General questions about NES clones

This is an archive of a topic from NESdev BBS, taken in mid-October 2019 before a server upgrade.
View original topic
General questions about NES clones
by on (#79651)
Which one is considered the most accurate?
Do any support s-video?
Are there any known upcoming NES clones that are better than the currently released ones?

by on (#79652)
Not really, a lot have mixed sound duty cycles and I doubt that have accurate PPU/CPU's. Don't run MMC5 or the powerpak mostly.

I don't know, probably. But a RGB toaster would be a 10x better investment.

Kevtris has a damn good everything clone. Can't wait for that to come out! :)

by on (#79664)
Right, the sound on the famiclones are off, the PPU isn't quite right, the video quality isn't quite as good as a deck loader and it will not run MMC5 games or any game that uses four-screen VRAM or name table ROM.

On the up shot I have never had any problems with games starting up on mine (it's a Yobo), the controllers it came with are actually pretty decent, it's never erased a save file (except when I kicked the power cord out of the wall), and it is very hard to crap out the game by bumping the cart or system.

All in all I'd say it's the best $20 I've ever spent. I highly recommend them as a good value. The quality is not the highest but the price is right :D

Now if you want to drop some real dough you could go get one of those VHS label maker NESes with the RGB output or something. I've never seen the point though.

by on (#83333)
qbradq wrote:
Right, the sound on the famiclones are off[...]


Weird, i already saw a licensed Brazilian clone from Dynacom, and it had perfect sound and etc... :?

by on (#83335)
IvanDSM wrote:
Weird, i already saw a licensed Brazilian clone from Dynacom, and it had perfect sound and etc... :?

Are you sure it had perfect sound? I have one of these Dynacom clones, and I never realized there was anything wrong with the sound until it was pointed out to me. I never had a good ear for fine details, but when listening to them side by side the difference is obvious. I'll try to capture some samples later.

The Dynacom consoles do appear to be better than the average Chinese Famiclone, but they're still not perfect.

by on (#83336)
tokumaru wrote:
IvanDSM wrote:
Weird, i already saw a licensed Brazilian clone from Dynacom, and it had perfect sound and etc... :?

Are you sure it had perfect sound? I have one of these Dynacom clones, and I never realized there was anything wrong with the sound until it was pointed out to me. I never had a good ear for fine details, but when listening to them side by side the difference is obvious. I'll try to capture some samples later.

The Dynacom consoles do appear to be better than the average Chinese Famiclone, but they're still not perfect.


Yes, it was one of these Dynavision Blacks. It sounded perfect to me. The salesman put Super Mario Bros. and it was just like the original. Except it was a small 16:9 TV it was plugged on, lol.

by on (#83370)
From what I've heard the NoaC are nearly perfect internally because they were just copies from looking at the internals of the Famicom and cloning them. The duty cycles got swapped apparently on some. Incompatibility with certain carts like MMC5 seems to be a wiring issue more than some internal issue. Things outside the very internals (CPU and PPU) may not be perfect as the wiring is external, so would the circuits that output the audio and video to the user. It would be great if someone had the source to the NoaC so that the duty cycle problem could be fixed as well as the wiring of the cartridge port to be done properly. It probably isn't that far from a perfect clone. It would be fantastic if it was fixed up. You can fit it on a tiny IC and would be perfect for making handhelds or just a small footprint console.

by on (#83387)
A proper decap and tracing of the PPU is probably the biggest thing that stands in the way of an open-source Verilog NOAC. I wonder what else could be shoehorned into the architecture, such as writing specific things to $2002 to unlock other registers.

by on (#83406)
If you're going to start enhancing the NES, another 6K of WorkRAM ($800-$1FFF) would be nice. DMA from CPU bus to PPU bus would be handy too. While we are at it, overclocking without altering audio generation would be very neat as well. Other useful things would be a built in PPU based IRQ, like MMC5's scanline counter functionally. Oh another 2K for 4 nametables total would be nice to include as well.

by on (#83407)
Yes, then let's call it the TurboGrafx 16.

by on (#83408)
I've seen literally hundreds of old (1990s) Famiclones, none of them had mixed duty cycles. So it is probably only related to newer ones.

by on (#83411)
Older clones, which had individual chips for the CPU and PPU rather than a NOAC, are know to be much more accurate. There are a few brazilian clones that have even been confirmed to work with a PowerPak.

by on (#83415)
I've seen three kind of Famiclones in first half of 1990s. They are probably 'generations', steps of increasing integration.

First kind is close to the original, with 2A03 clone, PPU clone (UMCs), SRAM chips, and some 74xx logic. Console had large two layers board with all these ICs and small one with 7805 and RF circuit.

Second kind had an epoxy bulb kind large IC, but also two SRAM chips. The bulb sometimes was on a separate small board that was soldered with pins into main single layer board with cart slot, SRAMs and gamepad connectors. These clones had two boards too.

Third kind is the NOAC, it had just an epoxy bulb IC, and almost no other components - quartz, few capacitors and resistors, rarely a transistor. These clones usually had three boards - one with gamepad connectors, other with cart slot and the chip (single layer), and third with 7805 and RF circuit.

More modern clones were like third kind, but the board was ultra-small. Originally it was similar to Famicom cart by size, and modern boards were like 2-3 times smaller.

None of these had the mixed duty cycles.

Sound and picture quality-wise, first and second kind were better, third kind had a lot of digital noise in sound and often vertical bars of bright, don't know how this called properly (noticeable on black background as array of wide vertical columns).

by on (#83423)
Shiru wrote:
and often vertical bars of bright, don't know how this called properly (noticeable on black background as array of wide vertical columns).

Are the vertical columns about 8 pixels (one tile) wide? If so, they appear to be called "jail bars" by fans of the redesigned NES.

by on (#83451)
Dwedit wrote:
Yes, then let's call it the TurboGrafx 16.


Indeed.

by on (#83459)
MottZilla wrote:
If you're going to start enhancing the NES, another 6K of WorkRAM ($800-$1FFF) would be nice. DMA from CPU bus to PPU bus would be handy too. While we are at it, overclocking without altering audio generation would be very neat as well. Other useful things would be a built in PPU based IRQ, like MMC5's scanline counter functionally. Oh another 2K for 4 nametables total would be nice to include as well.


If you really wanted to, (with the exception of overclocking) couldn't you do most of this on a cart with an unmodified NES?

by on (#83463)
I guess... Many mappers already have extra RAM and PPU-based IRQ. Some have more VRAM for nametables. DMA is probably doable too (it would be pretty cool to be able to transfer 1000+ bytes to VRAM per VBlank!).

Overclocking is the only feature in that list that would improve existing games, but it's also the only one that can't be put in a cart. You could have another processor in the cart though (a second 6502 maybe), running faster than the one in the NES, but only new software would take advantage of it.

by on (#83471)
tokumaru wrote:
[In a mapper,] DMA is probably doable too (it would be pretty cool to be able to transfer 1000+ bytes to VRAM per VBlank!).

Dual-ported VRAM/WRAM is possible on a cart (just ask any emulator author who has got MMC5 games to work), but it's probably too big to fit on a CPLD.

Quote:
Overclocking is the only feature in that list that would improve existing games, but it's also the only one that can't be put in a cart. You could have another processor in the cart though (a second 6502 maybe), running faster than the one in the NES, but only new software would take advantage of it.

Think of the essay one could write about how that would improve Super NES games like Kirby and Super Mario RPG.

As for the TG16: If only it were cheaper to get one around here.

by on (#83475)
tepples wrote:
As for the TG16: If only it were cheaper to get one around here.

What about "If only it had been popular anywhere outside of Japan so that it was more accessible to everyone nowadays"?

The TG16 is nice and all, but it's a rarity in most places of the world. The NES, on the other hand, was so popular everywhere that it's still really easy and cheap to find consoles (even if they are clones), carts (even if they are pirates) and accessories. I believe that's one of the greatest advantages about the NES when it comes to retro consoles.

by on (#83487)
You're correct about that. It's just a shame that many clones are poorly constructed with the mentioned problems.

About the TG16/PCE, it's a shame no one has yet cloned it. Probably due to its lesser popularity and also because it would be wise to clone the CDROM too given the number of great games that are CD or Super CD. If one were to dream though, cloning the DUO but with the Super Grafx hardware and Arcade Card built in would be the coolest.

by on (#83489)
How much does a CD drive cost compared to a cart slot? And how much would it cost to get a PCE-on-a-chip ASIC fabricated including the over 2 MB of RAM on an Arcade Card?

by on (#83511)
I don't know, but given that PC-Engine DUOs often go used for anywhere from $150 to $300 I don't think price is a huge concern. I forget exactly what I paid for my refurbished DUO-R with RGB mod, over $200. There are people that would pay for a PC Engine with SuperGrafx and Arcade Card compatibility built in, so long as the cost wasn't astronomical.

by on (#83512)
Any ASIC production involves hundreds of thousands $ and up to millions, so a FPGA-based clone is much more realistic. This would cost few hundreds $ per unit.