Finally got my XEGS to start, though lacks colour

[Macross_VF1]

After months of tinkering with an XEGS that only started to a black screen I've finally managed to get it to work!

I've replaced all the capacitors, transistors, both RAM chips, and the OS ROM and also desoldered all major chips in order to test them in other computers. In the end it turned out that the Freddie chip was the real culprit (last chip to be tested, obviously), though I'm fairly certain the RAM chips were faulty as well.

There is one niggle left however that I hope someone can help me with. Despite the XEGS now starting and running properly, it does not display any colour and I've tried both RF and composite. Any idea on what might be causing this?

 

[xrbrevin]

have you checked the associated components between GTIA pin 21 and the composite out?

 

[Paul Westphal]

Have you tried adjusting the color pot?

 

You didn't mention anything about that in your post.

Edited March 7, 2019 by Paul Westphal

[Macross_VF1]

have you checked the associated components between GTIA pin 21 and the composite out?

 

No I have not checked all the components, but with the the picture you've attached it will be much easier. Thanks!

 

 

Have you tried adjusting the color pot?

 

You didn't mention anything about that in your post.

I've tried it but it do not seem to do anything.

 

[xrbrevin]

HI,

here is the rest of the set - i assumed you already had it but well done for getting it going without it!

xegs.zip

[Macross_VF1]

HI,

here is the rest of the set - i assumed you already had it but well done for getting it going without it!

 

 

I'm pretty crap at reading diagrams but I am slowly getting better.

 

So I've checked pretty much every component that is between video out (both RF and composite) and the GTIA. I found one resistor (R80) that measures 1k Ohm instead of the described 2k and one diode (CR4) that might be faulty. I haven't been able to test the L5 coil.

 

Just to be sure, I desoldered the 4050 chip and tested it in a 130XE and could conclude that it's not the source of the problem.

[_The Doctor__]

I'd check everything around GTIA ANTIC, the crystal and transistor section of the board for bridges, breaks, and cold joints, make sure all pins are in their holes and you didn't damage any pads or traces..

 

 

double check monitor/tv cable as well...

[flashjazzcat]

Is the machine PAL or NTSC?

[Macross_VF1]

I'd check everything around GTIA ANTIC, the crystal and transistor section of the board for bridges, breaks, and cold joints, make sure all pins are in their holes and you didn't damage any pads or traces..

 

 

double check monitor/tv cable as well...

It does seem like I have a lot of checking to do...

 

 

 

Is the machine PAL or NTSC?

It's supposed to be a PAL console but this problem does seem like it could be an NTSC issue.

[Macross_VF1]

Here's pictures of the machine, let's determine which chips are NTSC or PAL, if any

Now since I have concluded that the XEGS Freddie is dead, I had to borrow one from the 130XE I own, and that's what the last picture is of.

[+Nezgar]

Alright, so that answers what I was wondering... (PAL vs NTSC)

CO14889 - PAL GTIA

CO21698 - PAL ANTIC

 

So they're both PAL, not mismatched PAL/NTSC. Nice work replacing all the sockets, but in the process of replacing them a trace or through hole might have been damaged... Back to checking other things as posted previously.

 

Have these two chips tested good in another system?

[1050]

I've tried it but it do not seem to do anything.

That might be the most telling symptom ever.

They always "do" something.

Put some Deoxit or other contact cleaner on it and

rotate it about while wet. Let dry and rotate some

more and then see if it does something now.

[Macross_VF1]

Alright, so that answers what I was wondering... (PAL vs NTSC)

CO14889 - PAL GTIA

CO21698 - PAL ANTIC

 

So they're both PAL, not mismatched PAL/NTSC. Nice work replacing all the sockets, but in the process of replacing them a trace or through hole might have been damaged... Back to checking other things as posted previously.

 

Have these two chips tested good in another system?

One at a time they have but just to be sure I swapped both with the chips from a 600XL. In the 600 the chips works great, while the 600 chips in the XEGS show the same problem, ie no colour. So it has to be something other than the chips.

 

 

That might be the most telling symptom ever.

They always "do" something.

Put some Deoxit or other contact cleaner on it and

rotate it about while wet. Let dry and rotate some

more and then see if it does something now.

Did just as you suggested and no difference. Nothing happens when I turn the dial, neither on RF nor composite. I've measured the pot and it gives the proper resistance and it is connected to the proper components so no broken traces. That said, there might be something connected to it that doesn't work as intended.

[1050]

I concur, that part of the circuit is dead and we don't

know why yet.

 

Might be worth replacing Q5?

Edited March 10, 2019 by 1050

[Macross_VF1]

I concur, that part of the circuit is dead and we don't

know why yet.

 

Might be worth replacing Q5?

At first I was ready to dismiss this advice as I had already replaced Q5. Then I started looking closer and discovered that there are TWO Q5 on the diagram. In fact when I take a deeper dive in the XEGS diagrams I was linked to earlier, I can see that there are several missing components compared to my machine. Could it be that my revision is different? It's a Rev A by the way.

 

 

HI,

here is the rest of the set - i assumed you already had it but well done for getting it going without it!

Could you supply me with a Rev A set of diagrams?

Edited March 10, 2019 by Macross_VF1

[flashjazzcat]

I don't like the soldering on the left leg of C48, and I see it is a non-standard part. Might we have an issue there?

[_The Doctor__]

you called it... looks nigh on empty and look at the trace leading to it.... that's something terrible... needs tlc for sure.

[flashjazzcat]

Unsure. It's just a cap to GND, but it has the potential to kill a signal if shorted, wrong value, etc. We'll see, hopefully.

[_The Doctor__]

You're like me in that respect, if it's dodgy, fix it up and/or replace it. easy to do and one less thing to worry about.

c53 lower leg looks to be a cold solder joint neat the crystal

 

empty holes near w2 empty near c40 as well

 

wonder what the bottom of the pcb looks like...

 

the color pot r13 is turned all the way down/ off in the photo... adjust to just shy of half way while it is on.. see if color comes and goes..

 

empty holes near c56 c57

 

touch up u2

 

verify orientation of electrolytic caps at top of board.. my eyesight isn't the best

Edited March 10, 2019 by _The Doctor__

[xrbrevin]

At first I was ready to dismiss this advice as I had already replaced Q5. Then I started looking closer and discovered that there are TWO Q5 on the diagram. In fact when I take a deeper dive in the XEGS diagrams I was linked to earlier, I can see that there are several missing components compared to my machine. Could it be that my revision is different? It's a Rev A by the way.

 

 

Could you supply me with a Rev A set of diagrams?

 

here is the source of the diagrams - hope you get a result

ftp://ftp.pigwa.net/stuff/mirror/jsobola/atarisch/

 

its a good site to wander around too

[1050]

Could it be that my revision is different?

Probably not. What we have is only one schematic by a

third party and he had a bad habit of putting the

wrong number on just about anything at random. So

combine that with the wishcasting of thinking this is

using the same numbers as the 130XE and is shades of the

800XL and 130XE duplicity where that train of thought is

valid, this creature however is another bag of worms entirely.

 

We are at the mercy of what we can figure out armed

with only the above knowledge. Don't believe the schematic

until you can PROVE it right. In the meantime trust your

eyes and verify, verify, verify.

 

So verification from an actual XEGS board shows the

proper suspect 3904 NPN is Q1 as silkscreened and quite

close to the color trim pot silkscreened R13. Cancel

previous advice for Q5 as per schematic and do Q1 as

per silkscreen instead.

[Mclaneinc]

I'm stunned it even runs with that many components missing..

 

Deffo the odd dry joint looking bits here and there as well..Some stuff is just flux but its been through the wars....

 

Best of luck retracing the missing stuff and replacing it all, hope its a winner after all that!

 

Paul..

[1050]

I've got more missing on my NTSC version of the same

board than he does and it works fine. As I said this

is a unique creature with on the fly updates Atari

didn't fill out when it wasn't necessary.

 

MIA here are:

C22

C23

C24

R28

R29

U10

C4

R70

R71

R72

C43

Q6

R73

L7

C51

C52

Y2

C53

 

I have jumper W2 where he is missing it and that would

be down to the PAL/NTSC difference. So I look at his

and see nothing too horribly wrong that's obvious.

 

I'm also missing R36 in the middle of the five devices

at the rear of Freddie. Due to an issue of difference

between the 130XE and the XEGS the resistor at the very

rear of Freddie is silkscreened underneath it as L8. But

it is instead green brown black resistor - 51 ohms.

 

One needs the 130XE schematic as well as Jerzy's to

figure out some of it, but it's not impossible and

all we can do anyway.

[flashjazzcat]

I eventually gave up looking for R70 (which has a strange solder blob under it in the photos) on Sobola's schematic when I noticed duplicates of Q5 and a number of resistors.

[Macross_VF1]

Time for a report. The sockets I've installed have been verified to be connected to the correct points. I've even tested that the chips make good connection with the sockets themselves. I have also spent some time looking over the entire bad for potentially bad solder joints and I did find a lot that I had to touch up.

You don't have to worry about the caps however, they were some of the first parts to be replaced and I quadruple checked the type, voltage, capacitance and polarity when I installed them. There are still two Non Polar ones I haven't replaced however since the replacements turned out to be about four times as large as the original ones.

Since we have concluded that the XEGS schematics are mostly incorrect, I've instead tried to manually follow the paths to and from the trimpot. So far I've identified that it connects directly to pin 17 on GTIA and through a few components to pin 3 on Sally. I don't know if that's useful information however. Should I perhaps try to replace it just to be sure it's not bad?

 

Probably not. What we have is only one schematic by a
third party and he had a bad habit of putting the
wrong number on just about anything at random. So
combine that with the wishcasting of thinking this is
using the same numbers as the 130XE and is shades of the
800XL and 130XE duplicity where that train of thought is
valid, this creature however is another bag of worms entirely.

We are at the mercy of what we can figure out armed
with only the above knowledge. Don't believe the schematic
until you can PROVE it right. In the meantime trust your
eyes and verify, verify, verify.

So verification from an actual XEGS board shows the
proper suspect 3904 NPN is Q1 as silkscreened and quite
close to the color trim pot silkscreened R13. Cancel
previous advice for Q5 as per schematic and do Q1 as
per silkscreen instead.

 

Here my normally excellent English knowledge fails me. I can't for the life of me understand what silkscreened mean in this instance. Anyway I've replaced Q1 and no improvement.