Defender Ram Failure Help

QuartrArcade

Active member

Donor 2011
Joined
Dec 16, 2005
Messages
1,110
Reaction score
15
Location
Pennsylvania
Working through some Defender CPU boards, have had decent success getting most running again, will post the repair logs later. I'm stuck on this one, board boots to ram failure error. New 4116 rams installed, voltages check out, decoder roms good, good rom board, etc.

When I run self test to test rams, it starts the test, briefly flashes Ram Failure again and then continues the rug pattern test, but doesn't report any bad ram.

I've checked the 74153s, 74ls374s and 74ls165s, as far as I can tell they are working. I have noticed that when it gets done testing, the WE line (WE3, I think) for the row of rams on bank 3 is stuck low (maybe high, I forgot, but not pulsing), while the other WE lines are pulsing. It does pulse while it is testing rams though.

Thoughts on where to look next?

Thanks in advance,
Anthony
 
sounds like the clock isn't working
or the crystal I forget what its called ya know sounds like a total labotomy
 
Possibly the decoder Proms ?

depends on the error code I guess. I wanna say I have a Defender that gives a ram error but doesn't explicitly say which one. or several. unfortunately Defender hardware isn't so cut and dry with the ram circuit. the drawing set online is a horrible scan so I can't tell, and my hard copy is at work.

the decoders are important however. and their sockets are crapping out left and right now in my experience. I would pull them and make sure the legs aren't corroded, cause that happens too. I use a wire brush on the dremel on low speed for cleaning that up. then I pull the plastic housing off the socket and bend the legs tight on the decoder and place it directly on the pins.

the decoder 2 will probably be the one to focus on, as that more directly relates to the ram circuit. you can check decoder 3 too though.

I wanna say I have a board that the write enables are stuck high on at least bank 3 or all of them. it's been a few months. I'll get around to looking at that again someday.
 
It's hard to tell from the information however the 7474s on the Williams boards are pushed pretty hard. A logic comparator will help when trying to search down these types of problems.
 
It's hard to tell from the information however the 7474s on the Williams boards are pushed pretty hard. A logic comparator will help when trying to search down these types of problems.

totally this, there's a couple that were factory socketed. the one nearest to the crystal I think actually drives it, because I had a board that was totally dead and I had to install a new one in.

I had a Defender tonight with ram errors and it would reset. just be persistent and keep going back into test mode and running the ram test with the middle switch in the Auto Up (not pushed down) position and it should eventually give you the error.

if you have a logic probe, test the 374 buffers below. the usual corner pins for +5 and ground should be high/low respectively and the rest should pulse. if you find a pin out of whack, dig out the drawing set and follow it on the schematic (it'll be on the back page for the ram section)

check out the 74153 multiplexers... it's been so long since I've found bad ones I don't remember their symptom, I think it should infinitely rug like bad shift registers would, but it might generate ram errors. pins 7 and 9 should pulse on all except the upper right, pin 9 should show low there.

what complicates your situation is you have all new ram, whereas I had so much bad ram the error code was illegible -- I had to replace all of bank 3 with known good rams, then chipped away at the remaining 7 that were bad. this incidentally was the one dead board that needed new 7474s.

pin 3 at the rams should also pulse, that's the write enable pin. bad rams I guess can break that I found, I didn't have any overly complicated fixes for once.

this was also a very dirty board, I hit each bad ram socket with contact cleaner before dropping the replacements in. the plastic housings pop off, if your board was in a bad environment and you have rust or corrosion, that becomes a nightmare for the single wipe sockets as the pins usually break off like chopping down a tree.
 
Thanks for input, I set the board aside, but will get back to soon and dig deeper.
 
One of the problems with Defender, other than the LEDs for error codes, is that the test routine uses bank 3 to store return information. When bank 3 has a problem, and the memory test goes bank 3, 1, 2 then the program gets lost, the watchdog barks and the game resets.

What does the rug look like? Are there any stuck vertical lines? There is a 74139 which selects the bank, be sure that it is pulsing properly. You might want to cut the watchdog pad so that the game doesn't reset.

It's been a few years since I've worked on one. 7410 comes to mind for bank enables. Check them and look for damage on the board. All pins on the data buffers (LS374) should be high/low or pulsing. No floating pins.
 
One of the problems with Defender, other than the LEDs for error codes, is that the test routine uses bank 3 to store return information. When bank 3 has a problem, and the memory test goes bank 3, 1, 2 then the program gets lost, the watchdog barks and the game resets.

What does the rug look like? Are there any stuck vertical lines? There is a 74139 which selects the bank, be sure that it is pulsing properly. You might want to cut the watchdog pad so that the game doesn't reset.

It's been a few years since I've worked on one. 7410 comes to mind for bank enables. Check them and look for damage on the board. All pins on the data buffers (LS374) should be high/low or pulsing. No floating pins.

yeah, I mean when I turned that on and I couldn't even read the text, I didn't even bother trying to run test mode, I just replaced everything in bank 3. I'm totally lost on the order in which Defender tests banks though... I thought it went 1-2-3, but that one starting throwing errors in the 2's first, then migrated its way to ram 16.
 
Ram Bank Testing Order

As far as I know, all of the early Williams stuff goes bank 3-1-2. The later ones do something very odd on bank 2 at the end. Seems to report 8 before 7. Populate a Stargate/Joust/Robotron RAMs a few at a time and follow the errors...
 
Last edited:
Back
Top Bottom