The Naked Atari 800

[rubber_jonnie]

While I'm waiting on some more parts to arrive I threw together a simple prototype board that I can at least quickly use for testing:

IMG20260118150236.jpg

IMG20260118150222.jpg

IMG20260118150214.jpg

Part of the idea being that I can test different configurations, so no USB bootloader, USB bootloader with no startup delay, different clockspeeds etc with the ATTiny boards I have with it being quick to then retest.

I also give me a nice visual wiring reference and works great. It's not something I'll use as a working model, more of a bench thing and I'll look to make something neater for the final iteration.

[PhilC]

Looks good @rubber_jonnie , I'll definitely try that clock gen out at some point.

[rubber_jonnie]

Looks good @rubber_jonnie , I'll definitely try that clock gen out at some point.

They are pretty amazing little devices, and if you perhaps only need 2 clocks rather than 3 then there is an even smaller Sparkfun device with just 2 outputs.

I do need to sort the bootloader part out though so have some more parts coming that'll hopefully help me get that done. No bootloader=instant start, so I'm just waiting on a USBASP programmer which should be here tomorrow and will allow me to push a bootloaderless config to the ATTiny85. Fingers crossed...

[rubber_jonnie]

So I got some DIL packaged ATTiny85's yesterday:

IMG20260122092643.jpg

So I should be able to program these without a bootloader with my TL866, so I'll try that later, although I do also have a USB programmer now so have options.

Theoretically without the bootloader it should power up and fire the config into the Si5351 immediately meaning no 6 second delay on power up. Effectively this will be an almost instant clock so no power on delay :)

On a slightly different subject, I have also acquired an NTSC CPU board from the same guy I got the 800 from. It's quite a bit different than the PAL CPU board:

IMG20260122085108.jpg

The top board is NTSC and the bottom is PAL. The NTSC colourburst is 3.579575 Mhz so it simply uses the system clock rather than a separate crystal.

The PAL board has one less logic IC and a 4.433618Mhz crystal for the PAL colourburst.

So I am going to look at trying the following:

1. Configuring the Si5351 for 3.579575 Mhz and fitting the NTSC CPU board to see if a quick conversion from PAL to NTSC will work.
2. Configuring the Si5351 for 3.546894 MHz and 4.433618 Mhz to see if I can optionally replace both crystals
3. If running using both frequencies works just fine I'll see if it'll work in another machine, for example an XL or 2600Jr

Fingers crossed with the ATTiny 85 configured without a bootloader this will work perfectly, then I will see if I can get myself started on a KiCad project to create a board that uses an ATTiny with no USB board connecting directly to the Si5351 :)

I'll build it on protoboard first to verify operation.

It's been a fun project.

[ijor]

On a slightly different subject, I have also acquired an NTSC CPU board from the same guy I got the 800 from. It's quite a bit different than the PAL CPU board:

Somebody replaced the 6502 on that NTSC board for a Sally custom chip. That would not work without some modification.

The PAL board has one less logic IC ...

Because it is one of the later CPU boards designed for Sally. It doesn't need the on board HALT logic that is already integrated in Sally.

Theoretically without the bootloader it should power up and fire the config into the Si5351 immediately meaning no 6 second delay on power up.

Btw, there are other similar clock generators with non-volatile configuration memory that don't require to be programmed on every power-up. I understand they are more expensive, though.

[rubber_jonnie]

Somebody replaced the 6502 on that NTSC board for a Sally custom chip. That would not work without some modification.

Yes, I grabbed one of 2 CPU's from an anti-static bag and happened to grab Sally rather than a standard 6502, so I need to swap it out.

Because it is one of the later CPU boards designed for Sally. It doesn't need the on board HALT logic that is already integrated in Sally.

Interesting information that :)

My PAL CPU board only works with a Sally BTW.

Btw, there are other similar clock generators with non-volatile configuration memory that don't require to be programmed on every power-up. I understand they are more expensive, though.

Yeah, I was aware but wanted to keep the cost down for anybody wanting to use these parts for themselves as they are cheap and easily available. As it goes, the Clockbuilder Pro app used to generate the register values is able to generate custom part numbers if you want to order SI5351's with specific frequencies from the get go. Apparently there are programmers that allow you to do this yourself, but I can't find anything related.

Even if you do have custom parts you can still feed them a different config if you need to change frequency.

[ijor]

My PAL CPU board only works with a Sally BTW.

Yes, of course. That's a Sally board. You can't replace Sally with a standard 6502. Not without adding some extra logic.

Btw, I remember you said earlier that your board didn't much the schematics. I didn't pay too much attention then and assumed it was because of the PAL extra clock. Didn't realize then it was a "Sally" board. For obvious reasons, this board is quite different than the "original", 6502 based, one. Seems that Atari itself never published the schematics for this CPU board variant with Sally. But it is included in the Sam's Computer Fact:

https://www.atarimania.com/documents/Sa ... ri_800.pdf

[rubber_jonnie]

My PAL CPU board only works with a Sally BTW.

Yes, of course. That's a Sally board. You can't replace Sally with a standard 6502. Not without adding some extra logic.

Btw, I remember you said earlier that your board didn't much the schematics. I didn't pay too much attention then and assumed it was because of the PAL extra clock. Didn't realize then it was a "Sally" board. For obvious reasons, this board is quite different than the "original", 6502 based, one. Seems that Atari itself never published the schematics for this CPU board variant with Sally. But it is included in the Sam's Computer Fact:

https://www.atarimania.com/documents/Sa ... ri_800.pdf

That is very helpful indeed, thanks for that, I had been finding it difficult to get the schematics for the board, makes a big difference when you can see what goes where :)

[ijor]

That is very helpful indeed, thanks for that, I had been finding it difficult to get the schematics for the board, makes a big difference when you can see what goes where :)

Note that it seems it's not same exact board you have. The schematics are for a NTSC Sally board. The actual PAL logic with the second crystal is missing.

As a matter of fact the schematics describes a Sally CPU board with CTIA, not with GTIA. Interesting, but I doubt such a board was ever shipped in that configuration. By the time that Sally was released, all boards probably shipped with GTIA, and not CTIA.

[rubber_jonnie]

OK, so excellent news from clock town.

I've now been able to program the 3 SMD ATTiny85's that I originall got so there is no bootloader, just the program to fire the Si5351 config over I2C and the results are spectacular.

The 800 now boots instantly with zero perceptible delay to the clock starting. The machine just starts like my other 800 and the 400, no wait, nothing, just a nice normal start.

Once running the clock is happy over resets as well, and running games and demos seems just, well, normal.

Now that pesky bootloader is gone it really is super quick, the only problem I had was that because I was using the SMD boards, I had to desolder the ATTiny, solder it onto a breakout board that I could use with my TL866, flash it and then reverse the soldering process.

Now I have some DIL ATTiny85's that I can remove from their boards for programming, so what I'll do next is build another test board for them since they have a different pinout to the SMD boards.

Once I have that done I can try programming them and see how they fare. I expect the results to be the same as with the SMD version.