Oxidized 800XL keyboard traces

[Urchlay]

Someone was kind enough to give me an 800XL the other day, so I'm spending the day cleaning it up and generally messing with it...

 

Everything works just fine except about half the keys on the keyboard are dead. I tried pressing them repeatedly, with no effect, so it was time to crack the case open.

 

It's the type of keyboard with a sheet of clear plastic (mylar?) with carbon traces, folded over itself. The most suspicious-looking thing is that the carbon traces on the keyboard's connector have turned black, about the first inch or so. 2 or 3 of the traces are almost totally black. The parts that aren't black are an off-white color. From what I can see, a few traces on the keyboard itself (under the keys) are mostly black as well.

 

Am I right in thinking that this is oxidation? I'm thinking the correct way to fix it is going to be with a circuitwriter (silver) pen. Basically I'll be following the 1200XL keyboard repair instructions I've seen posted here, except that the traces are a lot longer, and they aren't worn away (the 1200XL ones are dark grey/black, look like graphite, and on my old 1200XL keyboard they were actually missing from the connector).

 

While I'm at it, I'm cleaning the case and keys... I actually removed all the keycaps from the keyboard, soaked them in dish soap, scrubbed each one with a toothbrush, rinsed, and right now they're tied up in a couple of socks in the dryer on "low heat". The case won't be dry for a while either, so I'm in no hurry...

 

Anyone have any experience fixing an 800XL keyboard in this condition, and want to pass some of their wisdom this way? Any "gotchas" I should know about?

[Almost Rice]

One of the things I have done in the past is cut 1mm off of the ribbon. That has solved most of my keyboard problems

[Urchlay]

One of the things I have done in the past is cut 1mm off of the ribbon. That has solved most of my keyboard problems

 

Measure twice, cut once? It'd have to be a very straight cut, right?

 

I may try this, but I dunno if it'll help: the black parts of the traces are a lot more than 1cm. If I cut it short enough to get rid of most of the black, the cable would be too short to actually plug in.

[Urchlay]

Update:

 

I've got the keyboard completely disassembled so I can get at the mylar "sandwich" to draw traces on it... and now I'm not so sure I should. It looks like someone at some time must have spilled a drink in this keyboard: there's dried up gunk all over the mylar. It doesn't look like it's actually eaten through the traces (whatever the drink was, it wasn't cola), it's just keeping the key contacts from making good contact with each other, or maybe shorting them (depending on how conductive it is).

 

If I get this thing fixed, I'll write it up in the restoration thread...

[+bf2k+]

I had 1 like that once... a certain % of keys didn't work. it ended up being Pokey!

[+kheller2]

Try wiping down the inside of the mylar, it should open like a book. There should be a mylar seperation sheet as well. You can then just plug the mylar back into the XL and test pressing the contacts. Almost all my mylar ones are black but still function.

[Urchlay]

Try wiping down the inside of the mylar, it should open like a book. There should be a mylar seperation sheet as well. You can then just plug the mylar back into the XL and test pressing the contacts. Almost all my mylar ones are black but still function.

 

I tried this... opened the mylar, cleaned it up... but the keys that didn't work before, still didn't work, and the ones that did work before, still worked.

 

I also tried following the traces to the connector, then shorting the connector pins together. I may have counted wrong (need both hands to hold the mylar so I had to follow traces with my unreliable eyes), but I was unable to make any of the non-working keystrokes show up that way, either (got some of the working ones though).

 

If it's the POKEY, I think I have a replacement... but it's from a 400. Will that work OK?

[Urchlay]

OK, never mind on the 400 POKEY, I either don't have one or can't find it. I *did* however find my 800XL that got fried 12-15 years ago. Its POKEY is socketed, so I tried it in the new 800XL... and it's behaving exactly the same way (same set of non-working keys).

 

Sitting right next to the POKEY though, are a couple of CD4051's, which turn out to be muxer/demuxers... they're socketed in the new XL, but not the old one. I swapped the two 4051's in the new XL, and the keys that used to work, now are dead (and the ones that used to be dead, now work).

 

So my next step looks like it should be to desolder one of the CD4051's from the old dead XL and try it in the new one. Likely that'll fix it.

 

However... the old 800XL has a 256K RAM upgrade of some kind in it. I'd really like to get it working if possible. That XL was killed by over-voltage, and I'm told the 256K RAMs are probably bad. There's at least a chance that it would boot if it had good RAM chips... can someone tell me, what happens if I replace the 256K chips with (known good) 64K ones, in an upgraded 800XL? Does it boot and run as a 64K XL, or does it fail to boot, or does it fail to boot and blow up the 64K chips? What about trying the 256K chips in the non-upgraded 800XL? Do they function as 64K chips, in that case? Do I need to lift or ground a couple of address pins?

 

I do have a known good set of 256K RAMs in my upgraded 1200XL, but it's my main machine and I really don't want to take it apart

[Rybags]

Fairly sure you'll have to do something with extra address lines on larger chips else they'll get floating values and so not r/w to the same location each time.

 

Does the upgraded XL retain it's original RAMs?

 

There's every chance that provided they're good that the machine should boot up - the OS only does RAM checks in the non-banked contiguous area it can find from 0000-BFFF.

[Urchlay]

Fairly sure you'll have to do something with extra address lines on larger chips else they'll get floating values and so not r/w to the same location each time.

 

Does the upgraded XL retain it's original RAMs?

 

There's every chance that provided they're good that the machine should boot up - the OS only does RAM checks in the non-banked contiguous area it can find from 0000-BFFF.

 

OK, so ground the extra address lines I guess...

 

The upgraded XL doesn't have its original RAMs. It has 8 256Kx1 chips sitting in the original sockets that the 64Kx1's must have come out of... I never had the original RAM for this machine though, I bought it that way, in maybe 1989 or 90.

 

So what do the extra pins do on a 64K chip? (the ones that would be the top 2 address lines on a 256K chip), are they just "no connection"? In that case the banking hardware would basically factor out of the equation...

[Rybags]

Probably best to get the schematics with pinouts for your chips.

 

Don't later chips use the same lines and have selects for RAS/CAS? In which case things become a little bit more complex.

[Urchlay]

Was trying to find the pinout for these 256K chips... ChipDir doesn't have them. They're 81256's.

 

The 64K ones are 4164's, found pinout easily.

 

I found a pinout for the 41256, but I'm too dumb to understand it: it only lists A0 through A8, which (to my limited understanding) means only 9 bits of addressing, or only 128K capacity. Anyway I've no idea whether the 81256 would be similar or not.

[Rybags]

9 bits gives 512 combinations.

 

Which would clearly indicate a RAS/CAS situation - so tying down the msb should force a 256x256 = 64 kbit situation.

 

Haven't really played about with RAM expansions myself - looking/hoping to do a 1 meg SIMM upgrade probably on my XEGS sometime.

[Urchlay]

9 bits gives 512 combinations.

 

Which would clearly indicate a RAS/CAS situation - so tying down the msb should force a 256x256 = 64 kbit situation.

 

Hm. So it's a 512x512 "2D array" of bits, not a 256Kx1 "1D array"... is that pretty close to right?

 

If the 81256 pinout's the same as the 41256, pin 16 is A8... a 4164 has the same pinout, except pin 16 is NC (no connection). With any luck, the non-upgraded XL might already have that pin tied to ground. Will check.

 

Haven't really played about with RAM expansions myself - looking/hoping to do a 1 meg SIMM upgrade probably on my XEGS sometime.

 

Well, you obviously know more about it than I do

[Rybags]

Makes sense to have 1 NC on the 64 kbit chip. 8+8 selects gives 16 bits of address = 64K, 9+9 gives 18 bits = 256K.

 

I'd guess that the PORTB bank selection comes into play with RAS only - Problem is, if any of the logic chips involved in extended RAM are playing up, then you'll probably get random errors in certain address ranges.

Edited June 3, 2007 by Rybags

[re-atari]

Was trying to find the pinout for these 256K chips... ChipDir doesn't have them. They're 81256's.

 

The 64K ones are 4164's, found pinout easily.

 

I found a pinout for the 41256, but I'm too dumb to understand it: it only lists A0 through A8, which (to my limited understanding) means only 9 bits of addressing, or only 128K capacity. Anyway I've no idea whether the 81256 would be similar or not.

81256's are completely pincompatible with 41256's. I used them in my XL without a problem. They're made by Fujitsu, check for a large F on top of the DRAM's. Like some other manufacturers, Fujitsu had come up with the brilliant idea of fiddling around with the component numbers and put an '8' instead of the '4'. What are standards good for, anyway? Likewise Motorola called their DRAM's 62XXX, OKI 37XXX (but did use 41XXX as well), Texas Instr. 4XXX, to name but a few.

Check the ST's schematics if you can't find the pinout for the 41256, the early models used them. Later on Atari switched to 41426's (256k * 4bit).

The difference between the 4164's and the 41256's is the extra addressline. This pin is NC in the 4164's.

 

re-atari

Edited June 3, 2007 by re-atari

[Urchlay]

81256's are completely pincompatible with 41256's. I used them in my XL without a problem.

 

Ah! Thanks for that.

 

They're made by Fujitsu, check for a large F on top of the DRAM's. Like some other manufacturers, Fujitsu had come up with the brilliant idea of fiddling around with the component numbers and put an '8' instead of the '4'. What are standards good for, anyway? Likewise Motorola called their DRAM's 62XXX, OKI 37XXX (but did use 41XXX as well), Texas Instr. 4XXX, to name but a few.

 

Yup, they're Fujitsus. The nice thing about standards is that there are so many to choose from!

 

Check the ST's schematics if you can't find the pinout for the 41256, the early models used them. Later on Atari switched to 41426's (256k * 4bit).

 

Hmmm. I've got a 520ST whose keyboard is completely covered in ketchup (looks like dried blood). It was that way when I got it... wonder if I can borrow some of its RAM. Bet it's soldered, not socketed...?

 

The difference between the 4164's and the 41256's is the extra addressline. This pin is NC in the 4164's.

 

So if the un-upgraded 800XL happens to ground the 4164's NC pin, the 41256's ought to work fine and give me 64K (with A9 always low). If they don't, I'll know the 41256's are bad. And... if I put the 4164's in an upgraded 800XL, it ought to boot up as a 64K 800XL. If dead RAM is all that's wrong with my old dead 800XL, I might actually be able to fix it!

[re-atari]

Hmmm. I've got a 520ST whose keyboard is completely covered in ketchup (looks like dried blood). It was that way when I got it... wonder if I can borrow some of its RAM. Bet it's soldered, not socketed...?

I'm afraid reusing the DRAM from an ST will require desoldering, as the Tramiels never bothered to put in sockets (would probably have added a whole dollar in manufacturing cost per computer). Back in the day I upgraded a number of my friends 520's (I never owned one myself, lack of appeal and resentment of anything that had to do with Tramiel) to 1Mb. The procedure of piggybacking of 41256's and adding a few wires to the MMU. I've never come across a ST with socketed DRAM's.

 

So if the un-upgraded 800XL happens to ground the 4164's NC pin, the 41256's ought to work fine and give me 64K (with A9 always low). If they don't, I'll know the 41256's are bad. And... if I put the 4164's in an upgraded 800XL, it ought to boot up as a 64K 800XL. If dead RAM is all that's wrong with my old dead 800XL, I might actually be able to fix it!

A bit of additional and more precise info about the 4164's: they are not all created equal. A number of manufacturers built 4164's that require a refresh signal on pin 1, so in that case this pin is definitely not NC. Among them are Motorola (MCM6664), Siemens (HYB4164), Thomson (EF6665), Fairchild (F4164), Texas Instruments (TMS4164) and INMOS (IMS2600). Japanese manufacturs (forerunners of technological development of DRAM in those days) did not need this signal, as they had built the refresh circuitry on the die itself (also called 'hidden refresh'). This is of importance for the XL, as pin 1 is high, i.e. connected to 5V via R32 (3k), meaning that only DRAM's with refresh onboard will work. The DRAM on the 1064 memeory expansion unit is layed out the same way. Thinking back, I just can't remember if I ever saw 4164 DRAM's of a none-Japanese origin, like the types mentioned above, in XL's.

BTW: I don't know why, but in the 130XE pin 1 is connected to GND instead of 5V. Twas probably done by the Tramiels to save cost by leaving out a single resistor

As the development of the 41256 DRAM was done by the Japanese, it had the refresh circuitry onboard from the start. This was a plain necessity as well, as they were otherwise short of a pin to put the new addressline. Putting this on pin 1 also meant the layout stayed more-or-less pin-compatible with the earlier DRAM types. AFAIK every manufacturer followed this design.

To get back OT, I think your suggestion for a test procedure will work. If you define an addressline high level on the 41256's (A8 on 5V via R32), this line simply can not be addressed. In this case the XL should find only 64K. Same goes for the 4164's in an upgraded XL, in which case A8 can not be addressed because of the physical absence of this addressline inside the DRAM. Try it, you won't break anything.

 

re-atari

[re-atari]

If the 81256 pinout's the same as the 41256, pin 16 is A8... a 4164 has the same pinout, except pin 16 is NC (no connection). With any luck, the non-upgraded XL might already have that pin tied to ground. Will check.

In a short addition to my earlier reply, I presume you meant that in the 41256 DRAM's A8 is on pin 1 instead of pin 16? On both types pin 16 is GND.

 

re-atari

[Urchlay]

If the 81256 pinout's the same as the 41256, pin 16 is A8... a 4164 has the same pinout, except pin 16 is NC (no connection). With any luck, the non-upgraded XL might already have that pin tied to ground. Will check.

In a short addition to my earlier reply, I presume you meant that in the 41256 DRAM's A8 is on pin 1 instead of pin 16? On both types pin 16 is GND.

 

Yah, I counted wrong... It's the pin opposite from GND.

 

Well, I popped one of the 4164's out of the new 800XL and tried each of the old XL's 81256's in turn... all of them booted to a READY screen except one (black screen), so I only seem to have one bad 256K DRAM. Desoldering one chip from my ST won't be too painful.

 

Also, the 6502, POKEY, and ANTIC from the dead XL are OK (work in the new XL). The PIA is bad. The GTIA might be bad: the new XL with the old GTIA will boot straight to the self-test screen, with "audio visual" pre-selected. It might mean the GTIA is damaged, or just that it's the wrong revision for that 800XL.

 

I'm on my way out the door to pick up a donor 600XL someone has offered me, which hopefully has a good PIA, GTIA, and CD4051's.

 

One question: what's the purpose of the resistor that sits above the RAM chips? On the upgraded XL, it's missing, and one of its pads is soldered to a jumper that goes to the upgrade board (which has 3 74LS158's and more jumpers that go to some bent-up pins on the PIA, which I assume are for the extra PORTB bits). I couldn't get the dead upgraded 800XL to boot with the new XL's PIA and one of its 4164's...

[mimo]

save the 600!, or send it to me!!!

[Urchlay]

save the 600!, or send it to me!!!

 

Well, I don't even know whether the 600XL works yet (just got in the door from picking it up, and the guy who gave it to me doesn't know if it works). When I'm done stealing parts from it, it will be missing a PIA, a CD4015, and possibly a GTIA. I was planning to hang on to it for parts...

 

If it works before I steal parts from it, I'll want to get it working again someday and upgrade to at least 64K. A 16K Atari 8-bit is pretty useless to me except maybe as a game console, if I owned enough carts to make that worthwhile...

 

Fear not though, whatever I do with the 600XL, it's not going to the dump or anything

[Urchlay]

Update: the 600XL worked, though it had the worst-looking RF picture I've ever seen... every chip on its board is socketed, it was made in early 1983.

 

Stole one of its 4051's, and now the new 800XL is fixed!

 

Will steal the PIA to try and fix the old 800XL when I get a replacement 256K RAM chip for the one that's dead. Killing one 600XL to restore 2 800XL's seems like a good tradeoff to me.

[Urchlay]

A note to anyone who disassembles an 800XL keyboard: the key inserts (little white plastic things the keycaps attach to) are NOT all the same height! The top row (numbers) has the tallest inserts, and they get shorter as you move down the keyboard... meaning the top keys actually move farther when you press them.

 

The same applies to more than just the 800XL: I know modern crappy PC keyboards are made the same way.

 

Word of advice: don't remove the inserts at all, if you can help it. You don't need to, to clean the keys (actually, the keycaps are the visible part of the keys. The inserts are the actual keys). Sometimes, when you remove a key, the insert comes out with the keycap. If this happens, pull the insert out of the keycap and immediately put it back in the keyboard. There are tiny "hooks" in the corners... on my 800XL keyboard, the hooks go in the upper-left and lower-right corners.

 

Don't do what I did: 10 or 12 of my keys came out with the inserts still attached. I was going to leave them in the keys when cleaning them, but then thought water might get trapped between the inserts and their keycaps, so I removed all the inserts that were stuck to the keys... and threw them all in a cup, mixed together.

 

When I put the keyboard back together, those 10-12 keys were all different heights from their neighbors. The keyboard worked fine, but looks and feels kind of weird. Took an hour to figure out what was wrong and fix the mess...

[re-atari]

Also, the 6502, POKEY, and ANTIC from the dead XL are OK (work in the new XL). The PIA is bad. The GTIA might be bad: the new XL with the old GTIA will boot straight to the self-test screen, with "audio visual" pre-selected. It might mean the GTIA is damaged, or just that it's the wrong revision for that 800XL.

 

I'm on my way out the door to pick up a donor 600XL someone has offered me, which hopefully has a good PIA, GTIA, and CD4051's.

 

A donor XL always comes in handy as a supplier of spare parts Shops generally tend to be closed at the very moment that you need spare parts, preventing you from tinkering on...

If you want to resurrect the 600XL in the future, it's good to know that you're not restricted to using a 6520 as a replacement for PIA. A 6821 will do just as well (talking from experience). Considering the age of this component, it may be difficult to obtain new ones, though. I don't think it's in production any more. CD4051's on the other hand, are standard CMOS types that are readily available to this very day, costing just a few quarters.

Upgrading the 600XL to 64K is easy, though it may be difficult to find the 41464 (64k * 4bit) DRAM's. Cannibalizing an old EGA or VGA card may be the way to go.

 

One question: what's the purpose of the resistor that sits above the RAM chips? On the upgraded XL, it's missing, and one of its pads is soldered to a jumper that goes to the upgrade board (which has 3 74LS158's and more jumpers that go to some bent-up pins on the PIA, which I assume are for the extra PORTB bits). I couldn't get the dead upgraded 800XL to boot with the new XL's PIA and one of its 4164's...

 

That's resistor R32 (3k Ohm) I was referring to in my earlier reply. The left hand solder pad is connected to 5V, the right hand one to the 4164's pin 1. In the standard XL layout this resistor defines a high level on this pin (refresh). In the upgrade it has to be removed (well, just desoldering the right hand terminal will do as well) to provide a spot to link A8 to the 41256's. If you leave R32 in place, the XL can not pull A8 to a low level, which in effect means that it 'sees' only 64K.

 

re-atari