(re)build a *NEW* Atari Computer?

[JamesD]

As far as the projects I have seen... it's very easy for someone that knows a little VHDL to start but they don't know how to design stuff from scratch. Setting up the pinouts, register model, addresses for the hardware, adding a 6502 core, adding the existing POKEY core looks like it should be pretty basic stuff. But then they have so much to write from scratch that an inexperienced person won't know where to go from there.

 

The most surprising thing is that I have yet to see even a partial recreation of the Atari. I don't think getting it to come up and print a basic prompt requires a lot of the hardware to be implemented.

[nathanallan]

And i'm still hooked on the idea of a small Linux based machine as the next Atari.

Following that line of thinking, I've already made the next Atari computer then.

 

 

No, I'm not giving out the parts supplier list.

That's awesome! Tell me more about it! Is it a Linux box running an emulator?

 

Nathan

[Hornpipe2]

And i'm still hooked on the idea of a small Linux based machine as the next Atari.

Following that line of thinking, I've already made the next Atari computer then.

 

 

No, I'm not giving out the parts supplier list.

That's awesome! Tell me more about it! Is it a Linux box running an emulator?

 

Nathan

Looks like it. Although there are some truly fantastic Atari 8-bit laptop creations which you have undoubtedly already seen:

http://www.benheck.com/Games/Atari_800/Ata...00_laptop_1.htm (Ben Heck's first attempt)

http://www.benheck.com/Games/Atari_800/Ata...0_laptop2_1.htm (Ben Heck's second try)

http://www.atariage.com/forums/index.php?s...1&hl=1500xl (our own Beetle's laptop!)

[mimo]

Just get a Sony Viao FZ and run emulators on it J/K

[peteym5]

I have been mentioning this in the '65816 and Fast IC' thread. If you put multiple upgrades on an original Atari 8-bit motherboard. It will end up looking like spaghetti mess of wires. If you add your popular upgrades like, 32in1 OS, Dual Pokeys, S-Video, and 1MB, could you imagine what the computer will look like inside. Now add on a 65816 board and VideoBoard XE. That is going to be a lot of soldering. You are going to have some issues with such a system. Could easily have a short or it overheating. There comes to a point to where it might be better to have a new motherboard built or port to a FPGA. A new motherboard is a direction I would go with a (re)build a *NEW* Atari Computer.

Edited October 22, 2007 by peteym5

[almightytodd]

And i'm still hooked on the idea of a small Linux based machine as the next Atari.

Following that line of thinking, I've already made the next Atari computer then...

 

No, I'm not giving out the parts supplier list.

That's awesome! Tell me more about it! Is it a Linux box running an emulator?

 

Nathan

I totally agree with this line of thinking. If you don't buy into my idea of "Create a new Atari experience and save the world one child at at time" concept of supporting the OLPC project (...you know, I'm not sure that I even buy into it, the more I think about it...), then figuring out a way to develop a tiny Linux-based system that boots right into Atari 8-bit emulation mode would be the next best thing. I think the most important thing is that it looks like something Atari, and acts like something Atari.

 

I can totally love my FB2 without having to know about the gruesome details of the hardware (does it have a REAL 6507 CPU? Does it have separate TIA and I/O chips?). All I care is that it is playing the actual game ROMS and displaying them exactly as I remember them. Do the same thing with Intel or AMD chips running Atari emulation and have a USB port that you can plug a jump drive into with ROM files, and everyone who is into this stuff will be happy. Add a Web browser and maybe an IM client (with some kind of cool Atari-look skin) and it becomes a delightful bridge between today and yesterday.

 

If you're a hardware purist and you demand that any such machine MUST be composed of native 8-bit processing hardware - well, start buying parts and hack a Frankenstein system like that together for yourself. We've seen plenty of examples of where people have done that. But don't expect such a thing to become a viable, marketable product. There just isn't a large enough consumer base to support it. Plus, when you get too authentic, you start competing with the market of the actual classic systems, which are frequently found in the under $100 price range. The FB2 was able to compete with this market because you got the plug-and-play system, 40 built-in games, two joystick controllers, and the comfort of knowing that if you hooked it all up and it didn't work, you could take it back and exchange it the same day... ...and all for 30 bucks.

[peteym5]

Hey, I wrote to Steve Tucker over at AtariMax and got an email back. It said, if we designed a working system, he can help the manufacturing and such. It is possible to design a new main board with the several suggested upgrades built in. However, a lower cost solution is to add expansion headers for each chip that allow access to all its pins without removing the IC. This would allow anyone to design simple plug-in upgrade modules. Probably add some open sockets or card slots for small boards. Would also consider the SIMM socketed memory modification we have seen here in another thread. Best Electronics might have some un-populated main boards that may be useful. You can put in sockets for all the ICs, and add a SIMM socket in place of the D-Ram chips. (I think 65XE, 1200XL, and XEGS)

[MEtalGuy66]

Hey, I wrote to Steve Tucker over at AtariMax and got an email back. It said, if we designed a working system, he can help the manufacturing and such. It is possible to design a new main board with the several suggested upgrades built in. However, a lower cost solution is to add expansion headers for each chip that allow access to all its pins without removing the IC. This would allow anyone to design simple plug-in upgrade modules. Probably add some open sockets or card slots for small boards. Would also consider the SIMM socketed memory modification we have seen here in another thread. Best Electronics might have some un-populated main boards that may be useful. You can put in sockets for all the ICs, and add a SIMM socket in place of the D-Ram chips. (I think 65XE, 1200XL, and XEGS)

 

That garman guy, Mega-HZ already did all that crap. He made a complete replacement motherboard (for the 800xl) that had every common/feasible/desirable upgrade I can think of built in. And he works for a large PCB manufacturer, so his costs are well below that of what any third party could get. Ask beetle how to get in touch with the guy..

[peteym5]

Hey, I wrote to Steve Tucker over at AtariMax and got an email back. It said, if we designed a working system, he can help the manufacturing and such. It is possible to design a new main board with the several suggested upgrades built in. However, a lower cost solution is to add expansion headers for each chip that allow access to all its pins without removing the IC. This would allow anyone to design simple plug-in upgrade modules. Probably add some open sockets or card slots for small boards. Would also consider the SIMM socketed memory modification we have seen here in another thread. Best Electronics might have some un-populated main boards that may be useful. You can put in sockets for all the ICs, and add a SIMM socket in place of the D-Ram chips. (I think 65XE, 1200XL, and XEGS)

 

That garman guy, Mega-HZ already did all that crap. He made a complete replacement motherboard (for the 800xl) that had every common/feasible/desirable upgrade I can think of built in. And he works for a large PCB manufacturer, so his costs are well below that of what any third party could get. Ask beetle how to get in touch with the guy..

 

I have not heard anything about this project myself. What a few of us might looking for in a rebuilt new computer is more ram with separate Antic/CPU modes, hard drive storage, 32 in 1 OS, S-video out, dual pokey, maybe a few other things that came by over the years. Right now, I think developing a New FPGA board might end up being more costly simply because no one has one completed, let alone the extension I had proposed. All the functions have to ported onto a FPGA, it will take time and money to do it. Videoboard XE is also a nice extension, but also it has yet to be completed. There has not been any new information for months now and I am suspecting it might have fizzled out. I am going to leave discussion of a 65816 cpu in my other thread I have going.

Edited October 24, 2007 by peteym5

[simonl]

What a few of us might looking for in a rebuilt new computer is more ram with separate Antic/CPU modes, hard drive storage, 32 in 1 OS, S-video out, dual pokey, maybe a few other things that came by over the years. Right now, I think developing a New FPGA board might end up being more costly simply because no one has one completed, let alone the extension I had proposed.

 

Have you seen this?

 

http://www.digital-force.net/projects/hyperspeed/

 

The documentation makes interesting reading, it looks like a good basis for a more economical design using CPLDs not an FPGA.

[peteym5]

Lets not try to get off track here. What someone wants to do here is build a new Atari computer and try not to drive the costs up. There are several vender's out there that can sell the main board, ICs, CPU, etc. I suggested building them with 576k with a SIMM and the separate video access, keeping it compatible with existing games that use it. Using Simms also makes it easy to upgrade the ram. How many programs out there use the 1mb upgrade other than ramdisk? The 32 in 1 OS is good because that gives you the ability to run anything from the 400/800 to the XEGS. We have talked about Harddrive support because anyone collecting games would not have to go through hundreds of floppy disks. Would have to wait and see what goes on with the Motherboard mentioned and Videoboard XE.

[MEtalGuy66]

Lets not try to get off track here. What someone wants to do here is build a new Atari computer and try not to drive the costs up. There are several vender's out there that can sell the main board, ICs, CPU, etc. I suggested building them with 576k with a SIMM and the separate video access, keeping it compatible with existing games that use it. Using Simms also makes it easy to upgrade the ram. How many programs out there use the 1mb upgrade other than ramdisk? The 32 in 1 OS is good because that gives you the ability to run anything from the 400/800 to the XEGS. We have talked about Harddrive support because anyone collecting games would not have to go through hundreds of floppy disks. Would have to wait and see what goes on with the Motherboard mentioned and Videoboard XE.

 

Not to tkae away from the 32-in one OS upgrade, which in itself was a good value for it's time, but out of those 32 OS's I dont think there is 5 of them that I would use with any frequency. Whats alot better is(the latest incarnation of) the smartOS upgrade which uses 512k of battery backed static ram (NVRAM) amd lets you store all kinds of images of whatever size, and swap them "on the fly" via firmware control and have them show up in either OS ROM space, or BASIC ROM space.

 

This the kind of functionality I would go for if I was designing a new motherboard. Not 32 arbitrarily chosen hard-rom's images.

 

As far as using SIMMs for the Ram goes.. Well, if you are talking about 30 pin SIMMS (the only SIMMS that use an 8-bit DATA bus) then thats a decent idea. Even though they are getting somewhat obsolete/hard to find these days. I personally have a mountainous pile of them, as Im sure others do.. The thing is, that with DRAM, you need a DRAM controller circuit that handles refresh, etc. And that circuit has to be designed to handle whatveer size memory configuration your going to use.

 

72pin SIMMS are 32-bit DATA bus. This would required quite a bit more logic for the DRAM controller/access circuit to make it "speak in 8-bits"

 

The third consideration is that to make DRAM non-volatile, you have to "keep alive" not only the RAM when the power is off. but also the circuit that refreshes it, the circuit that generates the clock signal for THAT circuit, etc.. Why not just use STATIC RAM?

 

Static RAM requires no refresh.. can be made non-volatile with a single IC, and a small NiMH or LiON cell.. Or, they can even be purchased as NVRAM with all of this built into each individual memory IC..

 

Static RAM also uses non-multiplexed adressing, which means you dont have to have circuitry to create Columns and ROW adress setups/strobes.. Its just a straight data and adress bus and a read/write enable line.

 

If you use the newer SMT/PLCC package ICs, STATIC RAM is plenty cheap enough too.

[Tickled_Pink]

Unfortunately a great number of the components used in the original Ataris are now obsolete. You'd have a hard time getting the processor and memory, never mind having to redesign the custom chips. The last company I know of that produced 6502 processors were California Micro Devices. I managed to get hold of a couple of 65SC02 and 65C51 ACIAs from them not that long ago but they've stopped producing them. There are probably others who have licensed the technology, but they're probably some obscure Taiwanese manufacturer. I did come across one Far Eastern company which produced 6502s in a smaller form factor but had repackaged them with some extras, and in a smaller form factor.

 

In terms of memory, one of the things I've been toying with is coming up with some way of using a single static RAM instead of the eight DRAMs currently in the XL. If it can be done without too much hassle, then it might mean fewer parts (depending on any decoding logic). If it used nvRAM then it may even be possible to guard against the computer losing power - switch the machine back on and you're back where you left off ... in theory, anyway (there would be an issue with the status of the processor registers).

[peteym5]

Well either OS ROM upgrade that lets you select and store different versions of the operating system would be cool and gives you potential to run any program that was written for the 8-bit. The 32 in 1 seems to be in some sort of programmable eprom and wonder if Steve Tucker would be open to put on a few custom versions of the operating system.

 

I did think about the 30pin Simms becoming harder to find and pretty sure they are no longer in production, most of what be used would come out of old PC motherboards. Either way probably do something that is cheap and upgradeable. Can easily wire up a single 512k chip. Battery backup static ram is a good ideal.

 

Most of the original IC chips are still in large abundance in warehouses like Best Electronics and can easily obtain Antics, GTIA, Pokeys, Freddies, the custom CPU for a new computer. We still have an option of having new IC(s) or FPGA made that are backward compatible with some extra functions on them, but this might become more costly to produce.

[MEtalGuy66]

In terms of memory, one of the things I've been toying with is coming up with some way of using a single static RAM instead of the eight DRAMs currently in the XL. If it can be done without too much hassle, then it might mean fewer parts (depending on any decoding logic). If it used nvRAM then it may even be possible to guard against the computer losing power - switch the machine back on and you're back where you left off ... in theory, anyway (there would be an issue with the status of the processor registers).

 

SRAM is much easier to implement than DRAM..

 

If you want to use SRAM in the XL, just remove the 2 74LS158 ICs... SRAM has actual non-multiplexed adress lines which can be hooked straight nto the adress bus of the machine.

 

In otherwords, to get 65536 8-bit adresses, an SRAM chip has adress lines A0-A16. DRAM must have 2 sets of adress setups, multiplexed over the same lines (A0-A8) and precisely timed to Column Adress Strobe, and Row Adress strobe signals.. Not only that, whatever circuit is geenrating the refresh signals must support the number of bits of Rows employed in the amount of DRAM being used (this is why the pre-800XL ANTICS cant refresh 256k of memory. They only create 7 bits of ROW adress setups on their refresh cycles). With later DRAM ICs, you can do whats known as a "CAS before RAS" or "AUTOMATIC" refresh which negates the need for this, but still, the circuit needs to be designed in a way to assert a CAS before RAS event on the refresh cycle. ANTIC doesnt do this AT ALL, so we still need additional logic for the DRAM uprgrade to work.

 

So, in a nutshell, this is why when you design DRAM upgrades theres is typically some amount of added logic needed. When you design an SRAM upgrade, you can actually REMOVE alot of the existing DRAM control logic, as it is completely unneccesary. To effectively read SRAM, you just slam an adress on the adress bus, pull the enable low, and your data shows up on the data bus. To write it, you slam an adress on the adress bus, pull enable and r/w low, and slam your data on the data bus, and it gets written. end of story. Maintain +5v to the SRAM chip, and the data will be retained.. no refresh needed.

 

to make DRAM work, you are looking at some relatively involved and timing critical stuff by comparisson, to say the least...

[peteym5]

What do you think would be the easiest to get and build? SRAM, SIMMs, Static Ram, or something else. I have to go back and find the SIMM mod thread and ask to see if the guy is interested in selling the design or have someone manufacture it. Like I say, there may be alot of 30 pin SIMMs floating around on old PC motherboards. 72pin SIMMs and 168pin DIMMs would simply be too much for our needs. How much memory could we put into such a system? I would like to see 576k and maintaining the 130XE style of separated Antic/CPU access. I know few had talked about a different chip set to control memory bank switching and also have an option of using the 65816 CPU.

 

I did some checking on the 7800 TIA/Maria ideal. Appears that Maria might be able to be turned off along with its DMA like the Antic/GTIA set, it might be possible if you re-address the hardware registers outside of page-zero because the operating system and many of your programming languages use the same area in page 0. Other options is have OS not use the locations, or use a CPU with a Direct register that can change its page 0. We had talked about different operating systems on a chip, maybe have a few geared for the this chip.

 

On the CPU end, WDC does have 65C02S chips that are pin compatible, with the extended instructions like BRA. PHX, PLX, STZ, etc. See http://www.65xx.com/wdc/w65c02s-chip.cfm Has a 14mhz max speed. Can run it 1.79mhz, the cheapest and easiest option. Maybe faster and do the fractional thing for the IC chips, but have to put parts on different boards that run at different speeds. If I were to build a faster system for a faster speed, I'd just use a 65816. The extra instructions come in handy for doing more on a Display List Interrupts, takes less clock cycles to push the registers before doing color changes.

Edited October 28, 2007 by peteym5

[mos6507]

256KB or 512KB would work for traditional banked RAM, but there is no way you'll be able to get 16MB of SRAM for the 65816 at a reasonable price.

[peteym5]

Building it for 576k (64k main + 512k extended) should be enough for any 8-bit applications. The 65816 system would be more of a higher end design and most likely use a different type of memory for that, SIMMs and DIMMs. I have another thread specifically that subject. Looking here to keep costs down, for someone building a new design. Use some newer technology that can be obtained at a reasonable price.

[MEtalGuy66]

Ok, first of all, your not going to adress 16mb of extended ram in any sort of "standard" way with the 8-bt atari architecture. The only upgrades Ive ever seen that even proposed this were contemplating the use of a second PIA chip.

 

Secondly, go check the price and availability of 16meg 30pin SIMMS. They arent just "laying around" in PC motherboards. In fact, ive never heard of a PC of the 30 pin SIMM generation that could even use them. Some macs could use them, but they are few and far between to say the least nowadays.. Yes you CAN find them. No they arent cheap.. and a decent supply of them may be hard to guarantee.

 

For the scale of the ATARI, and for compatability with existing extended RAM upgrades sake, I'd say that 1 meg of RAM would be a pretty ample.

 

And I say again: SRAM can easily and VERY cheaply be made non-volatile without keeping half the logic of the machine alive by battery power..

[peteym5]

On anything with a 6502 based original atari system, 576kb or 1088kb of sram would work just fine, or do that SIMM mod that is here in another thread. The WDC 65c02s are available for less than $10 USD, (40pin) version. Honestly I think going over 1MB is ludicrous for a 6502 8-bit based Atari. Certainly cannot use 72SIMMS or DIMMS on such a system.

 

(I was referring to using 16mb on 65816 cpu which addresses 16mb with a custom motherboard. I don't think I mentioned that in this thread. However I have to see if someone is interested in building a motherboard to do these things. I think the most 30pin SIMMS had was 1mb per module, I remember I could go to 8mb on a 386 system with 8 sockets. For a 65816 system is where 72pin SIMM would be used. I did some checking and 72pin 16MB SIMMs available for like $5 USD. But like it has been said, you need additional logic for that. If you build a system with a 16+mhz cpu, would probably need the faster ram and logic anyway)

Edited October 28, 2007 by peteym5

[mos6507]

There is no sense going over 256K on the A8 since the only thing it would get used for is a ramdisk. And with a HD or compactflash drive, ramdisks aren't that important anymore. I think the only piece of commercial software that used that much was Paperclip XE. And I don't see anyone in a rush to write new programs that use more banked memory. There are plenty of programs that use 128K, though. So that much is useful.

[cbmeeks]

The best place to learn about console/computer creation is here:

 

http://www.xgamestation.com

 

Tell Andre' cbmeeks sent you!

 

cbmeeks

[peteym5]

There is no sense going over 256K on the A8 since the only thing it would get used for is a ramdisk. And with a HD or compactflash drive, ramdisks aren't that important anymore. I think the only piece of commercial software that used that much was Paperclip XE. And I don't see anyone in a rush to write new programs that use more banked memory. There are plenty of programs that use 128K, though. So that much is useful.

 

I think a few games and demos were made that used 128k and that Commando game requires 320k. But memory is cheap, so someone probably will max it out to 576k. Maybe 320K, because after that you start using bits in PORTB that turn Basic On and Off, or loose the separate Antic/GTIA access. Personally I was never exited about big Ram disks, rather see the extended memory used for applications and games.

 

Does bring up another point, maybe build it to use the Turbo Basic inside modification and not bother with Atari Basic.

 

The best place to learn about console/computer creation is here:

 

http://www.xgamestation.com

 

Tell Andre' cbmeeks sent you!

 

cbmeeks

 

This is great, but not sure how this component is related to what we are doing here. Can we make an Atari 8-bit system from it or use it with an 8-bit system?

Edited October 29, 2007 by peteym5

[JamesD]

512K of static RAM is one chip. You aren't going to save much by having less so you might as well include at least that much.

 

Why would you need more than 256K? With a proper DOS you could compile a lot of tools to run natively on the machine. The Apple IIgs is upgradeable to 8MB. That might be enough to run a modern optimizing C compiler natively rather than having to use a cross compiler. You might be able to port some of the older PC games that have public source to the machine.

[Noelio]

I have seen the one laptop per child, but I want a totally new piece of hardware to be developed. Once the cost is known to produce such an item, then we will know what the price is. But the aim would still be something that is affordable to most families. $200 is not out of that range, unless you are just cheap.

 

Sorry to state the obvious, but anyone who is buying an A8 is an Atari fanatic, not someone looking for a budget general purpose computer.

 

This is today's budget computer for the masses:

 

http://www.notebookreview.com/default.asp?newsID=3829

 

Now this is something to get excited about. That article is a bit old--is this thing already on the market?

This is definitely a peak into future tech.

Linux based mini devices such as the GP2x & Nokia N800 as well as the Asus offering are opening up a world of possibilities.

They will be the chosen platforms for emulation. Just recently I purchased a gp2x an a usb cradle for it. Not since my 130xe was bought new at a dept store have I felt this level of wonderment.

 

When I first started posting on this board, my first item of business this same topic we are discussing now!

 

After MUCH thought I came to realize that emulation and hardware are both equally important. With the right board, such as the via epia itx boards you could retrofit any of the 8 bit line in a "McGyveresque" way (see my 1200xp project) and with a good emulation you could capture the feeling of the real hardware while making something that can be relevant for other tasks. They have just recently come out with an even SMALLER board that could fit in a 600XL several times over!!! In a nutshell I'm touting emulation as the cheaper and more feasible answer for instant satisfaction. What we should focus on for the hardware side is an add on device either pci, usb etc for PCs that will allow full use of all atari periphs. Perhaps even make it a universal "80's docking station" full of Atari / Apple / Commodore common interfaces. Hrrrm USB would be the way to go such as the sio to pc USB version. As well a proper interface for using original atari keyboards with PCs might help keep that retrofitted atari more authentic... Especially with character sets etc. It could have a usb header connection that goes right into modern motherboards. I know there are ways now, but something as simple as plugging in the atari keyboard header would be nice especially for a retrofitting kit or something.

 

In a brand new machine I would opt for that Atari XL laptop or 1450XLD that our resident hackmeister has made, even if barebones, requiring a donor 800xl and atari disk drive!!!

 

Anyone have any idea how much it would cost to:

#1 Have the potential mold injection facility ceate the casts from a supplied shell.

#2 Have 100 "copies" made of a 1200xl or a 1450XLD for example.

 

I've got a few ideas that I believe I have to meditate on some as I dont know where to begin to describe. Hahaha.

Great to see the spirit of a new A8 alive and well.