Project: HDMI/DVI out for STFM

[troed]

So, since I have one (thanks Smonson for letting me beta test and play around with what it can do!) there's one thing I'd like to point out though. It runs off its own 32MHz clock, which means the ST CPU is running at 8.00MHz. That's no problem - most things will work just fine.

However, a real ST runs at 8.01MHz (old STF), 8.02MHz (PAL STFM, PAL STE) or 8.05MHz (NTSC STE). Since the motherboard has a separate clock for the MFP (the "timer chip") in the ST demo coders have for a long time known that they need to compensate for the difference between the timer clock and the CPU clock. This mostly affects the opening of the lower (and maybe top) border in demos.

A CPU clock of 8.00MHz is "unknown" and some demos will simply not work perfectly. I don't want to exaggerate this as an issue, just pointing it out for purists (like myself )

[Smonson]

So, since I have one (thanks Smonson for letting me beta test and play around with what it can do!) there's one thing I'd like to point out though. It runs off its own 32MHz clock, which means the ST CPU is running at 8.00MHz. That's no problem - most things will work just fine.

However, a real ST runs at 8.01MHz (old STF), 8.02MHz (PAL STFM, PAL STE) or 8.05MHz (NTSC STE). Since the motherboard has a separate clock for the MFP (the "timer chip") in the ST demo coders have for a long time known that they need to compensate for the difference between the timer clock and the CPU clock. This mostly affects the opening of the lower (and maybe top) border in demos.

A CPU clock of 8.00MHz is "unknown" and some demos will simply not work perfectly. I don't want to exaggerate this as an issue, just pointing it out for purists (like myself )

I wasn't aware of this until you mentioned it, but I've looked around on Mouser and those other clock values are easy to obtain. But what happens if you run a demo that was written for an 8.01MHz machine on an 8.05MHz one? Same problem?

[Smonson]

OK, they don't have exactly the right ones after all... These are the ST MHz values I found: 32.0424, 32.04245, 32.084988 (PAL), 32.215905.

Unfortunately Mouser's closest clocks in the right size are: 32, 32.11, 32.263.

If it's slightly out, will the top and bottom border openings develop a cumulative error, or will it be a stable error? If it's stable, it may be offscreen anyway - I'm only showing 20 extra lines on the top border and 20 on the bottom, so there are still quite a few lines that are outside the display area.

[troed]

I wasn't aware of this until you mentioned it, but I've looked around on Mouser and those other clock values are easy to obtain. But what happens if you run a demo that was written for an 8.01MHz machine on an 8.05MHz one? Same problem?

Yes - and some ancient demos do have such problems. Most demo coders (European) learned to sort out 8.01 and 8.02 though, but that was done in ways that won't work with 8.00. What will happen is that the lower border (and maybe top) will either flicker or not be displayed at all. And yeah, it means the clocks you found at Mouser won't help either - too far off.

One solution would of course be to run the FPGA off the system clock - but I understand you need to rewrite all timing code for that to work. And also, I guess you would need different code in the FPGA depending on which motherboard clock it is ...

I just wanted to raise it so that no one gets caught out when their system suddenly doesn't show a particular demo correctly

[dhedberg]

Awesome achievement! I'd be interested in buying one if the sync issues with opening the borders can be solved (I mostly use my ST to watch demos, so would hate to not have the possibility to watch old classics without switching back to the old Shifter).

Do you have any plans on doing something similar for the Falcon? Hijacking the RGB signals and bypassing the DAC?

[exxos]

I can donate a osc if needs be, not sure what they are, but got stacks of MB for breakers..

[troed]

I will discuss my interfacing of this into my doubleST here: https://www.exxosforum.co.uk/forum/viewt ... 8998#p8998

[Smonson]

One solution would of course be to run the FPGA off the system clock - but I understand you need to rewrite all timing code for that to work. And also, I guess you would need different code in the FPGA depending on which motherboard clock it is ...

I can donate a osc if needs be, not sure what they are, but got stacks of MB for breakers..

The problem is the FPGA needs a full digital clock signal, and the ST's board (mine, at least) only has the crystal, with part of the clock circuit completed inside the ST shifter. Maybe it's possible to add a clock generator circuit to the little PCB that plugs into the shifter socket. But it sounds like a bit of a pain.

Awesome achievement! I'd be interested in buying one if the sync issues with opening the borders can be solved (I mostly use my ST to watch demos, so would hate to not have the possibility to watch old classics without switching back to the old Shifter).

Right now most demos don't run properly because the shifter model doesn't behave exactly like the ST shifter that most machines had. I think it does behave like a real ST shifter - but an early, uncommon version (the one that Ijor reverse-engineered). Changing the model is no problem, but I need a demo coder's help to find out where the difference lies.

[exxos]

The problem is the FPGA needs a full digital clock signal, and the ST's board (mine, at least) only has the crystal, with part of the clock circuit completed inside the ST shifter. Maybe it's possible to add a clock generator circuit to the little PCB that plugs into the shifter socket. But it sounds like a bit of a pain.

The ST generates 32Mhz signal, can't you use that then ? All the shifter does is convert that to 16mhz for the MMU.

[Smonson]

The ST generates 32Mhz signal, can't you use that then ? All the shifter does is convert that to 16mhz for the MMU.

On mine it doesn't, the crystal connects to 2 pins on the shifter to create the resonant circuit, the clean digital signal is only inside the shifter. If you measure one of the pins off the crystal, you get a low-voltage sinewave.