Battlezone Test Mode - No Alphabet

fox_tango

Member
Joined
May 27, 2015
Messages
140
Reaction score
10
Location
Livonia, Michigan
My video gaming neighbor and I are working on a Battlezone.

The test screen was giving me the "L" and only numbers 1-6 on the bottom (no other numbers, and no alphabet)

Items repaired so far:
1. New sockets for every socketed chip on the main board, and for the large 40 pin chips on the mathbox (the 2901 bit slice processors and the 5th other 40 pin chip)
2. Reflowed the interconnect header (many cracked joints)
3. Reterminated the interconnect wiring with new trifurcon pins
4. Replaced main processor (not working) and one 2901 bit-slice processor (also failed)

Now, the test screen doesn't show any mathbox error, but I don't get the alphabet along the bottom like the manual says I should.

However, in game, it appears that all letters are working properly. The game plays and as far as I can tell is displaying all words and scores.

There is just one screenshot that I managed to catch though: Very infrequently and only for a split second, I'll get some alphabet garbage across the screen on the same line that is already showing words. In the last picture I've attached, you'll see the extra letters in the same line as the "GAME OVER". It has only done this very sporadically, and I can't seem to make it come and go with any sort of wiggle or tapping of the boards. I've also seen this in the red area next to the scores.

So my two questions:
1. Should I be seeing the entire alphabet on the test mode screen like the manual says?
2. Is the generation of characters (letters/numbers) performed on the main board or on the mathbox aux board?

NOte, it is probably obvious, but the monitor is on the bench next to me and not viewed via the mirror, so images are left/right reversed.

Any advice on where to look would be appreciated.
 

Attachments

  • Test Screen Clean no Alphabet.jpg
    Test Screen Clean no Alphabet.jpg
    494.9 KB · Views: 34
  • High Score Screen.jpg
    High Score Screen.jpg
    489.3 KB · Views: 30
  • Title Screen.jpg
    Title Screen.jpg
    483.7 KB · Views: 31
  • Game Over Screen with Garbage.jpg
    Game Over Screen with Garbage.jpg
    481.8 KB · Views: 28
Does anyone have an idea on how the letters are generated? Is the aux board involved in this? I can let it go, as the game seems to play fine, but I'm just worried that something really isn't right since the test screen isn't giving me a full line of characters on the bottom.
 
The mathbox is mainly used for 3D calculations, so I would not expect the issue to be there. If it were on my bench, I'd swap it with a known good aux to be sure, but I think you're pretty safe in this case.

Could be something like a marginal ROM or RAM, which isn't bad enough to trip a self-test error, but is failing under the right conditions during gameplay.

Probably not going to be an easy one to figure out, if it isn't ROM or RAM.
 
The character set must be stored in a particular ROM.
I assume the test screen is consistent for all ROM revisions?
 
I'll pop some spare RAM in and see if the behavior changes.

I also wondered if this was just a difference in the ROM version. I'm not even sure which version I've got, but I'll pop them out and look at the checksums in my EPROM burner. I'll burn the other set and see what I get.

Thanks for the tips guys!
 
There are only two ROM versions, and they only differ by a single ROM.

The 036414-01 ROM's checksum is 4263, and the -02 version is 412D.

Note that there are ROM files floating around the web that are incorrect, which say v2, but have the 4263 ROM. I have confirmed this against actual labeled original factory ROMs.

However the issues you are seeing are not due to revision differences. Both ROM versions are the same in all observable ways, as far as I have been able to tell. So it's actually still an arcade a mystery as to what the revision actually changed, and your issues are not due to that.

Also, Atari tended to remove score cheats and similar 'bugs' in revisions, which for other games has resulted in less interesting and more 'op friendly' ROMs. As a result, I tend to prefer the oldest ROM revisions, if I have a choice.
 
Interesting.... seems like the vector instructions at the end of the routine for drawing the number six send the beam to nowheresville (but doesn't crash the VSM).

Self test will run without the AUX board even attached, using just the vector ROM at B/C3 and the game ROM at N1.

The drawing routines for the character set are in the vector ROM, maybe starting right at address 3000 if I remember correctly. But that can't be your problem because you do get the rest of the characters during gameplay. But maybe that's because it's not drawing any sixes.

Or maybe it's the vector RAM that's corrupting the instructions for the VSM. But that can't be it because all the rest of the game works.

Maybe it's the stack / program counter or the VG address selector, but again... no crashes or graphic glitches.

So we've eliminated every possibility... congrats, your game is fixed.
 
Interesting.... seems like the vector instructions at the end of the routine for drawing the number six send the beam to nowheresville (but doesn't crash the VSM).

Self test will run without the AUX board even attached, using just the vector ROM at B/C3 and the game ROM at N1.

The drawing routines for the character set are in the vector ROM, maybe starting right at address 3000 if I remember correctly. But that can't be your problem because you do get the rest of the characters during gameplay. But maybe that's because it's not drawing any sixes.

Or maybe it's the vector RAM that's corrupting the instructions for the VSM. But that can't be it because all the rest of the game works.

Maybe it's the stack / program counter or the VG address selector, but again... no crashes or graphic glitches.

So we've eliminated every possibility... congrats, your game is fixed.



I'd lean toward stack and program counter as well, as a next possibility if it isn't marginal ROM or RAM.

I've seen boards with issues related to the program counter, where it didn't cause crashes/resets. (And he is getting 'glitches', at least in terms of the alphabet garbage across the screen during gameplay.)
 
Update:

No change in test mode behavior.

Swapped (with other used but working):
* 6502 processor
* All RAM
* All EPROMs replaced with CS verified fresh burns

Back in the beginning of this project, as part of getting this going I was the one who removed all the old sockets and installed all new ones. Next step I think will be to check over each one and make sure they are soldered cleanly and make sure there is no funny business between adjacent address or data lines.

With regard to the program execution, I have a scope with the ability to capture 16 logical channels. Can I trigger on a particular address bus value to see when it is drawing a 6 on the screen and then see what address and instruction is next? I'm not sure what I'd do with the data, but is there something to be learned here? Can I actually follow the program execution in this way? I think I understand the concept, but have never done anything like that, nor am I sure if there is value in that exercise.

For what it's worth, I played the game until I got a 6 in the score. I didn't notice any garbage characters at all. I think once the game board and monitor are reinstalled in the cab and my neighbor and I can get some play time on this, we'll see if we really have a problem or not.

It is just weird to get this behavior that seems to indicate a problem, but not really see any actual problems (aside from the very random and infrequent extra characters).
 
Back
Top Bottom