A few words about our departed friend Greg Newby [OC]

https://lemmy.world/post/39739282

Atari 2600 Cartridge Emulation

I’ve been messing around with the Atari 2600 and at some point stumbled across the “Picotari” 2600 cartridge by Nick Bild, which emulates a ROM cartridge for the Atari 2600 using a Raspberry Pi Pico.

You might not recognise the name Nick Bild, but you will be very familiar with some of their projects which seem to blend quirky concepts with some neat engineering solutions. Most of their interesting projects will make it to Hackaday and similar places.

The Picotari was developed to be able to create a pixelated “photo” slideshow on an Atari 2600 turning it into a low-res digital photo frame. One of their recent projects links an LLM into a vintage speech chip to recreate the WOPR talking in War Games. More of their projects here: https://www.youtube.com/channel/UCcUMG56v69cuzsbM0gaSsOQ

There are some details of the Picotari build as part of the project notes on hackaday.io: https://hackaday.io/project/202729-atari-2600-digital-photo-frame/details and it also references another one of their projects: https://github.com/nickbild/picoROM.

Although there are Gerber files provided for the PCB used, and a high level description of the photo frame project, I don’t really have the expertise to follow along fully. I was finding it quite difficult to pin down the actual details of how the project was working, so this is me using that as a basis and guide as I dig into the innards of the Atari 2600 cartridges and documenting what I found out on the way.

The Picotari PCB

The Picotari PCB looks perfect for what I’d like to do, so I decided to see if I could work out what is going on. I started by examining the Gerber files in KiCad’s Gerber viewer. Some points to note about the PCB:

• A 74LVC245A is used as a level shifter between the Atari’s 5V and the Pico’s 3V3. There are only 8 lines though, so I’ll have to work out which signals they are used for.
• A 74AC04 hex inverter is used.
• There are four layers of copper. One of the layers is a ground plane.

KiCad has an option to pull the Gerber layers back into the PCB editor tool, so I did that to make things a little easier to work with. Note that I had to make sure it included all four layers of copper and that they mapped through into sensible layers in the PCB editor too (I ended up with F, In1, In2, B copper layers with In1 as the ground plane).

From this, I can note the following:

• It looks like only one of the inverters is actually used. All other inputs are connected to the Pico’s VSYS line.
• The inverter that is in use is connected to the /OE pin of the 74LVC245, so presumably is used to enable the whole ROM somehow.
• All eight buffers for the level shifter appear to be connected, so presumably these support the eight data lines.
• GPIO 12, 13 and 14 on the Pico appear to be tied to GND.

From earlier browsing around I noticed that there was a pin definitions file in the PicoROM GitHub repository that maps GPIO to data and address lines as follows:

• A0-A14 -> GPIO 0 to GPIO 14
• D0-D7 -> GPIO15 to GPIO 22

And that seems to mirror what is going on here. PicoROM is designed to emulate 27C256 devices, hence requiring 15 address lines: 32K devices require an address space $0000-$7FFF, but with GPIO 12, 13 and 14 fixed to GND that reduces down to 4K.

Mapping all these back onto some existing KiCad footprints and adding some labels gives me the following.

So I can expand on my notes, now noting:

• As already mentioned GPIO 13, 14, 15 are tied to GND dropping the addressable space down from 32K to 4K.
• The Atari edge connector is usually labelled D1-D8, so these map to GPIO15-22.
• A0-A11 do seem to map onto GPIO 0 to GPIO 11.
• 12 address lines will support addressing 4K devices, with an address space $0000-$0FFF.
• A12 goes to the inverter, so A12 going HIGH will enable the active LOW signal of the 74LVC245. This puts it address at $1000-$1FFF. If A13-A15 are used then it will also repeat at all higher addresses $3000, $5000, etc.
• The 8 data lines go through the LVC245 level shifter with D1-D8 mapped to inputs A1-A8; and outputs B1 to B8 mapped onto GPIO 15 to GPIO 22.
• The LVC245 is powered from the edge connector 5V and DIR is also at 5V.
• The AC04 inverter is also powered from the edge connector 5V and that also connects to the Pico’s VSYS.

Looking in a bit more detail at the level shifting, I can see from the 74LVC245 datasheet:

With DIR tied HIGH, it would appear that whatever is presented on the A lines should appear on the B lines (at the voltage level of VCC). In fact the Adafruit product page describes it thus:

“We suggest checking out the 74LVC245 datasheet for details but essentially: connect VCC to your logic level you want to convert to (say 3.3V), Ground connects to Ground. Wire OE (output enable) to ground to enable the device and DIR (direction) to VCC. Then digital logic on the A pins up to 5V will appear on the B pins shifted down to the VCC logic.”

This seems the wrong way round for me. I’d have thought that as these are the (to be read by the Atari) data lines, these should be going from lower voltage Pico logic (on the B bus) to higher voltage Atari logic (on the A bus).

What is also curious is that there is no logic shifting on any of the address lines. I would have thought these would be the inputs (from the Pico point of view), but they seem to be running natively from 5V logic levels and are being fed into the Pico.

I do note that the original PicoROM breadboard used three 74LVC245 chips, but couldn’t find a schematic for either build: https://github.com/nickbild/picoROM

On going back to the original project pages, I noticed this:

I’ve not found any details of what the bodge actually was, but I think there is a good chance that the DIR pin is now connected to GND via that bodge wire to reverse the direction of the device. DIR = LOW means go from B (3V3) to A (VCC or 5V). That is quite a tricky bodge though as the pad for that pin of the chip has 4 tracks connected to it linked to VCC…

As I understand things, the LVC245 should run at the “output” logic level, which in this direction is 5V. But the data suggests that ideally VCC would be 1.8V to 3.6V, but it does say it will accept anything up to 5V as an input. The absolute maximum ratings for VCC (for an SN74LVC245 at least) are -0.5V to +6.5V, but the recommended values are as follows:

The point to note is the Input and Output voltages: the input is always 0 to 5V, but the output will be 0 to VCC.

So at this stage in my understanding, it seems like the 74LVC245 would be great for 5V logic in, 3.3V logic out (when powered by VCC=3.3V) but there are probably alternatives for going the other way. But as the absolute max for VCC is still more than 5V, using it in this direction is probably ok, but there aren’t figures in the datasheet for what high-level and low-level look like for VCC=5V. I guess using a 3.3V high for a VCC=5V must work ok though and the output will be at 5V.

That still leaves the puzzle about the levels of the 12 address lines used. It looks like the original PicoROM had level shifting on them all, but for this board the 5V from the atari goes directly into the Pico GPIO pins.

There is some interesting discussion about the 5V tolerance of the RP2040 here: https://www.derekfountain.org/zx_pico_5v.php

The summary is that the Pico can accept 5V as long as it is powered. But until its 3V3 power is active, it should not see anything more than 3.6V. So this is effectively a power-on race condition – will the Pico see 3V3 power before the Atari sends it some data?

At the time of writing, it has just been announced that there is a new version of the RP2530 which, as well as fixing the now well-known bug with the internal pull-down resistors, is also apparently now essentially 5V tolerant. That might be one option if I can get a Pico with the new version of the chip installed.

Or maybe I should be adding my own level shifting on the address lines.

To be honest, at this point I am wondering if it is worth designing my own PCB building on the above knowledge now gained. I’m just not sure I’m up to trying to design a four-layer board yet. It would be a lot to ask to try to route between the Atari connector and a Pico with just two layers I think.

Maybe I’ll just accept that it seemed to work, so just pick out a sacrificial Pico and leave it to take its chance with the potential 5V inputs…

This is my finished build. Note – I added some insulating tape over the DIR pad of the LVC245 and bent the pin from the socket out to allow me to solder a jumper wire to the GND link of the nearby capacitor.

Atari 2600 Cart Details

There is a lot of detail online about the design and implementation of cartridges for the 2600 system. There were developments of the basic system for the 400/800 and later systems, but they all remained essentially backwards compatible with what went before them for the 2600.

References:

• Atari 2K6 Specifications: https://problemkaputt.de/2k6specs.htm#cartridges
• Atari Age 2600 Schematic: https://atariage.com/2600/archives/schematics/Schematic_2600_Low.html
• Atari Cartridge Documentation Project: https://a8.jindrou.sh/acdp.html
• Specifically the technical details: https://a8.jindrou.sh/acdp_tech.html (although these are more the computers that followed the 2600 than the 2600 itself).
• Atari 2600 on a breadboard: https://chester.me/archives/2021/02/atari-2600-on-a-breadboard-part-2-reading-a-cart/

Key details:

• The 6507 CPU has an 8-bit data bus and 13-bit address bus, so can access up to 8K of memory.
• All data and address pins are present on the cartridge connector.
• A12 is used to select the cartridge, which occupies the top 4K of the address space.
• Weirdly, all the documents list the cart address as $F000-$FFFF when really it is (surely) $1000-$1FFF with the entire address space being $0000-$1FFF – hence 8K, using A0 to A12.
  • Update: yes, according to “Making Games for the Atari 2600” by Stevem Hugg, there are 8 equivalent repeats of the memory map: $0000-$1FFF, $2000-$3FFF, through to $E000-$FFFF. By convention a non-bank-switched cart is treated as address $F000. Bank switching can use some of the additional ranges.
• There are some 2K cartridges. These end up appearing mirrored in the address space at two locations: $F000 and $F800.
• Larger cartridges use bank switching, but I’m not worrying about that at present.
• A12 is usually used as the chip select/enable for any cart ROM.

There are some ROM switcher options out there that allow different 2K/4K ROM images to be stored in a larger EEPROM and presented to the Atari depending on the settings of certain jumpers.

Here are a couple of options:

• https://grandideastudio.com/portfolio/gaming/pixels-past/
• https://www.pcbway.com/project/shareproject/ATARI_2600_MULTI_GAME_CARTRIDGE_PCB.html
• In fact, Joe Grand seems to be the source of a number of PCBs now available: https://oshpark.com/profiles/joegrand/page/3

And of course there were a number of “ROM scanners” and switchers back in the day that allowed choosing between different cartridges. There is a great write-up of one of them here: https://www.the-liberator.net/site-files/retro-games/hardware/Atari-2600/atari-2600-marjac-rom-scanner-1983.htm

PicoROM Code

All the code to allow the Pico to respond as a ROM device for the Atari can be found here: https://github.com/nickbild/picoROM. There are a few things to note:

• The Pico is overclocked to 400MHz, but not on power up which has to be slower to allow access to the flash.
• The whole of the 32K ROM is read into the Pico’s RAM prior to overclocking.
• Data is read from the address lines, and written to the data bus in one go using gpio_get_all() and gpio_put_masked().
• The ROM contents is a set of 32768 rom_contents[idx] = value statements. Well, 4904 statements as it happens for this specific application. These are created by a python script.
• The sample application is not for an Atari 2600, but for his Vectron 65 6502 based computer.

The best reference for building code for an Atari 2600 is “Dr Boo’s Woodgrain Wizardry” here: https://www.taswegian.com/WoodgrainWizard/tiki-index.php?page=Start-Here

It recommends using the Stella emulator for ease of development. This would get to the point of having a binary ROM image which can then be built into the Pico using the script mentioned above, building using the Pico SDK.

The actual running code on the Pico is pretty straightforward:

void main () {
    while (true) {
        put_data_on_bus(get_requested_address());
    }
}

int get_requested_address() {
    // Return only first 15 bits.
    return gpio_get_all() & 32767;
}

void put_data_on_bus(int address) {
    // gpio mask = 8355840; // i.e.: 11111111000000000000000
    // Shift data 15 bits to put it in correct position to match data pin defintion.
    gpio_put_masked(8355840, rom_contents[address] << 15);
}

That is basically it. As fast as possible.

Recall that GPIO0 to GPIO14 are the address lines, hence only grabbing the first 15 bits from the gpio_get_all() call. And GPIO15 to GPIO22 are the data lines, hence masking off the correct 8 bits for gpio_put_masked().

Building PicoROM

First I updated my Pico SDK environment by doing a ‘git pull’ in the sdk, extras, examples, and picotool areas. All my pico stuff is in a src/pico area:

~/src/pico$ cd pico-sdk
~/src/pico/pico-sdk$ git pull
~/src/pico/pico-sdk$ cd ../pico-extras
~/src/pico/pico-extras$ git pull
~/src/pico/pico-extras$ cd ../pico-examples
~/src/pico/pico-examples$ git pull
~/src/pico/pico-examples$ cd ../picotool
~/src/pico/picotool$ git pull

Then I cloned the original PicoROM build and the Picotari project, so I had something to work with.

~/src/pico$ mkdir picorom
~/src/pico/picorom$ cd picorom
~/src/pico/picorom$ git clone git@github.com:nickbild/atari_2600_digital_frame.git
~/src/pico/picorom$ git clone git@github.com:nickbild/picoROM.git

Then I tried to build picoROM as a starting point. And failed.

I don’t know if things changed with a more recent SDK or I just don’t know what I’m doing (which is almost a certainty with the Pico SDK – I find it quite impenetrable and don’t use it enough to get past that right now), but in the end this is what I had to do to build picoROM:

• Remove the artifacts from the original build:

~/src/picorom$ cd picoROM
~/src/picorom/picoROM/picoROM$ rm CMakeCache.txt
~/src/picorom/picoROM/picoROM$ rm Makefile
~/src/picorom/picoROM/picoROM$ rm -rf CMakeFiles

Then I had to recreate some of the basic boilerplate files and entries, which I cribbed in my ignorance, from a previous project.

• Copy in the main Pico SDK “user project” include stuff:

~/src/picorom/picoROM$ cp ../../pico-sdk/external/pico_sdk_import.cmake .

• Add some common initialising instructions to CMakeLists.txt

cmake_minimum_required(VERSION 3.13)
include(pico_sdk_import.cmake)
project(picorom LANGUAGE C CXX ASM)
pico_sdk_init()

Note that without detailing that assembly language is required in the project() directive, you’ll get an error something like “CMake Error: Error required internal CMake variable not set, cmake may not be built correctly. Missing variable is: CMAKE_ASM_LINK_EXECUTABLE”.

Create a build area.

~/src/pico/picorom/picoROM$ mkdir build

Then I can finally run cmake and then attempt a build…

~/src/pico/picorom/picoROM$ cd build
~/src/picp/picorom/picoROM$ cmake ..
~/src/pico/picorom/picoROM$ make

This indeed does appear to give me a pico_rom.uf2 file. I can’t test this one as it is an application for his bespoke DIY computer, but now I know I can build the original project, I repeat the above for the Atari picture frame app…

To start with, there is no associated build infrastructure, so I just copy the missing files from the picoROM area and rebuild.

~src/picorom$ cd atari_2600_digital_frame
~src/picorom/atari_2600_digital_frame$ cp ../picoROM/pin_definitions.h .
~src/picorom/atari_2600_digital_frame$ cp ../picoROM/CMakeFiles.txt .
~src/picorom/atari_2600_digital_frame$ cp ../picpROM/slower_boot2_padded_checksummed.S .
~src/picorom/atari_2600_digital_frame$ cp ../../pico-sdk/external/pico_sdk_import.cmake .

The CMakeFiles.txt file needs a minor update to rename rom.c to pico_rom.c. Then I tried a build

~src/picorom/atari_2600_digital_frame$ mkdir build
~src/picorom/atari_2600_digital_frame$ cd build
~src/picorom/atari_2600_digital_frame/build$ cmake ..
~src/picorom/atari_2600_digital_frame/build$ make

And copied the resulting pico_rom.uf2 to the Pico on the Picotari PCB.

Amazingly I got a display! But it was off – there was obviously something odd going on somewhere…

Then unfortunately it stopped working. I thought I’d killed my 2600 junior as it wasn’t working even with a normal cart, so I dug out my original woody 2600 with an original cart but that wasn’t working either, so concluded it was the TV or at least the TV’s tuning. It turns out that the TV struggles to spot a signal from time to time on the analogue channels, but switching preset channels out and back seems to wake it up again.

The photo frame code is meant to return different bytes from the embedded graphic when a specific address is read. The cart itself “believes” it is just continually reading the same byte. At least that is my understanding of how it is meant to work.

So to me, this looks like a data ordering issue in the code somewhere. As I’m not particularly interested in the picture frame application, this was just a test with a supposedly known-good application, I’m happy that the hardware appears to be essentially working fine.

But just thinking about this for a moment. It could be an issue with scan line timings. It isn’t clear to me if the original project is meant for NTSC or PAL, but given the author is in the US (I believe), I suspect NTSC. It may well be that the Atari ROM code assumes the 30 frame/s update rate of NTSC rather than 25 frames/s for PAL and that messes with the timing of the data. The refresh rates are actually 60Hz and 50Hz but it is an interlaced display so two scans are required per frame of image. PAL images are also 625 lines compared to NTSC’s 525.

With the Atari, all this is pretty important as the CPU has to manage everything to do with the display, timing and all, alongside anything else it is trying to do (see the excellent book “Racing the Beam” for some of the amazing tricks people came up with for getting the most out of this system!).

Of course, it could also be nothing to do with this. At this point I just don’t know enough personally about Atari code to delve into it to see if I could correct it.

I’d like to know if I could get this working, but I suspect that may be a project for another time.

Game ROM

One final test – could I build in an existing game ROM and have that boot and run?

I grabbed a ROM collection off the Internet and found the ROM for space invaders, then built that into the original PicoROM code using a translate python script from the Atari repository.

~/src/pico/picorom/picoROM$ python ../atari_2600_digital_frame/translate_bin2rom.py spaceinv.bin > romdata.c

Then just had to replace the original rom_contents[] = XX lines from rom.c with the new lines in romdata.c and build. There are less lines in the space invaders ROM, but once built and plugged into the Atari, sure enough, I get to see Space Invaders up and running on the display.

I’m not getting sound at the moment, but that could be any combination of an odd ROM or slightly incompatible version (there were about 5 to choose from!) or the TV tuning, which I still have little confidence over.

But as the game does seem to boot and run and is playable, so I’m happy that the basic system seems to be working.

Update: It was TV tuning. Fiddling around and I finally got some sound! 🙂

Conclusion

This is a very clever project and I’m really grateful for Nick Bild for releasing the details of their build.

I feel like I’ve got a bit more of a handle on what is going on now and have ended up with a really useful starting point for a few of my own project ideas.

My next step is to work out how to build a custom Atari ROM executable of my own.

Kevin

#2600 #atari #picotari #raspberryPiPico #vcs

Hands-On With The NEW Atari 2600+ Pac-Man Edition

https://makertube.net/w/7dLaWhxSaN7vDwpE4GY6b4

Atari History, Teil 4 – Die Fans fragen, Klaus Ollmann antwortet

Jeder, der sich wie wir ernsthaft mit klassischen Videospielen beschäftigt, ist natürlich auch immer auf der Suche nach Informationen aus dieser faszinierenden Zeit. Besonders über die Deutsche Geschichte finden man heute kaum mehr brauchbare Quellen.

https://www.videospielgeschichten.de/die-fans-fragen-klaus-ollmann-antwortet/

#2600 #AlienVsPredator #Atari #AtariJaguar #AtariLynx #AtariHistoryReihe #CentipedeWeltmeisterschaft1983 #Commodore #ET_ #GuidoFrank #HansUeliHasler #Homecomputer #Interview #JackTramiel #Jaguar #Lynx #Saturn #Sega #SteveRoss #Sunnyvale #TonyBruehl #UnimexDuplicatorSP280 #USA #VCS #VideoMarkt #Warlords

Now YOU Can Beta Test The New Atari 2600+ 1.1 Firmware!

https://makertube.net/w/tybtYqKjiUmm9uPGEocm5Z

BIG Atari 2600+ Christmas Firmware News - Loads Of Issues Fixed!

https://makertube.net/w/4EU9GasvFcB7vy513pzG3F

Atari 2600+ Ultimate Test: CRT, 7800 PAL Games, 2 Button Controller, Flashcarts & Lightguns!

https://makertube.net/w/suGw1hH5b4W5SZPHjMHiAf

Atari 2600 Junk To "Museum Grade" Restoration & Upgrades

https://makertube.net/w/7F2WZpdvr8kuV36s9aKife

I Played All Of Time Extension's "Best Atari 2600 and 7800 Games Of All Time"

https://makertube.net/w/1NV1coLKhkPq1cKuEBPLPL

Atari 2600 Composite Mod Tutorial - For TV's Jason Bradbury!

https://makertube.net/w/vyc6oyjNQFPsRGSUCzJ4iF

Atari 2600 Tripple Restoration Part 2

https://makertube.net/w/raDK6NqcHMX27zxvKWszL2

Konopka had formed an “anarchist group of boys” called Realm of Chaos to do vandalism, computer hacking and others forms of destruction for “entertainment purposes,” court records said.

“Court revives lawsuit against ‘Dr. Chaos'” – Journal Gazette, 27th August 2003

I’m planning on doing a video about this eventually over on the YouTube channel, but in the meantime let’s look at Dr. Ch@os in a blog.

Dr. Ch@os (referred to as “Dr. Chaos” in the media at the time) is a former computer systems administrator who worked for an ISP called “Infinity Technology” in Green Bay, Wisconsin.

In 2002 Dr. Ch@os was arrested by the University of Illinois at Chicago police while trespassing with two teenage accomplices in a steam tunnel system beneath the UIC east campus, UIC had recently had a string of burglaries and so were on heightened alert. It was at this point that police discovered a vial of sodium cyanide in Dr. Ch@os’ pocket and a digital camera with photos of numerous Chicago buildings on it’s memory card. The campus was temporarily evacuated pending a full inspection of the steam tunnels.

During the course of the ensuing investigation police eventually discovered that Dr. Ch@os had founded a group comprised of an “anarchist group of boys” called “Realm of Ch@os” “the purpose of which was to take personal entertainment” (as described by FBI agent Leslie Lahr) from physical vandalism, hacking, arson, sabotaging of TV and radio broadcasts, destruction of power lines and software piracy.

Along with the litany of destruction that the police and FBI uncovered it was discovered that Dr. Ch@os had been living in a CTA (Chicago Transport Agency) storage room that was attached to the Chicago subway that he had put his own locks on. With access to various CTA employee’s keys and subway tunnels Dr Ch@os was able to traverse large parts of Chicago underground, undetected.

In one storage area of the subway it turned out that Dr. Ch@os had hidden 7 boxes of sodium and potassium cyanide that he had retrieved from an abandoned warehouse and the various keys he had acquired to areas associated with electrical, rail, subway, water reclamation and bridge infrastructure around Chicago.

Chicago’s subway system was shut down as it was searched for any further possible hiding places Dr. Ch@os might have stashed dangerous substances amid what became a major police operation. Once the FBI became involved Dr. Ch@os began admitting his part in an extensive campaign of vandalism along with his group, the Realm of Ch@os.

As a result of the interviews conducted the FBI visited the local Chicago 2600 meeting, meetings loosely affiliated with the hacker magazine 2600, as Dr. Ch@os had attended some of these meetups. I recently filed a FOIA request with the FBI for records related to the interviews they conducted at that meeting.

I had previously filed a FOIA with the FBI for “Realm of Ch@os” or “Realm of Chaos” that was returned with no records found, though I intend to make further requests as this is a fascinating piece of history.

You can find a Wired article with details of the FBI questioning of 2600 participants here.

As well as the 2600 meetings the FBI also briefly investigated some urban exploration groups that Dr. Ch@os had been a part of, Dr. Ch@os was like a living embodiment of hacking adjacent hobbies but very much on the dark side, lockpicking, urban exploration and that classic hacker BBS catch-all section “anarchy”.

The FBI noted that Dr. Ch@os had stolen two computers that he had stashed in the subway maintenance area he called home and was “wardriving” around Chicago to see if he could access sensitive networks.

Ultimately Dr. Ch@os spent about 20 years in prison, some of that time in Supermax with high profile prisoners like the Unabomber and Al Qaeda members, eventually being released in 2019.

You can watch part one of a fascinating WGN News series about Dr. Ch@os below, made after his release from prison, there are four parts and they are all about five or six minutes long.

The WGN series also details a radio tower Dr. Ch@os tried to burn down in his native Wisconsin and his destruction of electrical infrastructure which cut power to the town in which he lived, before he moved to Chicago.

He told WGN, when they eventually tracked him down that although being Dr. Ch@os had been an interesting chapter in his life that he doesn’t have “huge regrets” over, it was not something he wanted to get back into. He expressed interest in becoming an “ethical” hacker.

You can also find a podcast series by WGN on Dr. Ch@os here, although I have not had a chance to listen to it yet if it is as informative as their video series I would recommend it if you want to learn more.

https://realhackhistory.org/2023/03/24/the-curious-case-of-anarchic-hacker-mastermind-dr-chos/

#2000s #2600 #Chicago #DrChOs #DrChaos #DrChaos #FBI #hacker #hackers #hacking #Konopka #RealmOfChaos