Debugging PAL Atari 800XL with scope

[Dr Memory]

We have verified that Vcc and GND are at reasonable levels, and that it is getting Phi 2 from Sally as expected.  I'm not sure what else to check on it.  I probed each of the signal pins (4, 6, 8, 10, 12) and they all have signals but I couldn't tell you if they are correct.

[+DrVenkman]

If you’ve verified the Reset line as Bob has recommended above, the next step is to check for activity on the address and data buses. The object here is not to read each line but just to ensure none of them are stuck high, low or floating. 

[Dr Memory]

4 hours ago, bob1200xl said:

Make sure that RESET is working properly. It should go inactive about 1 second or so after power on.

 

sequence:

Power ON

RESET is active (at ground) and goes inactive (plus)

6502 wakes up and clears all of memory, including the hardware chips (POKEY and such)

Speaker 'pops' meaning init was successful. Video should also reset to nada. Screen should be black.

 

Bob

I'm not 100% sure what you mean by this.  I monitored pin 40 on Sally during boot and it immediately goes to 5V, does not hang out at ground at all.  Or is that not what you meant?

[Rybags]

Might be normal - you could test by pressing the RESET key after power on (not sure about 1200XL but all other XL/XE OS variants have a software generated delay of a fraction of a second)

It generates a Reset signal and given the equipment you have you should be able to interrogate that.

Edited February 28, 2021 by Rybags

[Mazzspeed]

4 hours ago, Dr Memory said:

I'm not 100% sure what you mean by this.  I monitored pin 40 on Sally during boot and it immediately goes to 5V, does not hang out at ground at all.  Or is that not what you meant?

What sort of scope are you using, is it fast enough to catch the state of reset on boot? The fact it goes high indicates to me that it's not the problem, if it stayed low that would be a problem.

 

From memory pins 2,4 and 6 should also be high, if they're low or floating they may stop the CPU from running.

Edited February 28, 2021 by Mazzspeed

[TGB1718]

8 hours ago, Stephen said:

This wasn't really directed at you.  I am into vintage audio, and people will have you believe that if you don't just bulk replace every cap in every device more than 5 years old,

Agreed, I was given a Technics complete stacking Hi-Fi system circa. 1990 not working, usual "probably the Cap's are all shot"

Turned out to be the power amp module, replace it and it's been fine ever since, I just love the sound quality.

[Mazzspeed]

24 minutes ago, TGB1718 said:

Agreed, I was given a Technics complete stacking Hi-Fi system circa. 1990 not working, usual "probably the Cap's are all shot"

Turned out to be the power amp module, replace it and it's been fine ever since, I just love the sound quality.

But did the power amp module contain caps?

[TGB1718]

11 minutes ago, Mazzspeed said:

But did the power amp module contain caps?

No, the module was a large IC mounted on a heatsink, all the Cap's were on the main circuit board.

[+DrVenkman]

3 hours ago, Mazzspeed said:

What sort of scope are you using, is it fast enough to catch the state of reset on boot? The fact it goes high indicates to me that it's not the problem, if it stayed low that would be a problem.

The Atari reset signal is fairly slow - you can monitor it with a multimeter. The signal is active-low. If it’s stuck high and never pulses after power-on or when the Reset key is pressed, that is a problem. As Rybag suggests, having a scope or meter on the pin and then pressing Reset on the keyboard should trigger it to drop for a moment. The XL keyboard connector cable is not terribly long so this might be too awkward to do easily. Another thing the OP might do is simply short the Reset pin to ground momentarily and see if that changes the system behavior. 

[Faicuai]

On 2/27/2021 at 12:49 AM, Dr Memory said:

I'm trying to fix a PAL 800XL with the black screen of death.  Sometimes the red screen instead.  I've done the vast majority of things I've seen described in the various threads and youtube videos:

Failing socket-contacts and/or broken tracks on the MoBo are pervasive failures, usually hard to catch.

 

Check first for any signs of prior desoldering work, if any, and individually test THERE first. 

 

Then, assuming there is a contact failure somewhere, start by pressing (firmly but not crazy, with uniform pressure) with your thumb ALL active components, one by one, in sequence. Press, turn on, wait, then turn off and repeat for the next component. If there is ONE socket where contacts are failing (especially in MoBo) chances are you will find the culprit. If it os more than one, well, that would be a daunting task.

 

After this, you can now troubleshoot pjn-by-pin, signal-by-signal.

 

Good luck!

[+DrVenkman]

If you want to catch the Reset signal to see if that’s working, turn the system on and calibrate your scope with one of the clock signals you already know are good - Phi2 on pin 39 of SALLY for instance. Once you have a clean waveform, expand the time scale (compress the clock traces together) so that the traces look like a vertical smear. 10-20 microseconds is good. The Reset signal pulse is a lot longer than the clock pulses. Next stick your probe on SALLY pin 40 - you should see a fairly solid line at or near the top of your vertical scale at 4.9-5.0V with the system powered. The line might look a bit smudged due to noise in the ground plane or whatever but it should basically be steady. Next set your scope Trigger setting to Single and Falling edge - that means that the next time that signal you are measure it drops below the voltage level for the trigger you’ve selected, the scope will capture a trace. Now hit Reset on the keyboard and you should get something more or less like this. 
 

Precise values may vary a bit depending on the system but it should be a sharp drop to ground lasting around 80 microseconds, followed by a return to a baseline of approximately 5V. 

 

Edited February 28, 2021 by DrVenkman
typo

[Dr Memory]

9 hours ago, Rybags said:

Might be normal - you could test by pressing the RESET key after power on (not sure about 1200XL but all other XL/XE OS variants have a software generated delay of a fraction of a second)

I've been off trying people's suggestions, so I'll have a couple of small posts here.  Sorry, I couldn't figure out how to put more than one quote in a single post.

 

For this one, I plugged in a KEYCON, one of the cool toys I bought from tf_hh recently.  This allowed me to hit the XL reset key without the keyboard assembly flopping around.  I then watched Pin 40 on Sally with a multimeter and powered up.  Sure enough, it starts at ground and quickly goes to 5V.  It does the same when I hit the KEYCON Reset instead of power cycling.  Not sure why I didn't see it on the scope before - operator error I'm sure.

[Dr Memory]

9 hours ago, Mazzspeed said:

What sort of scope are you using, is it fast enough to catch the state of reset on boot? The fact it goes high indicates to me that it's not the problem, if it stayed low that would be a problem.

Rigol DS1054Z.  It's more a question of operator skill than scope speed here. 

[Dr Memory]

4 hours ago, Faicuai said:

Check first for any signs of prior desoldering work, if any, and individually test THERE first. 

 

Then, assuming there is a contact failure somewhere, start by pressing (firmly but not crazy, with uniform pressure) with your thumb ALL active components, one by one, in sequence. Press, turn on, wait, then turn off and repeat for the next component. If there is ONE socket where contacts are failing (especially in MoBo) chances are you will find the culprit. If it os more than one, well, that would be a daunting task.

 

Sadly, no sign of previous work.  At first I thought there might have been a couple of jumper wires added but I've since found several pics of PAL 800XL mother boards showing the same jumper wires, so I think they are stock.  No puddles of solder, no flux mess...

 

All chips have been reseated many times.  In fact everything socketed has been removed and reinserted, and most of them have been tried in other machines (where they worked fine).  This morning I replaced all the RAM chips with some newer RAM chips that I had bought off of ebay and tested.  No change.  Which makes sense, as I had previously run the original RAM chips through my Arduino-based DRAM tester.

 

Still black screen, rarely red screen instead.

[Dr Memory]

I've been studying the PAL section of the video circuitry.  I managed to find a set of redrawn schematics on the Internet, that included the things changed for PAL systems (which aren't covered by the Service Manual nor the Sams book).

 

One interesting this is that there is an extra IC!  U21 is listed as "not used" in the other books but on a PAL system it is indeed present.  It is a 74LS74, which is a "Dual Positive-Edge-Triggered D Flip-Flops with Preset, Clear and Complementary Outputs".  I've been looking at the pins of this with the scope.

 

I don't have a nice chart yet, but to summarize:

 

On the A half of the thingie, CLK 1 has 3.55 MHz, D1 has 1.77 MHz, and the outputs are 1.77 MHz.  Interestingly the inverted output loops back to become D, one of the inputs.  So I'm thinking this thing is dividing the frequency on the CLK pin by 2 and outputting a signal at that frequency.

 

On the B half of the thingie, CLK 2 has 1.77 MHz, D2 has 886 KHz, and the outputs are 886 KHz.  CLK 2 comes from the non-inverted output from the A side, Q.

 

So the overall effect is to take in a 3.55 MHz signal and put out a 886 KHz signal (by dividing by 4).  The 355 MHz signal comes from the NTSC circuit, which makes sense as it is the frequency of the other crystal.

 

This was probably a rabbit hole, as the numbers seem pretty good, but I had a lot of fun figuring it out.  I still don't get why they went to the trouble of generating that 886 KHz signal.  I see 1.77 MHz all over the place on the board but not that.  I also still don't get why Y2 shows 1.77 on one pin and 886 on the other - really didn't think crystal oscillators worked that way - but at least now I can see where the 886 originates!

 

Oh well, back to the drawing board.

[Dr Memory]

Breakthrough!  I finally worked up the courage to desolder the Delay Line from a working Atari 800XL.  I replaced it with a socket and plugged it back in, and my system still worked.  Phew!  I know this is old hat to many of you but the last time I did something similar, which would have been something like 35 years ago, I killed the computer I was working on.  If I remember right I was trying to install an Omnimon upgrade in a brand new 130XE, one of the first to hit the local store, and it did not survive.  So I've always been worried this sort of thing.  But this time it went well.

 

I then swapped the Delay Line chip from the sick 800XL to the working one, and the working one stopped working.  Black screen 90% of the time, red screen the rest.  So that's proven.  I swapped the known good chip into the sick 800XL and for the first time since it arrived from Moldavia (via ebay), it worked and gave me video.

 

I know a couple of you had pointed out this as a possible cause, but I didn't have any 800XLs with it socketed except for the one I was trying to repair.  Well, now I do, and that was a good suggestion.

 

Many thanks to those that helped.  I don't entirely get how that little chip with only a few pins and capacitors can be so critical to the system, but here we are.  Time to buy a another, or maybe one of the replacement projects.

 

Success!

[_The Doctor__]

Congrats,

after initial scan of  simple stuff like bad cables, wires, ribbon headers, cold joints, solder bridges, and jacks/ports/pins

the current cycle of what normally is dead by occurrence

dead memory

dead mylar

dead decoder chip (keyboard related)

delay line

power problem(voltage/current)

mmu

cpu

bad socket

bad OS(xegs for the most part)

bad ANTIC

support chips (ls etc.)

bad GTIA (mostly PAL machines)

bad pokey

bad pia

bad freddy

color pot

power switch

 

feel free to add you data and adjust the list

Edited March 1, 2021 by _The Doctor__

[bfollowell]

Congratulations!

 

I also have a dead 800XL that I've been working on without much luck. I worked on it a little bit, then the summer heat ran me out of my garage workshop. Now that's cooler, but not frigid, I've started tinkering a little bit. I just removed the MMU which had been a suggestion. I installed a socket for it and put the MMU from my dead 800XL into another that I have, but it works just fine. The delay line chip was going to be my next try. I know it isn't any of the big 40-pin LSI chips. I was comparing waveforms from the good XL to the bad and I found a couple of suspect areas that I want to dig into a little further.

 

Anyway, seeing you successfully get to the bottom of yours gives me hope and has been interested in working on mine again. Thanks for posting!