And now I feel very silly. Thanks for enlightening me.
Help collecting pictures of CPU and PPU packages
Moderator: Moderators
- Individualised
- Posts: 310
- Joined: Mon Sep 05, 2022 6:46 am
Re: Help collecting pictures of CPU and PPU packages
Re: Help collecting pictures of CPU and PPU packages
Or maybe the specific PPU that Kevtris analyzed there was just defective for some other reason, and the reason that we've never seen any games that rely on this unique palette is that Nintendo didn't know they'd made it.
If pin 30 is the clock divider, you should just see M2 change from one frequency to another when you bring the pin high/low, so that should be easy to test.
As far as testing the tuning tables, let me go write a simple test that will run through them and compare the pitches that they "should" make to the pulse channels.
As far as the ones that are /RDY - that's trickier, because MOS wrote that (for the plain 6502) RDY must only change during φ1, and the PC10 hardware gets that exactly backwards. (As a result, the supervisor Z80 can crash the 2A03, but has to reboot the 2A03 to restart it)
Re: Help collecting pictures of CPU and PPU packages
I just tried playing around a bit with a /16 UA6527P (an 8916S), a PAL RP2C07-0, and a master clock at 26.601712 MHz (1.66 MHz CPU clock). PAL games seem to play reasonably correctly and everything looks good in terms of video output. The sound is mostly okay, but I thought I could detect a few effects that sounded a bit off, but then again that could just be because the CPU is a clone chip after all.lidnariq wrote: ↑Tue Sep 27, 2022 6:10 pm If pin 30 is the clock divider, you should just see M2 change from one frequency to another when you bring the pin high/low, so that should be easy to test.
As far as testing the tuning tables, let me go write a simple test that will run through them and compare the pitches that they "should" make to the pulse channels.
As far as the ones that are /RDY - that's trickier, because MOS wrote that (for the plain 6502) RDY must only change during φ1, and the PC10 hardware gets that exactly backwards. (As a result, the supervisor Z80 can crash the 2A03, but has to reboot the 2A03 to restart it)
I can test pin 30 a little later. Which specific revisions were you interested in me checking out?
Re: Help collecting pictures of CPU and PPU packages
Here's a very simple test for the periodic noise sounds.
It just endlessly plays pairs of notes: first the noise channel, then a square wave that should be almost the same pitch, then silence. The first sixteen pairs are the NTSC table, the next sixteen pairs are the PAL table. If the first sound mostly right and the second mostly don't, the chip is using the NTSC table. Vice versa, it's using the PAL table. If it sounds like white noise, it's a 2A03 without version (or a clone). If it sounds like tones but they're all wrong, who knows?
Only the first ten/eleven tones are in range for the NTSC/PAL tables respectively, afterwards I play the same notes at higher octaves. Doesn't use the PPU at all (there's no video output).
It just endlessly plays pairs of notes: first the noise channel, then a square wave that should be almost the same pitch, then silence. The first sixteen pairs are the NTSC table, the next sixteen pairs are the PAL table. If the first sound mostly right and the second mostly don't, the chip is using the NTSC table. Vice versa, it's using the PAL table. If it sounds like white noise, it's a 2A03 without version (or a clone). If it sounds like tones but they're all wrong, who knows?
Only the first ten/eleven tones are in range for the NTSC/PAL tables respectively, afterwards I play the same notes at higher octaves. Doesn't use the PPU at all (there's no video output).
- Attachments
-
- 2a03noise.zip
- (2.96 KiB) Downloaded 54 times
Re: Help collecting pictures of CPU and PPU packages
Results!lidnariq wrote: ↑Tue Sep 27, 2022 7:34 pm Here's a very simple test for the periodic noise sounds.
It just endlessly plays pairs of notes: first the noise channel, then a square wave that should be almost the same pitch, then silence. The first sixteen pairs are the NTSC table, the next sixteen pairs are the PAL table. If the first sound mostly right and the second mostly don't, the chip is using the NTSC table. Vice versa, it's using the PAL table. If it sounds like white noise, it's a 2A03 without version (or a clone). If it sounds like tones but they're all wrong, who knows?
Only the first ten/eleven tones are in range for the NTSC/PAL tables respectively, afterwards I play the same notes at higher octaves. Doesn't use the PPU at all (there's no video output).
These were all done with a system at PAL timings (26.601712 MHz clock), your provided ROM (2a03noise.nes), an EverDrive-N8, and a PAL PPU (RP2C07-0).
Code: Select all
CPU Code Table
--- ---- -----
RP2A07 7C4 39 PAL
RP2A07A 1GM 36 PAL
UA6527P (No logo) 9027AC NTSC
UA6527P (Old logo) 9025S NTSC
UA6527P (New logo) 9023 NTSC
UA6540 (Old logo) 8833S NTSC
But I am kind of surprised about the UA6540 also having the NTSC table, since it's actually intended to be used as a PAL chip, isn't it?
Re: Help collecting pictures of CPU and PPU packages
Hey, I found some new chips!
RP2C02E wasn't listed on the wiki at all, so I added it.
RP2C02E wasn't listed on the wiki at all, so I added it.
Re: Help collecting pictures of CPU and PPU packages
That is definitely surprising! But maybe they didn't expect people to care about the tonal noise?
Here's a new test that tests both the DMC channel and the noise rate. It only bothers with one in each table... (noise:NTSC test PAL, gap, dmc:NTSC test PAL)
Now I think about it, I should remake the first test with a DMC test too.
Re: Help collecting pictures of CPU and PPU packages
No idea, they were bought loose, but from someone who desoldered them.
Re: Help collecting pictures of CPU and PPU packages
Results are the same as above, in terms of whether the middle tone matched the pitch of the first tone (NTSC) or the third tone (PAL). None of the chips had a split result, aka the result of the first set of three tones always matched the result of the second set, for each chip in the list.lidnariq wrote: ↑Wed Oct 05, 2022 3:11 pmThat is definitely surprising! But maybe they didn't expect people to care about the tonal noise?
Here's a new test that tests both the DMC channel and the noise rate. It only bothers with one in each table...
whichapu.zip
(noise:NTSC test PAL, gap, dmc:NTSC test PAL)
Now I think about it, I should remake the first test with a DMC test too.
Re: Help collecting pictures of CPU and PPU packages
Wild! So it seems there's no difference between the 6540 and the 6527Ps without a - in their date stamp. How confusing...
Re: Help collecting pictures of CPU and PPU packages
That’s what I was thinking… are there other possible differences? I wonder if they use the same die? I probably don’t have the tools to check, but I’d be happy to donate some chips to find out.
Re: Help collecting pictures of CPU and PPU packages
Uh, prescaler, noise pitch table, DMC pitch table, frame counter, length counters ... that's basically all of the differences between the 2A03 and 2A07. Some bug fixes, but we don't know what they all are... only one that I remember is the 2A03E/G's DMA glitches were fixed in the 2A07 but I'd be shocked if this doesn't have them given that nothing else was changed.
My understanding is that for this level of decapping, you could use the "Chips a la Antoine" technique without any problems. But the dice of the UA6527 that org and others already made don't show highly visible differences relative to the 2A03 anyway.
Would also be interesting to see if the 2A07A has the APU test registers or if pin 30 is still /RDY. (i.e. does it crash if you pull pin 30 high, or do the joypads stop working)
My understanding is that for this level of decapping, you could use the "Chips a la Antoine" technique without any problems. But the dice of the UA6527 that org and others already made don't show highly visible differences relative to the 2A03 anyway.
Would also be interesting to see if the 2A07A has the APU test registers or if pin 30 is still /RDY. (i.e. does it crash if you pull pin 30 high, or do the joypads stop working)
Re: Help collecting pictures of CPU and PPU packages
Okay, so, who has a microscope?