Major Havoc Random Resets

On other games, (Star Wars comes to mind)... Self test will work fine and the game will not run when there is something wrong with the interrupt circuit. I'm not sure on MH. Interrupts are often disabled in self test.. Once more thing to look at..
 
On other games, (Star Wars comes to mind)... Self test will work fine and the game will not run when there is something wrong with the interrupt circuit. I'm not sure on MH. Interrupts are often disabled in self test.. Once more thing to look at..
I think you're on the money here.

I made an adapter to my bench power supply that only does power and video. And the test switch. Nothing else.
Imagine my surprise when I put the thing on the bench and it booted up immediately. Odd.

I put the board back into the cabinet. No boot.
Back on the bench. Boots.
Back in the cabinet. No boot.

At this point I start messing around and pressing on stuff more. I hit the service credit button. Board boots. I turn it off and on again. No boot. Press the service credit. Board boots. This is repeatable. Why??

Also, it is drawing a box around the playfield. I feel like that's not normal.

 
Or perhaps not. Looks like the service button jumper had fallen off and into the normally closed position. That explains the lack of boot in the cabinet.

However it's still drawing these intermittent squares on the window and it is not reliable. It frequently dies or resets. It can completely lock up also, where the start buttons stay illuminated and it will not reset into test mode until the cabinet is power cycled.
 
Or perhaps not. Looks like the service button jumper had fallen off and into the normally closed position. That explains the lack of boot in the cabinet.

However it's still drawing these intermittent squares on the window and it is not reliable. It frequently dies or resets. It can completely lock up also, where the start buttons stay illuminated and it will not reset into test mode until the cabinet is power cycled.

So @mhavoc is going to know for sure here but I believe the squares are on purpose to avoid activation of the spot killer circuit on certain screens. There's a work loop called waste time in the code and that's what I recall anyway.
 
So @mhavoc is going to know for sure here but I believe the squares are on purpose to avoid activation of the spot killer circuit on certain screens. There's a work loop called waste time in the code and that's what I recall anyway.
These are definitely on purpose. @Kaworu5 checked this for me on an FPGA and it does the same thing.

Now then.

Some steps forward, some steps back, mostly forward.

The Good:
After chasing my tail for hours/days/weeks, I have come to the conclusion that the runtime instability is the damn SEEQ eproms. No matter how clean and oxidization free they are. No matter the fact that they read good in self test and in the programmer. They. Freaking. Suck.
How I arrived to this conclusion. I burned an entire new set of EPROMS, and started pulling and replacing sub groups. Alpha Group, no change. Gamme Group, no change. ALL ROMS, and then it ran for an hour without resets. I thought about it more, and then replaced all the SEEQ ROMS only. That seemed to still be running good.

Yesterday, I ran the thing for five hours and it didn't crash or reset at all once it got going*. So I'm crossing fingers on the EPROM issues here.

It had some audio issues I noticed at the end of that run, with some incorrect sounds, but I identified that as a lose LS245. Sounds seem good currently.

The Bad:
For whatever reason, it's still freaking out when first turned on cold. The first ~30-60 seconds, it's kinda unstable. Takes a few seconds to boot initially, and then it can reset a few times. However, once it's "warmed up" it's rock solid. I ran it all afternoon yesterday with one credit on the counter and it stayed there. Same thing today. At the point that it's "warmed up" if I cycle power to the cabinet, it will come back up immediately and not reset or display any weird behaviors. So I'm wondering what's going on there.

I'm mad at this whole ROM thing. I'm an idiot and it cost me a lot of unnecessary time, work, and socketing that probably introduced more problems from just manipulating the board.

So:
Intermittent Box -> Normal
Runtime Instability -> SEEQ ROMS suck
Boot Instability -> ???
Sound Issues -> LS245 Bad Socket
 
Time for the ice cube trick.
The Saga continues.

Fiddling with chips, I noticed a LS245 (12P) was affecting the resets. If I pressed hard on it, it would boot fine. Replaced that LS245. The game is booting solidly and playing the right sounds* and will keep playing without shutting off.

So, I turned my attention to the LETA. The new (eBay) LETA replacement arrived today. Plugged it in, the roller seemed like it was moving jerkily for a few seconds, and then nothing. The light on the LETA is on, the voltages seemed safe (TTL levels) when I had checked on the bench earlier, and checked now, so I don't know what's up. Did the game fry another LETA?

What could even fry it? As I had mentioned before, the LETA/Roller failure occurred after I had replaced the big blue on the system. I did nothing else at that time. Replaced the big blue, tested the AC ripple prior to hooking up the board again, came in good, hooked up the board, roller died.

I did pop in another (untested) opto pcb from my shelf. Didn't seem to do anything different. Measured 5v at the Optos also. I didn't see any logic on the probe when messing with it connected to the game. So that's weird, but it was the same thing on both. I need to try bench testing it I think.


* The right sounds except for credit. For whatever reason instead of playing the two tone sound, sometimes it will play a *boop* sound effect. Not sure if this is normal or not - I've always had it on freeplay and never credit drop before.
 
It is in fact the opto on the Havoc.

I pulled the known good one from my tempest and everything seemed to work fine. The one from Havoc did not work in my tempest.
So I guess my spare was *also* a dud. Neat.

And for those keeping score at home, the original LETA is definitely dead. The repro works however.
 
Well... not so fast

Havoc (Dedicated) and Tempest use 2 differently configured opto boards. One uses a 150 ohm and the other uses a 100 ohm resistor. (I don't recall which ATM). You should validate what you have as a "spare". I can't recall if Tempest will work with either resistor configuration.

Your Havoc is running as a conversion correct? So that means you should be using a Tempest controller. This also requires the addition of a pull up resistor at the LETA or to GND the pin. The pull up resistor is on the LETA multiplier pin to make the controller more sensitive for Tempest games. In many cases Atari failed to stuff this resistor into the conversion boards. Sometimes the boards "worked" but many times the game sorta didn't and they actually got worse over time. On the MH OEM conversion boards you need the pull up or short the solder pad to GND.

See this.
 

Attachments

  • 1746365770254.png
    1746365770254.png
    291 KB · Views: 9
Well... not so fast

Havoc (Dedicated) and Tempest use 2 differently configured opto boards. One uses a 150 ohm and the other uses a 100 ohm resistor. (I don't recall which ATM). You should validate what you have as a "spare". I can't recall if Tempest will work with either resistor configuration.

Your Havoc is running as a conversion correct? So that means you should be using a Tempest controller. This also requires the addition of a pull up resistor at the LETA or to GND the pin. The pull up resistor is on the LETA multiplier pin to make the controller more sensitive for Tempest games. In many cases Atari failed to stuff this resistor into the conversion boards. Sometimes the boards "worked" but many times the game sorta didn't and they actually got worse over time. On the MH OEM conversion boards you need the pull up or short the solder pad to GND.

See this.
Mine does have the extra resistor which is mentioned in the service tip.

Mine is a conversion, but I have (until this point) been running it with an original roller (and opto PCB).
 
1746825302394.jpeg

Finally back in the arcade. No more resets. No more wiggly vectors. The Roller is working.

In summary:
Random Resets -> Failing (but not failed) SEEQ ROMs.
Boot Instability -> Bad Solder Joint on a GAMMA CPU ROM
Squiggly Vectors -> Bad connectors on the WG6100
Sound Issues -> LS245 Bad Socket
Roller Input not working -> LETA + Opto PCB both bad. Replaced both.
Roller input too fast -> Ground the pad next to the LETA to disable R186 with dedicated Roller

Result -> Seems to be working
 
Dedicated MH is more reliable than a Converted MH. (laughs) :giggle:
Nice list though ... very useful reference. Unusual that so much, on a previously working board, failed at once. Though bad connections do get worse over time. Humm.
 
Dedicated MH is more reliable than a Converted MH. (laughs) :giggle:
Well if you have an extra I'm all for an upgrade to put next to my I Robot and Firefox :)
Nice list though ... very useful reference. Unusual that so much, on a previously working board, failed at once. Though bad connections do get worse over time. Humm.
I don't think it all failed at once.

I think the sequence of events went something like this:

Random Resets -> This is what prompted me to start investigating the board in the first place. It was getting worse over time. I thought it was sockets because the ROMs would read fine and self test passed, but it turned out to be the ROMs themselves.

However, in the process of figuring that out, I did replace all the sockets, which caused the boot instability problem later. (Bad solder joint)

Squiggly Vectors -> Bad connectors on the WG6100 - This was probably a result of me fucking with the cabinet too much. Shook something loose and then oxidation existing on the connectors didn't allow me to get it back into the golden spot again.

Sound Issues -> LS245 Bad Socket -> This probably happened while investigating the LETA issue.

Roller problems. Best guess these failed at the exact same time. Not sure what the mechanism of failure was, but they both died after I replaced the big blue. In retrospect, I think the LETA may have been on its way out for a while, because I noticed that the reproduction LETA, once installed, was picking up the higher resolution of the MH Roller. The original wasn't, with no other configuration changed. So I had to disable R186.
 
I can't say I've ever seen a LETA fail. Perhaps others could comment on their experience?
Though I would suspect that statistically, compared to POKEYs, there are fewer LETAs on boards in circulation.
And if one assumes the complexity of the die is similar in both customs. Humm.

REF:

Atari LETA custom chip found in Atari games. Reference Atari P/N: 137304-1001 & 137304-2002.

  • 720
  • APB
  • BadLands
  • Blasteroids
  • Champsionship Sprint
  • Crystal Castles
  • Off the Wall
  • Rampart (x2)
  • Major Havoc 12N
  • Marble Madness
  • Super Sprint

And, yes, mess with the game at your own risk. LOL
 
Back
Top Bottom