Williams Defender Arcade Prototype

Yes it is a gold contact rom. i am not shure my fix worked. there is no difference on the screen. mostly this colourful dots all over, see picture. and sometimes really big blue pattern, about 6 inch wide
the rug pattern ran for about 12 cycles in MAME. lot longer than red label
 
nice find. I was in Germany in the early 80's and I played Defender almost every day. Couldn't get onto Asteroids because the older kid were always on it. So I got really good on Defender.

As for your problem:
Your last picture shows the normal test patter (rug test). On the green set (1st ver) it would sweep about 7 or 8 times (can't remember the exact times). I'm guessing the white ver is the same. If it doesn't get out of doing the sweep test, reach through the coin door and watch the screen as you press on the ribbon cable for any changes. It looks like the original ribbon cable may need to be replaced.

Another thing to check, if ROM #7 and/or #10 are bad it will just be stuck in an endless rug test pattern. Try reseating those ROM chips. I can't see the notch on ROM #10 in your picture, is it in correctly? The notches should all be pointing to the left in your picture (to the power connector side of the board). Don't go by the stickers.
The notches are correct, i have reseated them and i repleaced the ribbon cable. that made no difference. but good to know, these are the correct rug pattern. So he is stuck in endless rug pattern.
 
The notches are correct, i have reseated them and i repleaced the ribbon cable. that made no difference. but good to know, these are the correct rug pattern. So he is stuck in endless rug pattern.
read this https://www.arcaderepair.net/williams-rom-board-information/

I would remove roms 7 and 10, they shouldn't be gold contact roms, but make sure the legs are all intact and that they're straight. never remove chips from one side, it will leave bends in the legs on the opposite side. I always alternate between sides until the chip is fully out.

also @rhombus is apparently educated in the art of Defender ;)
 
The next thing I would do is make sure that ROM #7 and #10 are good. Just by reading the chips doesn't guarantee that they are. You need to get a good know chip to replace them. Maybe a little hard since it's the white label? (check with https://www.hobbyroms.com/) Also another thing that could cause the same symptom is a bad PIA chip (6821) on the ROM board. If your soldering skills are up for it? That maybe the best option for you at this point.
 
What is the Rug Pattern?
With this hardware architecture, anything that is in the RAM gets 'written' out to the display (independently from the CPU - sharing alternate clock cycles). The 'Rug' pattern is the visual representation of the RAM test. After the RAM, ROM tests complete, the game code(application) is loaded and controls what's in the RAM.

Watchdog?
There is another independent circuit that will trigger a complete board/CPU reset if it isn't satisfied withing a certain time period. The term Watchdog is a great analogy. It 'Watches' the hardware and software and will 'Bark' if not 'Petted' every so often.

For this architecture, there is a counter that starts at zero and begins counting up with each vertical blanking pulse. The refresh rate is 60Hz, so you get a pulse every ~16.7 milliseconds. Once the counter reaches 8 (~133 milliseconds), the reset is triggered and the whole boot process starts over (Looping).

To prevent this reset from happening, the programmers had to add code that would write out a specific value to reset the counter back to zero before it reached 8. This ensures the code is running as intended and isn't 'hung' or locked up. There are also additional hardware monitoring that the watchdog circuit monitors, but they all have to be present and working for this software reset to get through.

Rug/Boot Looping?
When we power on any game, the CPU is hard coded to look at a specific memory address for it's first instruction. This is called the 'Reset Vector' (Boot Sector). Williams put the RAM/ROM test program in this location and is typically on ROM3/12 (closest to the Ribbon cable). Once the tests are complete, the system attempts to load the game code. This code is spread across the rest of the ROMs and if any problem prevents or delays the application from running, the watchdog will trigger causing this looping.

Take it to the BANK...
The CPU has a limited number of pins which in turn limits the number of addresses it can communicate with. To expand this range, the engineers got creative and created a method to Re-Use addresses for different devices called Bank Switching. If the system in on Bank 0, it can 'talk' to the full address range with whatever hardware is in that section. The code can then switch over to Bank 1 and communicate with all the devices there. This comes with some limitations such as trying to move data between the banks and the time to switch back and forth. Williams architecture put all of the DRAM and a few of the ROMS on Bank 0. This space includes the Reset Vector, so the game can boot and run the test programs (Rug Pattern) without having to leave Bank 0. When the test completes and tries to load the game code (mostly in Bank 1). if the Bank switching has an issue and can't move to Bank 1, the ROMS in that space can't be read (watchdog will trigger).

~Brad
 
Last edited:
Ok, so many information. Thank you all very much.
So, i wrote an email to Dave Langley in uk, if he can help fix my boards. On the other hand i contacted hobbyroms if he can make me rom 7 and 10. tomorrow i will have closer look to the PIA 6821, if i will be able to change this.
 
Ok, so many information. Thank you all very much.
So, i wrote an email to Dave Langley in uk, if he can help fix my boards. On the other hand i contacted hobbyroms if he can make me rom 7 and 10. tomorrow i will have closer look to the PIA 6821, if i will be able to change this.
oh yes, if the rom board PIA is bad you'll get random blinking LEDs. does the game play sound when you turn it on? if it doesn't that's another sign of rom board PIA failure.

it's been a very long time but I think I've had Defenders in the past that wouldn't boot if the PIA was that bad. you have an AMI part, in the pinball collective those are apparently the ones that fail the most. :)

What is the Rug Pattern?
With this hardware architecture, anything that is in the RAM gets 'written' out to the display (independently from the CPU - sharing alternate clock cycles). The 'Rug' pattern is the visual representation of the RAM test. After the RAM, ROM tests complete, the game code(application) is loaded and controls what's in the RAM.

Watchdog?
There is another independent circuit that will trigger a complete board/CPU reset if it isn't satisfied withing a certain time period. The term Watchdog is a great analogy. It 'Watches' the hardware and software and will 'Bark' if not 'Petted' every so often.

For this architecture, there is a counter that starts at zero and begins counting up with each vertical blanking pulse. The refresh rate is 60Hz, so you get a pulse every ~16.7 milliseconds. Once the counter reaches 8 (~133 milliseconds), the reset is triggered and the whole boot process starts over (Looping).

To prevent this reset from happening, the programmers had to add code that would write out a specific value to reset the counter back to zero before it reached 8. This ensures the code is running as intended and isn't 'hung' or locked up. There are also additional hardware monitoring that the watchdog circuit monitors, but they all have to be present and working for this software reset to get through.

Rug/Boot Looping?
When we power on any game, the CPU is hard coded to look at a specific memory address for it's first instruction. This is called the 'Reset Vector' (Boot Sector). Williams put the RAM/ROM test program in this location and is typically on ROM3/12 (closest to the Ribbon cable). Once the tests are complete, the system attempts to load the game code. This code is spread across the rest of the ROMs and if any problem prevents or delays the application from running, the watchdog will trigger causing this looping.

Take it to the BANK...
The CPU has a limited number of pins which in turn limits the number of addresses it can communicate with. To expand this range, the engineers got creative and created a method to Re-Use addresses for different devices called Bank Switching. If the system in on Bank 0, it can 'talk' to the full address range with whatever hardware is in that section. The code can then switch over to Bank 1 and communicate with all the devices there. This comes with some limitations such as trying to move data between the banks and the time to switch back and forth. Williams architecture put all of the DRAM and a few of the ROMS on Bank 0. This space includes the Reset Vector, so the game can boot and run the test programs (Rug Pattern) without having to leave Bank 0. When the test completes and tries to load the game code (mostly in Bank 1). if the Bank switching has an issue and can't move to Bank 1, the ROMS in that space can't be read (watchdog will trigger).

~Brad
bravo sir.
 
Since i restored the reset circuit i have a sound when turning on
 
Just for info, my boards are going to england for repair. I will keep you informed.
 
Quick update, the ROM board has been fixed and it just needed 3 new EPROMs (3, 6 and 11) but I haven't had enough time to dedicate the CPU board, it starts doing the rug test and gets stuck about 1/8 way across the screen. There is bad data on VA11 from the video address generator which generates Count 240 and 4msIRQ but it isn't the 9316. Thats as far as I got so far!

I did also dump the Decoder ROM and this is a previously unknown one where many of the bytes are reversed.
 

Attachments

  • Screenshot 2025-11-04 at 9.44.42 pm.png
    Screenshot 2025-11-04 at 9.44.42 pm.png
    4 MB · Views: 32
This would be (at least) the 7th "pilot" Defender found, with three of those being in the UK. Not sure if the one that is/was in Utah is included in this list, so that could be an 8th.

Wonder how many are out there...

Here are the known pilot serial #'s from those found:
  • 438691 [438692]
  • 438733
  • 438762 (mine)
  • 438776
  • 438830 (this one)
  • 439641
  • 439642

Here are some interesting deviations from production Defenders:
  1. art package, of course
  2. bezel differs
  3. "TM" marks missing from art
  4. Molex KK connectors instead of IDC
  5. different vents (four on top of cab vs top of rear)
  6. coin door cut-in lower in the panel (for art)
  7. leaf switches, risers, and CP hold-down brackets differ
 
This would be (at least) the 7th "pilot" Defender found, with three of those being in the UK. Not sure if the one that is/was in Utah is included in this list, so that could be an 8th.

Wonder how many are out there...

Here are the known pilot serial #'s from those found:
  • 438691 [438692]
  • 438733
  • 438762 (mine)
  • 438776
  • 438830 (this one)
  • 439641
  • 439642

Here are some interesting deviations from production Defenders:
  1. art package, of course
  2. bezel differs
  3. "TM" marks missing from art
  4. Molex KK connectors instead of IDC
  5. different vents (four on top of cab vs top of rear)
  6. coin door cut-in lower in the panel (for art)
  7. leaf switches, risers, and CP hold-down brackets differ
I've been working on Williams early solid state games and was like "ok they replaced all the plugs", but that's just how they were made then. so molex came first and then around Defender time they switched to IDC.
 
Awesome find. The side art with the tie fighters is wild. Oh yeah, I can't believe it hasn't been mentioned, but hopefully the batteries and the battery holder have been tossed by now
 
Awesome find. The side art with the tie fighters is wild. Oh yeah, I can't believe it hasn't been mentioned, but hopefully the batteries and the battery holder have been tossed by now
this really is an exciting project I hope to see get running

This would be (at least) the 7th "pilot" Defender found, with three of those being in the UK. Not sure if the one that is/was in Utah is included in this list, so that could be an 8th.

Wonder how many are out there...

Here are the known pilot serial #'s from those found:
  • 438691 [438692]
  • 438733
  • 438762 (mine)
  • 438776
  • 438830 (this one)
  • 439641
  • 439642

Here are some interesting deviations from production Defenders:
  1. art package, of course
  2. bezel differs
  3. "TM" marks missing from art
  4. Molex KK connectors instead of IDC
  5. different vents (four on top of cab vs top of rear)
  6. coin door cut-in lower in the panel (for art)
  7. leaf switches, risers, and CP hold-down brackets differ
this is rad
 
Awesome find. The side art with the tie fighters is wild. Oh yeah, I can't believe it hasn't been mentioned, but hopefully the batteries and the battery holder have been tossed by now
There were no batteries when it arrived here for repair but there was some significant corrosion on the battery holder terminals, but corrosion had not progressed to the board.

The holder has been replaced with a nice clean vintage one.
 
Here are the known pilot serial #'s from those found:
  • 438691 [438692]
  • 438733
  • 438762 (mine)
  • 438776
  • 438830 (this one)
  • 439641
  • 439642
I also have just the interface board from 438758. Both this one and the one from the set I am working on have the 4-pin molex on the interface populated which I have never seen fitted elsewhere.

1762940378993.png
 
This would be (at least) the 7th "pilot" Defender found, with three of those being in the UK. Not sure if the one that is/was in Utah is included in this list, so that could be an 8th.

Wonder how many are out there...

Here are the known pilot serial #'s from those found:
  • 438691 [438692]
  • 438733
  • 438762 (mine)
  • 438776
  • 438830 (this one)
  • 439641
  • 439642

Here are some interesting deviations from production Defenders:
  1. art package, of course
  2. bezel differs
  3. "TM" marks missing from art
  4. Molex KK connectors instead of IDC
  5. different vents (four on top of cab vs top of rear)
  6. coin door cut-in lower in the panel (for art)
  7. leaf switches, risers, and CP hold-down brackets differ
One other deviation is, mine has no legs (levellers) and nothing to mount them

Mine has only two vents on the top and one strange closed hole or something
 

Attachments

  • cab1.jpg
    cab1.jpg
    213.8 KB · Views: 13
  • cab2.jpg
    cab2.jpg
    283 KB · Views: 13
Back
Top Bottom