A4000D onboard mem reduced from 16MB to 4MB

[dirkies]

Got my A4000 recapped, was very well done, but now I noticed that the 4x4MB SIMMs on the mainboard are recognized as 4x1MB.

Tested also adding 1 by 1 the SIMMs, and it always went up with 1 MB only.

Jumper J852 is still set to 4x1M, instead of 4x256K (but I tried it also the other way around just to be sure, but no effect, always only 4MB recognized)

Anyone had the same or heard about this phenomenon? Seems this jumper has no effect anymore, it defaults always to 4x256K

Perhaps someone knows which trace to check, maybe it is shorted somewhere on the board causing it to think the jumper is set to max 4MB RAM (?)

[bebek]

It maybe caused by dmaged trackes due to battery corrosion or faulty F245. What happens if you use only one 4mb memory?

[dirkies]

1 simm reports 1mb, 2 simms 2mb etc

[K.C.Lee]

Since all 4 banks have same symptoms, so ruling out bad sockets and bad RAM. Could be that the upper address bit (MA9) is not getting to the SIMM or RAMSEY doesn't know it is 4MB.

A4000 Schematic page 13.

Possible cause:
- bad jumper you used for J852? (mechanical part, so it might not make good contact. Swap in a new one)
- trace broken from J852 to RAMSEY (U890)? (check with a multimeter if you have one)
- R881 (connected to RAMSEY U890) broken? Traces broken between there and the SIMM? If the Amiga is unstable or fail memory tests, that could be the case.
- That trace shorted to ground or +5V? (check with a multimeter if you have one)

[dirkies]

Cheers KC, will check that out! Maybe something went wrong during the recapping process?

[dirkies]

ok, so what I did so far:

1) other jumper > same result, but original jumper was checked and works fine
2) middle pin of J852 goes to Ramsey pin 12, it connects!
3) J852 pin 1 is +5V and pin 3 is GND, measured.
4) on each of the simm sockets, pin 32 reads around 33/34 ohm to Ramsey pin 83
5) tried jumper in 1MB mode, but same result as expected, 1MB out of 4MB recognized
6) tried 16MB SIMM I had laying around, but again only 1 MB recognized

So I am really at a loss wtf is wrong here!

[nogginthenog]

If it helps I have exactly the same problem :-)

I was never too bothered because I have lots (50Mb? lol!) of RAM on my CyberStorm Mk II
Just resurrected my A4000 last weekend.

Will have a play this weekend and report back if I find anything!

[dirkies]

Although i have 128mb on the cyberstorm aswell, it still bothers me to lose 12mb ram for no reason, esp when it worked for years without a hitch...

[dirkies]

seems I can kiss the 12MB goodbye then. To ease the pain, got myself a Bigram+ 256MB card, should keep memory worries away !

But isn't it better to remove the onboard memory completely and rely purely on Cyberstorm/z3 RAM ? As I suppose the onboard 16MB is a tad slower than the CPU/Z3 bound RAM?

[zipper]

RAM is used in the order of it's speed. ZIII is the slowest I think.

[dirkies]

Z3 slowest??? I doubt that, remember the FastlaneZ3 SCSI with RAM, was a trailblazer back then, how could it be slower than onboard mem?

[hooverphonique]

Z3 memory is slower than mobo-ram, yes.. the cyberstorm ram should be much faster than mobo ram..

[dirkies]

damn Buster chip then! hmmm, I wonder how it affects overall impression of the system the moment the Z3 mem is used (which I doubt it will be possible, 128MB is already overkill on Amiga unless you pump a lot of data into ram (sound recording, video editing etc)

[bdb]

Trust Bebek

[voyager_1701e]

Have you tried an 8mb simm in bank zero?

I've rectified memory problems with repairs immediately around U891, where battery acid / capacitor leakage has eaten tracks and I think even 1 via had been damaged on 1 board. There are some tracks under the chip which go to vias, so you would need to check the continuity on the entire upper row of the chip and also pin 1. Alternatively you need to remove the chip and do a close examination....but thats a bit more work than the continuity check!

Couple of photos of 1 repair with damaged tracks/pads on U891...there was so much gunk around the U891 area so I removed the simm sockets and replaced as well. But so far I haven't encountered any damaged tracks under any of the simm sockets of the 4k boards I've repaired...been lucky in that regard!

[hooverphonique]

if you can find a simple 'memtest' tool which allows you to manually specify the address range, you might be able to figure out, if it's a SIMM address line problem (you will get gaps in the fast memory region between 0x07000000-0x07FFFFFF). You cannot use a tool that uses the memory region configured by the OS, as gaps will cause it to stop searching.

remember to boot without running setpatch, so the MMU is inactive, otherwise your 'illegal' memory accesses might be blocked..

[dirkies]

so I need to figure out:
a) where to find such a memtest tool
b) inbetween which points I should test continuity between U891 pins and ... "wherever" (?)
c) how I can check that all these 74F245's are OK? Remember this board has been recapped, cleaned and health checked. This issue is that last one remaining.

[voyager_1701e]

Originally Posted by dirkies

so I need to figure out:
a) where to find such a memtest tool
b) inbetween which points I should test continuity between U891 pins and ... "wherever" (?)
c) how I can check that all these 74F245's are OK? Remember this board has been recapped, cleaned and health checked. This issue is that last one remaining.

From the A4000 schematic pins 2-9 of U891 - U894 (892-894 are above the simm socket, to the right of Bridget) go to the simm sockets (you will need to check the schematic for which to which). Pin 19 and 1 on all 4 chips are connected respectively, so easy to check. Pins 10 and 20 are for GND and 5v respectively again easy to check. Pins 11-18 seems to be connected to some HCT245's connected to the IDE connector, so I'm not sure if these pins would be relevant to your memory problem, but worth checking all the same.

I would imagine the only way to check the operation of the F245's onboard would be using a scope....

Was the board damaged anywhere by leaking battery or caps?

[bdb]

I've had this problem with a motherboard not grounded in the metal case; one screwed back in, all the memory was there (?). I didn't look a gift horse in the mouth.

[A4000Bear]

I had a similar problem with two A4000s several years ago. Like Voyager, I traced the fault to U891, though in both cases the chip itself was faulty, presumably caused by the battery chemicals getting inside the chip itself.

Incidentally, if the chip next to U891 is damaged, the Amiga sometimes thinks it only has an ECS chipset.