Atari 1088 XEL SMD version

[Levas]

I bought original 1088XEL and had problems collecting BOM. Also, I love SMD chips. I ranted about it in other forum thread and soon received some angry (and motivating) responses - if you don't like it, do it yourself. And thanks to original author of 1088XEL, some other people who give me a link to Eagle schematics file I had starting material. After some time (only few hours on weekends and thanks to ... Covid) I have made Rev.0.0 of my version of PCB. It is only TWO layers only (much cheaper in board house) and very common SMD components. It must be 99% compatible with original DIP version with some modifications:

 

• All resistors, capacitors and common chips are SMD
• I removed quite expensive USB-RS232(TTL) converter and replaced with STM32F103 (source from Aliexpress ~1USD module). Software for it is standard example from STM32CubeMX package.
• Removed very specific audio connector.
• No more other components under the chips. Only one (optional) capacitor is under the chip.

 

I have 4 pcs spare PCBs, so if someone wants it, contact me. You pay shipping and package (from EU) and PCB with SRAM chip is free. The only request- please test the PCB with your fancy additional hardware.

All files (for REV.0.0) are available at my www (http://www.vabolis.lt/2021/03/27/atari-1088xel-smd-rev00/ ) , sorry, Lithuanian version only (for now).

I am not reccomending making boards from available gerbers as soon I will post Rev.1.0 with some modifications.

[+mytek]

Very nice ?

 

I do have a question: Since you were designing your board from scratch, why not include the ROM and MMU socket footprints directly on the PCB?

 

Because of the space saving enabled by the use of SMT components, there should have been enough room to have the option of either the U1MB or the conventional ROM and MMU. This would have eliminated the daughter boards I see in the photo for cases where the user only wishes to have a stock 64K machine (no U1MB).

 

But not to detract from your accomplishment,  it looks like you've done a fantastic job, and in a relatively short period of time. Congratulations .

 

EDIT: Check out this schematic for some other options to look at for Video and the P-Channel Power MosFET (SOT223): 576NUCplus_schema_only_3-3-2021.pdf

[+mytek]

2 hours ago, Levas said:

Removed very specific audio connector.

At first I didn't quite understand what was meant by this statement since I can still see the 3.5 mm stereo audio connector on the board. But now I realize it's the DIN-13 Audio/RGB Video connector that is missing. Perfectly understandable since that is rather specialized and not really required unless you plan on having either a Sophia RGB or VBXE upgrade, which a wired Dsub-9 could be substituted for instead.

[Levas]

Some ideas about PSU: I selected very old LM2596 only because it is very available. It is quite big. More modern one is an option. "High voltage" power on is only idea, I need to test it on some scrapboard.

 

[+mytek]

3 hours ago, Levas said:

Some ideas about PSU: I selected very old LM2596 only because it is very available. It is quite big. More modern one is an option. "High voltage" power on is only idea, I need to test it on some scrapboard.

So you opted for higher voltage input instead of using a 5VDC PSU as used in original XEL?

[Levas]

Currently my design is universal- just one jumper ant it is 5V, no jumper- it is "HV". Same with MMU. The only incompatibility is audio connector.

 

I don't know about "original ROM". My PCB is only 2 layers and "wide" traces. So it is quite overcrowded.

 

Currently I must pause PCB modification- I think one of my Atari chips is damaged- I can not find probem. So some software is running quite bad (too fast). Until I solve this problem, I can not modify board- maybe there is some error on it or maybe some logic chip is slow/bad- I used de soldered ones.

 

These crappy screenshots are from "original" 1088XEL, mine version is even worse. Anybody knows the problem?

Edited March 28, 2021 by Levas
pictures

[Levas]

I found error! Error was in schematics I received. Someone, who prepared Eagle schematics made very stupid error. And I was too stupid to notice. All PCB are scrap, unless someone is in love with Kynar wires. Some other pedantic error detected, but nothing fatal.

 

See GTIA chip data bus.

[+mytek]

22 minutes ago, Levas said:

I found error! Error was in schematics I received. Someone, who prepared Eagle schematics made very stupid error. And I was too stupid to notice. All PCB are scrap, unless someone is in love with Kynar wires. Some other pedantic error detected, but nothing fatal.

I'm sorry to hear this . Just a bit of unsolicited advice: you might want to look at the original published 1088XEL schematics on my website as a reference, since those were verified, and there are a multitude of working boards out in the wild to prove that out.

 

Edit: if the Eagle schematic was the one I also recently posted a link to on my website, please keep in mind that no boards were ever designed from that schematic (same goes for the KiCad one as well), so those should be treated as unproven. BTW, if your board layout had been perfectly correct on the first attempt, that would've been quite remarkable, but also not something I would have expected from my 1st attempt. Usually it takes me 3 tries to get it completely right in all regards (the 576NUC+ took 4 tries). 

Edited April 4, 2021 by mytek
added note

[ivop]

I checked my KiCad schematics, and they are correct in this regard. But, I could have made another mistake somewhere else I quadruple checked them with MyTek's original, and found no mistakes anymore the third and fourth time. And as MyTek says, no boards have ever been made. I probably never will. Instead, I'm cutting it up in pieces for Project Jenny, the Backplane Atari Computer project

 

Edit: I'll use all MyTek's designs, plus the original Atari design, as a basis for that. One of the reasons for doing a backplane computer, besides that it's cool, is to eliminate having to order large expensive four layer boards over and over if one has made a mistake. If I make a mistake on a plugin card, I only have to re-order a card that is between 10x5cm and 20x10cm, and two layers.

Edited April 5, 2021 by ivop

[Levas]

I am not blaming anyone, only myself. I cut traces and put 3 Kynar wires. But my one of the Atari Chips is damaged (not by this error)- this error only causes some sprites in the wrong positions. I disassembled another Atari 130 with 64K and I have a working chipset. So, I am testing it again. I soldered the STM32 chip and I can see the Atari as Atari in the Windows device manager (no, it is only a COM port). Very pity, but my USB is weak to power the whole computer.

I can play cartridge games, no SIO at this time. In photo, MMU is replaced with Altera CPLD to test some configurations- I want OS and BASIC in the same chip with minimal changes and full backward compatibility with original. Thinking about 128K RAM (like in the stock). As CPLD is more sensitive, it detects one forgotten pull-up. Also- published MMU PAL/GAL source code is interesting

 

I am thinking about removing LM2596 from my new board. I prefer pure China chips (FR9888, MP1482) - they are smaller, more efficient and very available (tons for 1$ in Ali or salvaged from PCIe risers after ETH madness).

 

Edited April 6, 2021 by Levas
english grammar

[Levas]

On 4/5/2021 at 5:38 PM, ivop said:

 is to eliminate having to order large expensive four layer boards over and over

My design is 2 layers only , €26.58 for 5 boards.  If it was 4 layers... I never would order it.

[ivop]

2 hours ago, Levas said:

My design is 2 layers only , €26.58 for 5 boards.  If it was 4 layers... I never would order it.

? It's nice to see an SMD version of the 1088XEL. And that's a good price! With SMD there's less need for full power and ground planes.

 

My goal is not make it smaller, but larger and make it easy to understand. All through-hole, and an oldskool backplane. The cards will all be 2 layers, 1.6mm, 1oz copper, but I consider doing the backplane at 2mm, 4 layers, 2oz copper and 1oz for the sandwiched layers. If that's not too expensive

[Levas]

Status Report: onboard USB-SIO adapter works (still only without flow control and not all disk images are working).

I know nothing about SIO, so maybe my interpretation of flow control is nonsense.

 

Found PCB bug again- now it is mine, with STM32 lines. Nothing fatal.

 

Now running some demos.

[+Stephen]

Nice progress - this is not bad for a first power up!

[Levas]

What is max size of SIO transfer in bytes? 256? I need for USB-SIO firmware optimization.

 

Current schematics attached.

 

 

1088XEL.sch.pdf

[Levas]

Source code and compiled binaries for onboard USB-serial adapter. Still not working with flow control. I don't know if it is firmware or windows generic serial driver problems.

Not direct link, but link to my web pages. Scroll down for download.

 

http://www.vabolis.lt/2021/04/17/arm21-stm32-usb-cdc-ir-hardware-handshake/

(lots of old man's growling in the post)

[ivop]

13 minutes ago, Levas said:

Source code and compiled binaries for onboard USB-serial adapter. Still not working with flow control. I don't know if it is firmware or windows generic serial driver problems.

Not direct link, but link to my web pages. Scroll down for download.

 

http://www.vabolis.lt/2021/04/17/arm21-stm32-usb-cdc-ir-hardware-handshake/

(lots of old man's growling in the post)

I love Google translate from Lithuanian to English.

...

Shit those authors.

...

Bishop I added fog?

...

 

[+Stephen]

3 hours ago, ivop said:

I love Google translate from Lithuanian to English.

...

Shit those authors.

...

Bishop I added fog?

...

 

See - Monty Python wasn't too far off with their "your hovercraft is full of eels" idea!

[Levas]

On 4/17/2021 at 8:18 PM, ivop said:

Bishop I added fog?

...

 

It is very interesting, because "google translate" used "bishop" here...  The only connection to bishop is sound of "biški". Word "biški" means "small amount" , and "fog" is real word. It is not directly translatable idiom (like "rats in the attic" , "lazy bone" or "running around the bushes"). In current sentence- "I wrote lots of words, but the idea is not clear".

And the "shit those authors" in real means "those authors are tiny peaces of shit (like shit of rabbit)" - Idea is that all authors in internet are copy-paste same example project and every one is writing that "flow control is useless in modern USB communication". But I am not very angry- so called them "tiny shit".

I love to use strange words in my blog, also, sometime speak weird- using ancient form of words.

 

Returning to Atari. New PCB (V1.0) arriving on Tuesday. I will populate it in this week and post the results. Don't worry, all files will be free and I will write all documentation in English language. I am in hurry because in summer I live in summer house and I am not doing any serious electronics at that time. Also, new EU law about taxation of small parcels will be active in month or two. I am not against taxes,  but our government made it impossible to pay taxes as only "special trained person with license to fill custom declaration" can do it. Currently  it is 30EUR for paperwork , 20% tax from everything (PCB, shipping, handling and ... paperwork). We are smart people and I think in few months someone will invent virtual address in other country to avoid theses taxes. Like aliexpress is sending items from some EU countries...

[Levas]

New firmware for USB-SIO, now with hardware flow control (RI and DSR). For CTS flow control write letters to Microsoft or directly to Bill G.

 

http://www.vabolis.lt/2021/04/17/arm21-stm32-usb-cdc-ir-hardware-handshake/#comments

 

[_The Doctor__]

DSR is capable of full speeds on real rs232 devices where as RI is not and has a different edge trigger. This is why most devices pick all the other control lines over ri, but there are some uses for RI so it's nice you have done the work for it.

[Levas]

So, there is V1.0 and it must be 1.01 because of very stupid error (do not trust chip package library from internet). Also, because I do not like how the power button behaves.

It works without any patch wire. It works from cartridges. It works from USB (disk emulator- with hardware or no flow control). The only addition to start it in "stock mode" - mount ROM in riser. Must work with "U1MB"- just remove ROM board and stock MMU (the connector is hidden under MMU, in original location).

If you have an "UAV"- do not populate the stock video part. Untested with "fancy hardware" - I don't have any.

 

It is working from any (6V - 19V) PSU (about 3.5W in "stock" mode). I think it is possible to connect an SD card as a disk drive (no software written, but there are mount pins for it or any other SPI and I2C device).

 

Free PCBs are available (you pay shipping and handling from the EU).

Documentation T.B.D.

 

[+mytek]

You really did a nice job on the layout ?

 

The excellent ground planing, trace routing, and component placement make this into a work of art

 

[Sugarland]

Yes it's beautiful. Great work!!

[Levas]

For testing, I modified MMU (PAL, PALCE or GAL) equations to use the same ROM chip for OS and BASIC. Basic now works, but I have a question- in the original atari, "option" key disables basic. In the original 1088XEL there is no basic at all- MMU pin is floating and everything is done with U1MB. In my version, Option key is not disabling basic!. Also, I can not start tape loading with the "START" key. But, if I connect GTIA's START or OPTION pins to ground, the computer starts tape play and/or diag properly. The problem is with TK-II PIC software.

These keys are working fine in games or OS diagnostics. I can see that OPTION and START lines go low to late on boot.

 

Maybe there is a special key press to enable this boot option from the PS2 keyboard?

 

New permanent page (permalink) for this project in English language: http://www.savel.org/atari-1088xel-smd/