7.16mhz 1200XL

[Gunstar]

 

I'm not sure I understand the theoretical strategy here which imposes these limitations. It's appreciated that the entire 'legacy' bus would need to run at 1.79MHz (which is to say, any base RAM and ROM, IO, etc), but surely this would not hamper the ability of RAM and ROM on a 65C816 CPU card to run at 7.16MHz, 20MHz... or whatever. This is the method employed by Rapidus, which can optionally accelerate segments of base RAM by mapping its own 'fast RAM' at the same address. Providing the accelerator board can supplant the system RAM in the same way as Rapidus, there is no need for high speed signalling to extend beyond the CPU card itself. When the CPU accesses anything which isn't actually on the accelerator board itself, access happens at 1.79MHz. When the CPU accesses RAM/ROM on the accelerator board, it happens at full speed. One could probably go further and combine the functionality of Incognito, VBXE and a 65C816 all on the CPU card itself, which would remove even more restrictions. Of course, such a device would likely be more expensive than Rapidus.

 

If the intention is to try and run the entire Atari bus at a multiple of the base clock, I admit to being quite surprised.

Though I couldn't have explained it like you anyway, I thought he must be joking on some level myself...I just decided to wait at let someone else comment first.

Edited March 31, 2019 by Gunstar

[+bob1200xl]

yep... he was joking. Sorry.

 

I really haven't looked at it, but I would imagine that we could run the 400/800 bus at 7mhz, for real. That would allow you to just make new cards for the CPU and RAM, with maybe a little jumper cable between them. Throw in a CF card adaptor.

 

Bob

[+bob1200xl]

 

I'm not sure I understand the theoretical strategy here which imposes these limitations. It's appreciated that the entire 'legacy' bus would need to run at 1.79MHz (which is to say, any base RAM and ROM, IO, etc), but surely this would not hamper the ability of RAM and ROM on a 65C816 CPU card to run at 7.16MHz, 20MHz... or whatever. This is the method employed by Rapidus, which can optionally accelerate segments of base RAM by mapping its own 'fast RAM' at the same address. Providing the accelerator board can supplant the system RAM in the same way as Rapidus, there is no need for high speed signalling to extend beyond the CPU card itself. When the CPU accesses anything which isn't actually on the accelerator board itself, access happens at 1.79MHz. When the CPU accesses RAM/ROM on the accelerator board, it happens at full speed. One could probably go further and combine the functionality of Incognito, VBXE and a 65C816 all on the CPU card itself, which would remove even more restrictions. Of course, such a device would likely be more expensive than Rapidus.

 

If the intention is to try and run the entire Atari bus at a multiple of the base clock, I admit to being quite surprised.

 

I was kidding.

 

Actually, you can use the Atari bus, you just can't use the old ICs on the bus. They jump on the bus and jump off the bus when they feel like it, not when you want them to. I can guess that Rapidus has gates that isolate the old circuits from the new.

 

With the existing 800, you could use the bus to access memory at 7mhz. Could be cool...

 

Bob

[ricortes]

I'm more confused then committed. Lots of stuff comes out of the original things you said Bob, just with what I think is appropriate ~30 year later updates.Some of my impressions are due to my history of fixing stuff for people & how price sensitive people are. i'm also thinking in terms of a one off hobby computer for myself, just that lots of stuff is derivative of Atari and applies.

 

There's some odd stuff too. For example, you can get a cheap ~$2 a piece CPLD that could be used to replace the 6520s. On the other hand, switching to the current 6522 from WDC would let you run at high speed too. Confuses me as to what would be the best way to go.

 

I've made some questionable decisions already which are exactly boosting my confidence. That video chip from VLSI, 130k RAM composite video, I spent ~$4.30 each, they have an evolutionary chip that has 4 of them in a package so it will do composite & VGA for only $6. Of course there will be people that criticize not using HDMI!

 

Anway, kind of weakly moving towards an 816, min 512k + video chip, arrange it as maybe 130XE below 64k with the rest of the junk & RAM above 64k address space as '816 linear RAM. Kind of a mess at the moment. Using obsolete Xilinx parts for their 5V tolerance.

[+bob1200xl]

I hate to say this, but looking at the schematic of the 800:

 

You could toss the RAM boards and the ROM board and just re-do the CPU board. You don't even have to use the bus at all. (at high speed).

 

Bob

[Kyle22]

Hey Bob. How about the possibilities of an 800 816 CPU card with 16M RAM on it? Is there room to add VBXE? I think the RAM could be easily done by logically filtering address bus requests on the CPU card so no address above $FFFF gets sent to the system bus. In other words, an access to $xD201 would never be seen by the native chipset. If there's room for a VBXE, then so much the better. Run it at 7.x Mhz, We got a winner. If this thing is Incognito compatible, it will be a great seller.

 

 

BTW, Thanx for your help back in the day with the IDE code. As you know, it turned out to be a great thing.

 

Edit: Damnit, Spelling.

 

Edit2: Bob, get on my BBS.

Edited April 1, 2019 by Kyle22

[DavidMil]

 

I did the XL7 (a 7.16mhz 65816) as a proof of concept. It worked for the most part so I jumped into an XL14. That turned into an endless sea of landmines. Too many features, critical clock mis-timings, IC compatibility... running at 14mhz was going to be quite a challenge. Then, I packed everything in 1000 boxes and moved.

 

Anyway, the XL7 was not thoroughly tested, so I can't say how stable it is. I would also like to add a CF card interface to it. We could knock out a few samples and have folks test them. Get an idea where we are.

 

Bob

 

I'd be willing to buy a XL7 from you. Either assembled or in pieces. I can solder it up, even if it has surface mount parts.

DavidMil

[flashjazzcat]

You could toss the RAM boards and the ROM board and just re-do the CPU board. You don't even have to use the bus at all. (at high speed).

 

This is exactly what I suggested in my earlier post.

[+bob1200xl]

Yes, you did!

 

We can use the empty slots for 80col/video cards, and CF adaptors, and audio things.

 

This is not a trivial project, however.

 

Bob

[+bob1200xl]

 

I'd be willing to buy a XL7 from you. Either assembled or in pieces. I can solder it up, even if it has surface mount parts.

DavidMil

 

The XL7 isn't even close to be ready for distribution. If I send you a sample, can you test it for me?

 

Bob

[+bob1200xl]

Kyle:

 

The way it should work is that any request to SRAM is at 7mhz. Anything else is 1.79mhz.

 

I'm not aware of any 5v SRAM at 16MB. Most of the big stuff is dynamic ram that fakes out an SRAM by reading big blocks at once. If you write code that jumps all over memory space, the system will choke.

 

Bob

[DavidMil]

 

The XL7 isn't even close to be ready for distribution. If I send you a sample, can you test it for me?

 

Bob

 

I'd be very happy to test the XL7. I must warn you, I won't be gentle. Do you still have my mailing info,

or do I need to PM you with it?

 

David

[ricortes]

IMHO: Over reaching can be just as bad as falling short. Even if it is wildly successful by our standards i.e. 200 pieces, it would not justify a huge amount of board space and manufacturing. Once again, price comes into play in that you could do the greatest mod ever created but if it took $200 to buy it, 90% of the enthusiast would balk at the price. This is one of the reasons why i think Lotharek's approach makes sense. Rigged demo check this photo out. Of course it would need more parts like 3.3V regulator, caps, et al. Point being if you choose a 40 pin device on the mother board that has all the signals, like the processor, you can pretty much make a plug and play expansion.

I don't want to give the impression I have problems with anybody else's approach, it is all good. It's just when I think about where it leads to I have my own preferences of cost and complexity.

For example, the 512k static RAM I chose was only $3-4. Would a 1meg static RAM be better? Of course! Would 16megs be better? Of course when you are talking about performance but then you take a price hit. But look at what 512k gives you: You could either continue using the lower 64k as is and make the entire address space 65816 limier memory or replace memory 130XE style and have 384k left over for '816 linear memory. This is as much memory as a low end Android cell phone has. It would take some work to adapt some of the existing software to work with it, but I imagine some of the older stuff like Orca C for the Apple GS could be made to work.

Likewise the video chip provides ~130k of video memory that will do everything up to 640x480. It has a blitter function and uses an internal PLL to upscale the system clock so you can get everything from 80 column word processing & terminals to paint programs or web browsers.

Another one of the chips is the ubiquitous 65816 in the 14MHz version. We wouldn't have to run that fast of course, but with the small board layout and parts it would approach that number. It is another good'nuff thing. You have at least 5X the available RAM, 7X the potential clock speed at 7MHz, if that doesn't make you happy, what will? The video chip operates w/o interrupting the processor like ANTIC and no DMA for RAM refresh if you don't want it.

The CPLD I chose was just because it has 5V tolerant inputs but runs at 3.6V so for the sake of argument, put the SRAM on the other side of it. I'm not sure what should be done with it. If I managed to burn a few brain cells I would think you could have the '816 running at 3.3V and do something really cute like make the CPLD into a SPI interface. That should give you everything from SD cards to video displays like the one used in the Arduino drive emulator.

I don't know. I think the chips so far are ~$20 for a set of 4. The problem I can't resolve to moving forward is Atari 40 pin devices like ANTIC or POKEY sell for $10 a pop. It just seems like time to move on if we can. Price performance, low price with high performance, seems to fit our mindset.

[Kyle22]

I would be perfectly fine with me if Lotharek would build it. I don't care it it in kit form or pre-assembled.

 

I like some of the cpld ideas, but we need ANTIC & GTIA for compatibility reasons. With modern parts and technology, I think it can happen.

 

This would the perfect companion to Incognito.

[+bob1200xl]

 

I'd be very happy to test the XL7. I must warn you, I won't be gentle. Do you still have my mailing info,

or do I need to PM you with it?

 

David

 

Don't want gentle!

 

I'm still unpacking my stuff. Give me some time.

 

Bob

[+mytek]

Looks like I stirred up this thread pretty good, and got Bob back in the game.

 

Mission accomplished .

[DavidMil]

 

Don't want gentle!

 

I'm still unpacking my stuff. Give me some time.

 

Bob

 

Oh man! I hate opening boxes anymore! I was back up (you know where) and saw a box marked shoes. I was curious to see what

30 plus years in an attic in Houston would do to a bunch of shoes. My wife's shoes looked kinda like elven shoes, you know with the

toes curled up. The tennis shoes had dry rotted and I could bunch a hole in the sides of them with my finger. But...

Under the shoes and an old rag was two 850 that looked almost new and one 835 modem that look good too. Also found a couple

of 835 boards that look to be complete. No power supplies, but I've got plenty of both types around here somewhere. I just hope

they're not in the ATTIC...

 

David

[mr-atari]

Time to dig-up my XL14 board again.....

 

:-)

[+bob1200xl]

No XL14... I am going to work on the XL7 first. It will have 512K of linear memory and a 7.16mhz clock, both or either can be disabled. Also have programmable OS, maybe an 8K cart, and a PBI interface.(ramdisk and CF card)

 

The objective is simple, cheap and doable. (minimal surface mount logic)

 

Any color you want as long as it's black...

 

Any system you want as long as it's a 1200XL...

 

Bob

[+mytek]

No XL14... I am going to work on the XL7 first. It will have 512K of linear memory and a 7.16mhz clock, both or either can be disabled. Also have programmable OS, maybe an 8K cart, and a PBI interface.(ramdisk and CF card)

 

Bob you might wish to emulate existing CF drive addressing in order to allow the Ultimate 1 Meg and flashjazzcat's BIOS firmware upgrade to access it like it presently does for the SIDE2 cart. In fact you and Jon may wish to compare notes to see what kind of overlap makes the most sense. Of course this assumes you are looking at a commercial release in the future. Otherwise if this is just intended for yourself and few other hobbyist's as a fun exercise of what if scenarios, then pay no heed to what I just said and carry on .

 

Glad to see you once again working on this project.

 

I'll take one in black

[danwinslow]

I'm in for testing and eventual purchase.

[DavidMil]

No XL14... I am going to work on the XL7 first. It will have 512K of linear memory and a 7.16mhz clock, both or either can be disabled. Also have programmable OS, maybe an 8K cart, and a PBI interface.(ramdisk and CF card)

 

The objective is simple, cheap and doable. (minimal surface mount logic)

 

Any color you want as long as it's black...

 

Any system you want as long as it's a 1200XL...

 

Bob

 

You can send it assembled or in parts. I worked at Compaq in one of the board production buildings (CCM4) for 4 years. I can

solder any type of IC I can see and reach with a soldering gun. I don't have an infared device so I can't solder anything with leads

under the IC.

You can always do like my youngest son did when he moved to a smaller home after living in a house for 10 years. That was

six years ago and all of his closets in the new house are still full of boxes to this day! My wife and I volunteered to help him go

through them, but he said no. He knew exactly what was in each closet and each box. I guess that is better than sending everything

to the attic!

 

David

[+bob1200xl]

 

Bob you might wish to emulate existing CF drive addressing in order to allow the Ultimate 1 Meg and flashjazzcat's BIOS firmware upgrade to access it like it presently does for the SIDE2 cart. In fact you and Jon may wish to compare notes to see what kind of overlap makes the most sense. Of course this assumes you are looking at a commercial release in the future. Otherwise if this is just intended for yourself and few other hobbyist's as a fun exercise of what if scenarios, then pay no heed to what I just said and carry on .

 

Glad to see you once again working on this project.

 

I'll take one in black

 

At the beginning of SIO, the 600XL/800XL OS queries the PBI devices and the XL7enables code space (at $D800-$DFFF) for that 'device'. The XL7 will have just enough h/w to do that - the rest is up to the user. Two devices will be included, a CF interface and a ROMdisk.. They run from SIO. The system does not know that they are not real disk drives. You can just turn them off, if you like. There will be two more devices that have no code or association to a device. You could use them for whatever you like - 80 column card, printer driver, or, my favorite, an old HP plotter. There are also interrupt functions in the PBI, but I'm not going to go there.

 

Jon? How do these things that Michael is talking about work? Obviously, the SIDE2 cart is not PBI...

 

Bob

[+Larry]

"Any system you want as long as it's a 1200XL..."

 

Hmm... Your project, your way, but what would prevent it from going in an 600XL or 800XL, assuming that you get a board shape so that it is compatible? The XL14 board layout was compatible with all 3 models, wasn't it?

 

That was an issue with the original XL7 -- the pcb hung over the OS chip sockets, eliminating the possibility of using a different OS that that is thicker. (Think of the 32-in-1 or Puff 3-way OS module, whose name I can't think of at the moment.) The original had no PBI (of course it was for a 1200XL), and was 7 mHz only, IIRC. so having it 1.79 or 7.16 MHz sounds like a nice step forward. With a pbi interface, you would mount a CF card adapter on the pcb?

 

Anyway, glad to see you are interested in improving the original. FWIW, 7 MHz is plenty fast for 99% of user projects. Most of the time, I ran the XL14 at 7 MHz.

[+bob1200xl]

It's going to be a while before I have anything to ship.anywhere.

 

When I do, remind me.

 

Bob