Thinking of getting a new Atari 800XL board made....

[reifsnyderb]

While waiting for my latest 1090XL board to arrive, I was looking at what 800XL system boards are available.  There's a few people selling the EZ-Contents Blog board.  But I have no real incentive to build one of those as I have a version that has Chroma connected, +5VDC on the PBI, and /HALT on the PBI at pin 33.  So, I got to thinking about what would be a little more interesting.

 

Some thoughts came to mind and I added some more changes:

 

The 600XL had the ability to use a 24 pin or 28 pin ROM for BASIC.  Why Atari never gave the 800XL that capability, I don't know.  So, I added that feature.

 

I re-added C56, the blur capacitor, as maybe somebody would want that capability.  Realistically, most people remove it.  But, why not add the location anyhow.  So, I added it.

 

Some upgrades use the PBI refresh line to, essentially, shut down the MMU.  Normally, I am not a fan of this as it shorts out ANTIC.  (I don't care how safe people say it is, this is a 40+ year old irreplaceable chip....I am NOT shorting out my ANTIC chip.)  However, if there was an open collector refresh line going to the PBI, it would be safe.  So, I changed out the MMU for an ATF22V10 and have it connected so that it could be used as an open collector I/O.  This would protect the ANTIC chip as the MMU could just ignore ANTIC if a PBI device sets Refresh low.  There would be no shorting it out.

 

Since I put a bigger chip in for an MMU, I got to thinking...Atari's Sweet 16 project had PBI RAM at $D600-$D7FF.  While it would be incompatible with some modern updates, it would be something Atari would have added...if they pursued it.  Since I updated the MMU, it was easy to add and there is a jumper to enable or disable it.  As the 600XL and 800XL were the closest to the Sweet 16 project, this change brings it even closer.

 

Atari's normal joystick connectors are all plastic and they tend to wiggle.  So, up upgraded the footprint to use "modern" metal DB-9 connectors.  As they are soldered to the board, they are a lot more stable.

 

Atari's ground planes were, on some boards, a foil at limited locations.  So, why not just put a copper ground plane covering both sides?  Done.

 

While I was at it, and since there can be some issues with Audio and Video, I figured I'd isolate the A/V circuits with their own ground plane and +5VDC.

 

I didn't bother with any other changes as there is this thing called a PBI in the back....

For my purposes, I'll just connect a 1090XL or 1091XL board and enjoy.   🙂

 

I just want to check the existing footprints to make sure I can completely install the 800 video circuit as well.  Another thing to look at is @mytek 's simplified audio circuit works well and I'd like to make sure it's possible to install that circuit in this board.  Otherwise, here's where I am at now....

 

 

 

 

 

 

 

[reifsnyderb]

I think the AV section is finished.  Below, you can make out where the AV ground plane is separated from the rest of the board.  Atari had all +5VDC going through C50.  I decided to remove the ground going to C24 and have that ground go through C50 as well.  The power from L5/C51 to C50 has been slightly modified.  Originally, it went from L5 to C50 with C51 to the side.  I changed it so C51 is now in the direct path of the supply voltage.  C50 was originally a 10uF.  Some upgrades specify putting a higher value capacitor in it's location.  Modern 100uF capacitors fit, so it seems like a good idea to put a 100uF capacitor here. 

 

Also visible are extra pads.  The section can be wired up as per Atari's original specifications.  However, there are some extra pads that can be used to wire up A800 video or @mytek 's audio.  For example, U1 has "C B E".  These are the pads that can, optionally, be used for the PN2222 transistor.  Pin 8, of U1, would be jumpered to the pad below it to supply power to the collector.

 

 

I looked up the Atari 1450XLD's "Carmen" chip.  This chip was a 40 pin MMU Atari was going to use for the 1450XLD.  In the Carmen chip, R/W was available as well as the programming for the PBI RAM.  So, I removed the jumper option for the PBI RAM and added the R/W line.  If the chip enable, for the ROM, is no longer active on a write, Axlon RAM could be added to the PBI.  I figured this MMU would be a "Carmen Light" MMU.   🙂    The tradeoff is that PBI RAM can no longer be disabled.  This is the way the 1450XLD is configured anyhow.  The MMU can be programmed to not use the PBI RAM, of course.

 

As far as the MMU is concerned, with the exception of the right 4 pins and ground, all of the other pins are the same.  Very little modification would be required to install a U1MB.  (Just break the trace to pin 10 and and connect it to ground.)

 

 

 

 

[Mathy]

Hello Brian

 

46 minutes ago, reifsnyderb said:

(Just break the trace to pin 10 and and connect it to ground.)

 

Why not use a zero Ohm resistor or a jumper wire?  That way the user/upgrader doesn't have to cut traces.

 

When Panos designed a new 130XE board, he added a couple of pick up point for signals often used in upgrades.  Maybe you can do that too.  A pick up point is easier to access/solder to than for instance a pin on a chip or a resistor, etc.

 

Sincerely

 

Mathy

 

PS +5VDC on pin 47 and 48 on the PBI?

 

Edited October 24, 2023 by Mathy

[reifsnyderb]

51 minutes ago, Mathy said:

Why not use a zero Ohm resistor or a jumper wire?  That way the user/upgrader doesn't have to cut traces.

The board is rather crowded at this location so it's hard to fit in a jumper section.

 

51 minutes ago, Mathy said:

PS +5VDC on pin 47 and 48 on the PBI?

Absolutely!

 

[reifsnyderb]

1 hour ago, Mathy said:

Why not use a zero Ohm resistor or a jumper wire?  That way the user/upgrader doesn't have to cut traces.

I found a way to stuff the jumpers under the MMU.  Also, I put an extra pad in place to restore Refresh to normal so a stock MMU could still be used.

[reifsnyderb]

New MMU with configuration jumper block underneath.  Changes/Improvements are listed on silk screen.  Also, the pad to "restore" refresh operation back to normal is below the right-hand pad of R950.

 

[Overange]

Wow this is so cool

 

I to have been working on my own 800XL Motherboard project since August, which I've named the '800XLi.' 

 

 

Unlike other available boards, I chose to follow my own unique vision, and here's where I currently stand. 

 

There's still much work to be done before I can begin building the prototype. Like yourself, I've incorporated advanced features from other systems into this board, such as:-

 

Multilayer
Single 64k SRAM on board
Optional basic eprom size
replaced 800xl Video circuit with the XEGS Video Circuit
Crystal circuitry to use latter XE crystal values to overcome PAL Crystal obsolescence
Power circuit being redesigned
Reduced component count
Headers for the most common upgrades

 

The 64k SRAM circuit works great, I built a prototype pcb that currently sits under the ANTIC.

The Pal crystal and colour burst circuit would of worked, if I hadn't put the wrong device footprint on the prototype board lol, but then I found a better crystal divider circuit, so waiting to start on that.

Hopefully I have a lead on a better modern power circuit, so it will be better than the original.

 

Either way, its like a sort of mid life crisis, some build motorbikes from a bag of bolts, we build a computer from a bag of 1s and 0s 

 

At present I do not intend on selling it as a commercial product, but maybe release the design, along with the prototype boards into the public domain.

 

Its cool that we are all trying different things and keeping the Atari 8 Bits alive.

 

But its also time, I got another project that has just been released, and need to close that one before this one recommences, as well as put my other designs out there.

 

 

 

 

 

 

[+Larry]

Is there any way to simplify some of the THT soldering of resistors and capacitors to simplify the build?  Something like resistor networks (that used to be seen on disk drives)?  Or surface mount packages that could be more easily soldered?  Just a thought -- the bending and soldering of a zillion small parts has always been a big turn-off for me.

[reifsnyderb]

1 hour ago, Larry said:

Is there any way to simplify some of the TTH soldering of resistors and capacitors to simplify the build?  Something like resistor networks (that used to be seen on disk drives)?  Or surface mount packages that could be more easily soldered?  Just a thought -- the bending and soldering of a zillion small parts has always been a big turn-off for me.

Parts count reduction can be done and a surface mount version can be created. 

 

Some thoughts for parts count reduction are:

1.  Replace the DRAMs with a single SRAM chip.  This will eliminate at least a dozen chips and a dozen capacitors.

2.  If doing a stock remake, the MMU can be combined with the 74LS138 with a ATF22V10.

3.  Eliminate the RF circuitry.

4.  Use @mytek 's audio circuitry.  (Cuts audio components in about 1/2.)

5.  Use a 74LS14 for the clock circuit.  (Cuts the clock circuitry in about 1/2.)

6.  Eliminate the ferrite beads on output lines.

7.  Use resistor networks.

8.  When possible, use CPLDs and PLCs.

9.  Combine the OS and BASIC ROM into a single chip.

 

@mytek has the 1088 and 576NUC schematics available that show how some of that is done.

 

I wouldn't necessarily say that surface mount soldering is a lot easier unless you are baking the board.  It's easy to have the component float on the solder and have bad connections.  Finding bad connections can be a very frustrating experience, as well.  Some components are horribly small.  I purposely use through-hole inductors, for example, because the surface mount versions are miserably tiny.  (I once compared positioning such an inductor to pushing around a piece of gnat sh*t with a dental pick.  It can be that bad.)

 

I was thinking about staying as authentic as possible.  However, if more people would be more interested in a modern, minimal, build it could be done quickly.

 

Best Regards,

 

Brian

 

Edited October 24, 2023 by reifsnyderb
Wrong chip number.

[reifsnyderb]

1 hour ago, Overange said:

Wow this is so cool

 

I to have been working on my own 800XL Motherboard project since August, which I've named the '800XLi.' 

 

 

Unlike other available boards, I chose to follow my own unique vision, and here's where I currently stand. 

 

There's still much work to be done before I can begin building the prototype. Like yourself, I've incorporated advanced features from other systems into this board, such as:-

 

Multilayer
Single 64k SRAM on board
Optional basic eprom size
replaced 800xl Video circuit with the XEGS Video Circuit
Crystal circuitry to use latter XE crystal values to overcome PAL Crystal obsolescence
Power circuit being redesigned
Reduced component count
Headers for the most common upgrades

 

The 64k SRAM circuit works great, I built a prototype pcb that currently sits under the ANTIC.

The Pal crystal and colour burst circuit would of worked, if I hadn't put the wrong device footprint on the prototype board lol, but then I found a better crystal divider circuit, so waiting to start on that.

Hopefully I have a lead on a better modern power circuit, so it will be better than the original.

 

Either way, its like a sort of mid life crisis, some build motorbikes from a bag of bolts, we build a computer from a bag of 1s and 0s 

 

At present I do not intend on selling it as a commercial product, but maybe release the design, along with the prototype boards into the public domain.

 

Its cool that we are all trying different things and keeping the Atari 8 Bits alive.

 

But its also time, I got another project that has just been released, and need to close that one before this one recommences, as well as put my other designs out there.

 

Looks like a nice concept!  Using the wrong footprint is a painful experience....I've been there too often. 

[Overange]

23 minutes ago, reifsnyderb said:

Looks like a nice concept!  Using the wrong footprint is a painful experience....I've been there too often. 

 Yeah, Kicad Foot prints can catch you out if you do not check check double check  

[+mytek]

Looks like people are having fun with these new mother board projects that'll still fit in place of the stock ones. I obviously chose an entirely different path towards miniaturization which required a lot of what has been talked about here (e.g., replace DRAMs with a single SRAM, substitute resistor networks for multiple individual resistors, ect.). Since so many people end up going the U1MB route, why not do as I did for the 1088 series and provide the footprint with header locations to allow it to be plugged directly in - no jumpers what-so-ever. You can still use that underneath space for the original ROMs and MMU.

[bfollowell]

Didn't someone put out Gerbers a year or two ago for a completely redesigned 800XL & 600XL with tons of fixes and upgrades built-in?

 

I remember the original remake board that I saw about three years ago and, while cool, it was basically just a stock remake.

 

Then the ones I'm thinking of were being actively developed a couple of years ago, definitely post-COVID. Does anyone else remember those or have the links to those threads?

 

So, this is now a third set of 800XL motherboard remakes?

 

I'm watching in anticipation.

 

[bfollowell]

12 minutes ago, mytek said:

Looks like people are having fun with these new mother board projects that'll still fit in place of the stock ones. I obviously chose an entirely different path towards miniaturization which required a lot of what has been talked about here (e.g., replace DRAMs with a single SRAM, substitute resistor networks for multiple individual resistors, ect.). Since so many people end up going the U1MB route, why not do as I did for the 1088 series and provide the footprint with header locations to allow it to be plugged directly in - no jumpers what-so-ever. You can still use that underneath space for the original ROMs and MMU.

 

I was thinking the same thing. Go ahead have headers to plug-in a UAV and use it as the graphics output circuit. Have headers to plug-in the U!MB and a VBXE. Do away with RF completely and use that freed-up board space for something else. That way you could build it more or less stock, even if it is with newer components, but you could easily add some of the modern upgrades.

 

[Overange]

24 minutes ago, mytek said:

Looks like people are having fun with these new mother board projects that'll still fit in place of the stock ones. I obviously chose an entirely different path towards miniaturization which required a lot of what has been talked about here (e.g., replace DRAMs with a single SRAM, substitute resistor networks for multiple individual resistors, ect.). Since so many people end up going the U1MB route, why not do as I did for the 1088 series and provide the footprint with header locations to allow it to be plugged directly in - no jumpers what-so-ever. You can still use that underneath space for the original ROMs and MMU.

I totally agree, and that is where I am going with my design, to have it designed to be a basic 800xl, but have options for fitting any of the available upgrades out there, but then I thought, if U1MB starts to play up, and its traced to my board causing the fault, it makes my board look bad.... so i took it off the board, but with my current design, you could fit it in the normal classic way, but if I find my board is good then it could be put back in  

 

Here is one of my previous designs.

 

 

With regards to video, I took the thought of give the user the choice, so in my design I will use the XEGS video design, and only have Composite output like the XEGS, but if you want super video or digital video you can, by simply adding your own UGV, UAV, Spectra, Shopia in the classic manner etc etc and just hook up to the headers as needed, then unlink the onboard Video, with that in mind, I will keep the RF pads so you can add SAVO or its future models in its place.

 

Also, I have been PMing Reifsnyderb, about my project and how I would not talk about my design in his post, whoops, 🫢 but any way, its good to see other members ideas and thoughts.

 

 

[+mytek]

@Overange you would need female jumpers on the motherboard in order to mate up with the male pin jumpers on the U1MB. And double check the space available for any component that will live underneath because there is stuff on both sides of the U1MB. I ended up having to use elevated female headers on my projects for this purpose.

[Overange]

1 minute ago, mytek said:

@Overange you would need female jumpers on the motherboard in order to mate up with the male pin jumpers on the U1MB. And double check the space available for any component that will live underneath because there is stuff on both sides of the U1MB. I ended up having to use elevated female headers on my projects for this purpose.

Yeah I know, thats just Kicad and more me being lazy not selecting the right 3D part  

 

I did think that height would be an issue, hence also why I took it off.

[reifsnyderb]

Boards can be made that have a pile of plug-in optional boards.  What I ran into a couple years ago was serious feature creep that resulted in something like a $150 parts outlay for the components and connectors.  This was before the cost of the board and Atari specific LSI chips. 

[Overange]

12 minutes ago, reifsnyderb said:

Boards can be made that have a pile of plug-in optional boards.  What I ran into a couple years ago was serious feature creep that resulted in something like a $150 parts outlay for the components and connectors.  This was before the cost of the board and Atari specific LSI chips. 

You may also encounter the potential for sharing some responsibility regarding the third-party plugin.

 

So do you have all the options ( even then you will not catch them all ) or common hook up points or none at all????

[reifsnyderb]

 

11 minutes ago, Overange said:

You may also encounter the potential for sharing some responsibility regarding the third-party plugin.

That point alone sells me on not adding an easy connection.

 

 

11 minutes ago, Overange said:

So do you have all the options ( even then you will not catch them all ) or common hook up points or none at all????

You end up chasing your tail designing the board.  You get options x,y,z added them somebody comes along and asks for w.  Now, you get to re-design 1/2 the board to get it to fit.

 

Well, I've released everything I know about designing 1090XL boards.  Maybe people will start designing their optional cards for it?   😀

 

 

 

[+mytek]

30 minutes ago, reifsnyderb said:

Boards can be made that have a pile of plug-in optional boards.  What I ran into a couple years ago was serious feature creep that resulted in something like a $150 parts outlay for the components and connectors.  This was before the cost of the board and Atari specific LSI chips. 

Aww yes that is true, but only if all parts of the board are populated. Having extra pads for headers and such doesn't in itself add extra cost. So having the plug-in U1MB option would just amount to adding the pad footprints for 3 headers. Adding zero extra cost unless a person opts to stuff the headers for the plug-in U1MB feature. At which point installation of a U1MB becomes a snap.

 

As you said you aren't planning on selling this, and will likely just release the gerbers and schematics. So if you personally don't intend to add a U1MB to your own assembled boards the cost is the same. You're also not responsible for whether having a U1MB causes unforeseen issues. In fact you aren't responsible for anything that occurs downstream since you aren't selling it, nor encouraging others to build it. My policy has always been that the DIY person accepts the risk, and if they get bothersome, that's what the ignore feature on this forum is about. However if they are reasonable, and treat me fairly I'll do the same in return, and always try to help if I can.

 

And last but not least this is your personal project, so feel free to ignore any suggestions that don't fit into your plans - including mine

[Overange]

19 minutes ago, mytek said:

that's what the ignore feature on this forum is about. However if they are reasonable, and treat me fairly I'll do the same in return, and always try to help if I can.

 

And last but not least this is your personal project, so feel free to ignore any suggestions that don't fit into your plans - including mine

You and your projects are the reason for the inspiration to get this going on my own project, mine does have a lot of 576nuc and 1088xld/xle features, that's were I got the foot print dimensions for the U1MB cough cough 😜, originally it was more 576nuc based, but later started to go back to more of an 800XL design with features also from other systems such as the XEGS.

Another thought I had was, based on your sdrive-simple-ii is to have a FUJInet that can be mounted internally above the SIO port and with just some simple holes in the rear for the Fuji nets SD, LEDS and Buttons. but not for me as I have to many other things to do  hint hint 🤪

 

 

[+kheller2]

128K XE Antic/CPU mode please.

 

[reifsnyderb]

2 hours ago, mytek said:

Aww yes that is true, but only if all parts of the board are populated. Having extra pads for headers and such doesn't in itself add extra cost. So having the plug-in U1MB option would just amount to adding the pad footprints for 3 headers. Adding zero extra cost unless a person opts to stuff the headers for the plug-in U1MB feature. At which point installation of a U1MB becomes a snap.

 

As you said you aren't planning on selling this, and will likely just release the gerbers and schematics. So if you personally don't intend to add a U1MB to your own assembled boards the cost is the same. You're also not responsible for whether having a U1MB causes unforeseen issues. In fact you aren't responsible for anything that occurs downstream since you aren't selling it, nor encouraging others to build it. My policy has always been that the DIY person accepts the risk, and if they get bothersome, that's what the ignore feature on this forum is about. However if they are reasonable, and treat me fairly I'll do the same in return, and always try to help if I can.

 

And last but not least this is your personal project, so feel free to ignore any suggestions that don't fit into your plans - including mine

True, adding a plug-in option wouldn't add any cost.  However, this board is all THT components and I don't know where it would go.

 

I may sell the bare boards, if there is some interest.  I've got to order 5 of them just to try one out.  So there will be extras.  

 

Another option is that when the U1MB comes out on PBI, it could just be plugged in.  I highly suspect the PBI U1MB uses the /refresh line so this board would be ideal as it protects ANTIC.

[reifsnyderb]

1 hour ago, kheller2 said:

128K XE Antic/CPU mode please.

 

Well, the PBI has the /halt line and the modern remake of the 1090XL (and 1091XL) supports the /halt line.  So, a 1090 board can be made.   🙂