• Please review our updated Terms and Rules here

Very Early 5150 Issues (No Post, RAM bad)

Lutiana

Lutiana

Administrator
Staff member
Joined
Mar 26, 2009
Messages
3,301
Location
Dublin, CA USA
So I am trying to get this 5150 to work, it is a VERY early model, with the '81 BIOS. It acts like it is dead, but as most of us know, this usually just means memory has some issues. So I threw in the Landmark ROM and booted it.

I get this:

Code: 
Memory Refresh    FAILED

Memory Error at address 000000
       7 .  6 .  [COLOR=#ff0000]5[/COLOR] . 4 . 3 . 2 . 1 . 0
                 X
Parity Error at 08000

Seems pretty straight forward, bad RAM. But I am not sure which chip this is indicating, the manual for the landmark rom is a little bit hard to follow.

So anyone able to interpret this for me? And where on earth do I get RAM chips for this beast?
 
How much ram is soldered to the Mobo? How much is socketed? How much on expansion cards, if any?

Iirc they can have a Max of 64k on the earliest boards, 256k on later boards.

08000 in hex is 32768 or 32 * 1024 (32k). That information is highly ambiguous, unless you have memory problems at 000000? and 08000. I would just install sockets where you don't have them and
sequentially pop and replace.
 
Last edited:
2icebitn said:
How much ram is soldered to the Mobo? How much is socketed? How much on expansion cards, if any?

Iirc they can have a Max of 64k on the earliest boards, 256k on later boards.

08000 in hex is 32768 or 32 * 1024 (32k). That information is highly ambiguous, unless you have memory problems at 000000? and 08000. I would just install sockets where you don't have them and
sequentially pop and replace.
Click to expand...

No add-in cards, all 64Kb onboard in 4 banks with bank 0 soldered on.
 
I hope you didn't read that, I read your reply wrong.

Forging ahead, try piggybacking chips on to the soldered ones and see if that changes anything. The IBM System 23/Datamaster came stock with redundant piggybacked chips.

Also you will have to reconfigure your board for the amount of ram you have installed, obviously, as you're going to have to eliminate a bank unless you have spares. I have a tube and a half of some kind of ram chips, not sure they're 4116s though.
 
2icebitn said:
That's good news. Try swapping chips in the lowest banks and see if anything changes. For starters. You have enough chips there to replace whole banks and reconfigure the board to recognize 256k or however low you can go.
Click to expand...

Hmm, this is a 16Kb - 64Kb board. So I am not sure where 256K comes into play. I just ordered 9 new DRAM chips and 9 sockets, so I think I'll just remove Bank 0, add in the sockets and put in the new chips and see how it goes. From there I'll swap in the old chips till I can ID the one that has failed (be it in Bank 0 or one of the other banks).
 
What's the serial #? A place I worked at in the late 80s had 1 of the first 500. And I was complicit in upgrading the Mobo to a Bullet-286. I'm sure the statute of limitations has expired.
 
2icebitn said:
What's the serial #?
Click to expand...

0127962 - No idea where that puts it on the spectrum of age, but it has no "B" on the back, the PSU is black and has a yellow sticker on it.

I have about a hundred 64K DRAM chips, but this machine would appear to only take the 16K ones, so I'll wait till I get them. Then I might do the piggy back thing till I work out which chip is bad.

From what I read on minuszerodegrees these boards sort of have to have 64Kb of RAM to function, but also the landmark ROM does not really care about the switches when testing. So I am guessing I am stuck with making all 64Kb work.
 
lutiana said:
From what I read on minuszerodegrees these boards sort of have to have 64Kb of RAM to function, but also the landmark ROM does not really care about the switches when testing. So I am guessing I am stuck with making all 64Kb work.
Click to expand...

So according to the bios it doesn't pass the piss test and won't boot w/o that whopping 64k. Who needs that much ram? :)
 
Quick side note. Does the piggyback testing of RAM chips actually work? Sidetracking the thread a bit, but I have a similar problem with a 16-bit ISA memory card which has one bad chip which I'm trying to locate. The chips are soldered on this board, so I'd rather not have to desolder dozens of chips to find the bad one. I actually did a bit of testing with the piggyback method in the most probable locations and a large number of others, but couldn't find any location where piggybacking made a difference. I'm talking about a card that has 5MB of soldered RAM in 256kx1 chips, so a total of 180 chips including the parity ones.
 
You must log in or register to reply here.
Back
Top