C014806-03 vs C014806-12

[adrm]

Hi.

 

I have just received my first Atari 8-bit computers; two PAL XEGS. One works, the other doesn't.

(I have ordered up replacement RAM for the faulty one.)

 

But my question is: What is the difference between the C014806-03 and the C014806-12 CPUs?

 

My guess is that the -03 is a 3MHz chip for use in PAL units (this is what I have in my XEGS') and that the -12 is the 2MHz version for use with NTSC units.

I'd be grateful for a clarification, as I like to order up a few spare parts to have on hand.

 

 

Cheers,

Tore

[+mytek]

All A8's use a 2 Mhz processor irrelevant of whether they are PAL or NTSC based machines. The suffix on the part numbers you showed has to do with manufacturing batch/date and nothing to do with speed. Either one should work.

[adrm]

Ahh, thank you.

 

The confusion came from a paragraph I found in a Service Manual for the Atari 400/800 computers which said:

 

The domestic 6502 operates at 2.0 MHz, and the PAL & UK 6502 operates at 3.0 MHz. The increased speed is required to operate within PAL & UK television standards.

[+MrFish]

All A8's use a 2 Mhz processor irrelevant of whether they are PAL or NTSC based machines. The suffix on the part numbers you showed has to do with manufacturing batch/date and nothing to do with speed. Either one should work.

 

I was just reading this thread the other day, which would point to otherwise. The 6502B's we see in 400/800's should be 3MHz.

 

Real Core of 6502C (Sally)

[+mytek]

Thanks MrFish I stand corrected on the speed rating that became a standard for all machines after the 400/800 series. Apparently they were 3 Mhz rated parts by the time the custom '02 version (Sally) was created for Atari (although the discussion you linked to kind of goes all over the map about that). These also got used in the later 400/800 series as well, and it looks like there was a mixture of standard 6502 2 Mhz and 3 Mhz parts in use prior to the custom version coming online. But only a 2 Mhz speed rating was ever really required for either a PAL or NTSC machine to work properly (don't let the 3.5x crystal frequency fool you, since that gets divided in half before clocking the CPU).

The interesting thing is if Atari had used the 4 Mhz 6502 as the basis for their custom version, there might have been an upgrade path for a considerably faster machine based on the 3.5x crystal frequency not being halved. Although that would have required a faster version of the other custom chips as well.

Anyways back to the OP's original question... either part you specified will work in a PAL or NTSC machine. And I don't believe the suffixes on those had anything to do with speed.

[adrm]

Thanks, both.

[leech]

Might the difference between -12 and -03 be that the -03 is made in Singapore and the -12 is made in Mexico?

 

At least that is what it says on the bottom of my chips.

[_The Doctor__]

I have had numbers of -12 's fail but almost no -03's have failed, that's the only difference to me....

[Kyle22]

That seems correct as far as I know. Mexicans can be BAD. Don't trust the -12.

 

[Kyle22]

65C802 is always a nice upgrade.

816 is better if you have the RAM.

-xxl and IllOps. Disgusting.

 

[leech]

3 minutes ago, Kyle22 said:

65C802 is always a nice upgrade.

816 is better if you have the RAM.

-xxl and IllOps. Disgusting.

 

Ha, well I just pulled a -12 out of my 1200xl and swapped it with the -3 to see if it was the CPU causing my issues (it was not).  So maybe when I send off the 800 w/incognito to someone more qualified than myself to fix, they can determine if the Mexican made CPU should be swapped again... would love a Rapidus in the 800 though...

[Kyle22]

Only if you have an original 800 CPU board can you plug in a 65c802. The 65c816 requires a jumper wire.

Ask Bob Woolley. He is an expert and a friend. He is @bob1200xl

 

What we need is a 65c816 with as much RAM as possible (16M?) in an original 6502 or SALLY form factor. There must be 2 versions. Other 6502 people will buy if a drop in 65c816 with high RAM is available, the other one is a drop in replacement for C014806 Sally CPU.

A SMALL form factor CPU + RAM would be really nice for all of us 800 users and those who don't need Rapidus  [YET].

Simply make sure that all hardware registers are blocked from high address ranges.  $0D01F <> $xD01F The first 64K must be handled separately.

 

Just a suggestion. Feel free to delete or whatever...

 

We didn't have the the technology or the cheap RAM to do this on the FTe Sweet-16. I wanted to, but the cost was prohibitive.

Sorry if some think that I de-railed the thread. I was only talking about CPU replacements and what is possible.

 

 

 

Edited May 26, 2020 by Kyle22
Typo

[Xebec]

9 hours ago, Kyle22 said:

Only if you have an original 800 CPU board can you plug in a 65c802. The 65c816 requires a jumper wire.

Ask Bob Woolley. He is an expert and a friend. He is @bob1200xl

 

What we need is a 65c816 with as much RAM as possible (16M?) in an original 6502 or SALLY form factor. There must be 2 versions. Other 6502 people will buy if a drop in 65c816 with high RAM is available, the other one is a drop in replacement for C014806 Sally CPU.

A SMALL form factor CPU + RAM would be really nice for all of us 800 users and those who don't need Rapidus  [YET].

Simply make sure that all hardware registers are blocked from high address ranges.  $0D01F <> $xD01F The first 64K must be handled separately.

 

Just a suggestion. Feel free to delete or whatever...

 

We didn't have the the technology or the cheap RAM to do this on the FTe Sweet-16. I wanted to, but the cost was prohibitive.

Sorry if some think that I de-railed the thread. I was only talking about CPU replacements and what is possible.

 

 

 

It would be nice to have a version of this with higher clock speeds / cache or similar ideally too..   I think even a clock doubled 6502 without cache would help for a good number of instructions that aren't waiting on the bus.. 

[drac030]

5 hours ago, Xebec said:

 I think even a clock doubled 6502 without cache would help for a good number of instructions that aren't waiting on the bus.. 

I had an occasion to test that long time ago on a 10 MHz accelerator board prototype (then called F7). Clocking the R/W operations with 1.77 MHz and the internal CPU operations with 10 MHz gave average speedup of more than 20% - basically it felt like the Antic DMA was off while it was on.

 

But the 65C816 has signals which allow to tell easily, if a clock cycle is a R/W cycle or an internal operation. The 6502 does not have such luxury, every bus cycle is a busy one.

[KLund1]

15 hours ago, Xebec said:

It would be nice to have a version of this with higher clock speeds / cache or similar ideally too..   I think even a clock doubled 6502 without cache would help for a good number of instructions that aren't waiting on the bus.. 

In fact Bob1200XL has already made several prototype versions that use the 816. He demo'd a working version at our SLCC table at VCF-West last year in one of his 1200XL. He keeps adding switchable features. I think he is working on the Antic/GTIA speed bottleneck too. PM him for details.