STE video chip

[exxos]

I had a quick look into this issue. It looks like the 16Mhz clock is to blame for the issues.

16Mhz clock C300588

IMG_0163.JPG (254.73 KiB) Viewed 3142 times

16Mhz clock C301712

IMG_0164.JPG (302.37 KiB) Viewed 3142 times

The C300588 16MHz clock actually looks better

Next up, 16MHz vs 32Mhz.


C300588

IMG_0166.JPG (313.36 KiB) Viewed 3142 times

C301712

IMG_0165.JPG (330.78 KiB) Viewed 3142 times

Falling edges on the C300588 look a lot better synchronised than the C301712 .

Thing is here, that even a 10X probe on the 16Mhz clock is enough to drastically change the results of the video for fault ( which also crashes the machine). So the slight skew on the clocks on the C301712 could may well just be enough to destabilise the machine.

[Badwolf]

This is very interesting, but I confess I'm a bit confused.

I thought the 'good' result you got was when using the -588 and putting a -1712 in a -588 machine caused glitching, your latest post seems to imply it's the other way around.

Trying to figure out from the schematic what pin is which. Is the 'MC' line to pin 53 the 32MHz one?

Screenshot 2022-01-10 at 22.27.38.png (134.76 KiB) Viewed 3109 times

Cheers,

BW

[exxos]

This is very interesting, but I confess I'm a bit confused.

I thought the 'good' result you got was when using the -588 and putting a -1712 in a -588 machine caused glitching, your latest post seems to imply it's the other way around.

ah crap, the problem is the 1712 is dated 1990, so I assumed it was the "good" chip. I will double check..

Trying to figure out from the schematic what pin is which. Is the 'MC' line to pin 53 the 32MHz one?

yep MC is 32Mhz.

[exxos]

Quickly retested.

C300588 is the GOOD one. Dated 1988.

588.JPG (326.61 KiB) Viewed 3093 times

C301712 is the BAD one - Dated 1990

1712.JPG (308.66 KiB) Viewed 3093 times

ste.jpg (184.4 KiB) Viewed 3093 times

Maybe Atari changed manufacture of the chip.. So actually the 1712 is 1991, the 588 1989. So a year between the 2 chips something happened..

I guess / assume that a buffer of like 5ns on the 16MHz line would at least get the clocks back in sync. As to if that would solve it or not is another thing..

Edit:
I've got 4 other 1712 which have the same problem, 3 are 1991 and one is 1992. Didn't think these chips were still in production in those years.

[sporniket]

My STE has a C301712-001 (1990) Shifter, with a C302183-002/B (1991) MCU, and I don't have this problem. (but I use it only occasionnally for short sessions, as I didn't recapped the power supply yet).

[Badwolf]

Quickly retested.

C300588 is the GOOD one. Dated 1988.
C301712 is the BAD one - Dated 1990

Aha, that makes more sense now.

To be clear, the 32MHz 'MC' line is an input to the chip and the 16MHz line is an output that's sent to the MCU, is that correct?

(No immediate use, but I'm interested as I have an STE and don't really understand how the shifter works and have been thinking about graphics recently)

Thanks,

BW

[sporniket]

@Badwolf Yes, it is correct. (p. 20 of my converted schematics)

The "MC" is proxyed by the selector U405(p 21 of said schematics), that select between a "32MHz" clock, either by a simple 32MHz oscillator on non-PAL systems (the exact frequency depends whether the it is for NTSC or not), or a 32 MHz oscillateur modulated by a line phase generated by a sub-carrier clock for PAL systems (see p.8), either by external clock (genlock) from the video port (see p. 21)

(EDIT : yes, shameless self quoting of my work )

[exxos]

My STE has a C301712-001 (1990) Shifter, with a C302183-002/B (1991) MCU, and I don't have this problem. (but I use it only occasionnally for short sessions, as I didn't recapped the power supply yet).

Maybe you could scope the 16mhz and 32mhz ? Even on x1 to see if it creates the problem,

[exxos]

To be clear, the 32MHz 'MC' line is an input to the chip and the 16MHz line is an output that's sent to the MCU, is that correct?

Yep.

[sporniket]

My STE has a C301712-001 (1990) Shifter, with a C302183-002/B (1991) MCU, and I don't have this problem. (but I use it only occasionnally for short sessions, as I didn't recapped the power supply yet).

Maybe you could scope the 16mhz and 32mhz ? Even on x1 to see if it creates the problem,

I don't have a scope, it's my plan for this year.