Replaced 1200XL keyboard mylar, now have more questions

[+x=usr(1536)]

Some background: the 1200XL in question is a machine that I've owned for around 20 years, but is also one that hasn't seen much in the way of actual use in that time, particularly in the last 15 years or so.  It's also the first and only 1200XL I've ever owned, so it's not a machine that I'm terribly familiar with.  As a result, any and all help is gratefully accepted.  Note that it was always stored in a no-UV, climate-controlled envrionment, so has had a relatively easy life.

 

Here's the story so far:

 

Pulled the machine out of storage at Christmas.  It booted with a new PSU from Best and output video (composite via monitor port), but didn't do much beyond go to the rainbow Atari screen.  Sticking a Star Raiders cartridge in it showed that the game would run, but it did not respond to keypresses.

 

Replaced the mylar this evening.  Per Star Raiders, the START and SELECT console keys are working, as is sound.  No response from joysticks or the rest of the keyboard.  Power LED is on.  Pressing HELP at the rainbow Atari screen has no effect.

 

Connecting a 1010 to the 1200XL and holding START to boot from tape works, but it errors out with 'BOOT ERROR' after approximately a minute.  The only button that works to make the 1010 load is the BREAK console key; none of the others have any effect.  The 1010 and test tape (Thorn EMI's Darts) work fine on both a stock 600XL and 256KB 800XL.

 

Can't get it to boot a known-good disk from a 1050 that also works on the other two XLs.

 

One thing I haven't done is to pop the RF shield to see if there are any modifications, but given that the twist-tabs holding it in place look untouched, I'm relatively certain that this machine is stock.

 

Any ideas?  I think I'm looking at a machine with multiple issues, but am not totally clear as to where to start.

[Sugarland]

You used the Best mylar? Did you thoroughly clean the gold fingers contacts when installing the new mylar? You used the tiny pcb? Have you tried swapping the two 4051 keyboard chips with known good? Does the self-test test okay on RAM?

[leech]

From my understanding the mylars don't actually go bad, you generally need to just lift and clean and remove the crappy spacer.  I am trying to remember if I ended up needing to use my pens to redraw the circuits on one of them...

But I would suggest, if you can, to go into the self test and look at the keyboard one.  As you press on keys, try to roll yiur finger around on it and see if that helps create an initial connection.  I know on the second one I have, I ended up doing that until all the keys worked.

 

https://www.amazon.com/dp/B07PB69R7Y

These are awesome for fixing mylar keyboards.

Edited March 20, 2021 by leech

[+DrVenkman]

No experience with Best’s replacement Mylars - I’ve fixed my own 1200XL keyboards using the tried and true method of removing the degraded paper tape at the contacts and repainting with conductive paint. 

 

As for the joystick issues, that points to PIA. The SIO issues point to POKEY (which may also contribute to the keyboard problems but hard to know for sure). The fact that two chips are implicated in the problems could also point to an issue with sockets. I’ve been down this road myself with one of my 1200XL’s, so good luck.  

[+x=usr(1536)]

8 hours ago, Sugarland said:

You used the Best mylar? Did you thoroughly clean the gold fingers contacts when installing the new mylar?

Yes on both.  Removed the adhesive strip with the carbon traces and gave the fingers a really good going over with isopropyl alcohol; also removed all residual material from the adhesive strip.

8 hours ago, Sugarland said:

You used the tiny pcb?

Correct, and I believe it's in the proper orientation.  Did notice that the keyboard PCB now bulges slightly where it's installed compared to stock when reassembled, but figured that was par for the course.

8 hours ago, Sugarland said:

Have you tried swapping the two 4051 keyboard chips with known good?

Are you referring to the two on the back of the keyboard PCB?  If so, then no.

8 hours ago, Sugarland said:

Does the self-test test okay on RAM?

Can't get to the self-test; HELP is one of the apparently non-responsive keys.  Unfortunately, because it also isn't booting anything successfully (including my FujiNet) except for cartridges, I have no good way to try to see what is and is not working on the keyboard based on the software that limits me to for testing.

3 hours ago, leech said:

From my understanding the mylars don't actually go bad, you generally need to just lift and clean and remove the crappy spacer.  I am trying to remember if I ended up needing to use my pens to redraw the circuits on one of them...

From what I've seen, 99.9% of the circuit fixes require redrawing the fingers.  Wouldn't be surprised if that was required (or may be here).

3 hours ago, leech said:

But I would suggest, if you can, to go into the self test and look at the keyboard one.  As you press on keys, try to roll yiur finger around on it and see if that helps create an initial connection.  I know on the second one I have, I ended up doing that until all the keys worked.

If it would let me get at the self-test, I'd be 100% in agreement with you.  As it stands, though, the only keys I can verify as working (because Star Raiders is a lousy keyboard test suite ) are RESET, START, SELECT, and BREAK.  That's four more than I had working previously, though, so there is at least some progress.

3 hours ago, leech said:

https://www.amazon.com/dp/B07PB69R7Y

These are awesome for fixing mylar keyboards.

Thank you for that suggestion; I've just ordered a set ?

24 minutes ago, DrVenkman said:

No experience with Best’s replacement Mylars - I’ve fixed my own 1200XL keyboards using the tried and true method of removing the degraded paper tape at the contacts and repainting with conductive paint.

I do still have the original mylar, so conceivably could go back to it.  The adhesive was absolutely shot, though - it peeled off with near-zero resistance.  That might make things interesting (but not impossible) to replace.

24 minutes ago, DrVenkman said:

As for the joystick issues, that points to PIA. The SIO issues point to POKEY (which may also contribute to the keyboard problems but hard to know for sure). The fact that two chips are implicated in the problems could also point to an issue with sockets. I’ve been down this road myself with one of my 1200XL’s, so good luck.  

Thanks; the pointer is definitely appreciated.  Sockets had occurred to me, but, as I'm out of contact cleaner until I pick more up later today, I didn't want to pop the RF cage and start poking around unless I could do something about it.  Still, I will definitely be checking for anything that's come unseated or has oxidation on it.

[flashjazzcat]

23 minutes ago, x=usr(1536) said:

I do still have the original mylar, so conceivably could go back to it.  The adhesive was absolutely shot, though - it peeled off with near-zero resistance.  That might make things interesting (but not impossible) to replace.

No adhesive is needed and I'm not sure why they put it on there in the first place. As long as the carbon fingers (or repainted contacts if the carbon is history) are clamped against the gold contacts, it'll work. I had to put a piece of electrician tape over the back of the Mylar after repainting the traces on the last one I fixed, though, since the plastic was a little wrinkled. Works perfectly now, though.

Edited March 20, 2021 by flashjazzcat
typo

[+bob1200xl]

You need to strip off all the old mylar and wipe all the gold key contacts with 91% alcohol. Use Goof-Off to clean the contact fingers, then wipe that with 91%. Do not use the Best spacer-thing. Just assemble the k/b with the new mylar. It will work just fine.

 

Bob

 

[+x=usr(1536)]

42 minutes ago, bob1200xl said:

You need to strip off all the old mylar and wipe all the gold key contacts with 91% alcohol. Use Goof-Off to clean the contact fingers, then wipe that with 91%. Do not use the Best spacer-thing. Just assemble the k/b with the new mylar. It will work just fine.

 

Bob

Thank you, Bob - I'll ditch the spacer PCB when I get back to it this evening and see where things sit.  FWIW, Goof-Off and 91% alcohol was exactly what was used to remove the old carbon strip, and the rest of the contacts were also alcohol-wiped (no Goof-off).

[+kheller2]

You have a working 600XL and 800XL.  Start swapping the chips from the 1200 into the 600 and test.

All are (mostly) socketed.  If the break key is working that means at least one of the 4051s (U1?) is working .. sadly the Help key requires both 4051s to work.

[BillC]

6 hours ago, x=usr(1536) said:

(because Star Raiders is a lousy keyboard test suite ) are RESET, START, SELECT, and BREAK.  That's four more than I had working previously, though, so there is at least some progress.

Star Raiders allows testing more than just the console keys, there are also 21 other keys that are used for control.

https://strategywiki.org/wiki/Star_Raiders/Controls

[+x=usr(1536)]

5 hours ago, kheller2 said:

You have a working 600XL and 800XL.  Start swapping the chips from the 1200 into the 600 and test.

Going to hold off on that unless it's completely necessary, but it's definitely on the menu if it comes down to it.  Don't want to risk potentially introducing problems into two fully-working machines when there are other known issues with the 1200XL (more on that below).

Quote

All are (mostly) socketed.  If the break key is working that means at least one of the 4051s (U1?) is working .. sadly the Help key requires both 4051s to work.

I'm glad you mentioned needing both 4051s in order for the HELP key to work.  That reinforces my suspicions regarding something I found while looking at the PCB earlier: there was a bad trace under a solder blob that looked suspiciously like someone else's repair.  The trace can be followed completely from the ribbon cable through both 4051s before returning to the ribbon cable on a different pin.  This is, at the very least, a contributing factor if not the current issue outright.

 

Going to work on that for a bit; will post photos once I've got the trace jumpered.

 

 

Edited March 20, 2021 by x=usr(1536)

[+x=usr(1536)]

2 hours ago, BillC said:

Star Raiders allows testing more than just the console keys, there are also 21 other keys that are used for control.

https://strategywiki.org/wiki/Star_Raiders/Controls

 

And that is a valid point.  Unfortunately, I received no response from any of them - just the four console buttons mentioned earlier.

 

Check my previous post, though - I think I may have found another issue that I was previously unaware of.

[+x=usr(1536)]

Victory!  For the most part, but I'll take it.  More on that later.

 

You'll want to zoom in on the photos to get a better idea of what's going on, but:

 

On inspection, I noticed that the via circled in the photo below had traces either side of it that were darker than the remainder of the trace.  Continuity check showed that nothing was passing through it; the solder blob on it was also clearly not factory.  Removed the solder blob, found that there was no pad (nor was it attached to the blob), and that the jumper on the back of the board being fed by this via wasn't even making contact with the solder blob on this side.

 

Decided to just bypass that entire part of the trace, so, armed with 22ga. wire and a soldering iron...

 

 

Here's what it looks like from the back.  Not my greatest work, but it'll do.  Ask me how I learned that you can't reassemble the keyboard properly if you run the jumper wire on the front of the PCB

 

 

Aaaand here's the end result.  Clearly, the HELP key is working:

 

 

All tests pass (including manual testing of the keyboard), and I never realised that the 1200XL alternates illuminating the L1 and L2 LEDs when running the memory test.  Neat.  Incidentally, this is with following @bob1200xl's recommendation to leave the Best mini-PCB out; everything's working fine, so I'll leave it that way.

 

As for why I say this is partial victory: still can't get the 1010, 1050, or FujiNet to boot.  One thing I did notice while testing the 1010 is that that particular TV is a noisy bugger in terms of RFI: the 1010 is picking up TV audio from <insert channel here> that is audible during loading.  The audio in from the data track on the tape is also much lower than on the 600XL or 800XL.  I'll relocate things to someplace where I can eliminate the RFI issue, but it's good to know that the test display is as noisy as it is.  Not sure why the data track on the 1010 is as low as it is on this machine, though.

 

One interesting thing: the fire button now works in Star Raiders.  This has me suspecting that moving the unit around has jostled something into submission; more contact cleaner was picked up today, so I'll take a shot at popping ICs and cleaning their legs / sockets later this week.

 

For now, however, I am very happy with how it's progressing.  Thanks to everyone for their help; it's been absolutely invaluable and greatly appreciated.

Edited March 21, 2021 by x=usr(1536)

[+x=usr(1536)]

Just to follow up: had some unexpected free time today, so pulled the RF shield and poked around.  About half of the ICs were semi-seated at best, so ended up pulling all of them, giving the sockets a good blast with contact cleaner, and wiping the legs on the ICs with the same stuff.  So far, the 1200 has now booted from a 1010, 1050, FujiNet, and has also loaded ModemLink from a 1030.

 

Joysticks are still in the same boat - fire button only.  Having said that, I did notice some oxidation on the pins of both connectors, so will take care of those with some fine-grit sandpaper.  And, with the FujiNet now working on this machine, I have access to better diagnostic tools that don't require external media to run, so should be able to troubleshoot it more effectively if the need should arise.

 

Calling it pretty much fixed for now.  Absolutely thrilled to have a working 1200XL at last

[+DrVenkman]

10 hours ago, x=usr(1536) said:

Joysticks are still in the same boat - fire button only.  Having said that, I did notice some oxidation on the pins of both connectors, so will take care of those with some fine-grit sandpaper.

The triggers on the controller ports are read by the GTIA chip, but the directional signals are read by PIA and paddle lines by POKEY. If it were me, I’d a common game like Pac-Man or Dig Dug and set up a two player game to methodically check all four directions on both ports. Then I’d setup a 4-player game of Breakout or Super Breakout so you can test both paddle lines on each port. In fact, since you know the trigger lines work, maybe check paddles next. If those work, then your issue is probably the PIA chip. Of course, clean the contacts of the port (Deoxit or any electrical cleaner should do) and reflow the connector pins if necessary. 

[+x=usr(1536)]

30 minutes ago, DrVenkman said:

The triggers on the controller ports are read by the GTIA chip, but the directional signals are read by PIA and paddle lines by POKEY. If it were me, I’d a common game like Pac-Man or Dig Dug and set up a two player game to methodically check all four directions on both ports. Then I’d setup a 4-player game of Breakout or Super Breakout so you can test both paddle lines on each port. In fact, since you know the trigger lines work, maybe check paddles next. If those work, then your issue is probably the PIA chip.

Good call, and will do.  Don't have any paddles at the moment, but I do have a KoalaPad that can stand in.  It won't test that both paddles on a port are functional, but it should at least be good enough to get an idea as to whether or not they're at least readable.

30 minutes ago, DrVenkman said:

Of course, clean the contacts of the port (Deoxit or any electrical cleaner should do) and reflow the connector pins if necessary. 

Yep, hit the pins with contact cleaner but the oxidation remained.  The pins seem to be slightly oversized due to the crud on them, which makes me wonder if there wasn't a liquid spill (probably soda) that hit them at some point in their life which wasn't properly cleaned up.  The interior is really clean, though, so I'm not 100% certain of that theory.

[+DrVenkman]

15 minutes ago, x=usr(1536) said:

Yep, hit the pins with contact cleaner but the oxidation remained.  The pins seem to be slightly oversized due to the crud on them, which makes me wonder if there wasn't a liquid spill (probably soda) that hit them at some point in their life which wasn't properly cleaned up.  The interior is really clean, though, so I'm not 100% certain of that theory.

It’s finicky work, but you could slide a jumper wire with a female Dupont connector onto each pin and then verify continuity to the corresponding solder pad under the controller jack if you want to be 100% sure the pins are clean enough. 

[+x=usr(1536)]

1 hour ago, x=usr(1536) said:

Don't have any paddles at the moment, but I do have a KoalaPad that can stand in.  It won't test that both paddles on a port are functional, but it should at least be good enough to get an idea as to whether or not they're at least readable

Self-correction: the KoalaPad uses one paddle for X positioning and one for Y, so it should test both on a single port.

 

The proper order of operations in the morning is caffeine, then post.

[+kheller2]

If you have a BASIC cart, there are a few programs floating around that will test the values being return from the joystick direction and another for paddles. 

[+x=usr(1536)]

2 hours ago, kheller2 said:

If you have a BASIC cart, there are a few programs floating around that will test the values being return from the joystick direction and another for paddles. 

Pretty sure I could quickly knock something together from memory in BASIC.  Unfortunately, Rev. C is currently in the mail

 

The peripheral tests that can be pulled via the FujiNet work well enough, though.  The advantage is that they'll also let me easily test edge-case stuff hanging off of the joystick ports, like lightpens, the CX85, ST mice, etc.  It's arguably overkill, but if the weird end of the spectrum is working as well as the more orthodox end, it's a more thorough indicator of port health.

 

 

[+x=usr(1536)]

Coming back to the joystick ports for a moment: ran Florian Dingler's perhipheral tester via the .ATR image on Mr. Robot's TNFS server against joysticks, the KoalaPad, and CX22 Trak-Ball.  More on that below.

12 hours ago, DrVenkman said:

The triggers on the controller ports are read by the GTIA chip, but the directional signals are read by PIA and paddle lines by POKEY.

FWIW, the pins on the controller ports were filed with a moderately-fine emery board, hit with contact cleaner, filed again, given another dose of contact cleaner, then blown out with compressed air before starting tonight's round of testing.

 

Triggers on all three controllers work fine.  Using the KoalaPad as a paddle substitute, it reads fine on both paddles; the resolution is off compared to a paddle, but that's expected behaviour.  Tested specifically as a KoalaPad, it's fine in both axes.  The CX22 only registers movement in the Y-axis in Trak-Ball mode, however, and in joystick mode it replicates the results of the joystick test performed with actual joysticks.  All controllers work without issue on the 800XL, so this is looking like a 1200-centric issue.

 

Findings are consistent across both ports, with the wild card being the CX22 - the Trak-Ball test only works on port 1.  No idea how it may have behaved on port 2 as a result, but I wouldn't expect any real difference given how the others behaved.

 

12 hours ago, DrVenkman said:

It’s finicky work, but you could slide a jumper wire with a female Dupont connector onto each pin and then verify continuity to the corresponding solder pad under the controller jack if you want to be 100% sure the pins are clean enough.

Yep, that'll be the next port of call.  I'm suspecting that your earlier hunch re: a bad PIA also being in the mix may have been correct.  Will definitely nose around between there and the controller ports next time I open this machine up, which will likely be later this week.