Homebrew Atari ST ?

[simongarstin]

Hi.

Does anyone know if someone has made a homebrew Atari ST..and where I can get one ...if available ?

[joska]

I think MIST is the closest you get to a homebrewed ST

[alexh]

Suska III too

http://experiment-s.de/en

[simongarstin]

I was thinking of making/designing my own Homebrew Atari ST.. keeping it simple and popping as much glue-logic as possible onto a CPLD , yet still keeping it Retro (old School) for nostalgia .

Does anyone know how many macrocells (for a CPLD) it would need to pop all the glue logic, DMA etc on ? I was thinking of using an Altera MAX EPM7512 (512 macrocells) but would this be enough ?

[siriushardware]

To try to answer the possible thrust of the OP's original question, I would say it is near-impossible to knock up a homebrew replica Atari ST without using original special purpose ST parts like the MMU and the Shifter, and in some cases these devices are surface-mount devices with very fine-pitch pins which would not be easy to use unless you have the facility to design and produce your own printed circuit boards. It's not like, say, the very earliest home computers and development systems which were all made from standard IC logic and standard microprocessor ICs and their respective peripheral IC families.

Even if you had all the resources needed to put a home made ST together it would be a little bit pointless to make a new ST PCB from scratch because there are so many surviving real examples (of ST PCBs). I just don't see the point. I have seen people make beautiful replicas of machines which are very hard or expensive to get hold of (ie, Sinclair MK14) but that's because you can't just decide to buy one of those the way you can buy an old ST or STe.

There are a few very clever people here who do have the resources to design and make PCBs and moreover use modern large scale programmable logic to replicate the operation of the entire Atari ST subsystem (and many others) virtually in one chip. That is essentially what MIST is.

The 'third' way' is not really hardware, but software. If you build a decent PC (Linux) into an enclosure which looks however you want your 'ST' to look and then install an ST emulator like Hatari or Steem - both of those emulators are very good indeed - then you have a pretty good ST or STe for most purposes. Even the Raspberry Pi (mini Linux computer) - especially the just-released 1.2Ghz Pi 3 - can run ST emulation at real ST speed.

[rpineau]

Also, things that were done for the Suska could be reused. 1 FPGA per Atari chip (and you may combine some in 1 FPGA) and still use the other one you can find out there (68901, CPU (68000, 68020, ..). I think the hard part is, as said above, the Atari custom chips.
Using just plain CPLD might not be possible for all chip but I don't want to discourage anyone from trying.
The VHDL code for the Suska is available so it can be a good start.
Rodolphe

[AtariZoll]

While doing some clone of 8-bit computer in old fashion was and is (?) still popular, with computer of ST complexity that's different.
I guess that Simongarstin had something like assembling it from regular and ST custom chips in mind, and not FPGA clone . Maybe some parts done in CPLD. I think that it could be interesting and having sense only if doing some changes to original design. STE with it's SMD MMU/Shifter is really not suitable for homebrew.
But maybe doing ST with full CPU/bus expansion slot - where can attach diverse add-ons like other TOS in ROM, RAM expansions, accelerators and like. I say it, because such bus was missing. There is expansion bus in Mega ST, but it could have more lines. Of course, you can add expansion bus to existing ST motherboards too Point is that doing larger PCB will cost lot of money and time, and you can buy some working ST for less money.

[simongarstin]

It sounds more difficult than I first thought , though the logic on the ST was less complex (Messy) than say on An Amiga! My original though for a homebrew design was :

68HC000FN20 CPU (possibly overclocked to 24MHz) PLCC
8Meg or so SRAM (512K SRAM x 16pcs) DIP
2 Meg EEPROM (512K x 4pcs) DIP
EPM7128SLC84 as Memory and I/O Decoder PLCC84
68681
6850 2pcs
1772 FDC replaced with 8255s ?? for CF Card & IDE interface (Hard Drive)
Real Time Clock
DMA Controller
EPM7160SLC84 as Video Generator giving original video modes + (640x200,16 colors, 320x200, 256 colors) possibly using 64K Dual Port Fast SRAM IDT7008 ?
A Second video processor (V9938) in superimpose mode could be utilised to give SPRITES.. something the original ST lacked.
YM2149 Sound Chip
Possibly a Second sound chip ?
MCU ( Atmel,PIC Intel ??) to read PS/2 keyboard

This is obviously quite a loose specification and rather 'Pie-In-The-Sky' at the moment !

[AtariZoll]

Some notes about ideas in post above:

You can not separate DMA chip from MMU - they work together. Same is with some other - like Glue and shifter , ++ .
Basically, what you described would be some much faster 68000 based computer, with almost 0 ST compatibility.
For instance replacing WD1772 is very bad idea - of course, you can do (and must) new TOS which will operate with new FDC chip, but what with SW which does direct FDC access ? And even worse - you would be not able to access some disks - like those formatted with Fastcopy Pro with fast flag set on.
Static RAM is fine, but you need then complete new memory and video control.
All in all, that will be almost nothing like ST and you will need much more time for TOS than for HW - and even then, only SW performing strictly TOS calls will work. Get MIST, and forget whole idea

[DarkLord]

I'm going to echo what Chris (Exxos) said over at Atari Age in a similar message thread:

Why not just buy a *real* Atari ST or STe? They are cheap and fairly plentiful...

[joska]

You guys don't get it The build itself is the interesting part - not necessarily the result. Go for it and build something cool. Even a "plain" ST clone is supercool if you actually build it yourself. Why do people build strange Z80 computers and write their own weird OS for it now that we have multi-GHz supercomputers in our pockets? Because it's cool.

[wongck]

Why not just buy a *real* Atari ST or STe? They are cheap and fairly plentiful...

no fun in doing that.
making it from scratch is the fun part even if it does not work well.

[arf]

Hi.

Does anyone know if someone has made a homebrew Atari ST..and where I can get one ...if available ?

There were homebrewed STs in the beginning, e.g. the IPB 190ST or a commercial version, the Rhothron ST - both in a 19" rack case, built from Europa-card-modules, with several extensions. The building of the IPB has been described in a magazine. A later version of it, run by a 68020 cpu, is described in the German http://stcarchiv.de/stm1991/06/190st-020 ST-Magazin 1991/06.

This article mentions two other clones, the ESTE 1 from Musik Konkret and the Mega Station from Hybrid Arts: http://stcarchiv.de/stm1989/08/datenaustausch

See also
https://randoc.wordpress.com/2015/05/17/ibp-190st/
http://wiki.newtosworld.de/IBP_190_ST
http://dieter.lerntux.de/index.php/comp ... -ibp190-st

[simongarstin]

Hi.

Does anyone know if someone has made a homebrew Atari ST..and where I can get one ...if available ?

I think I'll just try making a basic 68000 design loosely based on the Atari ST..and see how it goes !

[LaceySnr]

I've been intrigued by the idea of a small single board 68k computer, Raspberry Pi / BeagleBoard style. I guess to make something useful though you'd want to be able to clock it pretty high!

[arf]

I've been intrigued by the idea of a small single board 68k computer, Raspberry Pi / BeagleBoard style. I guess to make something useful though you'd want to be able to clock it pretty high!

See https://www.youtube.com/watch?v=SRdLlaUmmpM

[AtariZoll]

...
There were homebrewed STs in the beginning, e.g. the IPB 190ST or a commercial version, the Rhothron ST - both in a 19" rack case, built from Europa-card-modules, with several extensions. The building of the IPB has been described in a magazine. A later version of it, run by a 68020 cpu, is described in the German http://stcarchiv.de/stm1991/06/190st-020 ST-Magazin 1991/06.
This article mentions two other clones, the ESTE 1 from Musik Konkret and the Mega Station from Hybrid Arts: http://stcarchiv.de/stm1989/08/datenaustausch

I would not call it homebrew, even if end user assembles it self - doing those cards in that compact format is almost impossible at home. By that logic every PC, what people assembles from bought components is homebrew too

The problem here is complexity of whole project. It was relative easy to do some 8-bit from scratch, or some 8-bit clone - probably Spectrum was most popular since it screams for some extensions. But 16-bit machine is much more complicated. Not that it's not possible to make it with home made dual layer PCB, and that can be fun for sure. If you want to do it to see how it goes that's fine idea. I'm just afraid that it will be just another project never finished. And again: SW is what takes most of time if you build something new.

Raspberry is based on high complexity custom chip(s). And then, we are back at FPGA solution

And finally, my experiences with transferring games from original format/storage to disks, Flash cards: I dealt with Sinclair Spectrum - made floppy interface, 2 types of mass storage adapter - latest for CF cards is really very simple. Transferring most of games onto was simple, because was no need to take care about protection, in-game later data loadings - most was single shot. So, doing RAM dump, packing it + launcher was all what needed to do.
In case of Atari ST games, it is aprox 20-100 harder. Partially because size is bigger, but more because used nasty protections, diverse low level floppy access code, then problems with diverse TOS versions and RAM usage.
That I'm talking not bull, confirms fact that we still don't have 100% accurate ST emulator - 30 years after it's release. FPGA code is not perfect too, as we know. But that can be start point for someone - joining existing, alive project instead doing something just for fun.

[simongarstin]

The secret with homebrew stuff is keep it simple , simplicity has an elegance over complexity !

A Very simple homebrew 68000 machine could consist of a relatively small number of ICs

68000 CPU
68901 MFP (Multifunction Peripheral )
6850 ACIA Serial I/O
Video logic to generate display
glue logic for memory switching , I/O selection , WAIT state generation etc
YM2149 Sound Chip
1772 FDC (if you want to use floppies)
IDE interface
MCU (micro-controller for keyboard etc)

And ideally all on a double sided PCB (if possible) ..

[simongarstin]

68000 ST style System : My initial design thoughts are :

CPU
MC68HC000FN20 PLCC (clocked at 24MHz)

RAM
16 pieces of 512K SRAM to give 8Meg (KM684000,HM628512,BS62LV4006 etc 55ns) DIP32

ROM
2 Meg (4pieces) 39F040,29F040,27C4001 512Kx8 , 55ns or faster DIP32

68901 MFP
6850 ACIA 2 pieces

Video logic: (128K Dual Port SRAM for video, IDT7008 x 2 pieces)
EPM7160 generating video modes and memory switching

video modes
320x200 16 cols
640x200 4 cols
640x400 2 cols
320x200 256 cols
640x200 16 cols
640x400 4 cols
640x200 256 cols
640x400 16 cols

Video logic synchronised to a V9938 VDP via HSYNC and VSYNC V9938 signals to GENLOCK both video outputs. The YS signal on the V9938 would switch between both video outputs to give 'Superimpose' function. The EPM7160 would give RGB output via DACs this could then be fed into a CXA1645 PAL/NTSC encoder to give composite video output. V9938 would have 128K VRAM.

This would be the most complex part of build, all these are ICs are either DIP or PLCC.


Sound :
YM2149
SN76489
MIDI ??

Storage:
Compact Flash & IDE
SD Card ??
Silicon Disc using NVSRAM could possibly be implement.

FDC (floppy Discs )
possibly implementing FDC (1772s) ?

RTC (Real Time Clock)
An updated RTC ..with Calendar,leap year etc possibly DS12C887

Even possibly a Maths Co-Processor ?

A moderv MCU (microcontroller) to read keybaord and other functions.

Glue logic :
Possible one or two EPM7128SLCs (PLCC84) are these are cheap and have 64 I/O lines each.


All still on the drawing-board at this stage...still need to do costing etc

[exxos]

Why not use a full 030 instead ? 50mhz speeds are possible.

[spiny]

It looks like it would be a nice 68k computer, but surely it would have 0% compatability with ST software ?

[galax]

It looks like it would be a nice 68k computer, but surely it would have 0% compatability with ST software ?

It could be called something like FireBug or BusyBee.

[AtariZoll]

68000 ST style System : My initial design thoughts are :

CPU
MC68HC000FN20 PLCC (clocked at 24MHz)

RAM
16 pieces of 512K SRAM to give 8Meg (KM684000,HM628512,BS62LV4006 etc 55ns) DIP32

ROM
2 Meg (4pieces) 39F040,29F040,27C4001 512Kx8 , 55ns or faster DIP32

68901 MFP
6850 ACIA 2 pieces

Video logic: (128K Dual Port SRAM for video, IDT7008 x 2 pieces)
EPM7160 generating video modes and memory switching

video modes
320x200 16 cols
640x200 4 cols
640x400 2 cols
320x200 256 cols
640x200 16 cols
640x400 4 cols
640x200 256 cols
640x400 16 cols
....


All still on the drawing-board at this stage...still need to do costing etc

Well, I can only wish you good luck and long life with this project - doing HW, and make it alive (OS, some SW) - you will need both, especially later

[arf]

The problem here is complexity of whole project. It was relative easy to do some 8-bit from scratch, or some 8-bit clone - probably Spectrum was most popular since it screams for some extensions. But 16-bit machine is much more complicated. Not that it's not possible to make it with home made dual layer PCB, and that can be fun for sure. If you want to do it to see how it goes that's fine idea. I'm just afraid that it will be just another project never finished. And again: SW is what takes most of time if you build something new.

Then have a look at my Youtube link above: a 68000 on a handwired breadboard.

[arf]

The problem here is complexity of whole project. It was relative easy to do some 8-bit from scratch, or some 8-bit clone - probably Spectrum was most popular since it screams for some extensions. But 16-bit machine is much more complicated. Not that it's not possible to make it with home made dual layer PCB, and that can be fun for sure. If you want to do it to see how it goes that's fine idea. I'm just afraid that it will be just another project never finished. And again: SW is what takes most of time if you build something new.

Then have a look at my Youtube link above: a 68000 on a handwired breadboard.