Avoid conflicts between C64 keyboard and joystick #1

 

The Commodore 64 keyboard may display unexpected characters when joystick #1 is used.  There is a solution!  Avoid the keyboard when scanning the joystick, and exclude the joystick data when scanning the keyboard.  But somewhat easier said than done.

Tracing the source of the issue is that PORTB of CIA1 is connected to the keyboard and first joystick simultaneously.  Normally the C64 KERNAL IRQ will write zero bits to PORTA to select a row, and read from PORTB to read the columns.  And reading the joystick also is read from PORTB.  So with two devices both read from PORTB there's bound to be conflicts.

Reading from the joystick without conflict from the keyboard is the easiest solution.  Write 0xFF (all ones) to PORTA to turn off reading any of the column lines from the keyboard, and read PORTB.  The result will only be the lines from the joystick.  Any of the lower 5 bits which are zeros will mean that the corresponding joystick line is pulled low (switch is connected), meaning a direction is held and/or the fire button is pressed.

Accommodating keyboard reads ignoring the joystick is not as easy because the joystick cannot be logically disconnected in the same way, it is always connected.  The approach I took was to read the joystick, and then read the keyboard ignoring any lines active on the joystick.  This solution gives the joystick priority (as it is always connected), and skips supporting some keystrokes which would otherwise conflict with the joystick.

As inputs can change suddenly, there is debounce logic consistent with how the Commodore KERNAL already handles the keyboard.  Any reads of PORTB are followed by a comparison with PORTB, reading it twice, and making sure the values are identical.  If there is a change in values, the process is repeated without limit.  Both the joystick and keyboard in this solution both have this debounce logic.  And additionally, the joystick is read before and after the keyboard is read, so if there is a change in the joystick values, the process as a whole is repeated for that keyboard row.  As the joystick is repeatedly read, only the last value for the last keyboard row will remain until the next scan.

To implement a general fix for the BASIC Editor (which accepts keystrokes and draws characters on the screen), the IRQ scan key handler would ideally be updated to incorporate these changes.  But the challenge is that the IRQ does other things as well, and the scan key routine is not vectored.  But the key log routine is vectored, which is called when any key is pressed, so the IRQ can be intercepted to scan the joysticks, and the key log routine is intercepted.

The key log routine is responsible for interpreting raw key matrix presses, handle character set swapping, keyboard repeating, and putting keys into the keyboard input buffer.  The new code vectored here effectively throws away the old scan key handler by doing its own new scan key routine avoiding conflict with joystick #1, writing its own results to the current key scan matrix value and shift flag memory locations, and then returning to the existing key log routine to handle these filtered key presses.

The effect is a cleaner keyboard input without conflict from joysticks, and joystick values stored in memory locations 253 and 254.  The joystick values are inverted so they are 0..31 based on positioning (1=up, 2=down, 4=left, 8=right) and fire (16).

Instructions: load the machine code at absolute address, and SYS 49152 to activate it.  Only titles, copyright, and links will be displayed.  While active, keyboard and joystick #1 will avoid conflict, and joystick readings every jiffy will be available to peek from memory locations 253 and 254.  Press STOP+RESTORE to deactivate it.  When deactivated, normal operation including possible conflicts will occur.

Links: open source and d64

No name IPS 240*240 ST7789 TFT display with Arduino GFX Library and ESP32-C3 (RISC-V) development board

 

This is an alternative take on the previous article Adafruit ST7789 TFT display with Arduino GFX Library and M5StampS3.   This time with a generic display, and a different MCU (ESP32-C3).  This one has a single RISC-V core (instead of dual core Xtensa).

So excuse the repeated text, the content has been changed only slightly for the different hardware...

---

Attaching a display to a circuit can provide a lot of detailed info and graphical value.  While "a picture is worth a thousand words," an animated display can be entertainment value, textual information can be informative, and add an input device, and the system can be interactive with billions of possibilities.

The tricks in embedded development is choosing the right display, having the right library that supports the display, and figuring out to use it all together with your target embedded system.

Shown above is a minimal footprint ESP32-C3 development board (I purchased one from AliExpress for around two US dollars, very cheap!), with a generic IPS 240x240 TFT display (ST7789 based) I had purchased years ago on eBay.  An SPI interface is utilized for communications using the GFX Library for Arduino.  I am using a solderless breadboard to prototype the circuit.  Later I will implement as a more permanent circuit.

Besides power and ground, there are only 4 connections from the ESP32: SCLK, MOSI, DC, and Reset.  In this sample, the TFT select line is permanently selected by connecting to ground.

#include <Arduino.h>

#include <Arduino_GFX_Library.h>

Arduino_DataBus *bus = new Arduino_HWSPI(6/*dc*/, 7/*cs*/,
   2/*sclk*/, 3/*mosi*/, 10/*miso*/, &SPI,
   true/*is_shared_interface*/);

Arduino_GFX *gfx = new Arduino_ST7789(bus, 11/*rst*/, 1/*r*/,
   true/*ips*/, 240, 240);

void setup() {

  gfx->begin();

  gfx->fillScreen(BLACK);

  gfx->setTextColor(WHITE);

}

void loop() {

  int x = (int)random(240);

  int w = (int)random(240 - x);

  int y = (int)random(240);

  int h = (int)random(240 - y);  

  int color = (int)random(65536);

  gfx->fillRect(x, y, w, h, color);

  if (random(20)==13)

    delay(100);

}

The wiring corresponds to the code of the databus and gfx initialization parameters, specifically lines GPIO02 (SPI CLK), GPIO03 (SPI MOSI), GPOI06, GPIO11 of the ESP32-C3 connected to SCL, SDA, DC, and RES (Reset) of the display respectively.  Also both the VCC and BLK lines of the display are connected to 3V3, and GND is connected commonly between the display and ESP32-C3 board.  MISO and CS lines are defined for use in the code, but wiring them was unnecessary and not possible.

TFT   ESP32-C3

GND   Ground

VCC   3V3

SCL   SPI Clock (2)

SDA   SPI MOSI (3)

RES   (11)

DC    (6)

BLK   3V3

Not sure why triangles are sometimes displayed (maybe out of range data?).  That is why there is a 5% chance of a tenth of a second delay to pause for the viewer.

This is just a demo.  You should be able to find much better uses for the display that these random rectangles.

Build details:

• Arduino IDE 2.3.2
• esp32 boards 3.0.4
• GFX Library for Arduino 1.4.7

Important!  The board I have will only run its flash chip in DIO mode, so be sure to check all the board options in Tools menu of Arduino IDE before flashing, or if you run into trouble.  Also, updating board libraries may reset your options to default so if flashing after a time, check these settings again, and again.

Some models of the ESP32-C3 development board have a serial chip, and some utilize the C3's built in USB support.  Be sure to set USB CDC on Boot settings accordingly, and use the BOOT/RST buttons to put into bootloader mode for flashing if necessary.

I found it necessary to unplug and replug the ESP32-C3 board to get the graphic demo to work.  Your mileage may vary.

Adafruit ST7789 TFT display with Arduino GFX Library and M5StampS3

 

Attaching a display to a circuit can provide a lot of detailed info and graphical value.  While "a picture is worth a thousand words," an animated display can be entertainment value, textual information can be informative, and add an input device, and the system can be interactive with billions of possibilities.

The tricks in embedded development is choosing the right display, having the right library that supports the display, and figuring out to use it all together with your target embedded system.

Shown above is a minimal footprint ESP32S3 in the form of an M5StampS3 from M5Stack connected to 2.54mm pins, with a rounded corner ST7789 based 280x240 display from Adafruit.  An SPI interface is utilized for communications using the GFX Library for Arduino.  I am using a solderless breadboard to prototype the circuit.  Later I will implement as a more permanent circuit.

Besides power and ground, there are only 4 connections from the ESP32: SCLK, MOSI, DC, and Reset.  In this sample, the TFT select line is permanently selected by connecting to ground.

(The Adafruit display used in this example also includes a MicroSD connector and SPI connections for that as well.  Support for SD is beyond the scope of this article, and would require additional changes to the circuit and code.)

#include <Arduino.h>

#include <Arduino_GFX_Library.h>

Arduino_DataBus *bus = new Arduino_HWSPI(1/*dc*/,

  GFX_NOT_DEFINED/*cs*/, 7/*sclk*/, 5/*mosi*/,

  GFX_NOT_DEFINED/*miso*/, &SPI, true/*is_shared_interface*/);

Arduino_GFX *gfx = new Arduino_ST7789(bus, 3/*rst*/, 1/*r*/,

  true/*ips*/, 240, 320);

void setup() {

  gfx->begin();

  gfx->fillScreen(BLACK);

  gfx->setTextColor(WHITE);

}

void loop() {

  int x = 20 + (int)random(280);

  int w = (int)random(300 - x);

  int y = (int)random(240);

  int h = (int)random(240 - y);  

  int color = (int)random(65536);

  gfx->fillRect(x, y, w, h, color);

  if (random(20)==13)

    delay(100);

}

The wiring corresponds to the code of the databus and gfx initialization parameters, specifically lines 1, 3, 5, 7 of the M5StampS3 connected to DC, RT (reset), SI (serial in), and CK (clock) of the display.  Also the displays V+ line is connected to 5V, and both TC and G (Ground) are connected to ground common to the ESP32 S3.

TFT   StampS3

V+    5V

3V    NC

G     Ground

CK    SPI Clock (7)

SO    NC

SI    SPI MOSI (5)

TC    Ground

RT    (3)

DC    (1)

CC    No Connection

BL    No Connection

It appears there is an overscan issue, beyond the rounded corners.  The gfx object is defining 320x240 display instead of 280x240, and the position and sizes of the rectangles are also interesting here.

Not sure why triangles are sometimes displayed (maybe out of range data?).  That is why there is a 5% chance of a tenth of a second delay to pause for the viewer.

This is just a demo.  You should be able to find much better uses for the display that these random rectangles.

Build details:

• Arduino IDE 2.3.2
• esp32 boards 3.0.4
• GFX Library for Arduino 1.4.7

Wireless controller adapter for Commodore

 

It's August 2024.  I've just bought a new (to me) Vic-20 for US$50 to replace my childhood one.  I already have the PenUltimate Vic-20 Cartridge with a ton of games.  I also bought a Hyperkin Trooper joystick.  But I want to game wirelessly!

I've already had great success with a BlueRetro Nintendo GameCube controllers wireless adapter with my original Wii game console.  I purchased it on ebay for a good price and it worked great.

But I wanted to go back further to replacing an Atari joystick on my Commodore Vic-20 that has a DB-9 connector.  Searching the Internet, it looks like BlueRetro internally supports Atari 7800 with a 2-player mode, and there is a single player version available for sale on Amazon for about $25.  Also the ESP 32 firmware source code is available as open source.

Having experience developing with ESP32, and a plethora of electronic parts already at my disposal, I did what any enterprising consumer would do -- I purchased additional parts from AliExpress.  I mean, you can't have enough electronic parts, can you?  Okay, I could've done it completely from scratch but my time is valuable, I have only so much energy at the end of the day, and cheating my way to success is legal here.   So we're doing it halfway.

First I bought the BlueRetro Core.  It provides the base functionality with a DB-25 connector for all its pins.  It is designed to allow connector adapters to be wired to various console systems.  It is a very creative and functional design.  And there are adapters available for many of the common systems.  But I didn't have any luck finding Atari connectors.

But no problem!  I can build my own!  So I found a DIY screw terminal DB25 male connector with the correct optional mating screws, and the counterpart DB9 female connector with screw terminals.  The only trick left is to wire it up.

Steps

1. Connect BlueRetro Core to USB power adapter (don't require direct connection to computer)
2. Download latest firmware ZIP via link provided on BlueRetro support site
3. Unzip, and look for the parallel 2P firmware (1P should work fine too, skip the 3V3 versions.
4. Follow flashing firmware instructions from BlueRetro support site
5. Pair a supported wireless controller (I'm using an Xbox One Bluetooth compatible controller)
6. Plug BlueRetro Core into the DB25 breakout
7. Connect a voltage meter to ground on the breakout, and then test the voltage at each remaining pin with and without pressing a button (e.g. DPAD up/down/left/right, and A)
8. Write down the DB25 pin numbers and the names of the controller functions
9. Unplug the core, and power off for now
10. Search the internet for the DB9 pin numbers for Atari joystick connector pinout so you have their functions handy
11. Cut and strip ends of 8 wires of different colors long enough to connect the DB9 and DB25 breakouts
12. Be sure to first feed the wires through the included rings, and confirm their orientation.
13. Then connect the appropriate functions: up, down, left, right, fire, framing ground, signal ground, and voltage in  (VIN) to +5V supply from Commodore/Atari
14. Minimally connect, and test the functionality first with meter, then with Commodore/Atari console
15. Once verified working, finish connecting wiring stress-relief, bolt the enclosures closed, and otherwise button-up and tidy up the solution.
16. Play games!

DB25-1   to DB9-4   (right)

DB25-4   to DB9-3   (left)

DB25-GND to DB9-GND (frame ground)

DB25-7   to DB9-7   (VIN to +5V)

DB25-8   to DB9-8   (signal ground)

DB25-13  to DB9-6   (fire)

DB25-15  to DB9-1   (up)

DB25-17  to DB9-2   (down)

Another link to support this wiring is the BlueRetro Hardware Design documented over at hackaday.  There are schematics showing signal ground and VIN with the DB25 pinout.

I had fun with this effort, and should be able to expand this effort to support a second controller simultaneously connected to the same core, or wire up adapters for other systems myself.

The cost of the solution for me was about $22 with shipping, and as I was careful to choose items shipped directly from AliExpress and live on the west coast of USA, received the items quickly in about a week.  While this didn't save me the labor and only saved $3 from the Amazon solution, I learned a lot and feel comfortable leveraging the result to make solutions for other consoles and a second player.

(Any of you worried about 5V tolerance on the ESP32 or the buffer chip included with BlueRetro Core?  The Commodore is at 5V, and while the ESP32 is powered by 5V its IO is exclusively 3.3V.  It works for me without any additional consideration, but your mileage or concerns may vary.  You can usually get away with it for ESP32 as many of the IO are designed to survive 5V, and a 3.3V high signal does read as a high signal with most 5V systems.  So am I just getting away with it?  Maybe.)

But wait! Some keyboard lines are not working. I'm not done yet!

So with some prototyping, it turned out that the circuit to the Vic-20 needed to be open drain instead of push-pull. Pulling down was fine, but something was amiss with pushing 3.3V (or even 5V with other attempts) from the ESP32 BlueRetro Core circuit.

How does an Atari style joystick work anyways? There are 4 directions and one extra button. When any of them are active, the associated line is connected by a switch to ground. The BlueRetro Core simulates this by pulling to ground just fine, but when idle, it is pushing 3.3V. And the Vic-20 has its own 5V pull ups. That does sound messy. Somehow (without looking at the various schematics) this appears to cause a conflict so the 2, 4, etc. and F7 keys for example don't work anymore, they can't pull down to ground when this joystick circuit is attached.

So let's think how to correct this.  We want the connection to ground when the joystick is active, but we want high impedance (appear disconnected) when the no direction or selection is made.

The solution is a circuit that can perform this.  At first I'm thinking I don't have any FETs left over from previous projects, but finally after using one of my go-to buffer chips in breadboard circuits trying to solve this thing, I finally remembered that instead of being annoyed there are enable inputs for each buffer, the enable inputs can function as the inputs from the BlueRetro Core.   They are enable low, so when the joystick selection is made, the circuit will connect, and can take a grounded low input and buffer that to the Vic-20 joystick port.   So when the button or movement is made, the circuit will be grounded low, but otherwise that individual circuit is disabled (high impedance) as if the switch is not connected.  And this separates the 5V Vic-20 circuit from the 3.3V BlueRetro Core (ESP32) circuit.  And best of all, testing on breadboard, it works great!   The keyboard is now functioning normally with the joystick circuit powered and connected, and is still self powered from the Vic-20.

The revision to the above wiring is to connect DB25 joystick lines to OE/ lines of a SN74AHC125N, connect ground to the A lines, and connect the Y lines to the appropriate DB9 joystick lines. Also connect 5V power (14) and ground (7) to the appropriate IC pins.  And since there are five joystick lines, and this is only a quad (4) buffer part, we're gonna need two of them (and three if extending to two joysticks for dual player such as with a C64 or Atari 2600 VCS).  I recommend setting the unused enable lines for the extra buffers high (5V) to disable their outputs.

Again, these buffers are only for up/down/left/right/fire(select).   The voltage and ground wires should be wired between the DB25 and DB9 without buffering.

Vic-20 memory configurations supported by emulator

The C128 team invented the GO64 command for compatibility.  Years ago I claimed to invent the GO128 command, and GO20 and some others.

My Commodore emulator (unified branch) that supports Windows and LCD targets including compact watch form factor, now supports an argument to the GO statement to specify the total memory to configure.

The Vic-20's memory map looks roughly like this.

//   (from emuvic20.cpp)
//   $0000-$03FF Low 1K RAM
//   $0400-$0FFF (3K RAM expansion)
//   $1000-$1DFF 3.5K RAM (for BASIC) [or alternate screen address]
//   $1E00-$1FFF 0.5K RAM (Screen characters)
//   $2000-$7FFF (24K RAM expansion)
//   $8000-$8FFF (Character ROM)
//   $9000-$9FFF (I/O)
//   $A000-$BFFF (8K Cartridge ROM, or RAM expansion)
//   $C000-$DFFF BASIC ROM
//   $E000-$FFFF KERNAL ROM

The Vic-20 was manufactured with 5K RAM, with only 3.5K available for BASIC, not counting the color RAM nybbles (half-bytes).

There is a missing hole of 3K RAM in lower memory, cartridge sold separately.   And you could also purchase 8K or 16K RAM cartridges.  And later on 24K and 32K solutions.   Up to a total of 35K RAM can be added into the memory map.

So specify the amount of RAM you want to add, and rounding down, the emulator will populate the appropriate memory.

But only 27.5K is available to BASIC programs.  1.5K is reserved for system use, the lower 3K RAM is only available to BASIC in some instances, and an upper 8K is reserved for cartridges and otherwise not contiguous, so also not available directly for BASIC use.

The VIC video chip (named VIC) can only address screen memory on the main board, so when both 3K and 8K memory expansions are populated, screen memory is directly in the middle, causing Commodore BASIC to ignore the lower 3K memory to prioritize for the larger contiguous memory configured after screen memory.  The 3K lower memory is only used directly when it is the only contiguous RAM expansion, such as with the separate 3K memory expansion cartridge and with the Super Expander that included 3K RAM expansion, but stole more memory for the programmable function keys, and for the graphics screen.

GO 20        (default, 3583 bytes free)
GO 20, 5    (minimum, 3583 bytes free)
GO 20, 8    (5K, +3K, 6655 bytes free)
GO 20, 13    (5K, +8K, 11775 bytes free)
GO 20, 16    (5K, +3K, +8K, 11775 bytes free)
GO 20, 21    (5K, +8K, +8K, 19967 bytes free)
GO 20, 24    (5K, +3K, +8K, +8K, 19967 bytes free)
GO 20, 29    (5K, +8K, +8K, +8K, 28159 bytes free)
GO 20, 32    (5K, +3K, +8K, +8K, +8K, 28159 bytes free)
GO 20, 37    (5K, +8K, +8K, +8K, +8K, 28159 bytes free)
GO 20, 40    (5K, +3K, +8K, +8K, +8K, +8K, 28159 bytes free)

Note that many of these show the same number of bytes free even though extra memory has been added. This again is because only contiguous RAM can be configured as available to BASIC.  The extra RAM is available for PEEKs, POKEs, and other machine language programs.

If a number out of range is entered, it is effectively rounded down to a valid configuration.  For example GO 20, 64 will configure with 40K RAM.   But 5 is the minimum, so GO 20,0 will configure 5K RAM.

Technically the Vic-20 tests for and can handle memory added in smaller configurations, but that's not commonly present in reality, and not how I implemented it.   Also I only implemented adding 8K memory forward in the map (with the exception of the 3K coming and going based on math), and not configuring each section separately.  Going with a total RAM value is a simpler concept to implement and understand.

Please enjoy the Vic-20 with extra RAM!

Save capability added to C64 version of VWAS6502 monitor

Save has been added to my 6502 monitor program for the C64 version.  The video above demonstrates creating a very simple 6502 program to output printable characters to the screen, and saving that program to disk.

Loading is not implemented, but you can easily use BASIC for that task as demonstrated.  Not explicitly shown is the X command to exit to BASIC.

The new save syntax is in the following form:

1000.2000 "FILENAME" 08 S

Snapshot feature implemented for MINIMUM 6502 mode

 

This snapshot feature is for the minimum 6502 emulation target included in c-simple-emu6502-cbm (unified branch).  (This is a bare bones 6502 target with CPU/RAM/ROM/UART.  The UART is MC6850.  See SWEET-16 and vwas6502 previous discussions.)

Ooo.  The screen is so fancy.  F2 or F3 to bring up this screen.  Up/Down make a selection.  Return performs load or save for that slot, or return on the slot number allows up/down to change the slot number.  When loading, the current state is replaced by the state loaded from disk.  When saving, the current state is backed up to disk.  (The reason for the Up/Down/Enter UI choices is that it can be performed by three buttons or touch equivalents with the supported M5Stack devices.)

F2 defaults to SAVE

F3 defaults to LOAD

This is about the simplest way to provide a general load/save functionality without implementing a higher level monitor.  The state includes 6502 CPU, RAM, ROM, minimum settings (where MC6850 is mapped in memory), terminal state, and UART state.  The data is serialized to a binary blob and stored as a single file with the name state01 or such.   Pretty imaginative huh?

So, now you can use the machine language monitor, or Apple 1 BASIC, or whatever you have ported or implemented to this system, and save a current snapshot of the system to restore later.

Example:

6502 hello world

In the example screenshot, I have restored a snapshot from a previous session where I created a program to display the visible ASCII characters 32 through 127.

So in the future, I can expand on the program, save the current state, and resume where I left off another time.  Also the snapshots can be transferred between systems running the emulator.  In my case, I can run on Windows, or M5Stack CoreS3, or Sunton 7".   Switching is a breeze of copying files to/from microSD. 

I hope this feature is useful to others as well!