4MB Upgrade & 16MHz Booster progress

Troubles with your machine? Just want to speak about the latest improvements? This is the place!

Moderators: Mug UK, Zorro 2, spiny, Greenious, Moderator Team

Post Reply
User avatar
exxos
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 4933
Joined: Fri Mar 28, 2003 8:36 pm
Location: England
Contact:

Re: 4MB Upgrade & 16MHz Booster progress

Post by exxos »

Arne wrote:This is the shift I was referring to:
Image
This could just be a issue on mono mode aswell. Though looking again, I see the scanline is starting to late and actually overlapped down to the next line. I thought it was just shifted to the right., not wrapping around to the next line. Thats like the shifter is not outputting data soon enough for some reason. Then the vertical count gets clocked and it finishes off the last 40 off pixels on a new scan line.

I assume here the MMU keeps count of when the shifter needs to output data, as opposed to just blank border space??

It could just be a mono related problem. The GLUEs video timings will be different, so syncing to the 2mhz clock is probably not timing right for mono mode. Maybe just swap it for a 4mhz clock and see what happens ?!

EDIT:
I also just realised the NAND used is using a NAND to invert the output, So really that should just be a AND gate ?

I have been reading this which gives a lot of good info http://alive.atari.org/alive9/ovrscn1.php

The MMU uses the Vsync signal to reset the video counter with the contents of $FFFF8201 and $FFFF8203.....The DE (Display Enable) signal tells the SHIFTER when to display the border
and when it has to display the usable screen. It also tells the MMU when to send data to the SHIFTER:



Those values can't be right for some reason. Maybe the DE is being driven when its not supposed to so those counters are not being reset ? Maybe that 2mhz input to the AND should be ORed with the Vsync signal ?

It may help if (arne) can measure DE in relation to Vsync. I don't know if DE is supposed to be LO or HI during the vsync pulse. Possible it does not care eitherway...
4MB STFM 1.44 FD- VELOCE+ 020 STE - Falcon 030 CT60 - Atari 2600 - Atari 7800 - Gigafile - SD Floppy Emulator - PeST - various clutter

http://www.exxoshost.co.uk/atari/ All my hardware guides - mods - games - STOS
http://www.exxoshost.co.uk/atari/last/storenew/ - All my hardware mods for sale - Please help support by making a purchase.
http://ataristeven.exxoshost.co.uk/Steem.htm Latest Steem Emulator
User avatar
Arne
Atari Super Hero
Atari Super Hero
Posts: 651
Joined: Thu Nov 01, 2007 10:01 am

Re: 4MB Upgrade & 16MHz Booster progress

Post by Arne »

exxos wrote:You could try linking it out, feed MMU direct from osc 32mhz. Though it might kill the 32mhz as it isn't very "strong", but I think it does work.. just..
I pulled out the 74F04 from its socket and made a short-circuit between CLK32_IN and CLK32_OUT. ST startet as usual, ran Memtest and QuickIndex to be sure. At one point when I clicked [Exit] on QIndex it returned to desktop, but black & white changed i.e. the Desktop was inverted with (mostly) black A: and B: icons. Could not reproduce this effect.
exxos wrote:I thought it was just shifted to the right., not wrapping around to the next line.
The first 40 (or so) pixels in the upper left corner are black-white-black-white...
exxos wrote:Maybe just swap it for a 4mhz clock and see what happens ?!
can do. Soldering iron is still hot :lol:
exxos wrote:So really that should just be a AND gate ?
I exchanged the 74LS00 with a 74F00 and the faster propagation delay did not seem to have any effect.
So, Stefan may used the 74LS00 because he had it at hand.
User avatar
Arne
Atari Super Hero
Atari Super Hero
Posts: 651
Joined: Thu Nov 01, 2007 10:01 am

Re: 4MB Upgrade & 16MHz Booster progress

Post by Arne »

exxos wrote:Maybe just swap it for a 4mhz clock and see what happens ?!
Then I get this again: http://www.atari-forum.com/viewtopic.ph ... 00#p264642
exxos wrote:It may help if (arne) can measure DE in relation to Vsync. I don't know if DE is supposed to be LO or HI during the vsync pulse. Possible it does not care eitherway...
Wilco.
SteveBagley
Captain Atari
Captain Atari
Posts: 187
Joined: Mon Jan 21, 2013 9:31 am

Re: 4MB Upgrade & 16MHz Booster progress

Post by SteveBagley »

exxos wrote:My initial thought is to just tie DE to the 16mhz/32mhz switch. So when GLUE wants to increase the counters in the MMU, it does it all back at default speeds. There might be other issues with clocking the MMU at double speed while loading the shifter registers, but that can probably be downclocked aswell. I don't see a problem with the shifter registers being loaded twice as fast ?
But GLUE doesn't increase the counters in the MMU as I understand it, the MMU does that itself based on one of it's clocks, but it only does it while DE is high (it must be high since the hardware overscan mod replaces it with composite sync and syncs are active low signals).

Of course, that clock in the MMU is arranged such that it only increments it during the SHIFTERs slot and also ensuring that the correct address is put on the MADDR bus, and DCYC is pulsed so that the shifter gets the correct pixel data.

Judging by this 16MHz hack, and the overscan hack, I suspect that the MMU simply ANDs DE with the clock pulse used to update the video counters which has the result of stopping the MMU during the borders and why the overscan mod can extend the picture so easily. But that clock pulse within the MMU is still derived from the 16MHz (or 32MHz) input so if you change that clock the duration of the CPU and SHIFTER slots will change since the MMU won't be aware it's clock is changing -- you'd end up with the short CPU slot/long SHIFTER slot pattern I suggested earlier which will still clock data to the SHIFTER faster than normal.

Steve
Last edited by SteveBagley on Tue Jan 06, 2015 12:21 pm, edited 1 time in total.
User avatar
exxos
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 4933
Joined: Fri Mar 28, 2003 8:36 pm
Location: England
Contact:

Re: 4MB Upgrade & 16MHz Booster progress

Post by exxos »

Arne wrote:
exxos wrote:You could try linking it out, feed MMU direct from osc 32mhz. Though it might kill the 32mhz as it isn't very "strong", but I think it does work.. just..
I pulled out the 74F04 from its socket and made a short-circuit between CLK32_IN and CLK32_OUT. ST startet as usual, ran Memtest and QuickIndex to be sure. At one point when I clicked [Exit] on QIndex it returned to desktop, but black & white changed i.e. the Desktop was inverted with (mostly) black A: and B: icons. Could not reproduce this effect.
Its possible the 32mhz clock isn't strong enough, I know that already. But if it works without that inverter, then really we just need a buffer for the 32mhz clock. Years ago I feed MMU with seperate 16mhz clock input, and it worked. So phase relationships on the clock would drive over several seconds anyway. Only side effect there was the screen had a "wobble". But it made me think that inverter isn't really needed and it just needs a buffer there.

Arne wrote:The first 40 (or so) pixels in the upper left corner are black-white-black-white...
That would indicate video counter problems. Corrupt data there could be indication video start address isn't correct.

Arne wrote: I exchanged the 74LS00 with a 74F00 and the faster propagation delay did not seem to have any effect.
So, Stefan may used the 74LS00 because he had it at hand.
That is what I am thinking. Really that should just be a AND gate. Then one less gate is less delays also.

There is some mention in OVERSCAN3 documents of using H&V sync in conjunction to the DE line. Still hard to figure out this problem as we do not know which counters are wrong yet. As we only mess with DE then assume that. Though with those first few pixels, it suggests to me video start address is not correct.

I have ordered some chips, they should be here tomorrow. Though I am not a home then, so I should be able to start building my own on Thursday.
4MB STFM 1.44 FD- VELOCE+ 020 STE - Falcon 030 CT60 - Atari 2600 - Atari 7800 - Gigafile - SD Floppy Emulator - PeST - various clutter

http://www.exxoshost.co.uk/atari/ All my hardware guides - mods - games - STOS
http://www.exxoshost.co.uk/atari/last/storenew/ - All my hardware mods for sale - Please help support by making a purchase.
http://ataristeven.exxoshost.co.uk/Steem.htm Latest Steem Emulator
User avatar
exxos
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 4933
Joined: Fri Mar 28, 2003 8:36 pm
Location: England
Contact:

Re: 4MB Upgrade & 16MHz Booster progress

Post by exxos »

SteveBagley wrote:
But GLUE doesn't increase the counters in the MMU as I understand it, the MMU does that itself based on one of it's clocks, but it only does it while DE is high (it must be high since the hardware overscan mod replaces it with composite sync and syncs are active low signals).

Of course, that clock in the MMU is arranged such that it only increments it during the SHIFTERs slot and also ensuring that the correct address is put on the MADDR bus, and DCYC is pulsed so that the shifter gets the correct pixel data.

Judging by this 16MHz hack, and the overscan hack, I suspect that the MMU simply ANDs DE with the clock pulse used to update the video counters which has the result of stopping the MMU during the borders and why the overscan mod can extend the picture so easily. But that clock pulse within the MMU is still derived from the 16MHz (or 32MHz) input so if you change that clock the duration of the CPU and SHIFTER slots will change since the MMU won't be aware it's clock is changing -- you'd end up with the short CPU slot/long SHIFTER slot pattern I suggested earlier which will still clock data to the SHIFTER faster than normal.
Some good points. It doesn't sound good for the MMU to to base the count on its own clock..

I can knock up a small STOS program to output the values of those counters if someone knows exactly what addresses to poke at. I assume I can poke at the video start address in RAM. At least we can check if that value is correct.
4MB STFM 1.44 FD- VELOCE+ 020 STE - Falcon 030 CT60 - Atari 2600 - Atari 7800 - Gigafile - SD Floppy Emulator - PeST - various clutter

http://www.exxoshost.co.uk/atari/ All my hardware guides - mods - games - STOS
http://www.exxoshost.co.uk/atari/last/storenew/ - All my hardware mods for sale - Please help support by making a purchase.
http://ataristeven.exxoshost.co.uk/Steem.htm Latest Steem Emulator
SteveBagley
Captain Atari
Captain Atari
Posts: 187
Joined: Mon Jan 21, 2013 9:31 am

Re: 4MB Upgrade & 16MHz Booster progress

Post by SteveBagley »

exxos wrote:I can knock up a small STOS program to output the values of those counters if someone knows exactly what addresses to poke at. I assume I can poke at the video start address in RAM. At least we can check if that value is correct.
There's two sets of registers: $ffff8201, and $ffff8203 are the video address base and store the start address of the screen -- high and mid byte (low byte is always zero on an ST). And the counters which are (read-only) at $ffff8205, $ffff8207 and $ffff8209. I'm not sure STOS will be fast enough to read them though :)

Actually, I've just had a thought what might cause the problem. The MMU gets DE and CLK2 to it's DE input, but the SHIFTER still gets DE. If CLK2 is low when DE goes high then the SHIFTER will start outputting pixels before the MMU starts loading data to it, so the first few pixels on the screen will be random (I guess whatever is left in the SHIFTERs flip-flops).

We know (from the work investigating wakeup states) the relationship between GLUE/SHIFTER/MMU can vary so I wonder if this is the cause? Although this would suggest that DE's timings are not synchronous to GLUE's 2MHz clock…

Steve
User avatar
Arne
Atari Super Hero
Atari Super Hero
Posts: 651
Joined: Thu Nov 01, 2007 10:01 am

Re: 4MB Upgrade & 16MHz Booster progress

Post by Arne »

exxos wrote:It may help if (arne) can measure DE in relation to Vsync. I don't know if DE is supposed to be LO or HI during the vsync pulse. Possible it does not care eitherway...
Image
SteveBagley
Captain Atari
Captain Atari
Posts: 187
Joined: Mon Jan 21, 2013 9:31 am

Re: 4MB Upgrade & 16MHz Booster progress

Post by SteveBagley »

SteveBagley wrote:We know (from the work investigating wakeup states) the relationship between GLUE/SHIFTER/MMU can vary so I wonder if this is the cause? Although this would suggest that DE's timings are not synchronous to GLUE's 2MHz clock…
Although, MMU's DE is delayed through 2 NAND gates I wonder if that could be enough to knock the timings out sufficiently on occasion? Arne, what happens if you delay DE to the MFP and SHIFTER through the other 2 NAND gates in the 74LS00 as well (wired as 2 inverters)?

Steve
User avatar
Arne
Atari Super Hero
Atari Super Hero
Posts: 651
Joined: Thu Nov 01, 2007 10:01 am

Re: 4MB Upgrade & 16MHz Booster progress

Post by Arne »

SteveBagley wrote:Arne, what happens if you delay DE to the MFP and SHIFTER through the other 2 NAND gates in the 74LS00 as well (wired as 2 inverters)?

Code: Select all

              __             __
DE Glue -----|  |         +-|  |
             |& |o--------| |& |o- DE MMU
   CLK2 -----|__|         +-|__|

              __             __
            -|  |          -|  |
             |& |o-         |& |o-
            -|__|          -|__|
Ehmm... what exactly do you want me to do?

Last scope pic was taken from a "good" i.e. non-shifted picture. Shifted is the same.
SteveBagley
Captain Atari
Captain Atari
Posts: 187
Joined: Mon Jan 21, 2013 9:31 am

Re: 4MB Upgrade & 16MHz Booster progress

Post by SteveBagley »

Arne wrote:Ehmm... what exactly do you want me to do?

Last scope pic was taken from a "good" i.e. non-shifted picture. Shifted is the same.
This:

Code: Select all

              __             __
DE Glue -----|  |         +-|  |
             |& |o--------| |& |o- DE MMU
   CLK2 -----|__|         +-|__|

              __             __
           +-|  |         +-|  |
DE Glue ---| |& |o--------| |& |o-- DE SHIFTER
           +-|__|         +-|__|
Probably won't make any difference but is should be an easy test.

Also regarding the black and white pixels in the top-left -- I wonder if they are actually the bottom-right pixels from the previous frame, you should be able to test by moving the mouse pointer into that area.

Also, does your ST have a modulator?

Steve
User avatar
exxos
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 4933
Joined: Fri Mar 28, 2003 8:36 pm
Location: England
Contact:

Re: 4MB Upgrade & 16MHz Booster progress

Post by exxos »

Arne wrote:
SteveBagley wrote:Arne, what happens if you delay DE to the MFP and SHIFTER through the other 2 NAND gates in the 74LS00 as well (wired as 2 inverters)?

Code: Select all

              __             __
DE Glue -----|  |         +-|  |
             |& |o--------| |& |o- DE MMU
   CLK2 -----|__|         +-|__|

              __             __
            -|  |          -|  |
             |& |o-         |& |o-
            -|__|          -|__|
Ehmm... what exactly do you want me to do?

Last scope pic was taken from a "good" i.e. non-shifted picture. Shifted is the same.
I mean not using those NANDS at all. One is just invertering the output of the first NAND so its back to a AND function. It would need a 74F08, just ignore the second NAND gate and just wire DE and 2mhz to the AND inputs.

Steve has a point about DE also, you could try cutting the shifter DE line and linking that to MMU DE.
4MB STFM 1.44 FD- VELOCE+ 020 STE - Falcon 030 CT60 - Atari 2600 - Atari 7800 - Gigafile - SD Floppy Emulator - PeST - various clutter

http://www.exxoshost.co.uk/atari/ All my hardware guides - mods - games - STOS
http://www.exxoshost.co.uk/atari/last/storenew/ - All my hardware mods for sale - Please help support by making a purchase.
http://ataristeven.exxoshost.co.uk/Steem.htm Latest Steem Emulator
User avatar
Arne
Atari Super Hero
Atari Super Hero
Posts: 651
Joined: Thu Nov 01, 2007 10:01 am

Re: 4MB Upgrade & 16MHz Booster progress

Post by Arne »

Did this (74F08):

Code: Select all

                    __
       CLK2 -------|  |
                   |& |---- DE MMU
                 +-|__|
                 |
    DE Glue -----+  __
                 +-|  |
                 | |& |---- DE Shifter
                 +-|__|
No change.
DE Glue = DE MFP
SteveBagley wrote: Also, does your ST have a modulator?
No, it's an original 520 board. Used the HF-Modulator's space to place a 2.5" HDD there.
User avatar
Arne
Atari Super Hero
Atari Super Hero
Posts: 651
Joined: Thu Nov 01, 2007 10:01 am

Re: 4MB Upgrade & 16MHz Booster progress

Post by Arne »

Code: Select all

                    __
       CLK2 -------|  |
                   |& |--+- DE MMU
    DE Glue -------|__|  |
                         +- DE Shifter

Worsened it:
- good pic
- shifted pic
- no pic
- good pic with jail bars (like on the IMP Shifter), but only once till now
User avatar
exxos
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 4933
Joined: Fri Mar 28, 2003 8:36 pm
Location: England
Contact:

Re: 4MB Upgrade & 16MHz Booster progress

Post by exxos »

Arne wrote:

Code: Select all

                    __
       CLK2 -------|  |
                   |& |--+- DE MMU
    DE Glue -------|__|  |
                         +- DE Shifter

Worsened it:
- good pic
- shifted pic
- no pic
- good pic with jail bars (like on the IMP Shifter), but only once till now
Is that a AND with DE & Shifter linked together ? If so did you try the AND before changing the DE line ?
4MB STFM 1.44 FD- VELOCE+ 020 STE - Falcon 030 CT60 - Atari 2600 - Atari 7800 - Gigafile - SD Floppy Emulator - PeST - various clutter

http://www.exxoshost.co.uk/atari/ All my hardware guides - mods - games - STOS
http://www.exxoshost.co.uk/atari/last/storenew/ - All my hardware mods for sale - Please help support by making a purchase.
http://ataristeven.exxoshost.co.uk/Steem.htm Latest Steem Emulator
User avatar
Arne
Atari Super Hero
Atari Super Hero
Posts: 651
Joined: Thu Nov 01, 2007 10:01 am

Re: 4MB Upgrade & 16MHz Booster progress

Post by Arne »

Yes I used a 74F08 the whole day.
User avatar
exxos
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 4933
Joined: Fri Mar 28, 2003 8:36 pm
Location: England
Contact:

Re: 4MB Upgrade & 16MHz Booster progress

Post by exxos »

Arne wrote:Yes I used a 74F08 the whole day.
So Shifter has to be to GLUE directly for it to work the best.. Its going to be a tricky one to track down. I think the thing to do is to try this mod in LOW AND MED res, though looking like I will have to try that one out.
4MB STFM 1.44 FD- VELOCE+ 020 STE - Falcon 030 CT60 - Atari 2600 - Atari 7800 - Gigafile - SD Floppy Emulator - PeST - various clutter

http://www.exxoshost.co.uk/atari/ All my hardware guides - mods - games - STOS
http://www.exxoshost.co.uk/atari/last/storenew/ - All my hardware mods for sale - Please help support by making a purchase.
http://ataristeven.exxoshost.co.uk/Steem.htm Latest Steem Emulator
User avatar
exxos
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 4933
Joined: Fri Mar 28, 2003 8:36 pm
Location: England
Contact:

Re: 4MB Upgrade & 16MHz Booster progress

Post by exxos »

DSCN0986.JPG
Took find finding that motherboard did! Though that's my "start" to it all 8) Next up is to remove the CPU and put a socket in there to fit a 16mhz one. Though the simmboard is already up and working. got piles of 4MB 70ns and 60ns simms so good to go there :) Soon be going under the knife... :twisted:

Currently 16mhz CPU on internal ops, and 16mhz TOS, gave about 50% speed boost. 200% across the board with ST-RAM I think its misleading, as TOS is still running at slower speeds due to GLUE controlling DTACK based on 8mhz. Even so, 200% baseline speed would turn into 250% with fast TOS. Though that was TOS based on 8mhz access to RAM. So I project speeds of 300% when TOS is running on a 16mhz DTACK.

Of course its possible to run the CPU on 32mhz on internal instructions so that is my target after this ST-RAM mod. No reason why alt-ram cannot run along with 32mhz also. I can see a really awesome booster in the pipeline 8)
You do not have the required permissions to view the files attached to this post.
4MB STFM 1.44 FD- VELOCE+ 020 STE - Falcon 030 CT60 - Atari 2600 - Atari 7800 - Gigafile - SD Floppy Emulator - PeST - various clutter

http://www.exxoshost.co.uk/atari/ All my hardware guides - mods - games - STOS
http://www.exxoshost.co.uk/atari/last/storenew/ - All my hardware mods for sale - Please help support by making a purchase.
http://ataristeven.exxoshost.co.uk/Steem.htm Latest Steem Emulator
User avatar
troed
Atari God
Atari God
Posts: 1460
Joined: Mon Apr 30, 2012 6:20 pm
Location: Sweden

Re: 4MB Upgrade & 16MHz Booster progress

Post by troed »

exxos wrote:
Arne wrote:This is the shift I was referring to:
Image
This could just be a issue on mono mode aswell. Though looking again, I see the scanline is starting to late and actually overlapped down to the next line. I thought it was just shifted to the right., not wrapping around to the next line. Thats like the shifter is not outputting data soon enough for some reason. Then the vertical count gets clocked and it finishes off the last 40 off pixels on a new scan line.

I assume here the MMU keeps count of when the shifter needs to output data, as opposed to just blank border space??

It could just be a mono related problem. The GLUEs video timings will be different, so syncing to the 2mhz clock is probably not timing right for mono mode. Maybe just swap it for a 4mhz clock and see what happens ?!

EDIT:
I also just realised the NAND used is using a NAND to invert the output, So really that should just be a AND gate ?

I have been reading this which gives a lot of good info http://alive.atari.org/alive9/ovrscn1.php

The MMU uses the Vsync signal to reset the video counter with the contents of $FFFF8201 and $FFFF8203.....The DE (Display Enable) signal tells the SHIFTER when to display the border
and when it has to display the usable screen. It also tells the MMU when to send data to the SHIFTER:



Those values can't be right for some reason. Maybe the DE is being driven when its not supposed to so those counters are not being reset ? Maybe that 2mhz input to the AND should be ORed with the Vsync signal ?

It may help if (arne) can measure DE in relation to Vsync. I don't know if DE is supposed to be LO or HI during the vsync pulse. Possible it does not care eitherway...
Don't spend a lot time on this. It's easy to, on a completely unmodified ST, get that effect when switching between resolutions. It's due to the Shifter having "leftover" words in its internal registers when the MMU believes it to not be the case.
User avatar
Arne
Atari Super Hero
Atari Super Hero
Posts: 651
Joined: Thu Nov 01, 2007 10:01 am

Re: 4MB Upgrade & 16MHz Booster progress

Post by Arne »

troed wrote:It's due to the Shifter having "leftover" words in its internal registers when the MMU believes it to not be the case.
It never happens when I do a reset via button and it showed a good picture before. Only when switching on the machine. Then additional resets (via button) do not change anything. Only off/on cycle gives me another chance for a good pic.
joska
Hardware Guru
Hardware Guru
Posts: 4752
Joined: Tue Oct 30, 2007 2:55 pm
Location: Florø, Norway
Contact:

Re: 4MB Upgrade & 16MHz Booster progress

Post by joska »

Arne wrote:It never happens when I do a reset via button and it showed a good picture before. Only when switching on the machine. Then additional resets (via button) do not change anything. Only off/on cycle gives me another chance for a good pic.
But that doesn't change resolution.What happens if you disconnect the mono monitor (which cause a reset) and then connect it again without powering down the ST?
Jo Even

VanillaMiNT - Firebee - Falcon060 - Milan060 - Falcon040 - MIST - Mega ST - STM - STE - Amiga 600 - Sharp MZ700 - MSX - Amstrad CPC - C64
User avatar
exxos
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 4933
Joined: Fri Mar 28, 2003 8:36 pm
Location: England
Contact:

Re: 4MB Upgrade & 16MHz Booster progress

Post by exxos »

troed wrote:
Don't spend a lot time on this. It's easy to, on a completely unmodified ST, get that effect when switching between resolutions. It's due to the Shifter having "leftover" words in its internal registers when the MMU believes it to not be the case.
So its probably a fault with the ST design itself then not the actual mod causing it ?
4MB STFM 1.44 FD- VELOCE+ 020 STE - Falcon 030 CT60 - Atari 2600 - Atari 7800 - Gigafile - SD Floppy Emulator - PeST - various clutter

http://www.exxoshost.co.uk/atari/ All my hardware guides - mods - games - STOS
http://www.exxoshost.co.uk/atari/last/storenew/ - All my hardware mods for sale - Please help support by making a purchase.
http://ataristeven.exxoshost.co.uk/Steem.htm Latest Steem Emulator
User avatar
Arne
Atari Super Hero
Atari Super Hero
Posts: 651
Joined: Thu Nov 01, 2007 10:01 am

Re: 4MB Upgrade & 16MHz Booster progress

Post by Arne »

exxos wrote:I think the thing to do is to try this mod in LOW AND MED res, though looking like I will have to try that one out.
Just lately found a special monitor cable I made myself many years ago for a NEC Multisync GS. Perhaps that means I still have it somewhere in the cellar...
User avatar
troed
Atari God
Atari God
Posts: 1460
Joined: Mon Apr 30, 2012 6:20 pm
Location: Sweden

Re: 4MB Upgrade & 16MHz Booster progress

Post by troed »

Arne wrote:
troed wrote:It's due to the Shifter having "leftover" words in its internal registers when the MMU believes it to not be the case.
It never happens when I do a reset via button and it showed a good picture before. Only when switching on the machine. Then additional resets (via button) do not change anything. Only off/on cycle gives me another chance for a good pic.
exxos wrote:
troed wrote:
Don't spend a lot time on this. It's easy to, on a completely unmodified ST, get that effect when switching between resolutions. It's due to the Shifter having "leftover" words in its internal registers when the MMU believes it to not be the case.
So its probably a fault with the ST design itself then not the actual mod causing it ?
Many demo coders had a setup where we wrote the code in mono with an SM124 connected, and when we ran the demo it replaced the default VBL code and then switched to color - which would output to a TV through the RF. When we exited the screen we'd switch back to mono and then restore the VBL rout.

When doing this, the switch between resolutions could happen at any point during screen update, and the Shifter could go from one mode to another with everything from 0 to 3 of its internal registers (see Alien's guide) already filled with data. This can cause the screen to be offset from where it usually would be - and in some cases that mode is stable. The screenshots in this thread look exactly like that - and since I've continued Alien's work in documenting GLUE/MMU/Shifter communication (for fullscreen and syncscroll purposes) I'm quite positive this mod manages to generate the same effect.

So; When the mod is beginning to stabilize you _might_ want to look into synchronizing it so that the Shifter is in a "clean" state (zero internal registers filled) at boot - but until then I wouldn't worry about the screen effects you see.

(As to why some of the effects survive RESET it's because that signal doesn't reset the CPU/GLUE offset. Only power on/off can do that. It's the same thing with ST wakeup states - they survive RESET as well)

You'll find my CPU-GLUE writeup here: http://www.atari-wiki.com/?title=ST_STE_Scanlines

I've not documented additional information with regards to the Shifter (which is the important part for this thread) yet so go with Alien's writeup in the Alive magazine. The every-other-16-pixels-blank (which has a related different but similar effect in mono), shifted screens etc are Shifter effects.

/Troed
User avatar
Arne
Atari Super Hero
Atari Super Hero
Posts: 651
Joined: Thu Nov 01, 2007 10:01 am

Re: 4MB Upgrade & 16MHz Booster progress

Post by Arne »

joska wrote:But that doesn't change resolution.What happens if you disconnect the mono monitor (which cause a reset) and then connect it again without powering down the ST?
I switched on the 520. Had a shifted screen.
Disconnected / connected screen 10 times. It always stayed shifted. So no improvement this way.
Post Reply

Return to “Hardware”