the battery and capacitor, is to try using a MAD-1 donor instead of the '139 board you're currently using. The MAD-1 incorporates SRAM battery power and /CS management into the IC package, rather than using discrete components like the '139 boards do.
Edit: I'm confusing getafixx's situation with ehks'. getafixx, can you post high-res photos of your PCB (both sides) so we can see if there are any visually apparent mechanical issues?