Missing 2A03 Functionality

Missing 2A03 Functionality
by Great Hierophant on 2005-11-10 (#6437)

As I recall, there seems to be known revisions of the 2A03, the 2A03E, 2A03G and 2A03H. The first was used in early Famicoms, the Nintendobox, Playchoice-10s. The second is by far more common, seen in everything from R/F Famicoms to A/V Famicoms (like mine, manufactured after the N64's release.) The third is very obscure and I have never seen one outside a NES top-loaderor Famicom AV, and even many of those use the more common revision G.

Brad Taylor's 2A03 Technical Reference says that the 2A03E does not support the noise channel's 93-bit short mode. As far as we know no revision E chip was used in consumer based NES hardware, but what about Famicoms? If a game uses this mode and is using a system with a 2A03E, he will hear the 15-bit long mode, which sounds different. I would hope that not many games used the short mode, but someone else must say.

by tepples on 2005-11-11 (#6463)

The older chips might have been used only in the first batch of Famicom consoles that got recalled.

by kingshriek on 2005-11-11 (#6464)

As for a game that uses the 93-bit short mode, Solstice does in its main game theme.

by Quietust on 2005-11-11 (#6468)

I should note that Brad Taylor's 2A03/2C02 "Technical Reference" documents are nothing more than compilations, containing numerous pieces of information that were never properly tested and some of which are outright false. The best example is his claim that the MMC3 watched for 42 toggles of PPU A13 (between 0000-1FFF and 2000-3FFF) to run its scanline counter, despite the fact that neither PPU A13 nor /A13 are even connected to the MMC3 - it actually uses PPU A12 (between 0000-0FFF/2000-2FFF and 1000-1FFF/3000-3FFF) with a simple filter to detect the transition between background and sprite tile rendering.

The idea of earlier CPU+APU revisions lacking looped noise may turn out to be nothing more than a rumour, just as the idea that the Famicom lacked RAW PCM (which was clearly proved false, since many Famicom games use it).

by Great Hierophant on 2005-11-11 (#6469)

Quote:
The older chips might have been used only in the first batch of Famicom consoles that got recalled.

I know that the Famicombox uses it, as shown on Kevin Horton's site. I have also seen it used in a Famicom Twin (which had to have been released no earlier than 1986), as shown at
http://assembler.roarvgm.com/famicomtwin.html
Finally, it is designated in Nintendo of America's Playchoice-10 schematic.

By far the G revision is most common, but some Top-Loaders and Famicom AVs (fortunately not mine) use revision H. I am wary, why disturb a successful design?

by tepples on 2005-11-12 (#6502)

Quietust wrote:
I should note that Brad Taylor's 2A03/2C02 "Technical Reference" documents are nothing more than compilations, containing numerous pieces of information that were never properly tested and some of which are outright false.

But it's the most accurate compilation that many of us in the community know of. I know of at least two of us here who would like to see somebody go through Brad Taylor's docs, find all the specific inaccuracies, explain why they're inaccurate, and post the explanations to the wiki.

Or to put it another way:
NESTECH 2 > Marat's doc
Brad Taylor's docs > NESTECH
What docs > Brad Taylor's docs?

This is post #6502; is this an omen?