1200XL - boots to red screen

[toddtmw]

I'm really glad you figured it out!

[+DrVenkman]

Okay, update on my Ugly Duckling -

 

At Bob's suggestion earlier, I measured the voltage rise on C7 ... it goes from 0 to about 4.7V over several seconds. So that didn't really seem to be the issue. After everything last weekend, I decided to try swapping chips from my donor 800XL and 600XL. First chip I tried was the 6502C and whadday know - booted right up to the rainbow logo.

 

However, something is still not quite right. If I insert Star Raiders, I hear the sounds of the engines but until I press Select or Option, no graphics appear. If I boot to the logo and press help, the image freezes on solid colored "ATARI" (usually red or reddish-orange though I did once see green) and never actually goes into the self-test screen. I'm thinking OS roms or MMU. Sadly, the only examples I have of each to swap are now running happily in the "pretty" machine and I am loathe to tamper with what's working in the absence of further spare parts in case of mishap.

 

So anyone have suggestions?

Also: I'm peroxide-treating the case this weekend. These are the results on the top after a couple hours of late-afternoon sun and 40% liquid. Tomorrow I'm going to re-do the edges and spot-treat a few small places on top with cream. But for now, wow.

 

 

 

 

[ACML]

I had a similar problem with Star Raiders and it was a RAM timing issue. Try replacing the RAM chips one at a time. Start with the one on the left hand side and work your way to the right. Its usually the the first one. Take one from the 800XL. If it works, the old one will probably still work fine in the 800XL.

[+DrVenkman]

I had a similar problem with Star Raiders and it was a RAM timing issue. Try replacing the RAM chips one at a time. Start with the one on the left hand side and work your way to the right. Its usually the the first one. Take one from the 800XL. If it works, the old one will probably still work fine in the 800XL.

 

Good suggestion, thanks. Do you think that would also explain the fact that it crashes on trying to enter the self-test screen? I would think that might be OS or MMU, no?

[+DrVenkman]

Okay update again on the Ugly Duckling - I tried swapping out the ram chip by chip but no luck. I then swapped out some of the 74xxx logic chips just to check - 74LS08 first, and a few others. No dice. I also swapped out ANTIC, GTIA, PIA and POKEY, still no dice. Then for giggles re-flowed the solder on the main IC’s, the MMU, and OS roms. Also no dice.

I think that leaves me pretty much down to OS roms or MMU. Ugh. Gonna have to break down and add them to my list to track down after all. If I leave it on the rainbow logo, after a few minutes, little glitches start appearing on-screen and eventually it crashes. The only chips more than barely warm are the GTIA and 6502 (fairly hot, each, but both work fine in other systems) and the MMU (which is scorching - literally too hot to touch for more than a second or so).

 

But in better news, my case whitening is going great and Jon has promised to send me a spare acrylic to go over the LEDs. So the Ugly Duckling may yet turn into a swan someday.

[flashjazzcat]

I promise to box that up tomorrow!

[+DrVenkman]

More experimenting tonight on the Ugly Duckling ... BASIC boots up and shows the correct amount of free RAM. Action! boots (it's an AtariMax repro but still a 16K cart - don't know if it has the oddball OSS bank-switching scheme) but will show graphical glitches, such as inverse space (white block) somewhere down the screen. Joust boots but has immediately visible and obvious graphical glitches on the title screen and next screen (which shows difficulty, number of players, and copyright info) is corrupted and almost illegible. Attract mode for the game will start but crash almost immediately with discoloration on-screen. I'd almost think this was a bad ANTIC but I pulled the ANTIC from another working donor system and I had the same kind of problems.

 

Still leaning toward MMU and/or OS roms unless anyone has a better idea.

[DavidMil]

 

Nope, not really. The 6502 was pretty warm, the POKEY and ANTIC somewhat warm. The MMU was pretty scorching but as Bob says, I guess they seem to run hot. The GTIA was about as cool as the PIA, RAM or any of the random 74xxx logic chips.

 

And now after replacing the GTIA, the machine has been running over 6 hours straight without a problem.

 

Did Atari every put out a Repair Flow Chart for the 1200XL? All I've ever seen is a 'System Check List'.

 

DavidMil

[+DrVenkman]

Well, there is a Field Service Manual but I’ve never studied it in great detail. I was looking at it to see the schematics but they’re frankly terrible. Poorly drawn, badly reproduced ... almost easier to just trace connections on the actual board.

[+DrVenkman]

So, interesting update on my Ugly Duckling 1200XL ...

 

Never one content to leave well enough alone, I decided to examine the machine closely again tonight, under bright light and magnification. I have done this the last couple weekends but fresh eyes never hurt. When that revealed nothing, I decided to look at the IC sockets just sort of on a hunch. Upon removing the 6502, I noticed what looked initially like a bit of dark debris or dirt on one of the socket wipes (pin 19, for what it's worth). This machine was filthy when I got it so I was unsurprised that there might be dirt in an obscure place. So I put on my magnifier headset and probed the apparent debris bit with a jeweler's screwdriver - alas, the dark "debris" was apparently corrosion; the wipe broke clean off and dropped into the hole.

 

 

I've never replaced a socket before. I gather the easiest approach is to cut as much of the plastic away as possible, then desolder the pins, much like removing a failed IC. Any tips or pointers before I tackle this tomorrow?

[+bob1200xl]

Cut out the plastic cross-bars. Lift off the pin guide with a screwdriver under the bar. Do not damage the PCB when you do this - keep the blade flat against the joint. You should have all naked pins once you do this, that may be removed one at a time.

 

Bob

[+DrVenkman]

Well that was fun. Whee!

 

 

And done. Mostly. I still need to clean the board and poke out a few stubborn fragments that got stuck but ... progress!

 

[+DrVenkman]

Okay, socket replaced. CPU reinserted. It works!

 

Bad news is it still crashes on self-test. Thanks to Jon’s generosity, I have a new MMU and a set of ROMs coming my way, along with the acrylic for the case. Hopefully that will restore the Ugly Duckling to full functionality.

[+DrVenkman]

So, the good news is I got Jon's package today, safe and sound.

​The bad news is that neither swapping MMU nor OS roms invididually resolves the issue.

 

I've got a logic probe being delivered tomorrow. I'm going to have to see if I can narrow down what's going on by comparison with my working machine. If that doesn't work, well ... I know someone who claims to be a shaman. Perhaps having some chakras realigned or some sort of ritual sacrifice might work, lol.

[+DrVenkman]

So my Ugly Duckling is turning into a REAL head-scratcher. I've now swapped every IC on the board with examples from otherwise working machines including all the RAM individually one-by-one, all the 74xxx logic, the OS ROMs, the MMU and every main IC (POKEY, 6502, 6520, ANTIC and GTIA). Hell, I even swapped the 4050 hex buffer on the video circuit just for grins. The only thing I didn't swap was the CO60472 "XL Delay" component, as it's soldered to the board. Still no change:

 

1. It boots; without a cartridge, it goes to the rainbow logo. However, I noticed that booting has some odd distortion in the video (see the photo of my testing TV below - the "VIDEO" overlay from my TV is distorted; booting with my normal machines doesn't do that to the overlay).

 

 

2. Pressing HELP at the logo does not bring up self-test. Rather, the logo flashes a single color (usually blue, sometimes orange or pale green), then goes to solid muddy red. System Reset works and returns to the rainbow logo.

 

3. Some carts seem to boot okay, at least initially. Star Raiders usually boots with audio but no video until you either change the difficulty or start a game. Then video shows and the game superficially seems to work. BASIC boots and shows the expected 37902 bytes free.

 

3. Atariwriter does NOT work. See the screenshot below. Usually the distortion is not as bad as this but sometimes it is. Sometimes the title screen appears normal then before it goes to the menu, I get an all-blue screen and then it goes black. Sometimes the title screen has a few random characters on it before going blue and blacking out. Of course, Atariwriter works fine on my other machines, including my other 1200XL.

 

 

I don't think this is an 8K vs. 16K cart issue, as Joust seems to work, as do Pole Position and Final Legacy.

 

Anyone know enough about Atariwriter's internals to know why it cases such obvious symptoms while the others are more subtle if they show at all?

 

I would think this is an ANTIC or GTIA problem but I've swapped both those from working machines to test, and no change.

 

I've just today gotten a basic logic analyzer. Are there any obvious candidates anyone can think of to check for stuck or incorrect signals?

[+Stephen]

The only time I've seen something like the above image of AtariWriter, it was a bad ANTIC. I'm the farthest thing from a hardware guy though. I hope you are able to get this machine working.

[toddtmw]

If you've replaced all the replaceable parts, doesn't that narrow it down to a motherboard or solder issue?

 

That Atari writer screenshot feels like a memory issue to me. It's like the app is building the character set but it's not "taking".

 

What if you swapped ALL the memory chips? Maybe more than one is bad?

[toddtmw]

Maybe build one of these?

 

http://hackaday.com/2016/03/18/testing-dram-one-byte-at-a-time/

[+DrVenkman]

Maybe build one of these?

 

http://hackaday.com/2016/03/18/testing-dram-one-byte-at-a-time/

 

Interesting but I've never used an Arduino of any flavor before. I'd have no idea how to build something with one. I'm kind of astonished there isn't some kind of commercial DRAM tester already in existence - do any of the USB EPROM burners out there do DRAM testing?

 

EDITED TO ADD: I just realized I have an UNO Cart - assuming this machine will boot to the menu (not guaranteed!) would the SuperSALT cartridge help narrow down the issues? I've never used it before.

 

EDITED AGAIN: Rats. The UNO Cart menu boots up and I can select the rom, but when the cart image boots I first see a pattern of gray bars on the screen, followed a few seconds later by a badly scrambled screen full of random characters and pieces of the main menu text.

[toddtmw]

It's really easy to do the Arduino thing. This is a very simple project. You need a breadboard and some jumper wires and a couple of LEDs and an UNO.

 

Breadboard with jumpers:

https://www.amazon.com/microtivity-400-point-Experiment-Breadboard-Jumper/dp/B004RXKWDQ/ref=sr_1_13?ie=UTF8&qid=1498514350&sr=8-13&keywords=breadboard

 

UNO (I've seen these for a couple dollars less on sale at MicroCenter, but not sure if you have one near you):

https://www.amazon.com/HiLetgo-ATmega328P-Development-Microcontroller-Compatible/dp/B01AR7YJ3O/ref=sr_1_5?ie=UTF8&qid=1498514503&sr=8-5&keywords=arduino+uno

 

LED:

I'm sure you know where to get LEDs...

 

Then look at this schematic:

https://github.com/FozzTexx/DRAM-Tester/blob/master/schematic.jpg

 

It shows which pins on the chip need to be jumpered to which pins on the arduino.

It doesn't show the LED's. They need to be connected to pin 4 (status) and pin 19 (done) on the arduino. Just put the + leg on one of the rows on the breadboard and the - leg on the outside ground row and then run pin 9 and pin 14 to the same rows as the + side of the LED and run a jumper from ground on the arduino and to the outside ground row on the breadboard,

 

It should look something like this:

 

The chip you want to test will straddle the middle line right up against the top. The Author used a ZIF socket. You should be able to pit a ZIF socket on the breadboard if you want.

 

Then, download the Arduino IDE (not the web editor):

https://www.arduino.cc/en/Main/Software

 

Then run it and then cut and paste this code into the IDE (replacing whatever is there when the IDE opens):

https://raw.githubusercontent.com/FozzTexx/DRAM-Tester/master/Arduino-DRAM-tester.ino

 

Then under tools->board, pick Arduino/Genuino Uno

Then plug the arduino into a USB part and then from tools->Port, choose the one that lists your Arduino.

Then sketch->upload (it will ask you to save it, go ahead and save it), let it compile and upload. The red light will come on for a long time while it is "testing" and eventually the other light will come on when it is done.

 

Once the upload is complete, you can unplug it from your computer and plug it into any USB power source. Then unplug the uno from power, put in a chip and then power it up to start a test.

 

According to the web site, it sounds like it takes a while to run the test.

 

The author wrote this for Apple IIGS memory, I assume that is the same (or similar enough) that this will work on Atari memory.

 

I do not have any memory chips available to test with, so I haven't tested this with actual memory.

 

You might want to test with a "spare" chip first to make sure it doesn't damage the chip.

 

If this still feels like something you do not want to tackle, I'm happy to put one together and load the software and send it to you.

 

Todd

 

[+DrVenkman]

It definitely looks interesting and I'm sure I could build it if I spent some time to learn the ins and outs of breadboarding with the Arduino. However, at this point I'm loathe to throw money at building a tester when it may not even be RAM. Remember, I did already swap them one by one as a test.

 

​At this point I'm trying to find LEGIBLE schematics that detail the 74LSxxxx logic chips layout and how they interface with the RAM. The schematics in the 1200XL FSM are horribly illegible, and the "Jerzy Sobola" schematics from 2002 or so, while neat, have very obvious mistakes (such as mislabeling some of the IC locations). It's very hard to trust them given that I - a rank amateur with more tools and time than sense - can spot some of the mistakes. Bless his heart for undertaking to make them, but I just can't trust 'em.

[toddtmw]

Did you see my suggestion about swapping all the memory in case more than one chip is bad?

[+DrVenkman]

Did you see my suggestion about swapping all the memory in case more than one chip is bad?

 

Ah, I do remember now you said that. I'll try that tomorrow and see how it goes. I'm not gonna mess with my working 1200XL with this - the set of DRAMs is coming straight outta my third 800XL.

 

I'm not holding out hope, though. I really do think something is screwy in the 74LSxxx logic. I just wish there was a reliable and legible set of 1200XL schematics handy.

[flashjazzcat]

But you already swapped out every IC on the board, right? As I think Stephen said, it looks like a busted socket or broken trace on the board. What happens if you flex the board lightly during use? I found a bad trace repair on and 800XL the other day by prodding it while it was running.

[+DrVenkman]

Yeah I have swapped IC’s. Yesterday I began exploring the running system with a logic probe. As far as I can tell, all the pins on the main IC’s which are expecting to have signals do - Vcc is high, Vss is low, the rest of the address and data lines have signals, etc. The only ones that don’t are the ones not connected. And a few of the 74xxx chips, which is why I’m trying to find legible and accurate schematics of those chips. I’m not entirely sure all of those are connected/used on the 1200XL but the only way to be sure is to disassemble my working machine and compare, which is looking like I’m going to have to do.

 

As Todd noted, it’s possible that I have more than one bad DRAM, which swapping one at a time wouldn’t find. Easy enough to test that tonight.

 

Another oddity - when left at idle on the logo screen, while probing the machine will eventually crash, usually to a black screen though occasionally on a frozen solid color logo. A power cycle always restores it and it doesn’t usually immediately crash again, which makes me believe this isn’t heat related.

 

Anyway thanks for your thoughts and insights, folks. It’s appreciated.