Stuff and Things.

[rubber_jonnie]

So, my non U1MB 800XL has been running this all morning:

IMG20260108115433.jpg

In fact I've been running it off and on for a few days now on account of the RAM fail seen in the previous post and some Fujinet instability.

Today it finally got me a result :)

IMG20260108115444.jpg

So it's going to go under the knife and have U12 replaced, possibly even as soon as tonight :)

[stephen_usher]

At least it found the marginal memory in the end. I wonder why it took so long.

[rubber_jonnie]

At least it found the marginal memory in the end. I wonder why it took so long.

Not entirely sure, but going to run it again this afternoon see if it comes back the same.

[rubber_jonnie]

OK, so the 800XL kept failing differently so I pulled it apart to give it the once over and forgot it had 41256 RAM installed rather than 4164 on account of the Rambo adapter. I always forget the Rambo doesn't have it's own RAM and it's an upgrade with larger RAM in the original positions.

IMG20260108192822.jpg

Anyway, I took all the RAM out as I socketed it during the upgrade and tested all the RAM IC's individually and wouldn't you know I got this:

IMG20260108192158.jpg

One bad IC. If it were working the 4 LEDs on the tester would be magenta. The test runs, but then ends up like this, which is the powered on state for the tester. Anyhoo, long story short, the bad RAM is replaced and it's been running RAM checks for about 7 hours and it is OK thus far.

I'm going to leave it going overnight see if it's still good in the morning. If it is I'll use the built in RAM tester to see if that's ok too. Interestingly that machine has been idle for ages, yet the one time I power it on it's got bad RAM. It's all good though as I have a lot of 41256 RAM going spare.

[rubber_jonnie]

Well, the 800XL ran all night and threw up no further errors using Sys Check. Trouble is the Self Test gives me the same errors again..

However although the built in test has always been OK for me, since Sys-Check ran all night and found no issues I did consider things to be all good, however I decided to run some other tests and about 50% of them report an issue...

So since the RAM tester said all was good I'm going to consider it probably to be a quirk of the Rambo upgrade. It's currently running a 130XE demo (ShinyBubbles).

I'll try some more stuff later and hopefully I can button it up at some point today or at the weekend.

[stephen_usher]

Those small RAM testers, although good, don't catch everything unfortunately. They don't catch all refresh/timing issues and I'm not sure that they do a march test for addressing errors.

I guess that sys-info has better RAM tests?

[rubber_jonnie]

Those small RAM testers, although good, don't catch everything unfortunately. They don't catch all refresh/timing issues and I'm not sure that they do a march test for addressing errors.

I guess that sys-info has better RAM tests?

Yes, it could be a timing thing and poor testing methods but the manual states this:

Screenshot 2026-01-09 at 12.57.18.png

So I think it should be 'good enough'.

Sys-Check does at least ID bad IC's, but I do wonder if the extra banks throw things off. After testing with Sys-Check and at least 5 other types of software test and getting mixed results I do wonder how thorough the tests actually are.

I do have enough RAM to swap in another 8 ICs just to see if it makes a difference, so will likely do that since it's all socketed.

I have also tried some demo's that use extra banks, but I haven't done anything that uses 256KB yet, just 128KB, so that's on the cards too.

[stephen_usher]

OK, so not a proper test then. No "own address" or "march" testing so it won't pick up addressing errors at all.

[rubber_jonnie]

OK, so not a proper test then. No "own address" or "march" testing so it won't pick up addressing errors at all.

No, very basic, enough to say it works but not enough to say the IC is 100% in every respect.

[rubber_jonnie]

OK, so I seem to be moving forward with the 800XL

Using XRAM.COM, an extended RAM tester run on the Atari, there was one failure that was consistent. So I started replacing RAM one at a time until the error went away.

Guess how many I had to change...

IMG20260109165518.jpg

Now it is passing the XRAM tests just fine :)

So I've left it running the self test for now, and hopefully when I get home from Karate it'll be just fine.