Boosting STFM: 16MHz bus, 64MHz Shifter

[troed]

I think it just goes from GLUE to MMU & SHIFTER. So probably cutting the track closer to the GLUE might help ?

Yeah this is why it's so confusing. With the track cut (and it's really cut .. ) GLUE, MMU and Shifter DE still have continuity between them. However, the trace I cut obviously does something else (it goes upward and I lose it underneath the WD1772) since the computer stops booting. Repairing the trace fixes it again.

And it really is DE, I'm not cutting something else ..

Oh well. That's an investigation for another day ... I'm really eager to get the "Stefan-mode" working now and then see if resyncing the clocks can fix the floppy ...

[exxos]

Which motherboard is it ? I probably got one here I can have a poke around at...

[troed]

Which motherboard is it ? I probably got one here I can have a poke around at...

It's the C070859-001 REV 2.0 with surface mounted MMU and GLUE. This is the via I cut just above (bright green, smack middle in the image):

de_via.jpg (315.24 KiB) Viewed 5170 times

And this is where the schematics clearly show DE being an output to someplace else in addition to connecting Shifter, MMU and GLUE (but I haven't found in the schematics where it's then an input).

de.png (737.57 KiB) Viewed 5170 times

[exxos]

And this is where the schematics clearly show DE being an output to someplace else in addition to connecting Shifter, MMU and GLUE (but I haven't found in the schematics where it's then an input).

ahhh... 68091... pin 20..

[rpineau]

DE goes to the MFP for the HBL interrupt (instead of wiring HSYNC to the MFP , Atari used DE).

RoRo

[troed]

Thanks guys Mystery solved and I'll make my trace repair permanent ...

(Now I'll ponder how come HBL fires all scanlines even though DE only does something on lines with screen output enabled)

It seems I'll have to cut DE between Shifter and MMU on the bottom side of the motherboard if I want nice vias to solder to.

[troed]

Ah! Now I remember It's not HBL, it's Timer B in line count mode. And that's why it uses DE and not HSYNC since it's really for counting the number of displayed lines.

[rpineau]

Yes timer B, we all did some "HBL" raster line effect at some point in our demo .. using timer B

[troed]

DE cut properly and all three modes now up and running. "Some" calibration of the 2MHz AND to DE still to do in the original "Stefan-mode"

Now to re-sync the clocks in the GAL code.

2mhz_and_de_uncalibrated.jpg (112.42 KiB) Viewed 5133 times

[troed]

All modes producing correct graphics & GAL code changed to have all clocks synced on the falling edge.

Have to power cycle a few times for it to boot, no matter which mode, not sure why. Once it boots it's stable though.

Still no working floppy Have verified that bits are clocked out on the DMA databus. Will have to think, and have in mind that it went from working to non-working with the F161 version of the mod as well. Only thing I did in between was to swap out the (recapped) PSU for another (recapped) PSU.

(I have of course tried with two different drives )