Defender ram failure 32

andyman

always pull away from your face
Credits
1,059CR
Someone help me with this, its doing my nut in
smiley5.gif


Unmodified later series CPU, red romset & rebuilt linear

Power up, you get the Defender sound, double rug, ram failure. Into ram test, reports ram failure 32. I have so far, checked all voltages and all are within 0.2 of specified, replaced with good known working 4116, replaced the socket (all 24), checked for a lifted trace or pad and all is ok, no shorts.

I can remove any 4116 in bank 1 or 2 and it is correctly reported. I remove any or all in bank 3 and it reports 32 still. I have checked the addressing/multiplexing logic and everything *looks* to be ok.

Anyone have anything specific ?

TIA
 

tb lilley

5 Star
vacBacker
Feedback
2 (100%)
Credits
1,686CR
Something wrong there, and probably not the RAM .... There's mention on Bob Roberts site of the decoder ROMs sometimes being responsible for erroneous RAM reports despite otherwise working in general... Long shot but 5 mins to swap out if you have spares.
 

andyman

always pull away from your face
Credits
1,059CR
I did half come to that conclusion but wanted to see if anyone else mentioned it. No spares and I don't really want to mod it for 2716's, looks like I'll have to order some
smiley6.gif
 

cliff_poole

Active member
vacBacker
Feedback
6 (100%)
Credits
1,069CR
Have the batteries ever leaked ?. Those sockets are the closest to the battery holder and are usually the first to go green. Checking the RAM chip with a logic probe or scope will tell you if you have good contact of not.

Could also be a problem with the associated 74153s or 74165s.
 

andyman

always pull away from your face
Credits
1,059CR
No leaks evident, have changed both sockets on those 7474's closest to the batteries as they are usually crap anyway, also replaced the 165's that control bank 3 and the 153 probes ok, ram chip checked also, all appropriate lines pulsing as others in bank.

I thought it was a random ram failure originally until I forgot to replace another one in bank 3 and it gave the same error, has to be ras/cas issue really doesn't it but is that prom related or not
 

cliff_poole

Active member
vacBacker
Feedback
6 (100%)
Credits
1,069CR
You were already way ahead of me there then. Could be PROMs, I once spent ages on a Stargate set with a 1-3-1 error that turned out to be PROMs. Luckily the mod doesn't require any cutting of tracks so it can be put back to normal without any trace, has to be worth a go to rule it out ?.

I had no luck getting the mod to work with 2716s, 2732s worked first time though.
 

andyman

always pull away from your face
Credits
1,059CR
A 31 error would have instantly lead me to the proms, after sockets and ram change, but 32 I've never had. Suppose I could try the 27XX's
 

cliff_poole

Active member
vacBacker
Feedback
6 (100%)
Credits
1,069CR
I have a couple of spare, later series, CPU boards if you are interested ?. Not in brilliant condition but they are both fully working (tested a couple of months ago).

Can anyone chip in with a valuation ?.

Both have PROMs rather than EPROMs and 4116 RAMs. One has suffered battery leakage but has been cleaned up and repaired. They have also both had their AA battery holders replaced with CR2032 sockets.
 

chadsarcade

Lord of Vectors (tm)
Feedback
11 (100%)
Credits
443CR
andyman said:
Into ram test, reports ram failure 32.

Anyone have anything specific ?

I've seen this a couple of times before, and the fix was to replace the 7432 in the RAM control circuitry. I can't recall the chip location offhand, but the schems should make it clear enough.

I'll dig out the schematics and try to pinpoint it if you get stuck.
 

chadsarcade

Lord of Vectors (tm)
Feedback
11 (100%)
Credits
443CR
Mesmeric_G said:
Could also be a problem with the associated 74153s or 74165s.

Failure of 153's is usually visually apparent as the rug doesn't sweep cleanly across the screen.

The 165's just blindly output the content of RAM to screen and I don't think they can produce a RAM failure even if they fail - just the screen output would be affected.
 

andyman

always pull away from your face
Credits
1,059CR
chadsarcade said:
I've seen this a couple of times before, and the fix was to replace the 7432 in the RAM control circuitry. I can't recall the chip location offhand, but the schems should make it clear enough.

And it was again this time, cheers for that
 
  • Like
Reactions: trm
Top