What is the real core of ATARI's 6502C CPU

[Nickolasgaspar]

1 minute ago, ivop said:

I was talking about the die of Sally. I.e. what's inside the chip. Did Atari copy the MOS die and added some 74xx circuitry (on the die), or did they design their own NMOS die?

I think the answer is obvious since the early machines (400 800) came with a standard 6502 and those additional logic chips and the daughter board in this video can run on an XL machine with a standard 6502.

[ivop]

3 minutes ago, Nickolasgaspar said:

I think the answer is obvious since the early machines (400 800) came with a standard 6502 and those additional logic chips and the daughter board in this video can run on an XL machine with a standard 6502.

https://en.wikipedia.org/wiki/Die_(integrated_circuit)

 

This has nothing to do with daughter boards or external logic.

Edited November 4, 2021 by ivop

[+Stephen]

3 minutes ago, Nickolasgaspar said:

I think the answer is obvious since the early machines (400 800) came with a standard 6502 and those additional logic chips and the daughter board in this video can run on an XL machine with a standard 6502.

No it is not obvious - the 800 CPU card may have been pin compatible on the edge card.  I don't believe the 40-pin 6502 is pin compatible.  In fact, I know it is not.  You cannot swap the CPUs across both generations of cards.  You can't swap in a 1979 800 CPU in a 130XE.  I.E., the inside of the 6502 - the die, is different.

 

EDIT:

Damnit - beat me by a few seconds.

[ijor]

1 hour ago, Nickolasgaspar said:

I never said that our posts contradicted each other. In a discussion one can add to other people's comments.

 

Oh, I see.

Quote

There was also no reason to put binned cpus in Atari 8bit and C64 machines because it would affect the rest of their architecture. As far as I know the PCs were the first to allow such upgrades without the need of any daughter boards...and this was their recipe of success.

 

Putting "faster" chips (faster in the sense that they would match faster speed grades) wouldn't have affected the architecture unless you actually run the chip at a faster frequency. And as it was pointed out in this thread already, there are reasons why you would prefer faster chips even if you run them at the same slower frequency.

 

[ijor]

4 hours ago, ivop said:

It would be interesting to know if Sally is just the known MOS 6502 die with some extra logic tacked to it, or not

 

I haven't seen a Sally die shot, but I would bet it is a bog 6502 mask with the extra logic. But yes, we need to see die shots to confirm.

 

Quote

Did Atari copy the MOS die and added some 74xx circuitry (on the die), or did they design their own NMOS die?

I don't think Atari copied the die. Some of the manufacturers, that already have access and rights to the original MOS mask, probably just added the logic to the original mask.

Edited November 4, 2021 by ijor
Edit: Added and commented to a second quote

[+DrVenkman]

1 hour ago, ivop said:

I was talking about the die of Sally. I.e. what's inside the chip. Did Atari copy the MOS die and added some 74xx circuitry (on the die), or did they design their own NMOS die?

Well, it wouldn't have been a "copy" in the sense of a stolen design - they had a license from MOS to use it. As I've posted in this very thread, I have an actual MOS-produced SALLY from one of my 5200's. That - to me - strongly suggests it's the same basic die with a few extra logic gates lithographed into otherwise-unused space somewhere around the periphery of the die.

 

It might be fun to get someone like CuriousMarc and Ken Shirriff to de-cap a SALLY for Marc's YouTube channel and do a visual comparison.  

[Nickolasgaspar]

5 hours ago, ijor said:

Oh, I see.

 

Putting "faster" chips (faster in the sense that they would match faster speed grades) wouldn't have affected the architecture unless you actually run the chip at a faster frequency. And as it was pointed out in this thread already, there are reasons why you would prefer faster chips even if you run them at the same slower frequency.

 

That is not the sense I use the phrase "faster chips". When I say faster, I mean to run faster.

Faster chips means more money. That makes no sense in our economic system even back in the 80s.

[Nickolasgaspar]

6 hours ago, ivop said:

https://en.wikipedia.org/wiki/Die_(integrated_circuit)

 

This has nothing to do with daughter boards or external logic.

How a wiki link on dies is relevant to the modified version of a 6502?

[ClausB]

6 hours ago, DrVenkman said:

get someone like CuriousMarc and Ken Shirriff to de-cap a SALLY

I once heard Marc say he finds consumer electronics uninteresting, but Ken might be more open-minded.

[Nickolasgaspar]

8 hours ago, Stephen said:

No it is not obvious - the 800 CPU card may have been pin compatible on the edge card.  I don't believe the 40-pin 6502 is pin compatible.  In fact, I know it is not.  You cannot swap the CPUs across both generations of cards.  You can't swap in a 1979 800 CPU in a 130XE.  I.E., the inside of the 6502 - the die, is different.

 

EDIT:

Damnit - beat me by a few seconds.

Well I don't really understand how you "know it is not". You can literally  google the pinout of both versions and find out that Sally just uses two previously disconnected pins (#35,36)  to implement a specific function previously performed by 4 logic chips on the main board.

Who knows what other changes were made. We weren't members of the "board" during that time to "know" anything. The Default Position in Logic is defined by the available evidence.

The things we know about the "Liz" project, their identical pinouts, the ability to reproduce Sally's function by adding some external logic chips, the cost of designing and producing new CPU architecture while there was already a die available   and of course the use of the name "6502"...all provide sufficient and necessary evidence to dismiss further unnecessary assumptions.(Parsimony).

Maybe there were additional changes...but the moment to accept that claim is only after we can put our fingers on objective evidence.

A  C014377(400/800 models) is a standard 6502. The pins 35,36 are disconnected and there isn't a logic circuit on the XE mainboard to perform the needed function hence allow a 130xe to function.

[Nickolasgaspar]

12 hours ago, DrVenkman said:

You don't need to "reverse engineer" anything except to compare the two versions of the Atari 800 Personality Board - early versions have standard 6502B chips and external logic for the HALT line; later versions use SALLY chips. The boards are interchangeable so the edge connectors are identical between the versions. 

 

For that matter, the Atari 800 Field Service Manual has schematics for the whole system. Just read that. 

To be fair, in the episode 3 he admits he is using a 400/800 cpu board as  a blue print. Well he is Dutch so maybe the "reverse engineer" statement is a "language barrier" issue.

 

[+DrVenkman]

3 hours ago, ClausB said:

I once heard Marc say he finds consumer electronics uninteresting, but Ken might be more open-minded.

And yet he was restoring an IBM PS/2 on his channel just a few weeks ago. *shrug*

[Rybags]

ed - I posted this thinking I'd read the leadup but had missed the last page so inadvertantly repeted some stuff others have said.

I don't think Atari designed anything other than the additional logic to incorporate the /Halt function onto the die (and in saying that, they could have gotten the work-experience kid to do it in his lunchbreak).

It would have been mass overkill to design a new die and if you did so then why not include fixes and improvements?  And if you look at the pinouts they're almost the same so the bonding wiring wouldn't have changed much.

 

The added components I doubt would account for near 1% growth and likely would have fit without need to change the die size.  And they would have been external to the existing components only needing interaction with the Phi0 and /Halt inputs which happen to be conveniently close to each other.

I'm no expert on chip making but I do suspect that such "afterthoughts" could probably be implemented into existing masks without too much trouble.

Edited November 4, 2021 by Rybags

[+Stephen]

6 hours ago, Nickolasgaspar said:

16 hours ago, Stephen said:

No it is not obvious - the 800 CPU card may have been pin compatible on the edge card.  I don't believe the 40-pin 6502 is pin compatible.  In fact, I know it is not.  You cannot swap the CPUs across both generations of cards.  You can't swap in a 1979 800 CPU in a 130XE.  I.E., the inside of the 6502 - the die, is different.

Well I don't really understand how you "know it is not". You can literally  google the pinout of both versions and find out that Sally just uses two previously disconnected pins (#35,36)  to implement a specific function previously performed by 4 logic chips on the main board.

I think we're both saying the same thing.  The core 6502 is the same and not modified, but the overall die of course has extra logic in it to account for the halt circuitry.  It would definitely be awesome to see a decapped version.

[ClausB]

4 hours ago, DrVenkman said:

And yet he was restoring an IBM PS/2 on his channel just a few weeks ago. *shrug*

I saw that. Guess he has a soft spot for IBM (and a hard-on for HP).

[+kheller2]

Okay I have to say it... no one else has followed up with this meme:

 

"My god, it's full of stars"

[Roydea6]

On 11/4/2021 at 9:41 AM, kheller2 said:

"My god, it's full of stars"

[_The Doctor__]

a further tidbit, Atari 6502C Sally has been sold from China as a 6502, they mark them as a 6502 and sell them. You can use a NOP generator Test PCB to identify these remarked chips.

 

Getting a real Sally isn't cake but it's not as impossible as folks make it sound.

 

edit* add video about one such tester...

 

Edited March 18, 2023 by _The Doctor__

[_The Doctor__]

re labeled 6502C 'Sally' from China says its a 6502 A code 0222

others exist, time for folks to check and search for them maybe make a list of markings and pictures.