Forever DUO 852 Games of NES (405+447)

[krzysiobal]

I spotted accidentally this PCB. Price on aliexpress is ~19$ for 60pin and 24.5$ for 72pin version. First time see that Chinese distribute universal PCB using general use CPLD chip, not blob or ASIC.


Could be nice for adopting as repro donor (instead of canibalizing existing original cartridges).



Board markings: FC-01C DT:1810


Features:
* 240 Cells programmable CPLD EPM240 - enough capacity for example for MMC3+MMC1+few other chips at once, or almost any single mapper (except MMC5/JY Company)
* 128 MB (Mehabyte!) PRG-ROM onboard
* Battery backed 8kB PRG-RAM, disabled when voltage falls
* 128 kB CHR-RAM

Pros:
+ External header available for connecting JTAG programmer to reprogram CPLD
+ Nice & small layout, fits shell with either clips on both sides or single bolt in center
+ Whole PPU address bus (A0..A13) wired to CPLD so can emulate even MMC2/MMC4

Cons:
- every chip is powered from +3.6V but no voltage translation on data/address pins
- Only 128kB of CHR-RAM
- PRG-ROMS's /WE connected to +3.6V so can't be reprogrammed with Kazzo without rewiring
- Only CPU-A0,A1,A8-14 routed to CPLD, so not every mapper can be emulated

[NewRisingSun]

Does the menu actually reprogram the CPLD for each game mapper, or is it just programmed to run some kind of universal mapper, maybe even just enhanced MMC3, for which all games are modified?

[aquasnake]

due to the bit width of lower cpu address bus(a0, a1), i guess this is some variant/enhanced of mapper268.

[dink]

4 Nin Uchi Mahjong (first pic screenshot) is MMC5, so that board possibly emulates MMC5 - unless its a hack to a different mapper. (or miss-titled game)

[lidnariq]

Looks like NROM to me

[tepples]

Not to be confused with Shin 4-nin Uchi Mahjong, which is indeed MMC5.

[krzysiobal]

Does the menu actually reprogram the CPLD for each game mapper, or is it just programmed to run some kind of universal mapper, maybe even just enhanced MMC3, for which all games are modified?

It is programmed to emulate fixed universal mapper. It is not suitable for reprogramming after each game selection, because:
* According to datasheet, this chip can withstand up to 100 erase and reprogram cycles (which in my opinion is a little lowered, because the first flash-cartridge I built based on this chip was reprogrammed more that that and still works fine),
* That needs interfering the JTAG pins, which are here not tied to anything except the pin header
* During programming, this chip is not functioning (all its input/outputs) are in high impedance state, so there would be need of other "supervisory" chip (althought there is something called real-time ISP which can be wisely used to overcome this)

Here is dump of the $8000-$ffff of this ROM.

$8000-$ffff.nes

(32.02 KiB) Downloaded 147 times

Menu code writes:

Code: Select all

CPU W A3 5700
CPU W 0C A000
CPU W 00 B000
CPU W 00 C000
CPU W 01 D000
CPU W 01 E000
CPU W 01 F000
...
JSR B000

when I made exact write sequence in Kazzo, reading $8000-$ffff still returned the same data, so there must be some quirk in it.

4 Nin Uchi Mahjong (first pic screenshot) is MMC5, so that board possibly emulates MMC5 - unless its a hack to a different mapper. (or miss-titled game)

Cartridge contains the NROM game.

[dink]

Oops, my bad - I confused 4 Nin Uuchi with Shin 4 Nin Uuchi

[lidnariq]

* According to datasheet, this chip can withstand up to 100 erase and reprogram cycles (which in my opinion is a little lowered, because the first flash-cartridge I built based on this chip was reprogrammed more that that and still works fine),

100 cycles is the maximum number for which the chip is still guaranteed to retain data for the full rated number of years and across the full temperature and voltage range.

Some years ago, someone sold a bunch of kits to intentionally wear out a small serial EEPROM. Subsequent reports showed that these EEPROMs did not start generating programming errors until 11 to 14 times as many writes as the EEPROM was rated for.

[Ben Boldt]

100 cycles is the maximum number for which the chip is still guaranteed to retain data for the full rated number of years and across the full temperature and voltage range.

Some years ago, someone sold a bunch of kits to intentionally wear out a small serial EEPROM. Subsequent reports showed that these EEPROMs did not start generating programming errors until 11 to 14 times as many writes as the EEPROM was rated for.

I did a similar experiment in a way. I have a laptop battery where some of the cells went bad and the controller chip, made by TI, has latched a "permanent failure" mode after detecting the bad cells. The datasheet says the mode is stored in flash memory and there is no possible way to come back out from that mode, due to the access to that function being restricted when the "seal" command was presumably sent at the factory. It also says that the data retention of the flash is 10 years. Based on characteristics of flash memory I found online, it appeared that elevated temperatures should drastically reduce data retention time. So I removed the controller chip and taped it to a lightbulb from an old microwave, along with a thermocouple. I measured temperature at just over 100 deg C. I left this lightbulb on for about 4 months, really cooked it good. Hard to solder again after that but putting it all back together with good cells, the chip had not forgotten its permanent failure mode after this. I have no reason to believe that I retriggered the failure mode. The flash is just much more resilient than the minimum that they promise.

[tepples]

Flash memory manufacturers' underpromising and overdelivering on wear resilience reminds me of Nintendo products' reputation for being "Tonka tough." Game Boy systems literally go through the war and still work.


(Except for Nintendo DS shoulder buttons and hinges, that is.)

[pakosup]

This board was discussed some time ago on emu-land http://www.emu-land.net/forum/index.php ... msg1428464
Scheme was changed to work with Cluster's dumper and ROM builder. Also some mapper kits are available.

[aquasnake]

this cart is distributed by Blazepro, which chinese name is "超技".

[NewRisingSun]

Scheme was changed to work with Cluster's dumper and ROM builder. Also some mapper kits are available.

What does that mean? Is Forever DUO 852 Games of NES a Coolgirl cartridge, then? Because the bank register writes do not look like it.

[pakosup]

What does that mean? Is Forever DUO 852 Games of NES a Coolgirl cartridge, then? Because the bank register writes do not look like it.

No, but it can be modified (new wiring and updating FPGA program) to use ROMs that created in Coolgirl ROM builder.
FPGA is cheaper so only few mappers are available at once.