Another (new) 130XE boots to red screen

[Beeblebrox]

Sure,  at this stage with these things it's usually several issues now contributing. 

 

So you are happy ram chips are stable? Bad or malfunctioning ram can still cause all manner of odd behaviours. 

 

I can't recall do you have another A8 from which you can confirm the other 40pin ICs are working in one at a time?

 

I had a bad Pokey the other day which caused the blue screen basic prompt to freeze without the ready message and no way to type.  Sound was OK so I thought it was other things causing it.  Turned out the Pokey chip was part malfunctioning. 

 

Like me with that 130XE I fixed you get to a point where you have done a significant amount and for the moment maybe you are feeling like you'll not find the answers any time soon. I would perhaps wait to get your sys check ii as that is a great help in these instances. 

 

It is great you are getting an image.  I think aside some other issues Antic as well as the transistor and possibly Pokey and PIA are worth looking  at.  You mentioned a power surge the other day.  Don't be surprised if that hasn't caused more issues. 

 

 

Edited September 2, 2022 by Beeblebrox

[woj]

No other A8 yet, working on it. And no, I have not yet excluded RAM, though it seems it is something else. When I get a fresh spray can I am hoping to at least narrow down the board area. My actual bet that I cannot yet verify is actually the GTIA that might have not survived the desoldering. 

 

Fresh boot after chilled during the night, runs as long as about 2/3 of the RAM test.

Edited September 3, 2022 by woj

[woj]

A bit annoying that I cannot edit the last post after some time, so they are multiplying (I write all this also for myself as a log). I inspected the transistors on the board. One thing that I noticed is that Q3 is different brand / packaging than the other 2N3904 transistors on the board. It does not look like it has been replaced post production though... Since transistors are not that expensive and not that risky to replace, that will me one other immediate thing to try.

[+kheller2]

So you suspect the GTIA yet want to replace an odd looking transistor.  I think you should wait until you have some replacement chips and a syscheck.   By all means continue to replace every component if you are into that and have the time, but for me it sounds like you are grasping at anything to fix this before ruling out major chips. And that’s fine. But…. It’s a lot of wasted work, perhaps.  

[woj]

Yes, I am grasping a bit as I am a bit novice with this kind of thing. I am trying to be organized about this in my own way of doing things ;). Why not replacing a 10cent part that takes 5 minutes to replace and that shows some likelyhood of being the problem (or one of them) while I am time blocked with checking other things? As long as I have time for that...

[Beeblebrox]

Talking from personal experience it's true the more you subject the pcb to heat and more you do,  the more stress you are putting in under,  and you potentially increase the chances of introducing new problems,  especially if doing too much too soon. Also worth noting that having checked a component or IC in the beginning I've often found that later it has subsequently failed.  Again as you had a power surge the other day it's worth revisiting some things at some stage. 

 

I'd stop for now and wait for sys check ii and if poss source another socketed A8 to testing chips in.  I have a test 65xe I socketed for that very purpose.  Of course not everyone can do this so perhaps you have someone near you with one here on AA? 

 

Edit  BTW I was a novice when I started fixing A8's last year and for the most part I still am. As you can see from my signature linked threads I too did the same as you a lot of the time.  Best advice I had from the forum was to read up as much as poss and where possible only make 1 change,  then test,  make another,  then test. It can be frustratingly slow but actually it makes it a lot easier to get to the route of the machine's issue(s) at the end of the day. 

 

And a multimeter is definitely you friend.  Use it as much as possible and when socketing and initially removing an IC,  before  you pop the socket in clean up the vias and check all the traces and vias for continuity,  especially top side of the board whilst the socket isn't obstructing it.  Then after installing the socket check it again as much as you can. 

Edited September 3, 2022 by Beeblebrox

[woj]

Right, so huge progress, but not out of the woods quite yet, but I hope much closer now. Got the air can, started cooling chips one by one after a crash, that did nothing. The only thing I concluded was that it is the whole board that needs to cool down (after what came after I think I was right). 

 

Cutting part of the story short about inspecting voltage supply to the chips with an oscilloscope, I went to inspect the board visually piece by piece again. That gave nothing, but when I was about to put it aside I noticed that one of the resistors (R20) is sticking out a bit, so I touched it. To my total amazement it turned out it was not soldered on one end, just touching the pad, see the picture. I looked at the schematic to confirm that it definitely needs to be there and to confirm the measured value (3k). Had no new one so took that one out, cleaned up, and resoldered.

 

All happy went to my TV being convinced that this is it. Not yet. But, the computer booted up properly, self test was working, just showing a lot of memory errors. No screen flickering. After a while rebooting / resetting made it go directly to memory test, consistently for several boots. I started swapping the memory chips around, same thing. But I thought to give it time to run the test a bit and see if something else will go wrong (awaiting screen flickering). It crashed in the sense that the screen got blank green. Somewhere in the meantime I was pressing gently on the RAM chips and other things. I then power cycled it, it started a normal boot rather than going straight to memory test. OK, I thought, weird, let's reboot to self test again. And from that point on, believe it or not, the memory test ran fine for half an hour. 

 

At this point I concluded that I reversed the state of affairs, that the computer needs to be warm to run OK. The other option is that by pressing the chips I made the ones in the sockets sit better. Or, that the initial errors were due to the board or resistor still warm after soldering. I let it cool for 15 minutes. Booted nicely again, but I did get one faulty square in the RAM test during just one of the passes. After that, as I write it keeps running OK for another 20 minutes.

 

So, next thing is to swap memories around again, perhaps all of them are bad, but some less than the other. And then let it chill fully over the night and see what happens then.

 

But please, do tell me if there is anything else I should be considering at this point (apart from also inspecting the board yet again, that poor manufacturing quality is unbelievable).

 

EDIT: And swapping around RAMs got it back to a properly non working state, either does not boot (black screen), or straight into memory test. When did boot OK to Self test, memory test failed quickly too. What is going on....?

 

 

Edited September 3, 2022 by woj

[woj]

This morning it behaved the same way as last night, booted nice for the first time, but ran only for several seconds. In the meantime I figured out I can bridge the eMMU socket to use memory slots 3 and 4 instead of 1 and 2 (you do that by bridging pin 6 with 10 instead of 11) to eliminate the possibility of bad connections between the sockets. That did not change anything, it was now booting straight into memory test. 

 

And now! When I was lifting the keyboard to peek inside, (yet again) I slightly dropped it when it was vertical on the bottom edge of the board while the computer was running the self test. This made the screen go black (this happened before, but it never helped). After a power cycle it is alive again, now the memory test runs for good 20 minutes without a glitch.

 

This tells me that I really have to inspect the board carefully again for any more mechanical issues. This would also explain why I get some changes in the behavior every time I take the board for soldering and flip it around a lot. 

 

 

[Beeblebrox]

Crack in a trace or a weak socket connection somewhere. Also are all new sockets installed precision types or are there any old single or double wipe ones used? Double are OK but single wipe aren't great.

 

These XE boards are definitely not as robust as the XL ones.  

[woj]

GTIA is precision, everything else so far double wipe (for a reason). I pretty much eliminated the memory sockets as such by moving to the other two, but it does not mean that a bad contact is not caused by the chip legs bent or something.

 

I can't work on the board or even test it until much later today, but in the meantime I identified at least 6 more questionable solder points, and also noticed that two legs of capacitors mounted "vertically" near the SIO socket were leaning on each other. Will find out later if that has anything to do with it, and later magnifying glass and more board inspection.

[woj]

OK, I think I am there, but I need some serious answers, from someone that knows this stuff. It was not the hit shock that was resetting the computer, it is, as far as I can tell, touching the metal case of the keyboard to the ground plate. By experimentation, see the picture (this is under Freddie) I discovered that when I touch the two left legs of the resistors on the right (presumably 2.2M if I can read the coding right) the computer crashes in practically the identical manner, including screen flicker and going black and white. The bottoms of the two vertical resistors are connected accordingly to those on the right, touching there does the same thing. I reflew the solder points on all four, cleaned up, computer seems to run fine when I do not touch this. I am yet to try to suddenly touch the keyboard case to the board border, but I bet it will shit itself again. After reflowing it seems I can touch the left two resistors without consequence, but I am not 100% sure about that, coincidence perhaps. 

 

Now, I now of course looked at the schematics, I also noted the C32 marking under one of the resistors that already told me something is off. The schematics says these should be capacitors. Also the 2K2 resistor on the schematic is not so on the board, it is 1M (the other one is also 240 instead of 220, but this I can let go). Fine, I have little clue about circuits, so perhaps replacing capacitors with high valued resistors makes sense, is desirable, cheaper, what not... But why? Then there are two board pictures in the same manual, one of 65XE with 2 RAM chips, and the 130XE with 4 RAM chips, see pictures. The 65XE has the capacitors, the 130XE has the resistors. All the values of resistors I see on those pictures are the same as on my board.

 

Now three things:

 

1. Can someone give me a short lecture on this?

2. Should I replace those two 2.2M resistors with the corresponding (according to schematics) capacitors?

3. Can the whole odyssey be caused by the fact that I removed the RF shield after the first opening of the computer, which in this case is perhaps needed to maintain some sort of electrical equilibrium?

 

EDIT: Forgot to say, the computer now happily runs the memory self test for 30 minutes with all 4 RAMs, eMMU in 130XE mode!

 

Edited September 4, 2022 by woj

[xrbrevin]

glad you got it sorted!

my faithful old multimeter has expired so progress has halted on my XE repair. it seems to read the opposite for continuity but i at least need another meter to examine it..!

 

regarding your resistor/capacitors quandry, just be aware that Atari are known to have used capacitors that look like resistors. theyre light greenish - like in the last image

[woj]

28 minutes ago, xrbrevin said:

regarding your resistor/capacitors quandry, just be aware that Atari are known to have used capacitors that look like resistors. theyre light greenish - like in the last image

 

? OK, that would explain why I had problems measuring them with the multi-meter, doooh. But why would they have misleading stripe markings? Also they are both the same, the schematics says they should be different. I will have a bit more search.

 

Anyhow, I am not closing this yet, computer still runs, not a single crash, Lasermania happily playing its intro tune, but it needs to do so for a couple of days. If it needs the shield, the top one won't fit without cutting (GTIA fixer board), and later on U1MB (need to take a break before I dive into this...).

 

One last thing, the previous fix with the loose resistor, since that fix the CPU runs much cooler! It has the same temperature (warmish) as Antic and GTIA, it used to be hot in comparison. 

 

EDIT: Right, so now I found the pages about those capacitors disguised as resistors 😕

Edited September 4, 2022 by woj

[Beeblebrox]

@woj In the early days of my learning about fixing these I also fell for the caps in disguise as resistors. Very confusing.

 

BTW IMHO there is absolutely no need for the shielding nowadays. Almost evey A8 I've fixed I've discarded the shielding, especially if adding modern upgrades like U1MB. No need for them. If you look at FJC's vids he rarely puts the shielding back on XL and XE machines if ever.

 

I've a fairly large pile of these sheilds for 600 and 800XLs especially now. Damn they are lethal - sliced my hands open on at least two occasions in the early days.

 

BTW worth running the Sys check II tests for RAM and ROM for a good hour when you get it.

 

Well done for getting this fixed. Painful for you but I bet you are feeling pretty pleased also? 

 

EDIT: when you come to the U1MB install for this you should be fine. The OS  and MMU are removed/replaced so they'll need precision sockets and the 4 x hookup points for the CPU can be taken off vias on the board. Check out FJCs YouTube channel with this vid: 

 

Edited September 4, 2022 by Beeblebrox

[woj]

Yes! But I will declare victory once I stress test it for several days.

 

The thing I am most happy about is that it seems that my (de-)soldering is quite solid, I had doubts...

[woj]

The machine cold booted and ran fine this morning for a good hour or so, this now looks good.

 

Side question, while desoldering the GTIA and later on the socket to correct it, some traces got disturbed, I mean the coating peeled / rubbed off a bit here and there. What should I coat it with? I read nail polish, but is there something better / specialised for that? I will be ordering some small things from an electronics shop (sockets and such), I can get something while at it, but I am not sure what...

 

EDIT: Nevermind the last question, the store has this plastic spray made specifically for that (just larger surfaces I believe), though it is out of stock there at the moment.

Edited September 5, 2022 by woj

[Beeblebrox]

@woj not sure if it makes much of a difference,  aside aesthetically these days. I may be wrong of course. I'd personally leave it and close it all up and enjoy the machine for now.

 

I'd also run some 128k games like Prince of Persia to further stress test it....aside playing it simply because it is such a great game. 

 

Congrats on fixing it. If you have the motivation perhaps do a blog covering your repair.  

 

One thing.  When you come to do the U1MB - to improve stability replace the 74LS08 logic IC with a 74F08. I pretty much do this as standard now with my XEs.  Definitely if you plan to then use SIDE3.  The 74F08s are dirt cheap. 

Edited September 5, 2022 by Beeblebrox

[woj]

14 minutes ago, Beeblebrox said:

One thing.  When you come to do the U1MB - to improve stability replace the 74LS08 logic IC with a 74F08. I pretty much do this as standard now with my XEs.  Definitely if you plan to then use SIDE3.  The 74F08s are dirt cheap. 

 

74F means different shop for me (very attached to the one I use :)), what about a brand new 74LS, they have those. Yes, I know that F is different spec and supposed to be faster, but perhaps the stability is also in age / manufacture quality?

[Beeblebrox]

Not as far as I know.  Ls08 are apparently less efficient or something to that effect.  It's a well known and highly recommended practice to replace it with an F08 and has fixed many an instability.  Whilst doing the U1MB upgrade it's just one extra step.  

Edited September 5, 2022 by Beeblebrox

[woj]

I was about to write that all is still fine and show you some shots of stuff running, however, during the process it crashed on me temporarily again, after good one hour of testing, it was "uneasy" and booting to memory test for perhaps 30 seconds, what might have helped was a small tap against the table, it came back to life. So I will need to check again, it defo looks mechanical and I really want to believe it is in the area of those capacitors and the clock signal from Freddie, it got much much better after I worked around there.

 

What would help me tremendously if some of you having an opened 65XE or 130XE of this kind could check if touching at the same time those two capacitors below Freddie makes their machine go flickering. Just hoping here really

 

Shots coming anyhow. Too bad IK+ is not really finished... My TV also sucks in color reproduction, I think it has some really low quality up-scaler in there. 

Edited September 5, 2022 by woj

[Beeblebrox]

@woj Did any of the ram chips get really hot at any point during that time incidentally? Any other ICs? 

 

Just out of interest how are you connecting to your display at present and can you post an image of the self test menu? 

 

Have you played with the colour pot at any point as it might simply be it needs turning a little to get away from the blue colour hue range it seems to be displaying. 

 

The instability could be mechanical where a trace/solder via connection expands after a good while.  Or sign of a failing IC or maybe other passive component. 

 

Do you have your sys check ii yet? 

 

Edited September 5, 2022 by Beeblebrox

[woj]

The case was screwed together, so could not quickly check and it was also too late to start disassembling again. 

 

It's an svideo and composite cable I made from some old dvd player cables I had at home. It is connected to a scart input on the TV. I know the color issue is the TV, because I had the same problem ages ago with the said dvd player. Will post pictures later. In any case, through svideo it half decent, composite is tragic. And also I am not touching this until I sort the other thing out.

 

No, syscheck is still in transit.

[xrbrevin]

try this - highly recommended and minimal work 😁

simple XE display mod.txt

[woj]

11 minutes ago, xrbrevin said:

try this - highly recommended and minimal work 😁

simple XE display mod.txt 277 B · 0 downloads

Seems simple enough, thanks!

[woj]

In a separate post. My detective work so far:

 

1. It all started with a red screen boot after working on the GTIA socket.

2. After desoldering the RAMs, the eMMU, the Basic, and reworking the bypass capacitor on Freddie it got to the state of booting up and then crashing, working better or worse depending on the "weather". One thing that matters here is that that part of the board got some heat in the process.

3. Somewhere in the process I discovered that disconnecting the CAS line to memory also results in a red screen boot. So the initial problem was probably cut-off memory circuit (Freddie/eMMU/RAM chips).

4. At the same time, after 2. I never ever got a red screen on boot again.

5. Also after 2. I changed the bypass capacitors on memory from 220nF to 100nF. This seems to have nothing to do with anything. 

6. The oscillator circuit below Freddie is very sensitive to touching and causing identical symptoms. There were also some very poor quality solder points there. Reworking them (did I do it well?) seems to have helped considerably, almost to the point of declaring victory.

7. The said circuit is full of discrepancies between the schematics and the board. R26 is 1M instead of 2.2K, the capacitors are 22pF (if understand the markings correctly) instead of 10pF and 18pF.

8. The loose capacitor near CPU also made it better, CPU runs cooler after this.

 

Thus, I will revisit that area, see what else could be amiss, and resit the RAM chips and eMMU yet again, and clean the sockets once more. Spare RAM chips should arrive at some point too.

 

Next option is that Freddie is playing up, so the next step is going to be getting a replacement. 

 

The other thing I would consider is redoing the oscillator circuit according to the schematics. Would that make sense?

 

And of course running SysCheck as soon as I get it.

 

Do I miss something?