Versions of the NES

[JimDaBim]

No, it's nothing specific. I was just asking myself: Since there are some revisions of games (PRG0, PRG1), you might want to get a specific version because in it a certain bug from a previous version is fixed etc. And I wanted to know if the same is true for the NES console itself: That certain revisions of the console might be better than others because some consoles might treat certain games incorrectly or something like that.
That was my only motivation for asking it. I just wanted to know if something like that exists and if the consoles might differ in the same way two revisions of a game might differ.

[Dwedit]

I'm sure Nintendo would have had no problem doing cost-reducing slight redesigns.

[BMF54123]

I have a very old toaster NES (smooth plastic on the bottom, very low serial number) which has noticeably darker video output than all of my later models, close to that of my AV Famicom. I used to prefer using it over my other toasters because the picture looked so much cleaner (sharper edges and less fringing) on my CRT.

My PowerPak has also exhibited different glitches on different consoles; on my AV Famicom, it shows a duplicate cursor in the main menu that jumps around erratically. This may be due to the 72-to-60-pin converter I'm using adding resistance or otherwise changing the characteristics of the data lines, though (I have a bootleg Takahashi Meijin no Bouken Jima 3 cart that has similar issues in a toaster when using a converter).

[MottZilla]

Mine shows a dupe cusor but only for awhile after power up. If you move it around a bit it eventually goes away and there is only one cursor. I have no idea of the date on my NES console.

[thefox]

Another case of manual $2004 writes not working properly.

[tokumaru]

Another case of manual $2004 writes not working properly.

So we have evidence of this? Last time I mentioned having heard that manual $2004 writes were not reliable everyone looked at me like I was crazy!

[blargg]

I think I had some notes of mine that manual $2004 reads/writes only work reliably on a certain CPU-PPU alignment.

[lidnariq]

I asked about this a couple months ago. viewtopic.php?f=9&t=9912
Quietust had a few ideas as to what might be breaking (more OAMADDR oddness). There were a few confirmations, too.

Regarding alignment, at least for me, it seems to be more often a thermal issue. But "alignment" plus "thermal" would be code for "timing", since ICs get slower as they get hotter.

[Bregalad]

Alignment is, as far as we know, the only difference between a manual write to $2004 and use of $4014. So that might be the only reason manual writes doesn't always work.

[blargg]

Wouldn't it be the same as far as the PPU is concerned? Manual $2004 and $4014 writes involve the 2A03 doing $2004 writes to the PPU with the same timing.

[tepples]

Manual $2004 and $4014 writes involve the 2A03 doing $2004 writes to the PPU with the same timing.

Are you sure it's the same timing, even down to the half master clock? With manual $2004 writes, the 6502 is doing the writes. With $4014 writes, the DMA unit is freezing the 6502 and doing the writes itself. Besides, in a DMA, the cycle prior to the write was a read of the correct value, leaving the correct value on the data bus, while in a 6502-mediated write, the previous value on the data bus was more than likely $20 for the high byte of the destination address.

[lidnariq]

I am pretty certain that there is no mechanism for the M2 duty cycle to be anything other than 5/8 on the 2A03.

[tepples]

But that still leaves 1. when the address lines become valid, which affects the chip select going into the PPU, and 2. what value is held by bus capacitance before the write cycle begins. Perhaps in the STA $2004 case, the CPU isn't meeting the PPU's setup and hold requirement on some alignments.

[Bregalad]

Then you should see if doing any writes to $2004 make sprites use tile $20 at position ($20, $20) behind the background. If so, then it's possible that the PPU somewhat latches the data lines one cycle before the actual write and use this value ?

[lidnariq]

... You know, I tentatively think I saw that as one of the failure modes in my "just how cranky" thread.