nesdev.org

aquasnake wrote: ↑Fri Jan 03, 2020 2:12 am atcually if use flash rom only, which means do not use sram，we do not need level shifters in data bus，because the direction of signal is 3v to 5v.

That's only true if nothing else drives the data bus. So this works ok¹ with the SA-1, because the data bus there is unidirectional. But in the NES, both the CPU and RAM drive the data bus often.

¹Safely, but not reliably: the SA-1 requires 5VCMOS voltage thresholds, so a 3.3V signal isn't reliably received as logic 1.

Well it might be controversial but years ago I made a 60 to 72-pin adapter that specifically accepts the thin pirate famicom carts. Normal thickness carts don't fit in this adapter. Before you hate on me too much, I made it out of a damaged cart. I am always on the hunt for damaged carts for purposes like this.
I will definitely be getting some 74LVC245 and making this adapter work just like yours krzysiobal! I am thinking about where to put the 3.3V/5V selector. I might be OK just always running 3.3V with my collection of pirate carts, will have to try it. Also, Earthworm Jim 3 that I destroyed earlier is currently being super glued, outlook good. Will be interesting to see if that glob can run on 3.3V.

krzysiobal wrote: ↑Thu Jan 02, 2020 11:28 am

krzysiobal, why did you not level-shift PPU /A13? It looks like the Famicom drives PPU /A13 with 5V when high?

I guess the question is how many of the 3.3V carts add extra nametable memory...

Well, there are 2 free gates in one of the 245's to /A13 can be theoretically translated into 3v3, but because this chip it is quite far away from the /A13, I would need to add plenty of vias. I was making this PCB adapter in home using etching and driling, so I wanted to save time and effort.

Plus 99% of games do not use this line, modern consoles do not have this signal present in the cartridge slot. And finally, my goal was to protect the console, not cartridge. Because this signal is driven by the 368 buffer, not any of the processor itself, it should withstand the higher current.

-

I was experimening with powering the cartridges using 3.3V but most of them crashed after a few seconds. Maybe there is some logic that needs 5V power supply (like the mapper blob).

This adapter can even work with ordinary 5V cartridges. I can just think of one exception - if there are HC chips in cartridge, which need high level at around 4.5V or more, but they are rare (99% of chips are LS).

-

I also don't know why in some carts that use pirate DIP40 VRC2/VRC4, there are dioses in series to those chips.

-

If you wasnt to fix single cartidge, probably the fastest and cheapest idea is:
1) Cut a line using sharp knife through the pads you want to modify (probably all pads if you are lazy)


2) Solder 100R 0603 resistors for all lines (except GND, +5V, /IRQ, CIRAM-A10, AUDIO) - those lines can be shorted.

krzysiobal wrote: ↑Sat Jan 04, 2020 3:38 pm Because this signal is driven by the 368 buffer, not any of the processor itself, it should withstand the higher current.

And even if it doesn't, replacing that part is cheaper and easier than replacing the RAM, CPU, or PPU.

I was experimenting with powering the cartridges using 3.3V but most of them crashed after a few seconds. Maybe there is some logic that needs 5V power supply (like the mapper blob).

Probably just the ROM being slower at lower supply voltage... speed quadratic with voltage isn't atypical.

lidnariq wrote: ↑Sat Jan 04, 2020 1:47 pm I guess the question is how many of the 3.3V carts add extra nametable memory...

coolboy/coolbaby/pocketgames/mindkids they dont have ext-workram and ext-ntram, a13 is just from console to cart, the cart dont generate inter-ntram-ce or outer/ext-ntram-ce. the only /a13 is just from console side to itself by shortcutting two socket pins

coolgirl does need drive ext-ntram-ce and int-ntram-ce, these two signals are not simply the inverting relationship.

most of chinese cheap carts they donot concern these problems mentioned above, because they dont suppory 4 screen ntram function and dont support mapper5, 19 either

I get your point with /PA13 lidnariq and krzysiobal. If that signal is never going to drive any 3.3V logic in the cart, then why bother. I didn't think about the bigger picture like that.

I took apart my "The 100 Best Videogames" cart which in fact is just a disguised 143-in-1. Now that I have seen the inside of Earthworm Jim 3, I know how to open these without breaking them. (I just carefully open by hand and spudger instead of using the clamp method.) This cart is a little better quality than the 400-in-1 variety. It has a mixture of 5V and 3.3V chips inside, so it might actually be wanting to blow itself up! hahaha

32Mx8 Main flash: MX29GL256ELT2I-90Q (3.3V)
8kx8 Battery backed PRG-RAM (I think): LH5168N-10L (5V)
256kx8 CHR-RAM (I think): BS62LV2001 (3.3V / 5V support both)
Mapper chip, TQFP-48 instead of a glob (unknown voltage). Marking:

SMD133-
2016.04.29

I applied 4.999V to the edge connector and measured 4.215V coming out of a fairly fancy XC6206 LDO voltage regulator that powers the main flash. That's above absolute max (4.0V) in the flash datasheet... However, V(IH) max is defined as VCC + 0.5, so that sets it at 4.7V V(IH)... Seems they're up to something there, especially with the precision of that 4.2V. Shady at best of course. But hopeful -- it looks like they are at least recognizing the issue.
Pretty dumb that they don't key these PCBs at all. The battery gets in the way of a plastic post on this one but I think it would be easy to put some of these in backwards.

Ben Boldt wrote: ↑Sun Jan 05, 2020 11:48 am SMD133
2016.04.29

Prior evidence implies that's the MINDKIDS mapper IC.

The last cartridge Ben featured was rather shoddy and marginal in terms of the input the voltage regulator is supplying to the ROM chip, but what do you do when the board doesn't even have that voltage regulator? I have heard of some boards with 3.3v chips being fed only by 5v. What if you have one of these worst-of-the-worst cartridges? I don't think anyone wants to risk having a cartridge shorting out inside their vintage system because you may not be able to limit what that cartridge takes with it.

Great Hierophant wrote: ↑Sun Jan 05, 2020 6:25 pm The last cartridge Ben featured was rather shoddy and marginal in terms of the input the voltage regulator is supplying to the ROM chip, but what do you do when the board doesn't even have that voltage regulator? I have heard of some boards with 3.3v chips being fed only by 5v. What if you have one of these worst-of-the-worst cartridges? I don't think anyone wants to risk having a cartridge shorting out inside their vintage system because you may not be able to limit what that cartridge takes with it.

I think you have to open and modify those carts if you want to use them, and/or build an adapter. I have 2 400-in-1 carts somewhere that I am pretty sure are doing exactly what you describe. I am having trouble finding them. (As well as a whole Famicom I misplaced......) I will share whatever I do with them. krzysiobal's suggestion with cutting across the fingers and installing 100-ohm resistors is very good advice to protect the famicom. My solution might be different because I have a special adapter. In the adapter, I may level-shift the signals and pass-thru the 5V VCC, then in the 400-in-1 carts, take the VCC down to 3.3V.

I am not sure it would be good advice to lower the 5V VCC without level-shifting. It may be worse to have V(IH) too high respect to VCC than just having VCC too high. I don't know what kind of research or testing they have done that got them to lower VCC to 4.2V in my cart. I would say, qualitatively, if it can run a long time on 5V, it should be happier and healthier with 4.2V. And max V(IH) remains reasonable at 4.2 + 0.5 = 4.7V.

4.2V regulator, 100ohm resistors. That seems like the best approachable method to be safe using these terrible carts.

Color rendition went to crap when I undervolted my OEM NES. But other than that, things seem to work down to about Vcc=4.0V, give or take.

The PPU seems to be actually what complains about being undervolted, not incorrect (too late) data from PRG ROM crashing the CPU. Some counter just stops counting.

Because the NES is NMOS, and it has access to no voltage higher than 5V, its ability to pull signals up is limited by its NMOS superbuffers, and my previous measurements have Voh from the 2A03 and 2C02 at approximately 4V. So in combination with something to fix the color rendition, one could be reasonably confident of not damaging the 2A03 or 2C02 by loading them with 3.3V logic with overvoltage protection diodes.

The RAMs in the mainboard, however... they're CMOS, and will pull the data bus up to Vcc much more successfully. They'd need extra help.

krzysiobal wrote: ↑Sat Jan 04, 2020 3:38 pm If you wasnt to fix single cartidge, probably the fastest and cheapest idea is:
1) Cut a line using sharp knife through the pads you want to modify (probably all pads if you are lazy)


2) Solder 100R 0603 resistors for all lines (except GND, +5V, /IRQ, CIRAM-A10, AUDIO) - those lines can be shorted.

Could this solution work on other retro stuff?
I have a some MegaDrive and SNES cartridges just like this one, so...

Fisher wrote: ↑Mon Jan 06, 2020 1:27 pm Could this solution work on other retro stuff?
I have a some MegaDrive and SNES cartridges just like this one, so...

Yes, you could do this on other stuff, but be really careful calling it a "solution". Also, be sure to check your clearance into the edge connector, in case the resistors obstruct the insertion. This is intended to help protect the console -- it may still not be healthy for the console, and the cartridge could definitely still break itself being run on 5V. I think the only proper solution would be to design a new board with level shifters and a voltage regulator, and reuse the chips from the pirate cart. This is going to be a very extensive and complicated rework without printing a new PCB, hence why I am only willing to attempt it once inside of an adapter.