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, Greenious, spiny, Moderator Team

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

Postby exxos » Fri Dec 12, 2014 8:51 am

Does anyone know what lines actually control the interleaving patten between CPU and video ? I assume the CPU gets 4 clock cycles, then the video, then the CPU, but what actually sets this number and how is it controlled ?

I also think when the ROM circuit was half a cycle to late, that something must delay DTACK or something similar from reaching the CPU.. I am a little confused about this part. As if the CPU get DTACK to late, then (assume GLUE) switches to video cycles, then what happens to /AS ? I assume /AS is still LO on the CPU while it access video, and assume DTACK is ignored somehow until th eCPU gets its turn again. In effect, a late DTACK will arrive after 8 clock cycles. This would explain why I got a 25% drop in machine speed when DTACK was to late, as it has to wait 4 clock cycles before the CPU can sample DTACK.

I ask as I fail to understand why the CPU will not operate at something like 12mhz. 8,16 is fine, but clock speeds in between do not work. I can only assume that the CPU in effect gets time for 5 cycles, but it probably issues /AS again while the video is having its turn. I was reading a page about this, and they use a counter and some logic to make sure the CPU does not access the bus during video cycles. I think as the counter is "out" when CPU is running at 12mhz, that the CPU tries to execute the next instruction at the same time the video has its turn. Anyone any clues about how all these signals fit together ??
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

mc6809e
Captain Atari
Captain Atari
Posts: 159
Joined: Sun Jan 29, 2012 10:22 pm

Re: 4MB Upgrade & 16MHz Booster progress

Postby mc6809e » Fri Dec 12, 2014 4:01 pm

CPU doesn't tell MMU when to start an access. It's the MMU that's in control. The MMU controls memory interleaving.

I think the MMU simply checks once every 4 CPU cycles for *AS from CPU to see if the CPU is requesting memory. If the MMU sees that the CPU wants memory, then the MMU latches the address on the bus and then uses *DTACK to tell the CPU to complete memory cycle. If the MMU doesn't see *AS near the beginning of a CPU memory access slot, then the MMU ignores the cpu and doesn't send *DTACK. It waits another 4 CPU cycles and checks *AS again.

User avatar
Cyprian
10 GOTO 10
10 GOTO 10
Posts: 1685
Joined: Fri Oct 04, 2002 11:23 am
Location: Warsaw, Poland

Re: 4MB Upgrade & 16MHz Booster progress

Postby Cyprian » Fri Dec 12, 2014 4:43 pm

mc6809e wrote:CPU doesn't tell MMU when to start an access. It's the MMU that's in control. The MMU controls memory interleaving.

if I remember correctly GLUE (not MMU) is responsible for the bus control. and if needed it inserts a wait states for the CPU.
Lynx II / Jaugar / TT030 / Mega STe / 800 XL / 1040 STe / Falcon030 / 65 XE / 520 STm / SM124 / SC1435
SDrive / PAK68/3 / Lynx Multi Card / LDW Super 2000 / XCA12 / SkunkBoard / CosmosEx / SatanDisk / UltraSatan / USB Floppy Drive Emulator / Eiffel / SIO2PC / Crazy Dots / PAM Net
Hatari / Steem SSE / Aranym / Saint
http://260ste.appspot.com/

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

Postby exxos » Fri Dec 12, 2014 5:04 pm

So really 12mhz should work without a problem then ? I think there is multiple issues going on here which is just adding to all the confusion.

I can run 8mhz, 16mhz, 32mhz fine, but any other numbers do not seem to work. Of course its not that easy to generate synced clocked, but I have a PLL which has a 2.5x setting, so I can have 8mhz x 2.5 =20mhz, but that does not work. There could be problems relating to the PLL. Though even if I try a 16mhz external clock, the machine will not boot.

I have designed a circuit which will sync clocks, using flipflops, but the problem is, when /AS goes from HI to LO, there are 2 16mhz cycles when they needs to be 8mhz. So what I did was to add even more flipflops, to basically delay the CPU /AS from the bus for those 2 cycles. This way, when /AS is LO, it will only ever be at 8mhz. and 8mhz when /AS is LO is the only thing which works. Of course the 16mhz and 8mhz clocks also have to be in sync.

Though building the flipflop circuit is going to take some time to build. Though the problem is (again) While the circuit can run at any speed, 20mhz for example, I have already tried the PLL on 20mhz and the ST did not boot. So I am skeptical about spending many hours work on something which probably won't work as there it is unknown why 20mhz does not work in the first place. In fact, I have tried at some point 12mhz an 14mhz I think months ago, and they did not work either. Though the strange thing is, slower speeds do work. Like I ran the ST on 4mhz I think, and crushed the ST down to about 25% scores :twisted: That was with the PLL, so really it should work with 20mhz also. But it doesn't.

As darklord mentioned the other day, there are boosters at like 28mhz and 36mhz. So 20mhz should not be a problem. It is possible the CPU can't run at 20mhz at the time I was doing those tests, but I never got speeds between 8-16mhz working either. So clearly something odd going on but I do not understand what :?
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

mc6809e
Captain Atari
Captain Atari
Posts: 159
Joined: Sun Jan 29, 2012 10:22 pm

Re: 4MB Upgrade & 16MHz Booster progress

Postby mc6809e » Fri Dec 12, 2014 5:10 pm

Cyprian wrote:
mc6809e wrote:CPU doesn't tell MMU when to start an access. It's the MMU that's in control. The MMU controls memory interleaving.

if I remember correctly GLUE (not MMU) is responsible for the bus control. and if needed it inserts a wait states for the CPU.


There's a pullup resistor (R30 on the 520ST) that tries to keep *DTACK high, but *DTACK can be pulled down by both GLUE and the MMU. Since the MMU controls access to DRAM, though, which is shared by SHIFTER and the CPU, the MMU must be responsible for *DTACK when determining DRAM access.

GLUE controls access to ROM and handles bus mastering.

If the CPU is accessing ROM then GLUE pulls down *DTACK, but if the CPU accesses RAM, MMU pulls down *DTACK.

SteveBagley
Captain Atari
Captain Atari
Posts: 176
Joined: Mon Jan 21, 2013 9:31 am

Re: 4MB Upgrade & 16MHz Booster progress

Postby SteveBagley » Fri Dec 12, 2014 5:13 pm

This may shed some light on the issue: http://pasti.fxatari.com/68kdocs/AtariS ... nting.html

Steve

User avatar
DarkLord
Fuji Shaped Bastard
Fuji Shaped Bastard
Posts: 4336
Joined: Mon Aug 16, 2004 12:06 pm
Location: Prestonsburg, KY - USA
Contact:

Re: 4MB Upgrade & 16MHz Booster progress

Postby DarkLord » Fri Dec 12, 2014 5:37 pm

exxos wrote:As darklord mentioned the other day, there are boosters at like 28mhz and 36mhz. So 20mhz should not be a problem. It is possible the CPU can't run at 20mhz at the time I was doing those tests, but I never got speeds between 8-16mhz working either. So clearly something odd going on but I do not understand what :?


There is a 20mhz board in that T-series of accelerator boards as well...
Welcome To DarkForce! http://www.darkforce.org "The Fuji Lives.!"
Atari SW/HW based BBS - Telnet:darkforce-bbs.dyndns.org 520

User avatar
stimpy
Captain Atari
Captain Atari
Posts: 305
Joined: Mon Aug 22, 2005 2:47 pm
Location: Somerset, UK
Contact:

Re: 4MB Upgrade & 16MHz Booster progress

Postby stimpy » Fri Dec 12, 2014 5:50 pm

Just to chip in, I seem to remember looking at this and worked out that AS and DTACK need intercepting to synchronize the 68000 when running async to the system. And the 6800 series stuff needs a new clock generated and again, resyncing a few things. What you need to do is align the data moving between the 68000 and the system so you don't mess up any of the setup and hold times that the system requires.

I think that the CT60 uses a trick to speed up writes to the system RAM as you can possibly release the CPU early from the bus to do any internal processing instructions. Howver, that may have a whole bunch of stuff to keep the data and address valid, I'm not sure.

This approach is a lot cleaner than switching the 68000 clock speed, which is susceptible to temperature, motherboard revision and type of logic series chip used in the booster.
Last edited by stimpy on Fri Dec 12, 2014 5:52 pm, edited 1 time in total.
Netus-Bee,Repairs,Upgrades,EtherNEC,Eiffel

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

Postby exxos » Fri Dec 12, 2014 5:51 pm

DarkLord wrote:
exxos wrote:As darklord mentioned the other day, there are boosters at like 28mhz and 36mhz. So 20mhz should not be a problem. It is possible the CPU can't run at 20mhz at the time I was doing those tests, but I never got speeds between 8-16mhz working either. So clearly something odd going on but I do not understand what :?


There is a 20mhz board in that T-series of accelerator boards as well...


:cry: I really do not understand it. There must be some "trick" to getting it to work. Maybe its when clocks go out of phase, that it upsets something. Even so, that 20mhz T-series would have the same problem. I have been looking at the boosters on dev-docs etc, but they do not really seem to be doing anything special.
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

Postby exxos » Fri Dec 12, 2014 6:02 pm

stimpy wrote:Just to chip in, I seem to remember looking at this and worked out that AS and DTACK need intercepting to synchronize the 68000 when running async to the system. And the 6800 series stuff needs a new clock generated and again, resyncing a few things. What you need to do is align the data moving between the 68000 and the system so you don't mess up any of the setup and hold times that the system requires.

I think that the CT60 uses a trick to speed up writes to the system RAM as you can possibly release the CPU early from the bus to do any internal processing instructions. Howver, that may have a whole bunch of stuff to keep the data and address valid, I'm not sure.



I have routed /AS so it "waits" for the clock to start correctly at 8mhz. Though not sure about DTACK. The CPU cannot run on its own clock, even if its 8mhz. Its has to be in sync, though again there are boosters which use other clock speeds. I'm going to have to build this new circuit idea :roll:

abehfbcj.jpg

Problem is with sync clocks, is they use flipflops which delay the change a couple of cycles, though for what ever reason, the CPU must run at 8mhz when /AS goes LO. Though because of the 2 cycle problem, I added in a fix to delay /AS from the motherboard. So the ST only ever sees the CPU at 8mhz. Thats the only idea I have that could fix all these issues.

The other problem relaying to the 2 cycle count, is the CPU runs at 2 8mhz cycles before it runs in 20mhz mode. Though while this is bad, it will be compensated with a higher clock speed. It would be something like 16mhz equivalent speeds using this design would actually need 18mhz to run at the same speed overall. Though the good thing is, we gain speed at 20mhz over running at 16mhz. It also allows the freedom of trying any mhz speeds.

Sometimes the 2 cycle count is 1 cycle, that may not happen on real hardware, I had to simulate /AS which is only synced to 8mhz clock, so when /AS is actually synced to 20mhz, there probably won't be 2 cycles there, only 1. Though its hard to simulate all this.
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
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

Postby exxos » Fri Dec 12, 2014 6:14 pm

SteveBagley wrote:This may shed some light on the issue: http://pasti.fxatari.com/68kdocs/AtariS ... nting.html

Steve


Interesting. A lot going on. Though the CPU can probably run out of sync with the ROM, as its not accessing the ST bus, though second up, the CPU as to be in sync to access the ST bus. Though ontop of that. Its also the reason buffers are added to sync things up with RAM. I suppose it basically explains why it does not work. Maybe some boosters add the buffers to keep things in sync. There could be a couple of ways to do it, or just "pause" the CPU with delaying /AS while the clocks re-sync back to 8mhz which seems the easiest idea to me.
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
stimpy
Captain Atari
Captain Atari
Posts: 305
Joined: Mon Aug 22, 2005 2:47 pm
Location: Somerset, UK
Contact:

Re: 4MB Upgrade & 16MHz Booster progress

Postby stimpy » Fri Dec 12, 2014 7:46 pm

exxos wrote:
stimpy wrote:Just to chip in, I seem to remember looking at this and worked out that AS and DTACK need intercepting to synchronize the 68000 when running async to the system. And the 6800 series stuff needs a new clock generated and again, resyncing a few things. What you need to do is align the data moving between the 68000 and the system so you don't mess up any of the setup and hold times that the system requires.

I think that the CT60 uses a trick to speed up writes to the system RAM as you can possibly release the CPU early from the bus to do any internal processing instructions. Howver, that may have a whole bunch of stuff to keep the data and address valid, I'm not sure.



I have routed /AS so it "waits" for the clock to start correctly at 8mhz. Though not sure about DTACK. The CPU cannot run on its own clock, even if its 8mhz. Its has to be in sync, though again there are boosters which use other clock speeds. I'm going to have to build this new circuit idea :roll:

abehfbcj.jpg

Problem is with sync clocks, is they use flipflops which delay the change a couple of cycles, though for what ever reason, the CPU must run at 8mhz when /AS goes LO. Though because of the 2 cycle problem, I added in a fix to delay /AS from the motherboard. So the ST only ever sees the CPU at 8mhz. Thats the only idea I have that could fix all these issues.

The other problem relaying to the 2 cycle count, is the CPU runs at 2 8mhz cycles before it runs in 20mhz mode. Though while this is bad, it will be compensated with a higher clock speed. It would be something like 16mhz equivalent speeds using this design would actually need 18mhz to run at the same speed overall. Though the good thing is, we gain speed at 20mhz over running at 16mhz. It also allows the freedom of trying any mhz speeds.

Sometimes the 2 cycle count is 1 cycle, that may not happen on real hardware, I had to simulate /AS which is only synced to 8mhz clock, so when /AS is actually synced to 20mhz, there probably won't be 2 cycles there, only 1. Though its hard to simulate all this.


What I mean is run the CPU at whatever MHz you want but just make sure the control signals going to the bus are fixed to match what the motherboard/system is expecting. Like someone previously said you probably need to make sure AS is high for a long enough time between successive bus accesses. A CPLD should be fine to do this, perhaps even a GAL.
Netus-Bee,Repairs,Upgrades,EtherNEC,Eiffel

mc6809e
Captain Atari
Captain Atari
Posts: 159
Joined: Sun Jan 29, 2012 10:22 pm

Re: 4MB Upgrade & 16MHz Booster progress

Postby mc6809e » Fri Dec 12, 2014 10:08 pm

Maybe you could go complete async and just talk through a state machine running off the 8MHz clock that emulates the 68000 bus cycle.

s0 and not ASp -> s1
s1 and clk -> s2
s2 and not clk -> s3
s3 and clk -> s4
s4 and not clk and not DTACKmb -> s5
s5 and clk -> s6
s6 and not clk -> s7
s7 and clk -> s0

ASmb= not(s2 or s3 or s4 or s5 or s6 or s7)

DTACKp = not(s6 or s7)

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

Postby exxos » Sat Dec 13, 2014 11:57 am

Interesting. I have seen something like that on the PAK I think.

I realised my 12mhz tests even though were going via a PLL when the clock switches, it will still glitch. I also been thinking, the number of cycles the CPU does while /AS is HI might not be a problem. As sometimes the clock runs at mostly 16mhz when accessing fast-roms, and I have even seen 2x 16mhz cycles followed by a lot of 8mhz cycles. So the cycle count will always vary while its the cpu's "turn". At least so I am thinking now anyway.

So I will solve the clock sync problems then re-visit all the speed tests again. Will take some days, and its xmas time now. So probably no progress until after xmas now.
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: 507
Joined: Thu Nov 01, 2007 10:01 am

Re: 4MB Upgrade & 16MHz Booster progress

Postby Arne » Wed Dec 31, 2014 5:48 pm

I don't know if I get this right, but there has been a 16MHz Mod by Stefan Nitschke.
I had this Mod running on a 520ST with 4MB 70ns PS/2 SIMM, fast TOS-ROMs internal 40MB IDE HDD and RSVE IIRC, because I wanted to use that ST as an answering/fax-machine. It worked pretty well, had 38400bps without any problems running. The Mod accelerates the whole CPU/RAM to 16MHz. Not every chipset is able to handle that. I had the impression that old chipsets on 260/520 are more likely to be used way out of the specs than newer ones. I'll attach the original German description. If there is interest I can translate it. He has put in some ASCII-Art schematics, so you may get an impression of the mod without translation. I had QuickIndex running on it and AFAIR I got +190% on RAM access and Integer division. Used an SM124 in conjunction with that 520ST. Drawback was that pretty often the left edge of the screen seemed to be "bent around" to the back. Had to reset the machine a couple of times and then it was perfect until the next reboot.

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

Postby exxos » Wed Dec 31, 2014 10:08 pm

Interesting. I had seen that file before, but did not take much notice of it.

Looking at a translation, they are feeding 32mhz into the MMU, understandable, 16mhz CPU understandable. Also a flipflop to take the clock back down to 8mhz...

The interesting part is the DE from GLUE to MMU. It looks like they are syncing it to the 8mhz clock. Though not actually doing anything else. This sounds like what I was thinking about before, but that method suggests the shifter is being programmed at double speed also.

It also mentions some blitters might not run at 16mhz. I have seen odd issues relating to overclocking blitter, but its not fully explored yet.

I take it you no longer have this machine ?

The "bent around" thing sounds like a problem with the monitor itself. Normally a bad cap somewhere IIRC. Have had that happen on monitors in our workshop a fair bit in the past.

Thanks for pointing this document out, it seems like it should defiantly be worth replicating :) Some things don't translate to well, though I can get some logic chips and try it out as it looks simple enough.
Last edited by exxos on Wed Dec 31, 2014 10:24 pm, edited 1 time in total.
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: 507
Joined: Thu Nov 01, 2007 10:01 am

Re: 4MB Upgrade & 16MHz Booster progress

Postby Arne » Wed Dec 31, 2014 10:23 pm

I still have that 520ST board, but some wires are loose. Don't know if I can fix it soon.
Hope one of those 260/520 external power-supplies still is in working condition. Didn't use
them for 10+ years.
Shifter still gets the original 32MHz, but MMU is fed with 32MHz (inverted by the 74F04).
The text about the DE syncing with CLK2 says that it restores the original update rate of how often data
is transferred to the monitor (CRT). It seems that DE is doubled (4MHz ?) and now gets throttled
down to 2MHz again. Think there is a typo in Stephan's doc: Glue Pin 52 is GND, but CLK2 is on Pin 54.
Think he is talking about the latter one.
Yes, he mentions problems with Blitters and especially IMP chipsets.
With "bent around" I mean:
Imagine the rectangular 2D image of a SM124 as being a sheet of paper. Now take the left edge of the paper
and bend it slightly to the back. It affected about the leftmost 10-15 pixels IIRC.
The SM124 I used did not have this "feature" when conncted to an unmodified (i.e. standard 8MHz) ST or upgraded with
a PAK or something similar.

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

Postby exxos » Wed Dec 31, 2014 10:30 pm

It would be interesting to see the results in Gembench 4. Though is the machine isn't working then not so easy :(

Not sure why it would cause a monitor fault like that. I would assume GLUE is running double speed there also ? Its hard to tell exactly whats what. My idea was to run the MMU at double speed when not accessing the Shifter. Though that circuit seems to run always at double speed, but delays DE so it syncs with the shifter. Though If GLUE is generating the video syncs, then running that at double speed also would really screw up the video timings. Its hard to visualise in my head how all this is working.
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: 507
Joined: Thu Nov 01, 2007 10:01 am

Re: 4MB Upgrade & 16MHz Booster progress

Postby Arne » Wed Dec 31, 2014 10:51 pm

Paragraph 2) says that all other components still get 8MHz except for the Blitter which gets an inverted 8MHz clock. CPU gets non-inverted 16MHz.
In 3) he states that one can try to feed the Shifter with 64MHz and using a Videosequencer (IIRC it's a data table stored in an EPROM. The German ST Computer magazine published a 12MHz Busclock mod once which used such a table stored in an 27C256 or 27C512. Still have some PCBs of that mod. By accident the guy who engineered that mod lived just 2 km away from my parent's home) one can achieve resolution of 960x640 e.g.
What he doesn't mention is that noone provides the much greater screen memory - so there is some software help needed.
I used the mod with the standard 640x400 72Hz.

You aim for 16MHz with CPU/RAM/ROM and all other parts get the standard clocks?

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

Postby exxos » Wed Dec 31, 2014 11:10 pm

Not really sure why the clock gets inverted ? Unless its just covering up the delays in the logic.

Currently TOS and CPU are running at 16mhz, It would be nice to link to 16mhz RAM also. I already overclocked the whole thing with a 64mhz osc into the shifter a couple years back. But that seriously messed up the video.

How that mod looks, is like you say, it just seems like the MMU is getting the faster clock and the default 8mhz clock remains 8mhz. Though the MMU does output 4mhz which would become 8mhz, so not really sure why the route that though a gate to downclock to 8mhz...

Also it seems to be running the CPU at 16mhz all the time, we never got that to work, there was some talk about it a few weeks back in this thread. Unless the issues were just with CPU <>MMU issues. If both are running at double speed then it could be possible the CPU could run 16mhz all the time. But then is also a big question mark on the E-clock on the CPU also.

If you are saying this mod really works without screwing up the default video modes then I am willing to build it and try it out. I have a STFM with a simm adapter so I can plug in 70ns RAM in there. So all I would need to get is some logic chips and hardwire it all together. Though I must confess I am really sceptical about 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: 507
Joined: Thu Nov 01, 2007 10:01 am

Re: 4MB Upgrade & 16MHz Booster progress

Postby Arne » Thu Jan 01, 2015 9:02 am

Happy New Year :D

I will try to get my 520ST running again - cannot promise anything. Need an external FDD anyway to load software, haven't used my SF314 for 10+ years, too.
The E-clock only feeds the 6850? Maybe they are already capable of taking 2MHz?
As I understand Stephan's doc you must use this DE patch (syncing with 2MHz clock) oterwise you'll end up with a messed up video output as you wrote when feeding 64MHz into the Shifter.

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

Postby exxos » Thu Jan 01, 2015 1:25 pm

Arne wrote:Happy New Year :D
[smilie=greencolorz4_pdt_01.gif]
Arne wrote:I will try to get my 520ST running again - cannot promise anything. Need an external FDD anyway to load software, haven't used my SF314 for 10+ years, too.
The E-clock only feeds the 6850? Maybe they are already capable of taking 2MHz?
As I understand Stephan's doc you must use this DE patch (syncing with 2MHz clock) oterwise you'll end up with a messed up video output as you wrote when feeding 64MHz into the Shifter.


IIRC I had to use 66mhz for it to work, so that might be a indication of something, I don't know. I will order some faster 6850's. I think the stock ones used on the ST were 1mhz parts. It was talking about some posts back anyway. Maybe you could take a clear image of your board so we can see what chips are in there and how it basically was fitted together?

I know I tried running the MMU on a seperate 16mhz clock before, it worked, but video had a "wobble". That sync thing could fix that. Though it does half prove I think that the MMU and shifter can operate "out of sync".

The inversion of the clock I am pretty sure if down to delays on the gates. I know when I was looking at 32mhz a 20ns delay really caused huge problems. It sounds a bit of a bodge way of doing it if thats the case. It also means if the F04 was changed for a LS04, it probably wouldn't work. I did notice a mix of F and LS types on the circuit. So I will have to order both types to test that out.

I would have thought it would be better just to buffer the 32mhz master clock and just feed MMU and Shifter with that clock. Unless for some reason the Shifter inverts the 16mhz output (I don't think it does but need to double check on that also). So I am half sure the 04 isn't even needed. Also the LS74 is there to half the clock and work with blitter. Though I am not sure thats needed as the MMU has by default a 16mhz input and a 8,4mhz output. so there wouldn't be any need to half the 16mhz clock down to 8mhz, as the 4mhz line would become 8mhz anyway. But that could be a problem in itself as 4mhz is now 8mhz, So the 68901 will now run at double speed for starters. Which does not sound like a good idea. So if anything, that LS74 IMO should have had 8mhz clock input to downclock back to 4mhz. Someone else might be able to confirm is the 68901 can be double clocked. I think you mentioned faster RS232 so that could explain that.

In anycase, I don't think the LS74 is needed. So it only leaves the LS00 to control the DE line to shifter. This is where it gets a bit tricky. As the GLUE now looks to be overclocked also. If the CPU and RAM is running double speed then I would assume GLUE would have to be double clocked aswell. Like you say , pin 52 is GND, so if thats pin 54 its the 2mhz output line, which would then become 4mhz. So then DE then becomes synced to the master clock but at 4mhz. That is probably the point I got stuck on in my head last time, as I was thinking along the lines of pulse skipping or something to re-sync the Shifter, but the LS00 looks to basically be going something similar by waiting for the 4mhz clock.

Obviously if the GLUE was double clocked then it would mash up the H and Vsync also. I assume DE is just loading the data into the shifter at the correct time. Though if GLUE is double clocked, then the syncs would still be messed up surely ? When I coverclocked the whole thing to 64mhz (well 66mhz) the dot clock was double speed, so the H scanline would simply run out of data at 50% across, and then start on the next scan line for the remaining 50%. This would cause the video data to simply run out 50% on the vertical. I did post images of that in another thread a couple years ago. But in that circuit, the dot clock is still the same. So the H data should still be correct. Though if GLUE issues H-sync double speed, then Not sure what would happen there. Probably video data simply runs out 50% on the vertical. So effective LO resolution would be 320 > 640.



What I could try is a little test to just buffer the 32mhz clock, feed it into the shifter and MMU and just see what happens. Shifter wouldn't work right, but on that basis, the machine should still boot from floppy so I can listen to the drive to see if it booting or not. Then at least if I press ESC I know its running GEM and its accessing the floppy fine. The only problem might be the ACIA might not overclock, but not sure if that would cause a crash or not.

I think it would greatly help if you could get your ST working again to see how that mod is actually working. The only way it might work is if GLUE is not overclocked. Though the document isn't to clear on that. If you think the video still is all at default resolutions, then GLUE will still be at 8mhz. MMU & CPU at double speed only. The ACIA and 68901 would get double clocked also.
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: 507
Joined: Thu Nov 01, 2007 10:01 am

Re: 4MB Upgrade & 16MHz Booster progress

Postby Arne » Thu Jan 01, 2015 2:13 pm

Had a look at my broken 520.
It's not bad, but not in good condition either. I removed RSVE and the RTC-PCB to get rid of the Spaghetti wiring. One wire was loose. Unfortunately it connected to Pin 35 (CLK4) of the MFP. Cannot solder it back in place as the whole MFP pin is ripped away directly at the plastic body. I may try to use a drill tool to get to the MFP's copper layer. I wanted to get MFP spares anyway...
In that Shifter metal-cage/housing was an unused space for a DIL-14. I used that to put in a quartz oscillator. That one seemd to have broken, too. It emitted 260kHz - measured with my DSO! Shifter divided it to 130kHz. I already replaced that one.
I can see on my Eizo CRT that it (sometimes) tries to sync to 15625Hz HSync and 60Hz VSync. I would have expected 50Hz.
I connect the CRT with a home-made cable that pulls MONO_DET to low inside the 13pin Atari Monitor connector.
IIRC the switch to 31kHz/71Hz is done by software, but if the MFP isn't clocked due to a broken 4MHz line/pin... it cannot perform the switch.
As I understand the meaning of the DE line Glue tells MMU to stop any RAM access by DMA/CPU and let the Shifter get the data to be displayed.

The I saw that I used the 2nd FF in the 74LS74, too.
I connected:
- Pin 11 to Pin 5 (CLK8)
- Pin 12 to Pin 8
- Pin 9 to MFP Pin 35 (CLK4)
These additional changes are done with wire on the PCB, not in the layout.
Don't ask me why I did that 16 years ago...

I had designed a PCB with ScooterPCB for the three 74xxx TTLs (s.attachment).
You can get ScooterPCB for free at: http://www.scooter-pcb.de/scooter-pcb-atari.php
The Win16 version is also free by now, but naturally it won't run on Win x64.

You may divide the E-Clock by 2 with a 74xx74 back to normal 800kHz clock.

I will keep trying to get it into working condition again to measure clocks and - of course - do some benchmarking.
But I am absolutely sure that it achieved a higher score than those CPU-only-16MHz (like the old c't speeder or HBS210)
and even those STs with 16MHz CPU and 16-32KB cache (AdSpeed, HBS240, Mach16...).

By the way: do you own a TT?

SteveBagley
Captain Atari
Captain Atari
Posts: 176
Joined: Mon Jan 21, 2013 9:31 am

Re: 4MB Upgrade & 16MHz Booster progress

Postby SteveBagley » Thu Jan 01, 2015 2:21 pm

exxos wrote:Also it seems to be running the CPU at 16mhz all the time, we never got that to work, there was some talk about it a few weeks back in this thread. Unless the issues were just with CPU <>MMU issues. If both are running at double speed then it could be possible the CPU could run 16mhz all the time. But then is also a big question mark on the E-clock on the CPU also.


If everything is sped up to 16MHz then the relation between DTACK, AS, and xDS lines will still be the same so everything should work fine, the problem comes when some parts are running at 16MHz and others at 8MHz because then the timing relationships expected by the CPU don't match what it gets…

As for the 6850s, I wouldn't be surprised if the 2MHz and 1MHz parts are identical and that each batch of chips are sampled to see what speeds they reliably work at, and then the whole batch labelled accordingly. Often you can find that several of the 1MHz parts are happy to run at 2MHz, but are just labelled as 1MHz.

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

Postby exxos » Thu Jan 01, 2015 2:43 pm

Arne wrote:IIRC the switch to 31kHz/71Hz is done by software, but if the MFP isn't clocked due to a broken 4MHz line/pin... it cannot perform the switch.
As I understand the meaning of the DE line Glue tells MMU to stop any RAM access by DMA/CPU and let the Shifter get the data to be displayed.

The I saw that I used the 2nd FF in the 74LS74, too.
I connected:
- Pin 11 to Pin 5 (CLK8)
- Pin 12 to Pin 8
- Pin 9 to MFP Pin 35 (CLK4)
These additional changes are done with wire on the PCB, not in the layout.
Don't ask me why I did that 16 years ago...


I would assume there its to downclock the MFP back to default speeds. Though as the MMU outputs 4mhz, then it would be easier to swap the ST bus 8mhz clock for the MMU 4mhz clock, then when the MMU double clocked, 4mhz becomes 8mhz. That would clock the whole ST at proper speeds, wouldn't need the 74's at all then.

DE sounds like a interrupt line then. Not really sure why it would need to resync the DE to clock if the GLUE is running at correct speeds. Pretty confusing.
Arne wrote:You may divide the E-Clock by 2 with a 74xx74 back to normal 800kHz clock.

Yes. Though I think I would swap them out for 2mhz parts and save a extra gate in there.
Arne wrote:I will keep trying to get it into working condition again to measure clocks and - of course - do some benchmarking.
But I am absolutely sure that it achieved a higher score than those CPU-only-16MHz (like the old c't speeder or HBS210)
and even those STs with 16MHz CPU and 16-32KB cache (AdSpeed, HBS240, Mach16...).


If the MMU is being double clocked then RAM should be double clocked also. I have the 16mhz CPU only benchmarked posted in a few places. Though with faster RAM, the RAM should boost to near 200% speeds, though with 16mhz CPU , all other figures should gain a small boost in speed to.

Arne wrote:By the way: do you own a TT?


Nope. Only machines I really own are STFM. I do have a couple of STEs but thats as far as it goes.
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


Social Media

     

Return to “Hardware”

Who is online

Users browsing this forum: No registered users and 3 guests