Another (new) 130XE boots to red screen

[Beeblebrox]

5 minutes ago, woj said:

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.)

Can you circle C1 on pic? Don't have the schematixs to hand. 

[woj]

9 minutes ago, Beeblebrox said:

Can you circle C1 on pic? Don't have the schematixs to hand. 

I am on the phone "in the field", it's the vertical one in the very corner 470uF close to the power switch.

 

EDIT: I mean literary every single 130xe board picture I find on Google has it the other way round.

Edited September 18, 2022 by woj

[xrbrevin]

i have a 65XE PCB to hand and it has C1 orientated correctly - as per the silkscreen. yours is the opposite - good spot!

[Beeblebrox]

I thought this particular 130XE pcb wasn't like the usual XE boards, (eg 2 x ram chips,  slightly different layout).  So not sure comparing different Xe boards works. However just look at the plus and minus silkscreen on your particular board @woj after removing C1. 

 

One thing -  presumably the XE wouldn't have powered up at all with the cap round the wrong way,  so not sure why it did?  Unless you recently recapped after it initially worked!? 

 

Not had time to read all the info on this thread of late. 

Edited September 18, 2022 by Beeblebrox

[woj]

The board design in this place is the same as the older board, the capacitor was simply mounted wrong. In all honesty I was hoping that is it, but it is not (or not yet, read on). What triggered me to look at capacitors in the first place was that yesterday and this morning I ended up in a situation that I had it more or less stable without SysCheck (it was running self test happily), SysCheck was crashing on a second run at the latest. BUT! when power cycling it without too much delay it was booting straight to Basic, I mean immediately without trying to boot an SIO device. That made me thing something is holding a state after turning it off. (The Basic is freshly booted though, any key strokes on the screen are cleared.)

 

Now, this was the same as I have at the moment after replacing and putting C1 the right way round. The off period when it boots back to Basic without delay can be as long as 5 or even more seconds. But also non-deterministic, sometimes it just does the right thing. When booted into Self Test it runs happily in a loop for as long as I have patience. It also seems to be better with SysCheck, right now it made 2.5 full rounds before it crashed (the crashing is not only reporting memory errors, but the line that prints the currently tested address gets garbled with random characters).

 

So to me there is still some electrical problem. In all recent tests today I heard a couple of audio spikes, but not as much as before. Hearing them convinces me that something of a similar fashion is still hidden somewhere. I wonder if this flipped cap could have destroyed / worn the other electrolytic capacitors responsible for power?... And I  still attribute SysCheck being less stable than Self Test due to drawing more current from the system. I really need to solve this equation. 

 

EDIT: One more fact, before correcting C1 the resistance between + and - where ever on the board was 300+ ohms (this seemed low BTW, but what do I know). Now it is 600+, so the capacitor orientation changed that drastically. Can someone measure it for me on their computer, what should it be?

 

Edited September 18, 2022 by woj

[Beeblebrox]

 

 

11 minutes ago, woj said:

The board design in this place is the same as the older board, the capacitor was simply mounted wrong.

Sure.

 

So the fact it can be on for a little while at not crash leads me to also think along these simplistic lines:

 

• Either a passive component is on it's way out
• An IC is part failing/is unstable
• An IC's socket is unstable
• There is a solder via or trace connection which is unstable and when the board heats up it loses a connection.

or one or more of the above combined.

 

Have you tried FJC's trick of spraying alcohol on ICs whilst off and turning the machine on to see if it evaporates instantly or notably quicker on some ICs than others?  With the exception of the CPU which will run hotter in my experience, generally this is a good one to test if ICs are going bad or - in the case of outputs of extreme heat - have died.

 

 

 

 

 

 

Edited September 18, 2022 by Beeblebrox

[woj]

All chips are cold, nothing stands out. 

 

It would really help me if someone could measure resistance between +5 on the board (middle pin of the power switch) and ground when it is fully unplugged (no extensions / upgrades / cartridges). It seems to be vital, because after correcting C1 it went from ~340 ohms up to ~650, and now after cleaning some still dirty areas (crusted flux from production) on the right hand side of the board that I missed before it went still up to 730 ohms. All those values sound little low to me still and the fact that mere cleaning changes it by 100 ohms tells me I still need to search for something trivial.

[woj]

15 hours ago, Beeblebrox said:

So the fact it can be on for a little while at not crash leads me to also think along these simplistic lines:

 

• Either a passive component is on it's way out
• An IC is part failing/is unstable
• An IC's socket is unstable
• There is a solder via or trace connection which is unstable and when the board heats up it loses a connection.

or one or more of the above combined.

 

Inspired by FJC's case of a computer not working with cartridges and the corresponding MMU failure, though SC is not really a cartridge, looking at the schematics again it can indeed be absolutely anything. I for now exclude Freddie, so far the oscilloscope says it's good, now I have found more elaborate procedure on how to test it in another thread by Jürgen, so I will do that too. A functioning MMU I hope will arrive in an 800XL, otherwise if I hook up U1MB and it is the MMU I should also find out. Otherwise I just ordered a complete set of all LM/LS chips with sockets (I forgot the 555 timer for reset, but that I guess is not a problem) to check them one by one, if I read the forum right, these can also be a problem.

 

I am till a bit clueless what potential damage could have the wrongly mounted capacitor do.

 

[woj]

The computer seems to be happily running without SC, did not try games, but Self Test memory test loops happily. With SC it crashes on the first run.

 

Intrigued by the resistance between + and - (~750 ohm) I started removing the ICs I have in sockets one by one and see how it changes. Nothing changed for almost all of the ICs (CPU, LS04, MMU, eMMU, OS ROM, Basic ROM), but it did change when I removed GTIA, it went up to ~1090 ohm. So I measured the GTIA resistance over its +/- pins, it is 5M ohm. So that can't be it. So I started looking at neighbouring pins on the GTIA socket, next to +5V there is the HALT pin, and I see resistance of about 600 ohm between HALT and +5V and 1800 ohm to ground (or the other way round I am not tired and I put away the board). The only IC still attached to the HALT signal is Antic, otherwise the line should isolated. I even cleaned up around these pins and traces on the GTIA socket, SC still fails, resitances did not change much. Now, I know there is probably a pullup resistor inside Antic of some sort (I forgot to check if there is one on GTIA), but something generally does not add up here for me, yet today I cannot think straight anymore, and I do get confused whether I test against +5 or ground.

 

TBC.

 

EDIT: Keep you fingers crossed, the Self Test was running when i was writing this post, after posting I now ran SC again and it is currently on the full 7th successful round... This is very new. Perhaps this cleaning helped...?

Edited September 19, 2022 by woj

[woj]

So since yesterday the computer happily runs SysCheck tests endlessly. Yet again, I am not 100% sure what fixed it, I was nowhere near it with the iron this time, there are two options:

 

1. When taking out all socketed ICs, I reinserted one of them just right way, one that is very sensitive to the right seating.

2. The cleaning of the board after the measurements of the resistance of HALT line against +5 and ground.

 

It did not work on the very first run of SC, but from a second one it keeps going, also from a cold boot in the morning. Perhaps the IPA did not dry well yet when I turned it on for the first time, or not all of the flux crap was dissolved yet. Or, which would be a bit worse, I was also blowing air duster under all ICs and around the board, something got very cold that does not like to be and needed to warm up to work. One thing I think I know for sure, regardless of how suspicious these resistances were/are to me, after the cleaning they got very stable, before the cleaning they were floating, which to me says there was some unwanted conductivity somewhere. 

 

From the past experiences, I hold on with declaring any victories, but so far so good.

 

So a quick summary of what was wrong with the computer so far:

 

1. Unsoldered R20 resistor just leaning against the board, effectively depriving the O2 signal to the CPU of a pull up. Factory fault, probably the reason for red screen boot.

2. Totally flaky and dried out 500K pot. Just age problem, to be expected, all kinds of weird video issues on the way.

3. Wrongly mounted (opposite polarity) main C1 capacitor, from factory! Not actually sure what it was breaking, if anything.

4. Absolutely filthy board with a lot of factory flux residue. My own (de-)soldering added to it. 

5. (Wrong software setup on my Linux causing unstable SIO).

 

Apart from a lot of unnecessary (but educational) (de-)soldering, the loses so far are:

 

1. RF-modulator gone (not broken, just detached), meaning composite gone, meaning having crappy color on my TV.

2. One of the rectangular pads on the board for the RF-modulator casing destroyed in the process.

3. One of the via pads for memory chip by-pass capacitor peeled off. Luckily on the non-vital side of the board (no trace on that side). 

4. A lot of scratches on the board from the slipping soldering iron (not the most steady hands that I have) and the desoldering manual iron+pump. 

 

The winnings:

 

1. A working computer, or so it seems

2. Got the hang of using the oscilloscope

3. The computer is U1MB ready (BTW, it takes ages to get to me, something is wrong, I need to check with Lotharek)

4. Now have SysCheck that turned out to be vital, but in a slightly different way than expected. Will be very useful in reviving an 800XL that is also arriving at some point.

5. A bag full of spare RAM chips, different brands too.

6. Much bigger knowledge base about this little thing

 

 

Edited September 20, 2022 by woj

[Beeblebrox]

Well done - that was a real hardware fix gauntlet eh?

 

45 minutes ago, woj said:

1. RF-modulator gone (not broken, just detached), meaning composite gone, meaning having crappy color on my TV.

I've lost where you were with this. I vaugley recall your actual TV/display being the bad colour issue. Didn't you also do the S.X.E mod? Can you use the machine with an S-video cable?  

Edited September 20, 2022 by Beeblebrox

[woj]

9 minutes ago, Beeblebrox said:

Well done - that was a real hardware fix gauntlet eh?

 

I've lost where you were with this. I vaugley recall your actual TV/display being the bad colour issue. Didn't you also do the S.X.E mod? Can you use the machine with an S-video cable?  

 

There are still chances this is not yet fixed, we will see, but yes, cars are easier to repair / upgrade (something I used to play with years ago, only now I see how well car ECUs are built ).

 

Yes, my TV takes both s-video and composite through SCART and some built in crappy up-scaler, s-video is sharp / solid, but with really bad color and a lot of color artifacting, composite has (had) good color, but very grainy picture. So neither is ideal. I preferred composite with the right slider settings on the TV, now I am left with s-video and no slider setting to make it acceptably right. The SXE mod did not help much, but since there were so many other things wrong on the way, I plan to give it one more gentle go, especially the color line. It is also remotely possible that one of the SCART adapters that I have to use on the path has some "correcting" logic in them, I remember seeing things of this kind before.

 

[woj]

4 hours ago, Beeblebrox said:

Can you use the machine with an S-video cable?  

 

Extra thank you for asking this question. I looked on the YT for videos on those s-video to SCART adapters, and I now know that I need to at least open mine and see how it's connected. (a) I get poor color, (b) I got effectively B&W with the SXE video mod both tell me something could be not right in there. 

[Beeblebrox]

7 minutes ago, woj said:

 

Extra thank you for asking this question. I looked on the YT for videos on those s-video to SCART adapters, and I now know that I need to at least open mine and see how it's connected. (a) I get poor color, (b) I got effectively B&W with the SXE video mod both tell me something could be not right in there. 

 

also - and maybe you have already checked - but make sure the colour pot is adjusted ok just to eliminate the off colour output you are experiencing being caused by a misdialing of the pot. 

 

Also feel free to pop a few up to date images of the 130XEs board and also it working/plugged into your display if you like.

[woj]

Any experimentation with S-video output led me to nowhere, it is just my TV. In the first two pictures you can see the difference between what the TV does with the S-video signal and the composite signal (that was before I removed the RF-modulator). The S-video has absolutely messed up color, the composite has the single half pixels amplified that the upscaler in the TV decides to give two different colors in many situations, see also the third picture. 

 

So, with the RF-modulator gone I implemented a temporary solution of adding the 470pF capacitor inside my cable, the effect is that I have the composite back with decent color, yet the same single half pixel amplification as before. Effectively this current composite output is the same as I had it from the RF modulator. Before I either get a more decent S-video display or Sophia 2, this is what I get. Albert in composite is the last picture. (I took pictures of other things, but it was too dark and the pictures do not show the colors right).

 

Otherwise, the computer still works, I have some more cleaning to do, there is still one large filthy patch on the bottom side of the board that I saw yesterday with right lightning and viewing angle. U1MB is in the mail, I more or less identified how I want to have it mounted, so I am getting closer to being able to seal this up

[Beeblebrox]

Well done. 

 

Here is where I mounted the U1mb on all my XE machines. 

 

 

 

The other upgrade is the aki USB keyboard interface.  This pic is my 800XE with the gold board. 

 

 

[woj]

1 hour ago, Beeblebrox said:

Well done. 

 

Here is where I mounted the U1mb on all my XE machines. 

 

 

 

The other upgrade is the aki USB keyboard interface.  This pic is my 800XE with the gold board. 

 

 

More or less what I intend to do, my goal is to mount it on two screws, but before I get the thing, I can't know exactly if that is possible (I know FJC does that, but he does that too close to the edge for my taste ;)) . 

 

Question: that Dell screen, is that a DVI/VGA monitor? How is it connected to the Atari? How well does it cope with PAL signal?

[Beeblebrox]

I mount mine on one pillar because I have the RF module to rest the pcb on.  With a plastic barrier underneath the U1mb pcb of course to stop it from shorting out. If I recall I used an existing screw hole. 

 

So the monitor is an LG Flatron M227WD 21.5" digital TV  with various inputs include S-video,  composite,  scart,  vga,  hdmi.  I have a Hercules double shielded S-video cable from Hercules workshop in Canada.  (Din to mini S-video) . Fantastic output. 

 

I had initially run with various displays with various upscalers and it worked to a degree but then I got this display and it can accept the 50hz pal signal. 

 

I also have a 10" old crt monitor with s-video which is great too. Finally I have a B&O 22" MX4000 crt which is amazing. 

 

These days I also have an old school Extron S-video amplifier which allows me to hook up both the LG and the 10" crt at the same time to an A8.  Means I don't need to unplug either and just have one on mostly. 

 

 

 

 

 

 

[woj]

4 minutes ago, Beeblebrox said:

So the monitor is an LG Flatron M227WD 21.5" digital TV

 

So what's that DELL screen doing displaying the U1MB menu?

[Beeblebrox]

12 minutes ago, woj said:

So what's that DELL screen doing displaying the U1MB menu?

Ah heh heh,  sorry,  you are right.  That was my old setup. I'd quickly posted that image whilst on my mobile and hadn't really clocked. that was the setup I had before. 

So it was a Dell ultrasharp 4:3 aspect ratio monitor 2007fdp.  It had svideo,  composite,  svga and dvi inputs if I recall. I bought it because of the svideo in.  However I soon realised that it couldn't pick up on the PAL 50hz signal. If I was running 60hz NTSC it would have just been able to pick up on it.

 

So at that point after much trial and error I had bought a retroscaler upscaler box with the A8 going into that via a din to composite cable,  then an hdmi to svga converter coming out of that and going into the dell's svga. I should point out I tried the dvi and composite in with the upscaler but had no joy. 

 

So here is an image of my current LG flatron and 10" crt setup which I described above, (with my windows gaming lappy in the middle that shares the LG display) Excuse the mess . 

 

 

 

 

[woj]

I see, I will sort out the video output problem I hope to my satisfaction at some point.

 

I finally got a chance to explore the computer a bit, once it's working. Interesting stuff surfacing out already. Awaiting U1MB, SysCheck allows me to run different OS roms, I tried the HSIO patched one to see what I can push out of the SIO system. It happily ran with divisor 0 with Lotharek's SIO2PC with most of the stuff I tried, even though all caps on the SIO lines in my computer are soldered in. So that's good I guess. What nearly stopped my heart though is loading the earlier mentioned HT cracked Blinky. It wouldn't load with high SIO speeds at all, it would load and start with lower speeds, but wrongly, the game starts messed up in a specific way. At that point I thought the computer is playing up again. But this was fully repeatable. I switched the rom setting on SysCheck to fully stock, slow loading, same thing. But with SysCheck unplugged and stock built-in OS it loads fine. Interesting, HT's Blinky does not like SysCheck, I wonder if it will like U1MB/Side3 combo... I know for sure it liked the cartridge based Turbo2000 system I used to have 30 years ago.

 

It also seems that my stock OS is a sort of rarity, the Self Test runs the memory test on the 130XE extended memory, I have not seen that before, and the stock OS from SysCheck does not do it. Note to self, once it is out again, make a dump of it with the EPROM reader. 

[Beeblebrox]

Great you can at last start to use the XE.

 

Interestingly I had issues with some game loading the other day on my 800 with SIO2PC with RespeQt. 

 

So with my Incognito'd 800 I can load the Rewind demo and various other atr and XEX games via the loader off the Side loader/CF card.  However when I tried to load the very same XEX files via SIO2PC/RespeQt they just either crashed or loaded but had corrupt graphics or the music was completely off. 

 

I'd not noticed any issue with SIO2PC before in this regard and no settings have been changed, so not sure what is going on there but it appears data is being corrupted. 

 

 

 

 

[woj]

I got very lucky in a second hand shop today, picked-up an Orion 22 inch LCD TV with all possible inputs and old enough to have a good quality for analog signals, all for the equivalent of 27€. Now my s-video output is stunning compared to my living room TV, colors are good, picture is sharp, and there is no frame sync loosing. 

 

I also now know that Simple S-Video Mod will do some more good things, jumping the R116 with tweezers already shows promising results. So I will be looking into that tomorrow.

 

There is of course an issue though, I get awfully moving mesh noise, but that (it seems) depends on brightness and/or color. The start screen of Self Test has it horrible, see the first video, the audio test red screen has none, the blue Basic screen somewhat less, see the second video (probably not visible). I already figured out that turning down the brightness on the TV reduces it considerably, on some screens fully (for example the ST front screen). So that would probably mean the luma signal having the noise. My question: anything obvious here that I should be looking at? (My video cable has shielding, but it can probably be better too).   

 

 

 

[xrbrevin]

nice find! cheapo TV is superior haha

it is advisable to isolate the composite and Svideo outputs from one another. if you only use one, then permanently cut off the other. altho if you prefer you can add a switch

[Beeblebrox]

Yup, so related to this I install a switch all the time on my 800XLs as I use S-video. It turns off composite when using S-video to stop it causing the horrible interferance on modern LCDs. It only happens on LCD displays.

 

See here: https://forums.atariage.com/topic/332130-source-for-5-pin-din-to-svideo-cable-for-a800/?do=findComment&comment=5032598

 

 

On the stock 800XLs you lift one side of the C54 resistor, create a switch which then allows you to turn off composite when using an S-video cable on a modern LCD screens.

 

Circuit open, C54 not engaged, jailbars/zigzigs gone, composite not availabe but great S-video output:

 

        

 

 

Circuit closed, C54 engaged, jailbars/zigzags return, composite circuit available again:

 

    

 

 

The hidden microswitch:

 

 

There will be instructions somewhere on AA for XEs on which resistor(s) you would need to lift in either case I guess if you wish to create a switch to turn off composite if using S-video and visa versa.

 

Whenever I get an 800XL the two things I always do are hook up chroma and install this switch. 

 

 

Edited September 24, 2022 by Beeblebrox