17 replies to this topic

[per]

Most cartridge-based games that keep save-data use a battery to back it up. As we all know, batteries don't last forever and eventually they will run dry. The big question is when that will happen.

-------------------------------------------------
NES
-------------------------------------------------

In the case with NES games, the battery usually only powers a low-power (or a very-low-power) SRAM chip.
The battery used is in most cases the CR2032, which according to specs can provide about 220mAh at 25 degrees Celsius (which corresponds to approximately 790C). This value is when the voltage the battery provides has depleated to only 2V. Three things should now be noted:

• Very-low-power SRAM chips can usually retain data down to only 1.5V, which means that the battery may provide more charge before your savegame goes.
  
• The diodes blocking the batteries off during usage has a forward voltage drop-off, typically about 0.7V for most diodes, which means that the gain in charge from the previous note practically doesn't have any effect.
  
• The voltage-time graph in the battery specs shows that the battery's own drop-off voltage (when depleated) is sudden enough for none of this to matter.

We can then calculate the expected lifetime based on the SRAM specs. Most very-low-power SRAM chips will typically drain 0.3 to 0.4uA in standby mode. Dividing the charge on the drain therefore tells us that the average lifetime of a typically battery-backed NES cartridge is around 70 years (+/- 10) at 25 degrees Celcius. As all battery-backed NES cartridges I have seen still retains data, this seems to still be a valid estimate, and it's nice to know that they may hold data for another half century to come.

-------------------------------------------------
GameBoy/GameBoy Color
-------------------------------------------------

The other class of cartridge systems I have looked at is Gameboy games. After examining quite a lot of those, I see that two different batteries were used. Older games, usually from 2000 and earlier, uses a whimpy CR1616 battery cell with only about 55mAh (200C). Later cartridges has the CR2025 cell, with 150mAh (540C). There are no diodes, and instead a MM1134 Battery/Power switcher (or equalent) is used (0.3V voltage drop-off, 0.3uA drain). Again, due to the reasons above, the voltage drop-offs are neglected in my calculations, but there are some resistors to reduce the current drain (I estimate it to be around 0.4uA instead of the expected 0.7uA, but this may be quite inaccurate). Lastly, cartridges with a Real-Time clock function (Like 2nd gen Pokemon games) also needs the battery to power the MCB3 chip (which has an unknown but very signifficant drain). Calculating the charge over drain again, we get:

• Old MCB1 games with the CR1616 are expected to retain data for 15 years +/- 5 years. About now, in other words.
  
• Old games with the CR2024 are expected to retain data for 40 years, so still no worries there.
  
• I'm unable to find the expected life of cartridges with a RTC, but as a matter of fact; some of them died as little as only 7 years after manufacture. Because of this, I find that they must drain more than 2uA which is more than 5 times as much as the regular cartridges. With a replacement CR2032 in place of the CR2024, I find that these will still only last about 10 years.

-------------------------------------------------

Conclusion:
You don't need to worry about the batteries in your NES games yet, but you should be aware that the time is running out for the batteries in the earliest GameBoy games. We all know about the GameBoy RTC games already, and I think we all can agree that's it's a major and unfortunate design flaw.

I haven't looked at SNES, N64 or GBA cartridges yet, so I can't say anything about those.

Edited by per, Fri Jan 27, 2012 6:48 PM.

[MagitekAngel]

The only games I've ever encountered with dead batteries were Game Boy games. My original copies of Pokemon Gold and Silver (bought for me as a gift when they were new) have been dead for years), and a couple of years ago I had to return a copy of Link's Awakening DX I bought from a classic game store because the battery in that one was dead, too.

[thegoldenband]

Great post to start this thread, very informative. I look forward to reading your findings on the SNES and N64, and hope that you'll also look at Genesis and SMS games.

I've had a couple of Genesis games whose savefiles became corrupt; of these, I'd beaten one and it sat around for a year before I tried it again (Light Crusader), while another was brand new and crapped out on me right before the end of the game (Star Odyssey). I also bought a couple of Genesis RPGs/strategy games recently that indicated savegame corruption when I first tested them, and I'm hesitant to play them much until I know if the battery's the culprit.

[bomberpunk]

if you got a problem, yo special-headed screwdrivers & new batteries will solve it. check out the hook while the dj revolves it.

[thegoldenband]

Savegames ranged out like a ray gun / I grabbed my Nomad, all I heard was "SEGA-A-A"

(Alternatively: "Will it ever stop? / Yo, I don't know.")

[high voltage]

per, on Fri Jan 27, 2012 6:45 PM, said:

Most cartridge-based games that keep save-data use a battery to back it up. As we all know, batteries don't last forever and eventually they will run dry. The big question is when that will happen.

-------------------------------------------------
NES
-------------------------------------------------

In the case with NES games, the battery usually only powers a low-power (or a very-low-power) SRAM chip.
The battery used is in most cases the CR2032, which according to specs can provide about 220mAh at 25 degrees Celsius (which corresponds to approximately 790C). This value is when the voltage the battery provides has depleated to only 2V. Three things should now be noted:

• Very-low-power SRAM chips can usually retain data down to only 1.5V, which means that the battery may provide more charge before your savegame goes.
  
• The diodes blocking the batteries off during usage has a forward voltage drop-off, typically about 0.7V for most diodes, which means that the gain in charge from the previous note practically doesn't have any effect.
  
• The voltage-time graph in the battery specs shows that the battery's own drop-off voltage (when depleated) is sudden enough for none of this to matter.

We can then calculate the expected lifetime based on the SRAM specs. Most very-low-power SRAM chips will typically drain 0.3 to 0.4uA in standby mode. Dividing the charge on the drain therefore tells us that the average lifetime of a typically battery-backed NES cartridge is around 70 years (+/- 10) at 25 degrees Celcius. As all battery-backed NES cartridges I have seen still retains data, this seems to still be a valid estimate, and it's nice to know that they may hold data for another half century to come.

-------------------------------------------------
GameBoy/GameBoy Color
-------------------------------------------------

The other class of cartridge systems I have looked at is Gameboy games. After examining quite a lot of those, I see that two different batteries were used. Older games, usually from 2000 and earlier, uses a whimpy CR1616 battery cell with only about 55mAh (200C). Later cartridges has the CR2025 cell, with 150mAh (540C). There are no diodes, and instead a MM1134 Battery/Power switcher (or equalent) is used (0.3V voltage drop-off, 0.3uA drain). Again, due to the reasons above, the voltage drop-offs are neglected in my calculations, but there are some resistors to reduce the current drain (I estimate it to be around 0.4uA instead of the expected 0.7uA, but this may be quite inaccurate). Lastly, cartridges with a Real-Time clock function (Like 2nd gen Pokemon games) also needs the battery to power the MCB3 chip (which has an unknown but very signifficant drain). Calculating the charge over drain again, we get:

• Old MCB1 games with the CR1616 are expected to retain data for 15 years +/- 5 years. About now, in other words.
  
• Old games with the CR2024 are expected to retain data for 40 years, so still no worries there.
  
• I'm unable to find the expected life of cartridges with a RTC, but as a matter of fact; some of them died as little as only 7 years after manufacture. Because of this, I find that they must drain more than 2uA which is more than 5 times as much as the regular cartridges. With a replacement CR2032 in place of the CR2024, I find that these will still only last about 10 years.

-------------------------------------------------

Conclusion:
You don't need to worry about the batteries in your NES games yet, but you should be aware that the time is running out for the batteries in the earliest GameBoy games. We all know about the GameBoy RTC games already, and I think we all can agree that's it's a major and unfortunate design flaw.

I haven't looked at SNES, N64 or GBA cartridges yet, so I can't say anything about those.

You haven't looked at SNES, N64, GBA....why not mentioning Genesis, MSX or even Epoch Super Cassette Vision (Dragonslayer, first cart to have battery back up).

Edited by high voltage, Sat Jan 28, 2012 5:17 AM.

[per]

high voltage, on Sat Jan 28, 2012 5:16 AM, said:

You haven't looked at SNES, N64, GBA....why not mentioning Genesis, MSX or even Epoch Super Cassette Vision (Dragonslayer, first cart to have battery back up).

None of those had any signifficant popularity in Europe at all. Here the home-computers ruled the gaming market till well into the 90's. Allthough I do mention the NES and SNES, that's because I have had some interest in those systems at some points in time.

[high voltage]

a

per, on Sat Jan 28, 2012 6:45 AM, said:

high voltage, on Sat Jan 28, 2012 5:16 AM, said:

You haven't looked at SNES, N64, GBA....why not mentioning Genesis, MSX or even Epoch Super Cassette Vision (Dragonslayer, first cart to have battery back up).

None of those had any signifficant popularity in Europe at all. Here the home-computers ruled the gaming market till well into the 90's. Allthough I do mention the NES and SNES, that's because I have had some interest in those systems at some points in time.

You're joking right? , Megadrive (Genesis) was hugely popular in Europe, more so than SNES and NES put together.
SMS was very popular in Europe, as a matter of fact, SMS has a bigger European cartridge range than the USA, but you're right about the 80s computers most popular in Europe, eg Spectrum, C64, Atari 8-bit, the NES was not popular in Europe at all. The book Game Over states that by 1992 only 1.5 mill NES sold in Europe.

Edited by high voltage, Sat Jan 28, 2012 9:30 AM.

[per]

high voltage, on Sat Jan 28, 2012 9:25 AM, said:

You're joking right? , Megadrive (Genesis) was hugely popular in Europe, more so than SNES and NES put together.
SMS was very popular in Europe, as a matter of fact, SMS has a bigger European cartridge range than the USA, but you're right about the 80s computers most popular in Europe, eg Spectrum, C64, Atari 8-bit, the NES was not popular in Europe at all. The book Game Over states that by 1992 only 1.5 mill NES sold in Europe.

Please. I may have said Europe, but what I was reffering to was where I live in Europe. It may have been popular on mainland Europe or in the UK, but nobody I know where I live have/had one as of I know. The Amiga 500, on the other hand, seems to have been very common here. It may have been a little before my time anyways, so I'll admit I'm not exactly the right person to judge. I did state in my previous post that the reason I included NES was of personal interest and not due to it being popular or not.

I may have been a little narrow-minded with that last comment, but it was about 2 AM (local time) when I wrote it. I did absolutely not intend to offend anyone with it.

Edited by per, Sat Jan 28, 2012 10:55 AM.

[high voltage]

That's ok, I am still to hasty with answers as well. At 2AM I write the weirdest stuff.

[Punisher5.0]

Did you take into account the resistance of the battery causing a self drain? I would imagine after 70 years a brand new unused battery would be dead just from it self discharging through it's internal resistance.


Makes me wonder how much power the Saturn uses to store data. I remember my brother having to replace his battery in the 90's.

[Austin]

Punisher5.0, on Sun Jan 29, 2012 7:12 PM, said:

Makes me wonder how much power the Saturn uses to store data. I remember my brother having to replace his battery in the 90's.

I've always had to replace the Saturn batteries frequently. Like, give it or year or two on a fresh one and then it's dead.

[bomberpunk]

My Saturn battery doesn't even last a full year.

[awace]

I know you can hot swap batteries and that most 2302 batteries and c-mos batteries are desigened to last about 6 years

Edited by awace, Mon Jan 30, 2012 2:05 AM.

[GroovyBee]

Please show your calculations for NES battery lifetime. Mine come out to less than 1/10 of yours.

[Gentlegamer]

My Zelda II cart from 1989 still has my original save game... from 1989. I was pleasantly surprised when I put it in my NES recently.

[TheGameCollector]

I've seen more dead SNES batteries than NES batteries by far.

[per]

GroovyBee, on Mon Jan 30, 2012 3:50 AM, said:

Please show your calculations for NES battery lifetime. Mine come out to less than 1/10 of yours.

Total charge of battery:
220mAh from 3V to 2V
220mAh = 0.22Ah
0.22Ah * 3600s/h = 792C

Current drained by Very-Low-Power SRAM chip in standby mode (I've seen those in several NES cartridges, some may only have Low-Power SRAMs, though):
typically between 0.3uA and 0.4uA when continously in standby mode
0.3uA = 0.0003mA = 0.0000003A
0.4uA = 0.0004mA = 0.0000004A

Discharge time (in both cases):
792C / 0.0000003A = 2640000000s
2640000000s / 3600s/h = 730000h
730000h / 24h/d = 31000d
31000d / 365.25d/y = 84y

792C / 0.0000004A = 1980000000s
1980000000s / 3600s/h = 550000h
550000h / 24h/d = 23000d
23000d / 365.25d/y = 63y

Taking signifficant digits into account:
60y min for typical drain, 80y max for typical drain.

...In other words:
70 Years +/- 10


But as somebody mentioned, the battery itself and the diodes blocking off the rest of the power lines has some fractions of uA leakage (internal leakage of battery varies widely from battery to battery, and by temperature), so these calculations may be a little too optimistic. With 1uA of leakage (which is quite a lot for Lithium coin-cell batteries), the estimated lifetime will be less than a third of the calculations.

Edited by per, Tue Jan 31, 2012 10:37 AM.