Amiga 1200 runing on 1 Memory SMD

[rjd324]

I need to debug some memory chips. Can the Amiga 1200 run with only 1 memory chip, if so - which location must the single chip be in (if any)?


Thank you.

[Krashan]

It can't. A1200 has either 256k×16 bit memory chips (so it needs at least two to run, with 1 MB chip RAM then) or 512k×8 (in this case four chips are needed).

[rjd324]

Hi Krashan, thanks for the response. I have no idea what the reason is for this. Tell me to RTFM for sure, but I am wondering if you could explain why it needs all the memory chips. Is there something in the ROM code that tests that all 2MB are there?

What is the best way to find a misbehaving memory chip? You need to replace all 4 of them (if you have 4) and then swap one out one by one? I know that you can use DiagRom to test, but the issue is an immediate green screen with 10 flashing LEDs.

Thanks.

[Jope]

The machine has a 32bit bus, hence there must be some RAM for each bit of the bus when it is being accessed.

[Daedalus]

DiagROM will get you some output even when Kickstart fails to start with a green screen and flashing LED. Basically, the 32-bit bus is spread across multiple chips, so as Jope says, you need at least enough chips present to cover the full width of the bus, otherwise you'll be missing e.g. two bytes from every 32-bit longword. If you don't have the full width, code or data can't possibly use it. DiagROM can still run, but it won't be able to set up a screen for output in this case. Instead, you can get the output via the serial port, and the output there will give you some clues as to where the error lies. Between the pattern of the failures DiagROM reports and the relevant A1200 schematics, you can decipher which chip might be faulty, with the proviso that something else might be wrong, such as a data line shorted to ground, damaged PCB track etc.

[Krashan]

Originally Posted by rjd324

I am wondering if you could explain why it needs all the memory chips. Is there something in the ROM code that tests that all 2MB are there?

68020 processor has 32-bit data bus. It means it can read/write 4 bytes at once. This bus physically is just 32 wires. Older A1200s (rev 1D) have four chips of 16-bit data bus (256k × 16 bits). Two of them are the first 1 MB, other two are second 1 MB. This is because Commodore planned to sell low budget A1200 with only 1 MB chip RAM, extendable to 2 MB with a module. So 1D mainboard can run with 2 chips out of 4. Why 2? Because half of the data bus is connected to one chip, other half to the second. In other words, one chip stores bytes with addresses 0, 1, 4, 5, 8, 9..., the second one stores bytes 2, 3, 6, 7, 10, 11... Any access of 32-bit size (and also some 16-bit accesses) reads from two memory chips simultaneously.

When rev 2B mainboard has been designed, it has been clear, that nobody wants cost reduced A1200 with 1 MB chip RAM. Then possibility to only mount 2 chips out of 4 was not needed. Therefore designers used 4 chips with 8-bit data bus (512k × 8 bits). I guess they were cheaper or easier to obtain. This means that one 32-bit access reads from 4 chips at once. Bytes are interleaved as follows:
chip A: 0, 4, 8, 12, 16...
chip B: 1, 5, 9, 13, 17...
chip C: 2, 6, 10, 14, 18...
chip D: 3, 7, 11, 15, 19...
Because of this in 2B A1200 all the four chips have to be present and working.

[nogginthenog]

Originally Posted by Daedalus

DiagROM will get you some output even when Kickstart fails to start with a green screen and flashing LED. Basically, the 32-bit bus is spread across multiple chips, so as Jope says, you need at least enough chips present to cover the full width of the bus, otherwise you'll be missing e.g. two bytes from every 32-bit longword.

I discovered something about DiagROM recently.
If you look at the LO / HI images with a hex editor you will see Chucky has helpfully added in some text:

LO: U6B,U176,U180   OL :6U,B1U67U,81 0
HI: U6A,U175,U181   IH :6U,A1U57U,81 1

U175/U176 is for an A4000, not sure about the others.

[rjd324]

Thanks guys

[rjd324]

So what I am seeing is. 3.1 kickstart is green screen. 10 flash led on keyboard. Restart loop. Does same again. If I use diagrom then I get red/green screen and nothing.

Anyone can point me in the right direction as to what to check?

[Vypr]

Originally Posted by rjd324

So what I am seeing is. 3.1 kickstart is green screen. 10 flash led on keyboard. Restart loop. Does same again. If I use diagrom then I get red/green screen and nothing.

Anyone can point me in the right direction as to what to check?

Red/green flashes on the screen indicate problems accessing RAM, if you connect the serial port to your PL you should get text output of the RAM test

[rjd324]

Amiga DiagROM V1.2 - 13-Aug-19  - By John (Chucky/The Gang) Hertell

'????   ????????Code $ff - Start of ROM, CPU Seems somewhat alive
    Checking status of mousebuttons for different startups:

    Set all Interrupt enablebits (INTENA $dff09a) to Disabled: Done
    Set all Interrupt requestbits (INTREQ $dff09c) to Disabled: Done
    Set all DMA enablebits (DMACON $dff096) to Disabled: Done

Testing if OVL is working: OK
- Parallel Code $fe - Test UDS/LDS line
  - Test of writing word $FFFF to $400 OK
  - Test of writing word $00FF to $400 FAILED
  - Test of writing word $FF00 to $400 OK
  - Test of writing word $0000 to $400 FAILED
  - Test of writing byte (even) $ff to $400 OK
  - Test of writing byte (odd) $ff to $401 OK

Followed by all the reading/writing failures (as in.. all of them fail) and then:

NO Chipmem detected
- Parallel Code $fc - Trying to find some fastmem (as no chipmem found)
- Parallel Code $fb - Memorydetection done
- Parallel Code $80 - NO Chipmem detected
- NO MEMORY FOUND - HALTING SYSTEM

This could still be a lot of things? A bad trace, bad memory chip, bad ROM socket?

[rjd324]

So, I am just wondering what to check here. I guess I should check all traces that end up at the memory chips? As in, start from the memory chip pin and work my way back to make sure I have continuity using PCBExplorer to help.

Would that be a start?

[cpiac64]

verify all connections ram to ram
i repaired a board with some trace corroded from leakage

[rjd324]

Hi all.

I have used PCB explorer for the 2B motherboard and all the traces seem to be going from each 4 memory chips to their destinations with continuity.

To recap, I am seeing a green screen with 10 power LED flashes and then a reset: loop.

I used a logic probe and the only anomaly that I see is that DRD28, unlike all other DRD that read "1" until reset where they go "0"/"PULSE" and repeat, shows nothing on "0" and nothing on "1" but will "PULSE" at the same time as all others.

I have no knowledge, but perhaps this seems suspicious?

Regards.

[rjd324]

Update.
I also have another 2B motherboard. I used the probe on the "good" motherboard's DRD28 before switching on the power. Once the power is switched on the signal is immediately "1" whereas on the "broken" motherboard there is no LED on the logic probe for "1" or "0".

At this point I am thinking that U16 memory chip is either bad or one of the chips on that path (Alice, Paula, Lisa).

--

I just wanted someone else's opinion before perhaps trying to swap out the U16 with a different chip from the other board.

[cpiac64]

test it with Logica diagnostic

[rjd324]

Okay, got to the bottom of this one.

Things I will never forget: if you get a problem like this, it is always good to press down on certain chips to see if you get a result. In this case, pressing down on Budgie fixed things. I reflowed the left side of budgie since that was data line 28 and, sure enough, it works.

Seems to always be a budgie issue for me with 1200s.

There is probably reasoning for this happening to me though. I had this board in a towered amiga with a mediator + a blizzard 1260 with a massive heatsink. It means that the blizzard is resting against the tower and at a slight angle (I tried to mitigate the issue by using slightly longer standoff screws on the side of the mboard on the expansion slot side). On top of that, I was inserting and removing cards often from the mediator board. Since the budgie is in the middle of the board then I imagine it takes a lot of the stress.

Anyway, thanks for all the help. I could not be bothered to get the oscilli out, but the logic probe gave some good hints as to where the problem lay.

PS:

Another thing I learned. Just because you have continuity between two end points, it don't mean everything is a-okay because by pressing down on a leg of a flat-package type IC with the needle of the multimeter you can make a contact which shows that there is continuity!

Another saved Amiga 2B motherboard, recapped and timing fixes applied.

[retrofriends]

I have posted a video about this on YouTube. I encountered one of these V1 motherboards, and was able to repair it. The video goes into detail about finding the fault. BTW kudos to Krashan for his post above that explained how an A1200 can show only 1MB.

Video here:

[ Show youtube player ]