A4000 repair troubles only 1M chipmem

[Daedalus]

The CPU slot? You should be definitely getting lots of activity there! Very strange... Do you have the scope grounded with the ground of the A4000?

[glenn]

Yes, in one of the PCB mountholes..

I'm not really friend with this scope (hardly ever use it), but I do get activity on other places, ECLK and TICK for example.

[glenn]

I realised that a scope might be overkill, I have an old logic probe, so I decided to use that instead, and now I get activity, however, I see (or hear) no difference when I pass the 1M linit, so it looks like it never flip the A20 bit ?

What does that tell me ? broken Alice ??

[glenn]

Just to be sure I connected another motherboard with working chipRAM and did the same test, it sure was difference when it passed 1M..

But why does it never generate it ?

[Daedalus]

Hmmm, that's the tricky bit You are manually testing the memory above 2MB, yes? It could be damage to a chip somewhere, or a short on the address line somewhere, or a bad pull-up resistor somewhere.

If it's an automated test however, it won't generate on that line because it thinks there's nothing there (though it would still be used for other hardware further up the address space...)

[glenn]

Yes O have both ran test without shadow detection on 0-2M and a manual test 1-2M.

So, it never flip the A20 bit.. what is responsible for that ? Alice ?

[cpiac64]

verify all 33 ohm near (under) U216 and U212

32 data are ok, you need verify A0-A9 and ras/cas

[glenn]

All 33ohms are ok near U216 and U212, just checked again.

Also, A0-A9 are clear all the way from Alice to Ramsey and fastslot, also just checked.

Well RAS/CAS, I measured that many times now.. :/

[cpiac64]

very strange, the chip module is double side?

[glenn]

Yes, and the same module works on other A4000 motherboards.

This time I even measured directly on the 030-card, if there was any troubles in the connector, but no.

[Daedalus]

Yep, it's a very strange one alright. The A20 line will be driven by the CPU for RAM access, though it doesn't drive the RAM directly. I don't know the A4000 all that well, so maybe someone with more experience like Chucky would be better placed to go into the finer details. To me though it looks like Alice and Gary are both involved, working together via U212 to generate the RAS and CAS signals from the address bus. If neither Alice nor Gary are getting an A20 signal, there has to be something more fundamental wrong.

[Kin Hell]

The underlying problem will probably be poor tacks/damaged components in/on the mobo, as a result of prior battery leakage.

Depending on the extent of the original leakage & how severe it was, it's normal for Amiga 4000's to suffer Clock, memory & sound issues at some point of their life-span. Pretty much every component on the Battery side of the Mobo is at risk from acid leakage, and on to the CPU side if stored on their sides. For this though, the Clock side of the board usually stays free from Acid attack, but the ZORRO interface will spread the Acid leak all over the place.

You'll probably need to send the board away for a competent repair guy to fix.

[cpiac64]

use diagnostic from "Logica", you can find in "Amiga forever"

[Chucky]

Logica doesn't test memoryareas it cannot detect.. so it will only test 1MB.
my DiagROM you can set to test any memoryareas.

[glenn]

Originally Posted by Daedalus

Yep, it's a very strange one alright. The A20 line will be driven by the CPU for RAM access, though it doesn't drive the RAM directly. I don't know the A4000 all that well, so maybe someone with more experience like Chucky would be better placed to go into the finer details. To me though it looks like Alice and Gary are both involved, working together via U212 to generate the RAS and CAS signals from the address bus. If neither Alice nor Gary are getting an A20 signal, there has to be something more fundamental wrong.

Yes, it is strange, it is like it never even try to set it high. ..Hmm, maybe if something is holding the A20 down, shorting it to ground or something.. Iäll take a look and see if I can find anything. ..On the other hand, then the fastmem shouldnt work either ?

Originally Posted by Kin Hell

The underlying problem will probably be poor tacks/damaged components in/on the mobo, as a result of prior battery leakage.

Yes, thats the most common fault, that and damages from leaking capacitors.. however, nothing near the chipram seem to be damaged here, and I measured everything.

Depending on the extent of the original leakage & how severe it was, it's normal for Amiga 4000's to suffer Clock, memory & sound issues at some point of their life-span. Pretty much every component on the Battery side of the Mobo is at risk from acid leakage,

Ys, RTC and sound is very common, even fastram, but not chipmem usally.

You'll probably need to send the board away for a competent repair guy to fix.

Will not happen, especially not after a friend sent a board to a well known amigacompany and got it back a year later in worse condition, more broken and they wanted money for it too..

I'm fully capable of doing this kind of repairs myself, I just never had any problem exactly like this before, but i learned a lot about both DRAM and how chipmem is detected now

Originally Posted by cpiac64

use diagnostic from "Logica", you can find in "Amiga forever"

I dont think the logica ROM is better than Chuckys DiagROM at any point nowadays ?

[cpiac64]

i use both

[Chucky]

Just curious.. what is the reason to use Dialogica today? so I can change that

[Kin Hell]

Ah well. Best of Luck.

[TTD]

Well no expert on electronics just what I understand on the schematic but I give it a try could it be fat gary problem, what are values on r77 and r78 should point to siz0 and siz1 is it 2.7k ohm?
Values on r171 and r172 is these 1K ohm?

[Chucky]

siz0 and siz1 have nothing to do with memorysize.. it is a way of telling the cpu (or bus) if it contains a byte, word och longword sized chunk of memory