Another (new) 130XE boots to red screen

[xrbrevin]

i wonder if the simple XE mod is less effective in the 4x DRAM PCB? ive had great results in all the ones ive done but they are all 16x DRAM boards...

anyway,  glad you got a reliable outcome for all your efforts 😁

[woj]

Honestly I think it's just my TV that resists all this s-video mod. It takes the s-video signal through an s-video to SCART adapter, then through SCART to proprietary Samsung mini connector, then uses a built in crappy upscaler. It can't even handle the PAL refresh rate right, on smooth scrolls you can see it jerking every couple of seconds in the attempt to resync. Until I get another s-video display or Sophia all effort here is probably useless.

[xrbrevin]

sounds like tight-arse samsung had a limited budget for backwards compatibility..!

[woj]

The computer still works without a glitch, today I successfully socketed the 74LS08 and replaced it with the F kind, and the base MMU socket, all in preparation for U1MB. In the process I also discovered that the manual de-soldering iron / pump kind of sucks (well yes, it is supposed to, but that's not what I mean :D), it does the job better than the iron plus manual pump, but it tends to scratch the board and spits solder around like a small baby. I get much better results by using the wick carefully. It's not as fast, but then I am in no hurry, I will just need to buy some wick-making-factory shares. It also minimized the amount of time I had to spend with the heat gun on it, practically down to a few seconds. 

 

I also did some research on using an external video up-scaler for a possible solution, it turns out that without spending big money I should not be expecting good or even working results. So let it be.

 

There is a question at the end, correct me if I am wrong: on this machine, if I desolder the RF-modulator, I will loose composite, correct?

[Beeblebrox]

20 minutes ago, woj said:

There is a question at the end, correct me if I am wrong: on this machine, if I desolder the RF-modulator, I will loose composite, correct?

Yes,  on the XE and XL line the composite is generated by the RF modulator AFAIK. 

 

Oddly with the XEGS it doesn't. 

 

 

[ZuluGula]

I think only XE units are affected by removal of modulator. Someone on http://www.atari.org.pl/ forum created litle board that replaces components responsible for composite signal.

[xrbrevin]

34 minutes ago, woj said:

I get much better results by using the wick carefully

you can boost the wick performance by pre-dosing it with extra flux

[woj]

My very next suspect for my initial problem and what followed is the OS rom. All was fine for some days now, including some soldering work to do sockets. Today I socketed the OS rom. At first it started fine and passed full memory test, so I was all happy. But like before with the GTIA, the next boot and memory test crashed, and the computer kept crashing in an identical fashion as before, sometimes in the first row of RAM test, sometimes boot straight to Self Test. I also now could test that removing the OS rom altogether makes the computer boot to red screen. Cleaning and streightening the rom pins for the second time and re-seating it in the socket at first did not help, but now it seems to be happily running again on the second attempt after a short while (again I used the contact spray that needs some time to do its magic).

 

So either the chip itself, or traces nearby. All changes in behaviour (whetever they were) were happening after soldering in the vicinity of the chip (GTIA / Basic sockets). Of course I checked all traces from the OS pins visually and for continuity before I soldered the socket. 

 

In any case, SysCheck is now somewhere close to me, so is the spare RAM. Hopefully this will finally resolve itself soon.

[woj]

I am totally lost again and the computer is playing up big time. Since yesterday's adventure it was working fine again, so I took the last U1MB preparation step and socketed the CPU so that I can have a clean U1MB connection with a proper adapter, see the pics. I did that extremely carefully, no heat gun was in use. This all worked fine, self test was running happily again (for several full rounds), all was fine, kind of relief as this was the last soldering job for the time being. 

 

Then the SysCheck was finally dropped into my mailbox. Hooked it up, ran the test, it passed no problem, then I hit start to run an endless loop as advised here. This lasted until the middle of the next memory test, SC reported both RAM chips bad. And then it started crashing, screen flickering, weird things. Essentially it is like that since then. After reading the manual for SC I ran it without OS rom, Basic, or RAM chips. At first there was hope, because it just reported the missing RAM chips as bad and kept displaying its screen. I decided to leave it on to confirm that all apart from those removed chips is OK. It is NOT! It crashed soon after, currently it is almost impossible to make it boot properly, I mostly get either black screen, or flickering screen (the attached picture is one of those, but it can be different each time). I reseated the GTIA and the CPU, still the same. I even tried bending the board very gently in different directions, still nothing. 

 

Either a huge coincidence, or SysCheck killed something (Freddie?) off with brutal testing. I eliminated a lot of things by now, yet I am still clueless on what it could be (and I am still short on replacement parts).  I also thought perhaps dirt in the cartridge / expansion ports, gave it good blow with air duster, nothing (but I perhaps need to revisit this option).

 

Any constructive remarks?

[woj]

If it helps, I also get static / noise in the audio that seems to be in somewhat of a sync with what's on the picture. Also, when I measure voltage on the bottom leg of diode CR3 it seems to add to the static (or, touching the leg with the probe causes additional static in the audio). In general (after a lot of measuring / touching / etc.) it seems that it could be something in the video or audio circuitry on the left of Antic / CPU that bothers GTIA which in turn bothers the bus (it can of course be the other way round). Something similar was going on and off before when I did the video mod and then undone it...?

 

I also measured the current usage (to check for shorts). The computer alone uses 0.59/0.6A, with the SysCheck in that goes up to almost 0.9 (0.88 IIRC). I wonder if that increased current flow could have triggered something...

 

And another thing, I said earlier that I was seeing 9 volts across the legs of the 500K pot. (I am not sure anymore if I used the same meter as now, and I suspect one of them is a bit shot). Today with the new pot it was 4.7V or so (pot as such did not seem to be contributing to all this mess). 

[Beeblebrox]

Just a suggestion but might be worth removing some of the transistors in the video circuit and either replacing them again or at least test them out of circuit. 

 

AFAIK sys check is unlikely to have caused any damage. 

[woj]

10 hours ago, Beeblebrox said:

Just a suggestion but might be worth removing some of the transistors in the video circuit and either replacing them again or at least test them out of circuit. 

 

AFAIK sys check is unlikely to have caused any damage. 

 

Right, so that was already done as these were my primary suspects from the schematics. Also the CR3 diode as it looked very badly soldered and totally mangled. (Would have replaced CR2 too, but had only one new one.) All the same. EDIT: Now on the big screen I looked again at the schematics and the reference board picture, I clearly see one point that needs revisiting, it's around R51 / R116 (I soldered there before), there is a trace from C48 very tightly packed there, and that one is supposed to smooth pot signal to the GTIA.

 

Not saying that SC could do harm by design, but by simply putting more current load on the whole system (the difference is substantial), perhaps some weak spot did not like it. 

 

What I am also bothered about is that regardless of what I have said earlier, the CPU seems to be running much warmer than the rest of the chips. I now have an 800XL in an unknown state on the way to me, will take 2 weeks or so again before I get it. In the meantime I will take some rest and try looking at things again fresh. I have seen videos on YT where similar problems turned out to be the power supply, but now I checked 3 different ones, nothing. For this, the only other option is that the power socket / switch is bad in some way, do Ataris have a history of that...?

 

One other thing, couple of posts back when I was suspecting the OS rom, I also realized that earlier literally none of the crashes happened when the game / demo was loaded in. Those are known to typically switch off the OS rom and run with all RAM happily, hence I was suspecting the OS rom even more. SC behavior says otherwise.

 

Edited September 15, 2022 by woj

[Beeblebrox]

The cpu generally does appear to run warmer than other ICs AFAIK. I was working on an 800XL the other day and noticed this.  It is running fine but cpu is definitely one IC that does run warmer than the other 40pin ICs. If it is so hot it hurts to touch for more than a few seconds I'd be worried of course. 

 

Reason I mentioned the transistors again is I had replaced and checked the video circuit transistors in A8s and they'd checked out at the time.  But later on after I'd been working on the board fixing other issues some had failed again.  I guess you can't rule it out.  The display issues you are having could still in part be down to a resistor gone bad.  Worth replacing some as they are dirt cheap. 

 

Does Antic check out BTW?  All the chip pins definitely making contact with the socket? 

 

 

[woj]

Not sure about Antic, I recheck the IC to socket continuity from time to time. As I just edited in the previous post, I defo have more careful checking to do. 

[Beeblebrox]

Sure. Can really test ya patience sometimes fixing these machines.  I've had to give up for now on fixing my XEGS which has a red screen.  Completely stumped. 

[woj]

51 minutes ago, Beeblebrox said:

Sure. Can really test ya patience sometimes fixing these machines.  I've had to give up for now on fixing my XEGS which has a red screen.  Completely stumped. 

Keeping looking at the schematics (this is really a good lesson) it can be so many different things, including the RF modulator...

 

Note mostly to myself, if I unhook C23 I should be able to figure out if the static is caused by GTIA or by the audio / video circuitry. 

[woj]

First, inspired by one YT video where a dirty cartridge connector gave similar results on an 800 XL and realizing how clean those things have to be I cleaned up the CPU and GTIA yet again and reseated it one more time. Nothing changed.

 

Then jumped a ground to one area of the video circuitry which otherwise is connected by only a thin long trace to other ground plates (this was in fact my primary suspect for toady). Still nothing.

 

Then unhooked the C23 on one side (this connects the S3 of GTIA to the audio circuit and provides keyboard click). The static noise was still there and still crashing. OK, so GTIA is not the source of the noise (I suspect it's actually a noisy/shorting ground).

 

Also cleaned up the board here and there, most notably around the solder points for the RF modulator where I still noticed tiny solder splatter. Did not have enough time to clean the whole board and inspect it carefully, but that's one thing that has to be done.

 

Last thing for today, I desoldered U1 to disable the audio all together (at first I was hoping there would be one power line to it somewhere, but I could not figure it out). The static was gone (I'd be really surprised if it stayed to be honest), yet the computer crashed. If something, I could say that it made things worse, but it could be in my head. 

 

Leaving it as it is for today, tomorrow I will study the schematics again and try to see what these experiments mean logically. 

 

[woj]

On the attached picture found on the Internet I see a jump cable exactly the same as I was trying yesterday, only I used crocodile clips, perhaps I should try again, I definitely had a good idea.

 

One other thing I can notice on this picture (its the older generation board) that the soldering work is like 100x times better on this one than mine. I honestly start to think that this NOS batch of computers I bought this one from was factory rejects from quality control or something...

 

EDIT: Looking at other pictures it seems this was factory equipped (the jump wire) on the older boards. There were also other jump wires mounted by factor (GTIA pin 32 to LS08 pin 8, and pin 10 of LS08 to pin 15 of U2). Clearly these were added for stability? Or was there no routing for these on the older boards?

 

Edited September 16, 2022 by woj

[woj]

Right, so now I know what the GTIA jumper cables were for, these I do not need on my board. This ground cable on the picture is also for machines that do not have a corresponding trace on the board. Yet, the trace on my board looks very poor and insufficient, so I will be looking at this again. I do have a good plan on what to check next

[woj]

I soldered the U1 back in, cleaned here and there, checked some things with the multi-meter, unhooked R9 to see if that would remove the audio static. And... I did not get a chance to see that, because with it connected or not there was silence. I also noticed there is now 9V again across the 500K pot legs, so I guess it should be like this. So this particular thing seems to have cured itself... This is again worrying, because I do not know if that was incidental or I did something today to get rid of it. 

 

But, the computer still crashes after running for several seconds, and SC refuses to accept 3 (!) different sets of RAMs (they all now arrived), now right up front testing 0000-01FF. When I allow it to stay one with the message to replace RAMs it starts crashing with the screen all messed up. Also without RAM altogether. 

 

Now, this is the same kind of crash that I get when I disturb the Freddie crystal signal by touching pins 1 & 2. So my next suspects are the poor Freddie and the crystal (we have been there, perhaps what I did to the capacitors / resistors woke it up for some time0. Not sure how likely the crystal is to fail (tell me 😀), but I can see it is going to be a bit of a hassle to find one (but I think I might have found just the right place for that). For Freddie, well Lotharek has the replacement part available. Too bad none of these two things can be sourced out of an 800 XL...

 

EDIT: Actually, looking at the schematics, when the test starts and I get a properly running screen, but failed memory test, that could mean that the OSC signal going out of Freddie is fine (until it crashes), yet Freddie refuses to process RAM requests properly. That would make Freddie the first candidate (if the problem is in that "area"). 

 

My really last thought is to try to go back to the 220nF capacitors for the memory chips as they were when the computer arrived, now I have 100nF in there... 

Edited September 16, 2022 by woj

[xrbrevin]

FREDDIE gerbers.zip BOM.txt

[Beeblebrox]

9 hours ago, woj said:

Too bad none of these two things can be sourced out of an 800 XL...

As per Xrbrevin's post above you can build your own Freddie or get one of the modern replacements ordered.  I got mine from Ebay UK which came via Poland. 

 

Incidentally Atari 800XLF boards have Freddie chips. 

 

 

[woj]

I took out the oscilloscope and committed to sort out how it works, and I think i managed. The state of the computer today is that crystal / clock signals on Freddie look totally healthy (check pictures), yet the memory testing fails, once it run all the way to the end and reported only one chip bad. I also measured the signals on the RAM and Freddie somewhat haphazardly, they visually looked OK. But I'd really need to do pin to pin comparison between the memory and Freddie to see if the stuff that goes out gets in OK. 

 

I also isolated the very spot where the crystal circuit is sensitive to touching. It's between the leg of one of the resistors and the leg of one of the capacitors. These legs are actually connected with a trace! When I touch / press them both, the amplitude of the wave signal going to pin 2 of Freddie decreases somewhat. The OSC out clock on Freddie goes out of sync then and that garbles the image. Since working with the iron around this area helped once, I need to revisit and check that trace carefully. I also have replacement capacitors for this, so I will try that too. I simply want to eliminate something trivial before I commit to replacing Freddie. 

[woj]

This computer has definitely some electrical issue, and I should not proceed before I identify what it is. The oscilloscope helped me to identify that the audio spiking is appearing on leg 1 of U1, voltage jumps from stable 4.8V to the maximum power 5.2V (what comes out of the power supply). That pin is directly connected to the audio pin of the RF modulator, and otherwise to the audio output. No input (leg 2, 3) of this output pin 1 of U1 indicate any jumps on that side. I concluded that the RF modulator at this point should go out just to be on the safe side. That did not help. 

 

The computer seems more stable in terms of garbled screen, it stays in tact (in SC), but it still fails the memory test (twice today it almost made it to the end), and there is occasional spiking coming out of the speaker. (Yes, I did check the cable, when unplugged I still can see the spikes on the oscilloscope). For this spiking, it actually leaves me little options of what it could be. It also seems that all / most problems revolve around the C +5V line on my schematic from the PDF linked here earlier. 

 

One of my next plans is to remove Q2 (the picture should go B&W) and see what that would do...

[woj]

I am away from my bench, please tell me I am not hallucinating - the C1 capacitor is connected the wrong way round?! (This is a picture of the board as it came from factory, when recapping I connected the new one exactly the same way.)