New project: ATX motherboard for the Atari ST

[Mikerochip]

Interesting.

Ah, so, the pcb pinout isn't an IDC standard numbering, (Odd on top, even on the bottom) for those 3 ports, but is for the FDD/IDE HDD ports, I guess.

Does the VGA port, out of curiosity?

No harm done though!

Just nice to know, especially if we're making a DMA port PCB, too. (Which I guess, just needs to be remembered!)

[kodak80]

As I am using the 40 pin IDE header for my hard disk, I have not used or tested the ACSI connector although Angel_X confirms that it does work.

Looking at the schematic for the ST ATX board, the ACSI port has alternating pins top then bottom like the parallel and serial connectors. Based on the DMA external port pinouts running top: 1-10 bottom: 11-19 the standard UltraSatan IDC cables won't work with the ATX ACSI connector.

Here is a comparison of the Exxos H5 header and UltraSatan pinout:

H5 US DMA schematic.png

ATX ACSI pinout:

ATX DMA schematic.png

Clearly, the pins are all different and the H5 pinout works with a straight through IDC ribbon cable to and UltraSatan.

So if you plan to connect an UltraSatan, I would say that you will need a fix PCB for this header as well. It would be good to know if anyone has the ACSI connected and using a hard disk?

[kodak80]

I tested the v1.1 of the ISA cart adapter yesterday and it works which is great considering I had to move from Eagle to KiCad for v1.1 due to the size of the PCB. It was too big when I moved the cartridge port outside the ATX casing and Eagle (free version) would not allow the size board needed.

I have spent time today working on the next version (v1.2) of the ISA cartridge adapter and have added in:

• a 40 pin internal header to allow an EtherNEC adapter to be connected using a 40 pin IDC ribbon cable

• Forget-Me-Clock II components to allow a real-time clock

Cart Adapter RTC 40 pin Header-front.png

Whilst I was working on this, I also deciphered the EtherNEC adapter information to create the 8-bit ISA board to connect up to the above new cartridge 40 pin header which should allow network connectivity to the mATX Atari ST.

EtherNEC2022-front.png

And I have also created a fix adapter for the ACSI header that should work

ATX ACSI 20 pin Header Fix-top.png

ATX ACSI 20 pin Header Fix-bottom.png

I have placed an order for these PCBs and will get them tested once they arrive.

[Mikerochip]

Wow! That's *fantastic*.

[matej]

Kodak80: will You sell cartridge adapters?

[Mikerochip]

Boo. I thought there was something wrong with my FDD, so, I got a gotek, and connected it to the mATX pcb.

But, the gotek is saying "Is the floppy cable on the wrong way around?" And nothing attempts to boot from floppy!
TCH!

OK!

Desoldering the 7406N from the ST, and seating in the the mATX pcb solved it! I can boot from floppy!
Just to try it, I reseated the (new) 74LS06N that I had previously, aand. Also booting from floppy fine.

So, if I'd just reseated the two 74 logic chips, I probably would have been fine... FFS!

Anyhow! That marks the first time my micro atx booted from floppy and ran software!
So, I'm happy!!

The mouse PS/2 socket, doesn't appear to be working, but the keyboard works perfectly.

[viking272]

Great work mate!

[kodak80]

Kodak80: will You sell cartridge adapters?

Not sure there is a big market for selling them. I will release the gerbers once tested so others can get some made.

[Mikerochip]

Hmm. Maybe the problem is my PIC?
I used a PIC16F876A-I/SP

What's everyone else got?

I swapped out my MC6850P, aand it still made no difference.

But, strangely, Mouse doesn't work in either OS, but, Keyboard only works in TOS 1.4, but not in 2.6. (Using the TOS images on the site)

[Mikerochip]

OK. The problem may have been mine! I re-seated the PIC several times, but, no joy.
And, I re-flashed it, once, as a final last-ditch before I ordered a new PIC ... and it's working perfectly.

Either I flashed it as the wrong PIC the first time, or, there was a problem with the XGecu T48 software, at the time.
EITHER WAY! Mouse+Keyb+Joysticks all working fine, in both 1.4 and 2.6

[frank.lukas]

Hello @Mikerochip and @kodak80 , can you please put the additional hardware (Gerber) on http://ataripcb.pl/ or somewhere else on the web where everyone has access? Some Users from the Atari-home forum do not have access to atari-forum.com and cannot register either ...

[Mikerochip]

Not sure I have any Gerbers (yet)
I'll sort out that ACSI-BluePill shortly, to sit directly on the MOBO though.

Also, I stuck those (windows) batch files, to split TOS roms for ST/STE/TT/this MicroATX EEPROM set and the PAK68/2+3 up on Github.

https://github.com/M1kerochip/romwak

For anyone who wants them, and, so I don't forget.

[kodak80]

I have created a new page on my website which has the hardware gerbers and information:

https://chillichai.com/atari-st-micro-atx/

[frank.lukas]

Thanks ...

[Mikerochip]

I presume, too, we should be building an IDE hardware swap cable?

[frank.lukas]

Hello Mikerochip, some users have problems programming the Eiffel PIC and setting everything up correctly on the Atari. Please describe exactly how to proceed in individual steps. Should we use the normal 4mhz pic or is the "A" type also possible? Which programmer with which settings is used?

[Mikerochip]

Hi. I bought
PIC16F876A-I/SP
From Mouser.ie, https://www.mouser.ie/ProductDetail/Mic ... F876A-I-SP
I also bought some from Ebay. https://www.ebay.ie/itm/191740036295
The mouser ones actually never came, I see from my notes, so that's why I bought the Ebay ones.

If you look in the photos XAngel posted, I believe that's what's in his board, a PIC16F876A-I/SP

I programmed it with the XGecu T48: I picked '16F876A' and wrote the file 'eiffel_i.hex'
I changed no defaults.

That's it really. It wrote, verified, and then sat it in the machine.
After I built the machine, the eiffel was behaving strangely, so, I popped it out, re-wrote it, and it was fine for a few days.
After another day or two, it was misbehaving again, and I popped it into the T48, and just did a verify, and found 8 bytes were different.
So, I wrote it again, and it's been fine since.

I'm not sure exactly what's going on, but, I suspect those 8 bytes were bad the first time.

When it was misbehaving, the keyboard wouldn't work in TOS 2.6, but would work fine in TOS 1.4. The mouse worked in neither.
The second time it was strange, the keyboard kept repeating.

I can record a video, if you want to see the settings etc. I've changed nothing, btw, from the defaults.

What problems are other users having?

[frank.lukas]

My experiences at the time back in these days were that the "A" type didn't work. After programming the PIC, only the mouse worked, reflash on the Atari and then loading the Keyboard tables and everything was fine working ...

[Mikerochip]

How do you reflash on the Atari, frank.lukas?
Using the Eiffel utility?
I can't run it, atm, since it only works in high resolution, and I'm stuck on Low/Med atm.
(This is my first time I've ever had an eiffel)

[frank.lukas]

I run it on ST-High but should work may be with ST-Med, try it?

Czietz make a patch of the Software ...
https://forum.atari-home.de/index.php/t ... #msg263604
... original it run only with TOS4 I think.

[Mikerochip]

One hurdle would be the initial study of the schematics. I am porting those valuable but crappy scans into Kicad, and having the real thing besides me is required, to check some values (the '6' and '8' tends to be quite similar). I almost finished 4 sheets of the 6. Then hopefully, someone wanting to recreate the STE would have a usable foundation to bootstrap such a project.

(you can follow my progress here and clone the project there)

Hi sporniket, kodak80 has said the midi in of this atx mobo isn't working.
Looking at your STE schematic, pin 5 of the OKI 4N33 is going to GND, and pin 4 to RX.
but the micro atx,
Pin 4 is going to GND, and pin 5 going to RX.

Am I reading that right?

[sporniket]

yes, on my STe schematics, RX comes from pin 4 of the octocoupler, and GND is pin 5.

Now, I just found this reference electronic implementation on Sparkfun's tutorial about MIDI : https://learn.sparkfun.com/tutorials/mi ... ementation

midihw.gif

We can see that the Atari is just following the reference implementation by connecting 4 to rx, and 5 to ground.

[Mikerochip]

This is the micro ATX one.

So, I need then, to cut that trace, on pin 4 and 5, and manually fix pin 4 to rx.

[sporniket]

So, I need then, to cut that trace, on pin 4 and 5, and manually fix pin 4 to rx.

I guess that pin 4 on the ATX is tied to a ground plane (enable visibility of copper pours to check), so you will end up with rx grounded. We really have to "switch" the connections to the pins.

If the chip is not socketed, I believe that I would snip the two pins and lift them (less violent : try to unsolder them, but pin 4 with ground plane will be troublesome -personnally I struggle anytime there is a copper plane-). If the chip is socketed, unplug it, lift the two pins, replug.

Then wires the two lifted pins to their correct pad.

[Mikerochip]

Yeah, mine is socketed. I like the idea of cutting the trace, but, it's a pain in the ass when one of the 'traces' is gnd, as you say.

I find bumping the iron/desolder gun up from 300 to 450 when dealing with gnd, usually sorts it for me.