Jump to content

Photo

Finally got my XEGS to start, though lacks colour


65 replies to this topic

#26 _The Doctor__ OFFLINE  

_The Doctor__

    Flux Capacitor Master Craftsman

  • 6,744 posts
  • Location:10-0-11-00:02

Posted Mon Mar 11, 2019 7:00 PM

the lettering on the pcb that identifies components etc...   it is applied using a process whereby ink or another agent is pressed through a screen that once in history was made of silk. The screen is covered by a material that blocks the liquid ink paint what ever...  the lettering and graphics are cut out of the sealing material and removed, the screen is still there.. the the screen frame is pressed or laid over the pcb or t-shirt or whatever you want to apply the graphic to. It is held or clamped in place and the liquid is smeared and pushed throught the exposed area of the screen. You do this over and over very fast to mass produce a crap ton of t-shirts or chips locations, pin1, or components and their orientation.

 

that's the short version of it....


Edited by _The Doctor__, Mon Mar 11, 2019 7:00 PM.


#27 1050 OFFLINE  

1050

    Stargunner

  • 1,100 posts

Posted Mon Mar 11, 2019 11:15 PM

Don't think it's a language skill anyway, it's more
to what you've been accustomed to. A lifelong experience
with circuit boards would eventually have you knowing
that silkscreen means anything printed on a circuit
board such as the Revision number, Atari logo and
especially here, the component numbers in white ink
that were placed there by passing ink thru a silkscreening
process. Hence the word. That same process is used to
apply the mask where the copper is etched away from and
then even to paint the copper left behind with the green
fingernail polish giving the board it's overall look.

So the schematic isn't as bad as all that. It's just
a bit wrong and only sometimes. Most of the circuit
laid out as such would agree with what you can trace
out visually at your end. We have for an input signal
to base of Q1 Phase 1 of the system clock. It's used
mainly as an otherwise unused clock signal source to
drive the Q1 amplifier and the output to be voltage
doubled by CR2 and CR3 into a voltage higher than 5
volts as GTIA may need that much to yield the proper
colors. In the 800 this same circuit can do 12 volts.
This is the voltage adjusted by the R13 pot.

The secret to living with Jerzy's XEGS schematic is
to first understand and recognize that he just cut,
copy and pasted these from his 130XE series. So look
them up and notice that they are the very same there.
Further the 130XE stuff came from the 800XL stuff so
one could even have those schematics on hand too.

For example I know in one of the Sam's Photofact
schematic series there is a proper location look up
index by component number. FJC (FlashJazzCat) is
looking for resistor 70 so I look first to sams 130XE
but the table is not there. It is there for the sams
800XL schematic but R70 does not exist it would seem.
But there is a reason, it's only in the PAL 2nd
oscillator circuit which would not be the pictured
NTSC machine so no need to draw an arrow to it
when it's not there. Jerzy just leaves it out
entirely and that's why no find in his collection.

R38 is the color adjust pot on the 130XE as shown here
in the sams 130XE schematic.

try1.JPG

Atari schematics are at the end of the Field Service
Manuals from the site below.
Get the sams photofact series of schematics here:

http://www.atarimani...uments_3_8.html

Unfortunately they never produced one for the XEGS and
neither did Atari. Jerzy is all we have.

Of note in my fumblings around is that 130XE C45 is
XEGS C16. I had no reason to double check everything
so there may well be other pink elephants in the general
area. Please note that on the 130XE schematic above is
shown voltage levels. You should confirm those volts
on your machine which of course requires a voltmeter.
You should have one just to check your power supply
voltage and any power supply that you come into contact
with as a bad one outputting too much voltage will
kill your ram chips at the very least.

#28 Macross_VF1 OFFLINE  

Macross_VF1

    Star Raider

  • Topic Starter
  • 61 posts

Posted Tue Mar 12, 2019 12:26 PM

So that's what silkscreen means. Thanks for explaining it to me, much appreciated.

Anyway, back to the machine. I think I've identified all the components between pin 3 of Sally and pin 17 on GTIA ie the trimpot and its associated circuitry. 

C44 on the Atari 130XE diagram is C19 on my machine.
R45 do not seem to exist on my board.
C45 = C16 
R35 = R14 (obscured text, between CR2 and R15)
CR2 = CR2
CR3 = CR3
C46 = C14 
C47 = C15 (obscured text, between CR3 and R12)
R47 = R12
C48 do not seem to exist on my board.

That should hopefully make it slightly easier to figure out what's what. 

Of note is that I've checked some of the voltage levels displayed in the Sams schematics and it's not looking good.

Between Pin 3 of Sally and C44 I'm getting about 2.5V and should be 4.8 
The trimpot's range should be between 0-8.5V but is between 0-6.9V
I should be getting 5V between C45 and CR3 but instead I get 6.7V

Also of note is that R36 is marked as 680 Ohm in the schematic but is 680k Ohm on my machine. Might it possibly have been upscaled to compensate for the removal of R35? With R35 gone it's the only source of +5V for this part of the board. 

As for the PSU, I'm currently using a USB charger that provides a very steady 5V, instead of the brick PSU I had originally borrowed from my 600XL. I have pretty much ditched the old PSU after being told it's unreliable as fudge.
 



#29 Macross_VF1 OFFLINE  

Macross_VF1

    Star Raider

  • Topic Starter
  • 61 posts

Posted Tue Mar 12, 2019 12:27 PM


Double post, please disregard.


Edited by Macross_VF1, Tue Mar 12, 2019 12:28 PM.


#30 Macross_VF1 OFFLINE  

Macross_VF1

    Star Raider

  • Topic Starter
  • 61 posts

Posted Tue Mar 12, 2019 2:45 PM

Of all the boneheaded, stupid rookie mistakes one can make... Not only did I set my multimeter to the wrong setting I also put the common probe to the wrong place! 

Here's the updated and correct measurements. 

Between Pin 3 of Sally and C44 I'm getting about 1.6V and should be 4.8 
The trimpot's range should be between 0-8.5V but is between 0-7V
I should be getting 5V between C45 and CR3 but instead I get 5.5V
Also, directly after the collector leg on Q1 there's a 2.4V indicator. I don't know if it's relevant but the voltage I got from that point is -0.6V.



#31 ChildOfCv OFFLINE  

ChildOfCv

    Moonsweeper

  • 335 posts

Posted Tue Mar 12, 2019 4:20 PM

O1 is a clock, so it won't give you a constant voltage.  If you're taking a DC reading, the multimeter will likely give you an average of its on/off time unless you have a MAX function.  But all you need to check for working at the moment, is the voltage at both pins of C46, which ideally would be 10V, but practically 9V or better should be awesome (due to loss of voltage at the diodes).  R37 will decrease that a little at the trim pot.

 

You mentioned that C48 is dodgy.  I don't know exactly how the delay line works, but intuition says that the TIA (or GTIA) will discharge the capacitor when it's ready to start counting for output, then let the trim pot charge the capacitor until it reaches a threshold voltage, then start its internal quadrature oscillator.  So C48 must have a good connection to ground on one leg and a good connection to the trim pot and pin 17 on the other leg.  And of course the capacitor itself must not be internally shorted.

 

If you have any voltage at all at GTIA pin 17, you're probably good enough to get SOME color, even if it's the wrong color.

 

The color output goes from 21 to the other "Q5", and that's about the only thing blocking color output (unless that capacitor "C44" is bad or broken).  Did you verify whether there's any color output from pin 21?  Does "R54" have DC voltage across its pins?



#32 1050 OFFLINE  

1050

    Stargunner

  • 1,100 posts

Posted Tue Mar 12, 2019 7:12 PM

O.o What is your replacement transistor by exact number
as read off it? Who made it as well if known. Sometimes
they don't put the base in the middle which can lead to
another round of calling one's self unsuitable names.
Wrong setting will plague us all among many other
foolish things. It's ok - we are just going to grin a
little bit - any more than that is tempting our own fate.

 

Between Pin 3 of Sally and C44 I'm getting about 1.6V and should be 4.8 


Those are scope diagrams and are only relevant if you
are using a scope. A multimeter will be time averaging
that and giving a wrongful reading. point 6 is the
voltmeter reading on the base of Q1 that you should have.
 
 

The trimpot's range should be between 0-8.5V but is between 0-7V
I should be getting 5V between C45 and CR3 but instead I get 5.5V


That much seems to indicate it is working, but not
strongly enough perhaps?
 
 

Also, directly after the collector leg on Q1 there's a 2.4V indicator. I don't know if it's relevant but the voltage I got from that point is -0.6V.


-0.6 volt to me might be the base instead of the collector?
Replacement transistors will state they are suitable
but don't go into details of how they moved the base
from the middle leg and now it's opposite #1, the emitter.
As counted/read left to right.
 

Also of note is that R36 is marked as 680 Ohm in the schematic but is 680k Ohm on my machine. Might it possibly have been upscaled to compensate for the removal of R35? With R35 gone it's the only source of +5V for this part of the board.

This is an excellent find but with some confusion,
R35 is not missing, it's XEGS R13. But XEGS R14 at
680K is a thousand times too much and is certainly
keeping the circuit from generating adequate power
for the voltage doubler part. Blue, gray, brown for
blue, gray, yellow here too - good eye. You've repaid
me all I've invested in you already, thanks.

This is likely a typo error, somebody said K when they
shoulda shut up already and stopped at plain 680 on
the read back.

C48 MIA might not be an issue at all as that particular
value is the standard by-pass cap value and they are
scattered like rice at a wedding. Noisy chips in
particular are supposed to have one close to their
power supply to shunt noise spikes to ground and in
general its good practice to do that. But in this case
there may be no need or an engineer decided upon his
own that there was no need. Doesn't matter - add one
if you want one or leave it out. But for further
reading I like to point people to this paper while I
have a bag of .33 uf chip style at the ready.

Selecting Decoupling Capacitors for Atmels PLDs
Attached File  appnote0484.pdf   59.34KB   7 downloads

On a sad note, the original url for the above pdf is
a redirect to microchip which means Atmel has been
bought out.
Hold on, it's working now?
http://ww1.microchip...tes/DOC0484.PDF

http://www.atmel.com...ges/doc0484.pdf

Must be so recent a change they are working on it
actively, I swear it could not be found even searching
microchip for it. So now it's here too, and I've got a
copy salted away as well.

Other possible of interest revelation is the Blue ESR
meter that is effectively a low ohms digital meter that
is so accurate as to knock one's socks off. To be used
in the determination of a decent capacitor or one that
is about to go tits up and mess the bed too.
Search for Blue ESR meter on ebay - you can't miss it.

#33 Macross_VF1 OFFLINE  

Macross_VF1

    Star Raider

  • Topic Starter
  • 61 posts

Posted Wed Mar 13, 2019 4:14 PM

O.o What is your replacement transistor by exact number
as read off it? Who made it as well if known. Sometimes
they don't put the base in the middle which can lead to
another round of calling one's self unsuitable names.
Wrong setting will plague us all among many other
foolish things. It's ok - we are just going to grin a
little bit - any more than that is tempting our own fate.
 
-0.6 volt to me might be the base instead of the collector?
Replacement transistors will state they are suitable
but don't go into details of how they moved the base
from the middle leg and now it's opposite #1, the emitter.
As counted/read left to right.

 

Here's the thing with all the transistors. Before I started this topic I had already replaced the old ones. When we started to discuss the possibility of Q1 being faulty, I actually switched back to the original one. I had kept them just because I'm aware that the leg order sometimes change despite using supposedly perfect equivalents. The old Q1 is marked as 751 2N3904 and the replacement as 2N3904 R1 C. Unfortunately I don't know who made the new ones.

As for the 680k resistor, should I simply replace it with the correct one? Unfortunately I don't currently own a 680 ohm resistor, the closest one I have is 1k.

I don't know if this is in any way relevant but I checked all the diodes as there are only four of them. I don't know much of the inner workings of diodes but two of them gives of a strange reading. All of them gives 0.6V in one direction, which is pretty standard. CR4 and CR3 however gives 1.8V and 1.5V respectively in the other direction. 

 

O1 is a clock, so it won't give you a constant voltage.  If you're taking a DC reading, the multimeter will likely give you an average of its on/off time unless you have a MAX function.  But all you need to check for working at the moment, is the voltage at both pins of C46, which ideally would be 10V, but practically 9V or better should be awesome (due to loss of voltage at the diodes).  R37 will decrease that a little at the trim pot.

 

You mentioned that C48 is dodgy.  I don't know exactly how the delay line works, but intuition says that the TIA (or GTIA) will discharge the capacitor when it's ready to start counting for output, then let the trim pot charge the capacitor until it reaches a threshold voltage, then start its internal quadrature oscillator.  So C48 must have a good connection to ground on one leg and a good connection to the trim pot and pin 17 on the other leg.  And of course the capacitor itself must not be internally shorted.

 

If you have any voltage at all at GTIA pin 17, you're probably good enough to get SOME color, even if it's the wrong color.

 

The color output goes from 21 to the other "Q5", and that's about the only thing blocking color output (unless that capacitor "C44" is bad or broken).  Did you verify whether there's any color output from pin 21?  Does "R54" have DC voltage across its pins?

 

O1? I honestly can't find anything marked 01. I have checked C46, which gives 5V, unless you refer to C46 on my board, which would be C14 and that gives 7V.

I most certainly get voltage to GTIA pin 17 and it varies with the trimpot.

How do I verify if I get any colour from pin 21? R54 gives 0,23V

 



#34 ChildOfCv OFFLINE  

ChildOfCv

    Moonsweeper

  • 335 posts

Posted Wed Mar 13, 2019 5:02 PM

O1? I honestly can't find anything marked 01. I have checked C46, which gives 5V, unless you refer to C46 on my board, which would be C14 and that gives 7V.

I most certainly get voltage to GTIA pin 17 and it varies with the trimpot.

How do I verify if I get any colour from pin 21? R54 gives 0,23V

 

O1 is the output from pin 3 of Sally on the schematic, into "C44".  It's  going to be a clock signal.  The proper designation is φ1 or ϕ1 depending on who's writing it, but try to do either with an American keyboard....   Anyway, it's constantly flipping between 0 and +VH so your meter will always see something in the middle on a DC reading.  Or an AC reading for that matter.  (VH is the voltage output of a high, which depends on logic technology... may be about 5V, may be about 3.4, etc).  And yeah, I am placing the component designators in quotes due to the mentioned inaccuracy of the schematics.  By the way, do you get AC voltage on pin 17?

 

Does an AC mode measurement of "R54" give you any appreciable voltage?  0.23V (DC, I assume?) is really low and may even be bleed-off from the luma signal.  If the 130XE is a good model, that should be closer to 2V (and looking at the schematics, they are pretty much identical there).  Note that the voltage across "R54" should be just a little less than the voltage on the base leg of "Q5 east".  Both AC and DC readings of both of those components might shed some light on the problem source (GTIA or "Q5 east")... that is, if it's being told to turn on but it isn't, vs not even being told to turn on.



#35 ChildOfCv OFFLINE  

ChildOfCv

    Moonsweeper

  • 335 posts

Posted Wed Mar 13, 2019 5:05 PM

 

As for the 680k resistor, should I simply replace it with the correct one? Unfortunately I don't currently own a 680 ohm resistor, the closest one I have is 1k.

Even 1K is a major improvement.  Should probably get you a lot closer to the 8.5V expected.



#36 Macross_VF1 OFFLINE  

Macross_VF1

    Star Raider

  • Topic Starter
  • 61 posts

Posted Thu Mar 14, 2019 9:40 AM

O1 is the output from pin 3 of Sally on the schematic, into "C44".  It's  going to be a clock signal.  The proper designation is φ1 or ϕ1 depending on who's writing it, but try to do either with an American keyboard....   Anyway, it's constantly flipping between 0 and +VH so your meter will always see something in the middle on a DC reading.  Or an AC reading for that matter.  (VH is the voltage output of a high, which depends on logic technology... may be about 5V, may be about 3.4, etc).  And yeah, I am placing the component designators in quotes due to the mentioned inaccuracy of the schematics.  By the way, do you get AC voltage on pin 17?

 

Does an AC mode measurement of "R54" give you any appreciable voltage?  0.23V (DC, I assume?) is really low and may even be bleed-off from the luma signal.  If the 130XE is a good model, that should be closer to 2V (and looking at the schematics, they are pretty much identical there).  Note that the voltage across "R54" should be just a little less than the voltage on the base leg of "Q5 east".  Both AC and DC readings of both of those components might shed some light on the problem source (GTIA or "Q5 east")... that is, if it's being told to turn on but it isn't, vs not even being told to turn on.

I've done some further measurements based on your suggestions though I admit that I'm not well versed with AC voltage. I pretty much only get 0.01V AC no matter where I take my readings though it just might be that I don't know how to do it properly.

Pin 3 of Sally gives 1.5V DC
 

Q5 Collector - 5V DC
Base - 0.73V DC
Emitter - 0.23V DC


Edited by Macross_VF1, Thu Mar 14, 2019 9:43 AM.


#37 ChildOfCv OFFLINE  

ChildOfCv

    Moonsweeper

  • 335 posts

Posted Thu Mar 14, 2019 11:08 AM

Yeah, taking an AC voltage reading is the same basic process as taking a DC voltage reading.  So getting close to 0 means that not much is happening at that location.

Q5 Collector - 5V DC
Base - 0.73V DC
Emitter - 0.23V DC

So Q5 is doing what it's supposed to do then.  But GTIA isn't sending color.

 

When you were swapping chips, did you check that the GTIA worked in the other system?

 

And JOOC, is there an AC voltage on pin 17 of GTIA in your working system?

 

For that matter, (it's not common for passive components to fail, but considering what else has happened to this board, apparently...)  check "R87" and "R55".  Both the DC voltages and the resistances.  Also make sure "L5" has almost no resistance across it.


Edited by ChildOfCv, Thu Mar 14, 2019 11:26 AM.


#38 Macross_VF1 OFFLINE  

Macross_VF1

    Star Raider

  • Topic Starter
  • 61 posts

Posted Thu Mar 14, 2019 12:47 PM

I have at least one bit of good news, I've received my new Freddie chip so I no longer need to borrow one from my 130XE.

All the chips I've desoldered (GTIA, PIA, Sally, ANTIC, Freddie and 4050) have been tested successfully in other computers. The only major chip I haven't desoldered is Pokey but that chip really shouldn't be an issue here.

It turns out that Jerzy's schematic of the XEGS is pretty darned accurate when it comes to the video output from GTIA so I've used it to visualize my findings.

Red are resistors with a faulty value, though I haven't checked yet if the rings on the resistors match the value I'm getting from my multimeter.
Yellow is just a typo, the resistor is named R78 instead of R53.
Blue are the voltage readings I get from all the transistors
Green are other voltage readings, DC if nothing else is written.
I also used red to show the resistance across L5.

 

T8hcBKk.jpg



#39 flashjazzcat OFFLINE  

flashjazzcat

    Quadrunner

  • 14,498 posts
  • Location:United Kingdom

Posted Thu Mar 14, 2019 1:05 PM

If you're testing these resistors in circuit I would not necessarily be surprised to see low readings.



#40 ChildOfCv OFFLINE  

ChildOfCv

    Moonsweeper

  • 335 posts

Posted Thu Mar 14, 2019 1:14 PM

+1 on flashjazzcat.  Also, resistors have a tolerance spec of some percentage from the stated reading (that's what the red/gold/silver/brown band at the end is).  In short, if the actual reading is within a percentage of the resistor's stated value (based on the final color) then it's okay.

 

Anyway the resistance from +5C to Q5 base should be about 6.6K.  If so, then that's apparently good.

 

Now, on the PAL pin...  On the 130XE, this pin is connected to pin 20 28 on NTSC versions, or a separate PAL oscillator on PAL versions.  So that answers what that pin is for:  The color clock.  I think you determined that yours is a PAL version, right?  Anyway, proper color depends on getting clocks on GTIA pin 16 too.  If you take a DC reading, it ought to be in the same mid-range you saw on Sally 3.  An AC reading should be significant too.


Edited by ChildOfCv, Thu Mar 14, 2019 1:16 PM.


#41 Macross_VF1 OFFLINE  

Macross_VF1

    Star Raider

  • Topic Starter
  • 61 posts

Posted Thu Mar 14, 2019 3:59 PM

+1 on flashjazzcat.  Also, resistors have a tolerance spec of some percentage from the stated reading (that's what the red/gold/silver/brown band at the end is).  In short, if the actual reading is within a percentage of the resistor's stated value (based on the final color) then it's okay.

 

Anyway the resistance from +5C to Q5 base should be about 6.6K.  If so, then that's apparently good.

 

Now, on the PAL pin...  On the 130XE, this pin is connected to pin 20 28 on NTSC versions, or a separate PAL oscillator on PAL versions.  So that answers what that pin is for:  The color clock.  I think you determined that yours is a PAL version, right?  Anyway, proper color depends on getting clocks on GTIA pin 16 too.  If you take a DC reading, it ought to be in the same mid-range you saw on Sally 3.  An AC reading should be significant too.

 

 

In most cases I certainly agree with you on the resistors. The only one I would find any major fault with is R80 and even then, I don't think it's serious. I just figured it would be better to include the values, instead of leaving them out. 

Speaking of resistors. I've changed the 680k resistor to a 1k and the voltage across the trimpot now go from 0-8.7V. Sadly no colour despite that bit of improvement.

 

Resistance from +5C to Q5 base is indeed 6.6k Ohm. Spot on.

Oh yes, this is most certainly a PAL machine. Pin 16 gives 4.5V DC and the same low 0.01V AC. I saw that Q6 is also closely connected to pin 16 so I figured I should take those readings as well. 

Q6 C 4.5V
Q6 B 0.2V
Q6 E 0V

Also part of this particular corner of board is Y2 which I suppose is the oscillator you mention. Any tips on how I should measure it?


Edited by Macross_VF1, Thu Mar 14, 2019 4:01 PM.


#42 ChildOfCv OFFLINE  

ChildOfCv

    Moonsweeper

  • 335 posts

Posted Thu Mar 14, 2019 4:21 PM


Oh yes, this is most certainly a PAL machine. Pin 16 gives 4.5V DC and the same low 0.01V AC. I saw that Q6 is also closely connected to pin 16 so I figured I should take those readings as well. 

Q6 C 4.5V
Q6 B 0.2V
Q6 E 0V

Also part of this particular corner of board is Y2 which I suppose is the oscillator you mention. Any tips on how I should measure it?

Okay, now we're on to something.  No oscillator = no color.  Of course we're also off the schematic since the one posted doesn't even seem to have this circuit.  So, I'll refer to the 130XE schematic.

 

So, the transistor Q6 you're referring to has three resistors connected to the base.  The schematic for the 130 labels these as R71, R70, and R69.  Look for the equivalents on your board.

 

R70 is 10K and should have your 4.7V at one side and *should* have close to 2V on the other side (which is the base of Q6).  R70 also connects to the collector.

 

R69 is the voltage divider to ground and is 6.8K, so 2V on the base side and 0V on the ground side.

 

R71 is the input from the frequency divider and crystal and should be 470.  It ought to be oscillating on at least one side.

 

Well, actually, Q6 base to emitter is a diode to ground, so it ought to be around 0.5-0.7V.  0.2V is still too low, but you won't get 2V due to the base to ground diode.  But anyway, see if at least one side of the 470 resistor has AC input.  That side will likely be close to 0V DC.

 

 

I also might as well mention U21, or whatever it will be on your board.  It's a 74LS74 that feeds your crystal with a frequency division of the main clock.  Pin 9 will be the output that goes through the crystal on its way to Q6.  Make sure that pin has AC output too, and will be around 1.5VDC if it's working.


Edited by ChildOfCv, Thu Mar 14, 2019 4:28 PM.


#43 ChildOfCv OFFLINE  

ChildOfCv

    Moonsweeper

  • 335 posts

Posted Thu Mar 14, 2019 5:49 PM

Thinking about this some more, voltage difference between base and emitter should be between .5 and .7V when positive.  And when it's positive, that means the transistor is turned on, so the voltage at the collector ought to be a lot lower too.  So unless there's something I'm forgetting here, (and someone feel free to point it out if so), it appears that Q6 is both shorted out and dead.  Well, this is predicated on the notion that R70 is connected correctly.


Edited by ChildOfCv, Thu Mar 14, 2019 5:50 PM.


#44 1050 OFFLINE  

1050

    Stargunner

  • 1,100 posts

Posted Fri Mar 15, 2019 9:02 AM

Sorry about the absence, the bombogenesis ate my homework.
Then stole my power the other night and most of the
following day too.
But we all survived anyway.

PAL Color clock.JPG

Y2 is the PAL crystal itself, no real way to measure it,
but they rarely are a problem. XEGS U10 is XE/XL U21.

XEGS U10 signal source is 14.8757 MHz clock
signal and divides that by 4 to tickle the PAL color
clock crystal Y2 at 3.7 MHz which is enough to ring
it like a bell such that it oscillates at 4.433618
MHz of the crystal in spite of that mismatch.

The half of the crystal circuit that involves Q6 is
feeding GTIA pin 16 directly. Two active components
that can go bad more often than passive components
are then Q6 and XEGS U10 which is a LS7474 dual
JK Flip-Flop configured to divide a clock's frequency
by 2 twice which is same as 4. Next on my suspicion
list would be XEGS L7 (XE/XL L8) either shorted or
open would render the circuit very much challenged.

Forgotten part is bias resistor network (R69, R70)
for Q6 should have it at 1.6 volts at the base,
disregarding power consumption across R72 AND an
active working Q6. We don't have that so Q6 is working
and it's base voltage isn't half of the time based
1.6 volts the bias network would have it at when
turned off in the 50% duty cycle we know it has.
It's not oscillating, as soon as it does, base voltage
rises to .8 or so.

I think my money is on a hinky LS7474 out to lunch.

Of note is that R72 is 1K and powering yet another
3904 with a grounded emitter. Just like the other
circuit is now set up as, I like it.

I have the feeling we are about to strike paydirt
finally.

#45 flashjazzcat OFFLINE  

flashjazzcat

    Quadrunner

  • 14,498 posts
  • Location:United Kingdom

Posted Fri Mar 15, 2019 10:27 AM

I've seen a couple of machines with a dead crystal, one with a bad transistor on this circuit, and one with a bad LS74.



#46 Macross_VF1 OFFLINE  

Macross_VF1

    Star Raider

  • Topic Starter
  • 61 posts

Posted Fri Mar 15, 2019 11:23 AM

Paydirt indeed!

Since you guys suggested that U10 (LS7474) might be a possible culprit I promptly desoldered it and tested it in my 600XL as every chip is socketed in that machine. And sure enough, no colour when using my 600XL with the XEGS LS7474! Finally!

Unfortunately I don't have a 14 pin socket to put in my XEGS though considering I have to order a new LS7474 anyway, that's not much of a problem.

While I feel it's unlikely that this particular unit will have more challenges to present to us as just about every part of it has been scrutinized under a magnifying glass and/or been replaced, I will hold my breath until I have the new parts. Still, it has been a pleasure to work with you, and I've learned more about troubleshooting and reading schematics than all of my previous 37 years of life combined. I'll be sure to return if I have any other problems.



#47 ChildOfCv OFFLINE  

ChildOfCv

    Moonsweeper

  • 335 posts

Posted Fri Mar 15, 2019 11:45 AM

Forgotten part is bias resistor network (R69, R70)
for Q6 should have it at 1.6 volts at the base,
disregarding power consumption across R72 AND an
active working Q6. We don't have that so Q6 is working
and it's base voltage isn't half of the time based
1.6 volts the bias network would have it at when
turned off in the 50% duty cycle we know it has.
It's not oscillating, as soon as it does, base voltage
rises to .8 or so.

As an exercise to try to reawaken those old semiconductor class brain cells, I did a DC analysis on the Q6 part of the circuit and calculated about 2V at the collector and 0.7 at the base, when not oscillating, of course.  I also found one of those "free" online simulator sites and tried it there with similar results (yay for me I guess).  So it still seems that there is something wrong with the circuit around Q6 if it's reading 4.7V at the collector, and 0.2V at the base.  It's confirmed now that U10 is dead, but I still wonder if Q6 contributed to the murder.



#48 1050 OFFLINE  

1050

    Stargunner

  • 1,100 posts

Posted Fri Mar 15, 2019 1:47 PM

My numbers were just shooting from the hip and I'm not
surprised to be way wrong doing that either. Jon tells
us of dead crystals which I wouldn't have thought very
possible. And my bad on the part number as my grey cells
failed me it's 74LS74, I knew there was an extra 74 in
there somewhere that was getting left off for convenience.
And shows how my faulty brain works too.

As to a guilty party, I want to lay this at the feet of
Atari. There is no good reason at the time this was put
together to have bad silicon around so they were buying
marginal cast offs that chip manufacturers had an unusual
level of failures with and were offered at a discount rate.
Atari probably had to then test them individually not
really caring that a defect might show up later even with
the ones that passed as working at that time. We have ROM
failures of an extraordinary rate with the XEGS too, again
Atari shopping from the trash bin can be the only real
excuse in my way of thinking. I've never heard of a crystal
going bad, didn't know it was a thing and Jon's repair
skills are top of the line so I have no choice but to buy
his suggestion at face value. Leave it to Atari to stock
up with junk for parts at assembly time. Even with a 680K
resistor where something much less used to be, the vast
majority are still working today with no issues, it's
not all bad news. I also enjoyed helping on this one and
learned new things from new people too, thanks to all.

#49 ChildOfCv OFFLINE  

ChildOfCv

    Moonsweeper

  • 335 posts

Posted Fri Mar 15, 2019 10:50 PM

...and the plot thickens.  I redid the DC analysis out of curiosity, assuming that the other side of R71 is shorted to ground.  I got 0.2V at the base of Q6... just as Macross measured.  With the base voltage this low, the transistor would be off, so that leaves about 4.56V at the collector (which is also close to what he measured).  This is significant because it should be nothing but capacitors between R71 and ground, which means there's no normal way for that short to be there.

 

Could you take a resistance measurement from that 470-ohm resistor (other side from Q6 base) to ground to confirm whether or not there's a short here?  If so, "C61" is suspect.  On the good side, that would also mean that Q6 is good.


Edited by ChildOfCv, Fri Mar 15, 2019 10:52 PM.


#50 Macross_VF1 OFFLINE  

Macross_VF1

    Star Raider

  • Topic Starter
  • 61 posts

Posted Sat Mar 16, 2019 10:48 AM

...and the plot thickens.  I redid the DC analysis out of curiosity, assuming that the other side of R71 is shorted to ground.  I got 0.2V at the base of Q6... just as Macross measured.  With the base voltage this low, the transistor would be off, so that leaves about 4.56V at the collector (which is also close to what he measured).  This is significant because it should be nothing but capacitors between R71 and ground, which means there's no normal way for that short to be there.

 

Could you take a resistance measurement from that 470-ohm resistor (other side from Q6 base) to ground to confirm whether or not there's a short here?  If so, "C61" is suspect.  On the good side, that would also mean that Q6 is good.

 

 

Ok, so this turned out to be a lot more difficult than I had first anticipated. It seems that my XEGS is vastly different around Q6 than compared to the schematics. I've tried to do my own diagram, though I fully admit that I might have missed something. 

*Edit* Just saw that I missed marking which leg is the emitter on Q6. It's the leg connected to ground.

FXVNmQ7.jpg

 


Edited by Macross_VF1, Sat Mar 16, 2019 10:49 AM.





0 user(s) are browsing this forum

0 members, 0 guests, 0 anonymous users