Trak-Ball Atari 2600 Trak-Ball Games

[Andromeda Stardust]

Good points. Missile Command and Colony 7 operate differently in that you have to move quickly to a target onscreen. Speed >> precision in these games, especially that there are no enemy sprites to hit with your cursor if you overshoot the target. The faster you can move the cursor, the better off you are. With the 'pedes, You are navigating tight spaces between frantically moving enemies and must react quickly and precisely. Overshoot in Centipede or Millipede can be fatal. So in these style games where you are dodging enemies, speed << precision. Speeding up the movement could actually make motion clunkier and less precise leading to more frequent deaths.

 

I haven't played the SW game yet and am unfamiliar with gameplay so I'm not sure how movement affects them.

[alex_79]

 

I mentioned before on PM that the ST version feels a little slower to me when moving on the vertical axis compared to the cx22. Now I did some more tests and my impression is confirmed, but only when moving upwards. In fact if you spin the trak-ball slow enough, the cursor doesn't move up at all, while you are always able to move down (and on the X axis) no matter how slow you spin.

I wired an adapter for my trackball and tested the amiga version of SW Arcade. As expected, it behaves exactly like the ST one and it shows the same issue when moving in the UP direction.

 

[Thomas Jentzsch]

Confirmed (Amiga TB and mouse)

[Omegamatrix]

I will have to build a separate rom that only moves a dot on a screen to isolate the problem.

 

I know you said the problem also occured in Stella. I did want to mention I found a physical problem on my cx80 (cx22 wired) trakball on the weekend. All directions except up were working at slow speeds. Up would start working if i applied pressure to the ball. I found out after taking it apart thar one of the bearings wasn't to good. Still have to clean and lubricate that.

 

I have a feeling this problem might be in how i make the plus minus differential. Will look later.

[Thomas Jentzsch]

I have added double speed (-1) CX22 variations for Centipede, Reactor and Challenge of Nexar to the first post.

 

Now those games should be as responsive and precise with a CX22 as with a CX80/ST mouse or an Amiga mouse.

[alex_79]

Thank you for the new roms!

The control is smooth and almost identical to the cx80/st versions. Anyway, in the case of Centipede and Reactor I somewhat prefer the slower movement of the previous cx22 versions which is easier to control when dodging the enemies in Centipede and to manouver in the reduced play area when the core expands in Reactor. So my vote is to return to the previous cx22 version and to halve the speed in the cx80/st versions for these two games (I guess you should increase the value by 1 before halving, else very slow movements which only cause 1 bit change per frame won't be detected).

Challenge of Nexar and Missile Command play great with the latest cx22 double speed (and cx80/st single speed) version, and Colony7 should have the same treatment IMO.

Thank you again and keep up the great work!

Edited December 17, 2015 by alex_79

[txnull]

I currently have an Atari Trackball from a friend here.

It was in an unopened original CX22 box.

It has two cables, one with male and one with female connector.

When I turn it 90º clockwise it works like a ST mouse.

The box has no markings suggesting is it an CX80...

 

But it's some sort of CX80?

 

Does other CX80 Trackballs behave like a ST mouse without rotating?

[funknflow5200]

I will have to build a separate rom that only moves a dot on a screen to isolate the problem.

 

I know you said the problem also occured in Stella. I did want to mention I found a physical problem on my cx80 (cx22 wired) trakball on the weekend. All directions except up were working at slow speeds. Up would start working if i applied pressure to the ball. I found out after taking it apart thar one of the bearings wasn't to good. Still have to clean and lubricate that.

 

I have a feeling this problem might be in how i make the plus minus differential. Will look later.

 

 

I have added double speed (-1) CX22 variations for Centipede, Reactor and Challenge of Nexar to the first post.

 

Now those games should be as responsive and precise with a CX22 as with a CX80/ST mouse or an Amiga mouse.

I just wanted to thank you both so much for your work! These trak-ball conversions are among the best games to play on the 2600 now. I'm a bit biased since I inherently love trak-ball games, but hey, whatever!

[Omegamatrix]

Anyway, in the case of Centipede and Reactor I somewhat prefer the slower movement of the previous cx22 versions which is easier to control when dodging the enemies in Centipede and to manouver in the reduced play area when the core expands in Reactor. So my vote is to return to the previous cx22 version and to halve the speed in the cx80/st versions

 

I vote that too.

[Thomas Jentzsch]

OK, I will create versions with half speed for those two games. Then people can chose what they prefer.

 

BTW: The vertical speed in Centipede is already half speed compared to horizontal speed. Should I further reduce it for the half speed version?

[Eckhard Stolberg]

By the way, apart for the very old posts by Eckhard Stolberg in the stella mailing list (see here), I've found no other mention of cx80 trackball with (presumibly) factory installed ST mouse encoding. Could those have been the results of a modification made by a local distributor instead of Atari itself? There are several aricles on late '80s, early '90s mailing lists or magazines explaining how to turn Atari trackball into Amiga or ST mice and I found a few pictures of home-modified ones (but it was clear by the pictures that the mod was home-made and in those cases the joystick-mode functionality wasn't preserved). If anyone reading this thread has a cx80 trackball with working TB-JS switch that behaves like a ST mouse, and can post detailed pictures of the pcb board, it would be really appreciated.

I bought that trakball in the late 1990 in a small store that specialized in used and NOS Atari ST stuff. They were doing repairs in the back, so it is very well possible, that my cx80 was modified there. Or it might have been modified by the previous owner, since it was a used device. This is a picture of the board inside my trakball. I hope it's detailed enough.

[alex_79]

BTW: The vertical speed in Centipede is already half speed compared to horizontal speed. Should I further reduce it for the half speed version?

I'd say yes: If the cx22 has 1/2 speed on the vertical axis, then the ST/cx80 one should use 1/4 speed to keep the same responsiveness.

 

[alex_79]

I currently have an Atari Trackball from a friend here.

It was in an unopened original CX22 box.

It has two cables, one with male and one with female connector.

When I turn it 90º clockwise it works like a ST mouse.

The box has no markings suggesting is it an CX80...

It's probably an home-modified unit. Rotating the axis allows to use both buttons with one hand and maybe the female connector is used to connect the joystick with the switch on the base repurposed as a joystick/mouse selctor. Anyway I highly discourage to try connecting a joystick to it without tracing the board to verify this theory first, as you could damage the controller and/or the computer otherwise.

 

[alex_79]

I bought that trakball in the late 1990 in a small store that specialized in used and NOS Atari ST stuff. They were doing repairs in the back, so it is very well possible, that my cx80 was modified there. Or it might have been modified by the previous owner, since it was a used device. This is a picture of the board inside my trakball. I hope it's detailed enough.

trakball.jpg

Thank you! Your trackball seems to have the exact same mod as the one described in this post http://atariage.com/forums/topic/243453-atari-2600-trak-ball-games/page-12?do=findComment&comment=3391862(so much that they may have been both modded by the same person), and confirms the assumptions I made on the parts which weren't clearly visible on previous pictures.

I'm not an expert, but I think that your controller left the factory as a standard atari trackball (that is, compatibile with the cx22 model) and it has been modified later in its life. The solder joints for the jumpers and for the cable (which has been replaced to allow for the 2 independent buttons) show flux resin residue which is not present in the rest of the board.

It's plausible that the store specialized in ST stuff modded an old stock of trackballs to make them compatible. After all, there is more software compatible with ST mouse than with the original (cx22) trackball even on the 8-bit/XL Atari computers. http://www.atarimania.com/faq-atari-400-800-xl-xe-what-programs-have-a-track-ball-mode-or-support-a-mouse_70.html

 

Edited December 19, 2015 by alex_79

[Thomas Jentzsch]

I'd say yes: If the cx22 has 1/2 speed on the vertical axis, then the ST/cx80 one should use 1/4 speed to keep the same responsiveness.

 

Probably I was not exact enough:

 

In Centipede the X axis was 100% and the Y axis 50% (200/100% for CX22). If I now half the speed for the X axis, should I again half the speed on the Y axis? Or is the X axis precise enough already? Result would be 50/50% or 50/25% (for CX80, ST/Amiga mouse).

[alex_79]

If I understand correctly, this is the situation:
Centipede (Xaxis/Yaxis speed):

cx80/st version: 100/50%
older cx22 version: 100/50%
latest cx22 version (1.3): 200/100%

I'd say to try 50/25% for the cx80/st (so, having again the speed on Y) and to return to the previous (100/50%) cx22 version.

[Thomas Jentzsch]

I have added new archives for Reactor and Centipede to the first post, containing full speed and half speed versions. Plus source code.

[BigO]

Once I get a working trackball back together, I need to figure out how to even play Reactor. If the goal is to kill off the player as fast as possible, then I guess I have it figured out already.

[Lynxpro]

Thank you! Your trackball seems to have the exact same mod as the one described in this post http://atariage.com/forums/topic/243453-atari-2600-trak-ball-games/page-12?do=findComment&comment=3391862(so much that they may have been both modded by the same person), and confirms the assumptions I made on the parts which weren't clearly visible on previous pictures.

I'm not an expert, but I think that your controller left the factory as a standard atari trackball (that is, compatibile with the cx22 model) and it has been modified later in its life. The solder joints for the jumpers and for the cable (which has been replaced to allow for the 2 independent buttons) show flux resin residue which is not present in the rest of the board.

It's plausible that the store specialized in ST stuff modded an old stock of trackballs to make them compatible. After all, there is more software compatible with ST mouse than with the original (cx22) trackball even on the 8-bit/XL Atari computers. http://www.atarimania.com/faq-atari-400-800-xl-xe-what-programs-have-a-track-ball-mode-or-support-a-mouse_70.html

 

 

That's an interesting list. But there were only 3 commercial titles back-in-the-day on A8 that supported Trak-Balls? Missile Command, Slime, and Shanghai? I'm trying to picture playing Shanghai on a 5200 with the Trak-Ball controller. Actually, that's sad that there's only 3 commercial A8 games that officially supported them [not counting all of the 5200 versions Glenn-the-5200-Man back ported to A8.

 

I know Dan said back-in-the-day, he officially took some Trak-Balls over to EA to give them a demonstration pitch to make games compatible with them. I guess it would've been for A8 because I'm unaware of EA having done anything for the 2600 or 5200 back then nor even considering it. I guess Atari's management decided if more publishers supported the Trak-Balls, then more owners would buy them and then Atari would score a great profit of each one sold. Remember, they weren't cheap in Atari Inc's days. They were only cheap in Corp's later days since they were all surplus at that point. Just a thought, and a side note.

[alex_79]

I have added new archives for Reactor and Centipede to the first post, containing full speed and half speed versions. Plus source code.

Thank you! Tested these on real hardware and found no issues so far. I confirm that I prefer the half speed version for both games.

[Lynxpro]

Tachyon - because it's based upon the GCC/Atari coin-op Quantum - was apparently another Atari 8-bit (A8) game that was supposed to use the Trak-Ball. There's a discussion over at Atari Museum on Facebook about Tom Hudson trying to extract the source code to it.

 

I'm guessing Thomas and Omegamatrix don't do A8/5200 or 7800.

[alex_79]

Here is a a small test rom I wrote which displays the trackball movement as numerical coordinates. My goal was to estimate sensitivity of a trackball without taking it apart and measuring the mechanics (roller diameter and number of gaps in the encoder wheel). I'd feel better giving feedback about the hacks in this thread If I could make sure that the response of my hacked trackball is similar to the original Atari hardware.

tb_test_NTSC.bin

tb_test_PAL.bin

When started, it displays a flashing screen. Spin the trakball on both axis until the test screen (3 lines of extremely blocky text and some colored squares...) appears.

The first line displays 2 characters for the encoding mode detected:
TB = Atari Trak-ball (cx22)
ST = Atari ST Mouse
AM = Amiga Mouse
The following two lines display the X and Y axis coordinates (in blue and green respectively).
Values on both axis range from -5000 to +4999 and then wrap-around.
The trackball is polled continuosly with a max interval between two consecutive polls of about 4 scanlines (0.27 msec). Each variation of the input bits cause the count relative to that axis to increase or decrease by 1 unit.
Pressing the button (left button on ST and Amiga mouse) resets coordinates to 0,0.
Finally, the little squares on the bottom display the current state of the 4 direction bits (In yellow: from left to right they are SWCHA D7,D6,D5,D4 which corresponds to pins 4,3,2,1 on the 9 pin plug) and buttons (In red: INPT4, INPT0 and INPT1 which are pins 6, 5 and 9 of the plug). The middle and/or right buttons can only be detected with some ST or Amiga compatible trackballs and mice.
The RESET switch on the console can be used to go back to the flashing mode-detection screen, in case you need to test a different trackball (or if the autodetection failed).


To estimate your trackball resolution, place a ruler on the trackball so that the '0' of the scale is in the center of the ball, reset the coordinates by pressing the button and slide the ruler along the horizontal or vertical axis for a determined distance (again, stop when the final distance on the scale is in the center of the ball). Take note of the value displayed (sign doesn't matter) and repeat a few times (also try the other axis and both directions on each axis using the same distance). It's important to avoid slipping to have accurate readings, so move slowly and try to move on-axis and not diagonally (The value of the other axis should stay as close to 0 as possible). Placing some tape on the ruler can increase the friction and applying a little bit of pressure with the other hand while sliding it will help too. If the trackball is in good working order (rollers are clean and bearings are lubrificated), you should only have little variation between readings on all axis and directions. Divide the (average) counts value by the distance you used to get the resolution in counts/cm (or counts/inches)


These are the results for my trackball (Sony GB7 for MSX computers, modified to output ST mouse encoding, plus pass-through adapters for cx22 and Amiga mouse):

distance used: 10cm (3.937 inches)
Counts (average):
144 (cx22) -> 14.4 counts per cm / 36.6 counts per inch
288 (st-am) -> 28.8 counts per cm / 73.2 counts per inch

Of course these values aren't of much use without something to compare them with, therefore if anyone want to try this and post the results, it would be much appreciated.

 

[Thomas Jentzsch]

Of course these values aren't of much use without something to compare them with, therefore if anyone want to try this and post the results, it would be much appreciated.

Very good idea. Unfortunately I am on vacation for the next 12 days, so my results will have to wait.

 

BTW: You could also measure how many counts you can create per frame (60 and 50 Hz frame rate).

[BigO]

Modded CX-53

CX-22 mode: 170

ST Mouse mode: 340

Do any of us have original Atari hardware?

[alex_79]

BTW: You could also measure how many counts you can create per frame (60 and 50 Hz frame rate).

 

Here is a version which displays the highest counts per frame reached so far for each axis. Like the coordinates, it is zeroed by pressing the button.

The max value that can be detected is limited by the polling rate to 78 counts for PAL and 65 for NTSC. (about 3900 counts/sec)

tb_test_v2_NTSC.bin

tb_test_v2_PAL.bin

 

 

Modded CX-53

CX-22 mode: 170

ST Mouse mode: 340

 

Thanks! You didn't mention how far you moved the trackball to get those counts. Did you use a the same value (10cm) as I did?

Edited December 23, 2015 by alex_79