6507 and TIA clocks

[powlow]

Hi everyone !

 

I’m quite new to this list and been lurking around for a while, building little by little my own knowledge of the VCS programming art

 

I would have a little question about the synchronization between 6507 and TIA : Why is the CPU clock divided by three from TIA clock ?

Would have it been possible to “overclock” the 6507 (or maybe a 6502B) by providing full 3.58MHz ?

 

By digging into Stella ML I’ve found an interesting post further on this topic, it’s about getting a WDC 65C02 @14.32MHz and have 4 machines cycles for one color clock ! Unfortunately I’ve haven’t found any follow-up on this topic.

 

Anybody tried something similar ? does it sound feasible ? I like the idea of a Turbo2600

[EricBall]

If you look at the 2600 schematics, you'll see that the TIA Phi0 is connected to the 6507 Phi0in. So the TIA is generating the clock signal for the 6507.

 

The divide by 3 is to keep the 6507 clock within the maximum clockspeed of the part.

 

The problem with overclocking the 6507 it generates the Phi2 clock signal used to access the TIA registers and the RIOT. In fact, the 7800 runs it's 6502 at 1.79MHz (50% faster), but downshifts the clock to 1.19MHz when accessing the TIA (used for sound & paddles) and RIOT.

[+batari]

Hi everyone !

 

I’m quite new to this list and been lurking around for a while, building little by little my own knowledge of the VCS programming art

 

I would have a little question about the synchronization between 6507 and TIA : Why is the CPU clock divided by three from TIA clock ?

Would have it been possible to “overclock” the 6507 (or maybe a 6502B) by providing full 3.58MHz ?

No, because the 6507 would not have been able to handle it. The 6507 should have been good for 1.79 MHz, however, but there must have been a technical reason for running it at 1.19 MHz. Perhaps the TIA or RIOT couldn't deliver data fast enough, or they simply wanted to be able to use the 450ns EPROMs common in 1977.

By digging into Stella ML I’ve found an interesting post further on this topic, it’s about getting a WDC 65C02 @14.32MHz and have 4 machines cycles for one color clock ! Unfortunately I’ve haven’t found any follow-up on this topic.

 

Anybody tried something similar ? does it sound feasible ? I like the idea of a Turbo2600

There was a drop-in chip for the Apple II that ran at 10 MHz. I forget the name. It was a drop-in replacement for the 6502 that had its own cache, to help compensate for the slower memory bus of the Apple II motherboard. So it is "feasible" from a computation standpoint, but of course none of the existing games would work anymore. Also, programming would be difficult since cycle counting would be hard due to the processor's cache, which may or may not write its contents to main memory at the time you want it to.

[powlow]

Thanks a lot for the insightfull answers I like learning those many subtleties !

 

Anyway, I'll go back silently studying VCS until my next brain bug