Dead New-in-Box 600XL?

[flashjazzcat]

OK: expert help required on the Antic problem. Before I think about replacing the socket, here are the symptoms

 

 

And with index finger touching CPU, approximately covering pins 24-28 (A14-15, and D5-D7):

 

 

Can't find any shorts or breaks, but if I understood why my body's capcitance or conductivity was fixing the problem...

 

The only alternative is to require the insertion of body part into the case for normal operation!

[Mclaneinc]

Looks like a freaky earthing issue..

[Rybags]

I get something just like that since I put the 32in1 OS into my machine.

 

Although nowhere near as bad, maybe a dozen speckles per second and it comes and goes.

 

I suspect with mine it's some sort of connectivity issue too... when doing my VBXE I botched some traces to Antic.

 

I've not looked at it properly yet - the machine doesn't crash although reads to $E000 ROM area sporadically return the wrong data.

[+bob1200xl]

That is ANTIC latching data ($FF) when nobody is sending it to the buss. ANTIC is different from the other chips in an Atari. He does his own address decoding directly from the buss, whereas all the other chips get decoded from the MMU or LS138. I get 'birdies' like that on the 7mhz 65816 if I don't allow enough Phase02 overlap into the next cycle. In this case, the fast memory drops off the buss before ANTIC latches the data. ANTIC decodes A15-A8, so touching A15 and A14 will skew the decode just enough to prevent the problem.

 

Yeah - so?

 

We can't 'see' inside ANTIC, so it's just guesswork. I have found differences in ANTIC chips vs. 65816s, as well as differences in 6502s. You may find a set that works together, or you may not. The 1200XL has problems with cartridge selection because of poor ground sourcing and loose decodes on $D5xx. You see factory fixes on 800XLs for power distribution. Some (all?) 130XEs have re-driven GTIAs.

 

Bottom line is that the Atari runs twice as fast as an Apple, barely... Most circuits are not timed, just relying on the 'speed' of devices on the buss. Old, 250ns ROMs and such could be de-selected and you could count on them holding their data on the buss for tens of ns. New flash storage drops its data 'right now'. Noise on ground could not be seen by old, vintage circuits. Address transitions were too fast to trigger anything. What you may see is newer build chips being too fast for their own good.

 

Anyway, try some original chips from a 400/800. See if they are better behaved. Your Atari is running on the edge.

 

Otherwise, wait for a hardware fix. Are you sure that your 600XL boards are identical?

 

Bob

 

 

 

 

OK: expert help required on the Antic problem. Before I think about replacing the socket, here are the symptoms

 

 

And with index finger touching CPU, approximately covering pins 24-28 (A14-15, and D5-D7):

 

 

Can't find any shorts or breaks, but if I understood why my body's capcitance or conductivity was fixing the problem...

 

The only alternative is to require the insertion of body part into the case for normal operation!

[flashjazzcat]

That is ANTIC latching data ($FF) when nobody is sending it to the buss. ANTIC is different from the other chips in an Atari. He does his own address decoding directly from the buss, whereas all the other chips get decoded from the MMU or LS138. I get 'birdies' like that on the 7mhz 65816 if I don't allow enough Phase02 overlap into the next cycle. In this case, the fast memory drops off the buss before ANTIC latches the data. ANTIC decodes A15-A8, so touching A15 and A14 will skew the decode just enough to prevent the problem.

 

Yeah - so?

 

We can't 'see' inside ANTIC, so it's just guesswork. I have found differences in ANTIC chips vs. 65816s, as well as differences in 6502s. You may find a set that works together, or you may not. The 1200XL has problems with cartridge selection because of poor ground sourcing and loose decodes on $D5xx. You see factory fixes on 800XLs for power distribution. Some (all?) 130XEs have re-driven GTIAs.

 

Bottom line is that the Atari runs twice as fast as an Apple, barely... Most circuits are not timed, just relying on the 'speed' of devices on the buss. Old, 250ns ROMs and such could be de-selected and you could count on them holding their data on the buss for tens of ns. New flash storage drops its data 'right now'. Noise on ground could not be seen by old, vintage circuits. Address transitions were too fast to trigger anything. What you may see is newer build chips being too fast for their own good.

 

Anyway, try some original chips from a 400/800. See if they are better behaved. Your Atari is running on the edge.

 

Otherwise, wait for a hardware fix. Are you sure that your 600XL boards are identical?

Good points. The other board (i.e. the one I originally started this thread about, which has a RAM problem) from which I'm borrowing chips is a Rev. A, while the board with the Antic problem is Rev. B. I have indeed found that the Antic from the "other" board doesn't work well at all in the Rev. B board (virtually blank screen with some junk on it). Also, swapping the MMUs causes similar problems.

 

In any case, the Antic problems arose after socketing the chip and removing the RF modulator. I'm going ahead with the rest of the chip relocation/desocketing work necessary to get the VBXE fitted until I nail this problem. At least that's going well. I had to mount that ugly-ass EP8212 chip on the underside of the board.

 

I have a few Antics and 6502s which I can try swapping. What puzzles me is: what's changed? Could the loss of the RF mod have affected the overall capacitance of the circuit?

[+bob1200xl]

I don't know about 600XLs, but removing the modulator on a 1200XL requires you to restore the ground circuit that was incorporated in the 1200XL modulator. You have the shield amd the ground pin that need to remain connected.

 

Although, I did the modulator on a 600XL and I do not remember any such issues.

 

Your 600XL is not socketed? I thought they all were...?

 

Bob

 

 

 

That is ANTIC latching data ($FF) when nobody is sending it to the buss. ANTIC is different from the other chips in an Atari. He does his own address decoding directly from the buss, whereas all the other chips get decoded from the MMU or LS138. I get 'birdies' like that on the 7mhz 65816 if I don't allow enough Phase02 overlap into the next cycle. In this case, the fast memory drops off the buss before ANTIC latches the data. ANTIC decodes A15-A8, so touching A15 and A14 will skew the decode just enough to prevent the problem.

 

Yeah - so?

 

We can't 'see' inside ANTIC, so it's just guesswork. I have found differences in ANTIC chips vs. 65816s, as well as differences in 6502s. You may find a set that works together, or you may not. The 1200XL has problems with cartridge selection because of poor ground sourcing and loose decodes on $D5xx. You see factory fixes on 800XLs for power distribution. Some (all?) 130XEs have re-driven GTIAs.

 

Bottom line is that the Atari runs twice as fast as an Apple, barely... Most circuits are not timed, just relying on the 'speed' of devices on the buss. Old, 250ns ROMs and such could be de-selected and you could count on them holding their data on the buss for tens of ns. New flash storage drops its data 'right now'. Noise on ground could not be seen by old, vintage circuits. Address transitions were too fast to trigger anything. What you may see is newer build chips being too fast for their own good.

 

Anyway, try some original chips from a 400/800. See if they are better behaved. Your Atari is running on the edge.

 

Otherwise, wait for a hardware fix. Are you sure that your 600XL boards are identical?

Good points. The other board (i.e. the one I originally started this thread about, which has a RAM problem) from which I'm borrowing chips is a Rev. A, while the board with the Antic problem is Rev. B. I have indeed found that the Antic from the "other" board doesn't work well at all in the Rev. B board (virtually blank screen with some junk on it). Also, swapping the MMUs causes similar problems.

 

In any case, the Antic problems arose after socketing the chip and removing the RF modulator. I'm going ahead with the rest of the chip relocation/desocketing work necessary to get the VBXE fitted until I nail this problem. At least that's going well. I had to mount that ugly-ass EP8212 chip on the underside of the board.

 

I have a few Antics and 6502s which I can try swapping. What puzzles me is: what's changed? Could the loss of the RF mod have affected the overall capacitance of the circuit?

[flashjazzcat]

I don't know about 600XLs, but removing the modulator on a 1200XL requires you to restore the ground circuit that was incorporated in the 1200XL modulator. You have the shield amd the ground pin that need to remain connected.

 

Although, I did the modulator on a 600XL and I do not remember any such issues.

 

Your 600XL is not socketed? I thought they all were...?

Well, there's something bloody funny going on with it. I'm nearly finished the VBXE prepping work (getting into the flow now; only damaged one trace out of seven ICs), and the status-quo remains... still the old finger on Sally trick.

 

Re socketing: note, I said "desocketing". There are seven socketed ICs which get in the way of the VBXE board in a 600XL. The sockets have to be removed and the ICs refitted, either straight to the top of the board or underneath. The socketing in XLs is usually a nice thing, but soldered-to-the-board ICs I find much easier to remove.

 

Anyway: nearly done, and nicely warmed up to continue somewhat more confidently with the SIO2SD job. You can see a nice photo of a VBXE in a 600XL in Candle's Ultimate 1MB thread. I'll post a pic of mine when it's done.

 

Still unsure whether to just go for broke and replace the Antic precision socket. However, I don't want to go through all that and find the problem persists. Logic dictates that there are no bad contacts on the socket.

[flashjazzcat]

Glad that part's over:

 

 

 

I tried another Antic, still the same crazy display...

Edited March 20, 2011 by flashjazzcat

[AtariGeezer]

Have you checked any capacitors inline with pins 24-28 (A14-15, and D5-D7) from the CPU to ground? maybe a dry cap or a cold solder connection?

[flashjazzcat]

Have you checked any capacitors inline with pins 24-28 (A14-15, and D5-D7) from the CPU to ground? maybe a dry cap or a cold solder connection?

Having totally lost faith in the Antic socketing work, I removed the socket and painstakingly repaired the damaged traces, before fitting a new precision socket. Sadly there's now no picture at all, so it looks like this board's heading for the bin.

 

Forewarned is forearmed, so I'll just have to be more careful next time (assuming I ever get hold of another working 600XL or can fix the other one). Having done lots of trouble-free work on XE boards, I was assuming the XL traces to be as robust, and they aren't.

[flashjazzcat]

Fixed it. Turns out I had connected HALT to REF when fixing the traces. Jumper in Antic pin 8 instead of 9... just as well I realized that before I threw the board out of the window. Actually I was doing point-by-point tracing using the schematic by this stage..

 

 

Not pretty: 20 traces needed repair after the second time of removing the socket. The soldering's a little thick, since I'd already reflowed the joints a couple of times before I realized the mistake with the Halt line.

 

Regardless of the hours I spent on this, it's been a valuable experience. They say the hallmark of a good musician is the ability to recover from a mistake... Well, at least I'm prepared for the worst now

 

I should add that the screen garbage problem has gone as well. That first precision socket was obviously just fatally botched. I think it's safe to continue dremelling that 1064 now...

Edited March 21, 2011 by flashjazzcat

[mimo]

nice save, another 600xl lives to fight another day

[flashjazzcat]

nice save, another 600xl lives to fight another day

And so it does!

 

 

Just needs Ultimate 1MB now, and it's finished!

[flashjazzcat]

Pleased to report this machine plays nicely with the KMK/JZ HDD interface - and without any timing mods. I just fitted an IntSDX, and for extra RAM I'm gonna flash the VBXE core which emulates the Compy shop upgrade. Turning into a really sweet little Atari.

 

BTW: anyone confirm the modifications necessary to turn the BASIC socket on the 600XL into a second OS-ROM socket (apart from the obious soldering in of a 28 pin socket)? BASIC's pins 18, 20, 21 and 24 do not correspond to the relevant OS ROM pins (20, 22, 23, 26). I understand (or have deduced, at least) that BASIC pins 18 and 20 will need to be connected to pin 15 of the MMU to duplicate the OS ROM functionality. Pins 21 and 24, though, I'm not so sure about.

 

Bob - surely you know this.

 

Why? Because I want to put a stereo Pokey board alongside the IntSDX (until such time as the Ultimate 1MB arrives), and although the two won't co-exist with the IntSDX adapter in the OS ROM socket, they will if it's in the BASIC socket. The loss of BASIC is not an issue, and there are other reasons why I want to tinker with this arrangement.

 

After the weekend's shennanegins, though, I want to be doubly sure I know what I'm doing.

[sloopy]

just be sure to use a option swapped OS, or hold option when booting... it does make a difference...

 

sloopy.

[flashjazzcat]

Thanks - it's probably worth bearing in mind. But these other two pins... ???

[+bob1200xl]

The 600XL has jumpers that set either 24 or 28 pin ROMs. You seem to be referencing pins by the 24 pin ROM labels, where Atari lists them on the schematic by their 28 pin counts. So, what you're asking refers to pins 23 and 26, not 21 and 24, OK?

 

The jumpers are hard to find and not clearly defined in the schematic. They look to be near the u/r corner of the ROM. You will have to meter them to be sure. One jumper (W6) connects A13 (not present on the BASIC 8K ROM) to +5v. This jumper will have to be removed and A13 will need to be connected to its pin. (pin 26) The other jumpers will have to be 'moved' to their alternate, 28 pin wiring.

 

If you are careful, you should be OK as long as each pin on the new 28 pin socket (where BASIC used to be) corresponds to the same pin on your old OS socket. Just meter them.

 

I've never tried it, myself...

 

Bob

 

 

 

Pleased to report this machine plays nicely with the KMK/JZ HDD interface - and without any timing mods. I just fitted an IntSDX, and for extra RAM I'm gonna flash the VBXE core which emulates the Compy shop upgrade. Turning into a really sweet little Atari.

 

BTW: anyone confirm the modifications necessary to turn the BASIC socket on the 600XL into a second OS-ROM socket (apart from the obious soldering in of a 28 pin socket)? BASIC's pins 18, 20, 21 and 24 do not correspond to the relevant OS ROM pins (20, 22, 23, 26). I understand (or have deduced, at least) that BASIC pins 18 and 20 will need to be connected to pin 15 of the MMU to duplicate the OS ROM functionality. Pins 21 and 24, though, I'm not so sure about.

 

Bob - surely you know this.

 

Why? Because I want to put a stereo Pokey board alongside the IntSDX (until such time as the Ultimate 1MB arrives), and although the two won't co-exist with the IntSDX adapter in the OS ROM socket, they will if it's in the BASIC socket. The loss of BASIC is not an issue, and there are other reasons why I want to tinker with this arrangement.

 

After the weekend's shennanegins, though, I want to be doubly sure I know what I'm doing.

[sloopy]

just wire up all the address lines to where they should go:

 

 

   Basic        ROM OS         New OS Socket
  - N/A     28 - Vcc        Pin 24 of BASIC socket
  - N/A     27 - /PGM       Pin 24 of BASIC socket
24 - Vcc     26 - A13        Pin 26 of OS socket
23 - A8      25 - A8         same
22 - A9      24 - A9         same
21 - A12     23 - A11        Pin 18 of BASIC socket
20 - CS      22 - /OE        wire to an OS select switch and pullup resistor
19 - A10     21 - A10        Pin 19 of BASIC socket
18 - A11     20 - /CE        wire to an OS select switch and pullup resistor
17 - D7      19 - D7         same
16 - D6      18 - D6         same
15 - D5      17 - D5         same
14 - D4      16 - D4         same
13 - D3      15 - D3         same

 

 

sloopy.

[+orpheuswaking]

o_O sounds complicated!!

[Mclaneinc]

Nice save Jon...I hate hearing that an Atari has gone to the great database in the sky.

[flashjazzcat]

The 600XL has jumpers that set either 24 or 28 pin ROMs. You seem to be referencing pins by the 24 pin ROM labels, where Atari lists them on the schematic by their 28 pin counts. So, what you're asking refers to pins 23 and 26, not 21 and 24, OK?

 

The jumpers are hard to find and not clearly defined in the schematic. They look to be near the u/r corner of the ROM. You will have to meter them to be sure. One jumper (W6) connects A13 (not present on the BASIC 8K ROM) to +5v. This jumper will have to be removed and A13 will need to be connected to its pin. (pin 26) The other jumpers will have to be 'moved' to their alternate, 28 pin wiring.

 

If you are careful, you should be OK as long as each pin on the new 28 pin socket (where BASIC used to be) corresponds to the same pin on your old OS socket. Just meter them.

 

I've never tried it, myself...

Thanks Bob. These are the jumpers you're talking about, with their connections to the BASIC ROM socket (using 28 pin numbering):

 

 

So presumably it's a case of cutting those two traces and inserting the relevant jumpers. Just got to take a moment to work this out...

 

just wire up all the address lines to where they should go:

 

 

   Basic        ROM OS 		New OS Socket
  - N/A 	28 - Vcc        Pin 24 of BASIC socket
  - N/A 	27 - /PGM   	Pin 24 of BASIC socket
24 - Vcc 	26 - A13        Pin 26 of OS socket
23 - A8      25 - A8 		same
22 - A9      24 - A9 		same
21 - A12 	23 - A11        Pin 18 of BASIC socket
20 - CS      22 - /OE        wire to an OS select switch and pullup resistor
19 - A10 	21 - A10        Pin 19 of BASIC socket
18 - A11 	20 - /CE        wire to an OS select switch and pullup resistor
17 - D7      19 - D7 		same
16 - D6      18 - D6 		same
15 - D5      17 - D5 		same
14 - D4      16 - D4 		same
13 - D3      15 - D3 		same

 

 

sloopy.

Thanks: maybe that will help me make sense of the jumpers...

Edited March 23, 2011 by flashjazzcat

[sloopy]

those 5 holes are:

 

o - A12

o - Pin 21/23 (28/24) BASIC socket

o - A11

o - Pin 18/20 (28/24) BASIC socket

o - Pin 20/22 (28/24) BASIC socket

 

 

sloopy.

[+bob1200xl]

The W6 jumper is right next to pin 28 of the Basic ROM. Have you made up a set of instructions yet? Post what you are doing and I'll try it on one of my 600XLs.

 

Bob

 

 

 

The 600XL has jumpers that set either 24 or 28 pin ROMs. You seem to be referencing pins by the 24 pin ROM labels, where Atari lists them on the schematic by their 28 pin counts. So, what you're asking refers to pins 23 and 26, not 21 and 24, OK?

 

The jumpers are hard to find and not clearly defined in the schematic. They look to be near the u/r corner of the ROM. You will have to meter them to be sure. One jumper (W6) connects A13 (not present on the BASIC 8K ROM) to +5v. This jumper will have to be removed and A13 will need to be connected to its pin. (pin 26) The other jumpers will have to be 'moved' to their alternate, 28 pin wiring.

 

If you are careful, you should be OK as long as each pin on the new 28 pin socket (where BASIC used to be) corresponds to the same pin on your old OS socket. Just meter them.

 

I've never tried it, myself...

Thanks Bob. These are the jumpers you're talking about, with their connections to the BASIC ROM socket (using 28 pin numbering):

 

 

So presumably it's a case of cutting those two traces and inserting the relevant jumpers. Just got to take a moment to work this out...

 

just wire up all the address lines to where they should go:

 

 

   Basic        ROM OS 		New OS Socket
  - N/A 	28 - Vcc        Pin 24 of BASIC socket
  - N/A 	27 - /PGM   	Pin 24 of BASIC socket
24 - Vcc 	26 - A13        Pin 26 of OS socket
23 - A8      25 - A8 		same
22 - A9      24 - A9 		same
21 - A12 	23 - A11        Pin 18 of BASIC socket
20 - CS      22 - /OE        wire to an OS select switch and pullup resistor
19 - A10 	21 - A10        Pin 19 of BASIC socket
18 - A11 	20 - /CE        wire to an OS select switch and pullup resistor
17 - D7      19 - D7 		same
16 - D6      18 - D6 		same
15 - D5      17 - D5 		same
14 - D4      16 - D4 		same
13 - D3      15 - D3 		same

 

 

sloopy.

Thanks: maybe that will help me make sense of the jumpers...

[flashjazzcat]

Yes: I see the W6 jumper at the top left of the socket. That's one I'd missed.

 

Busy trying to figure out just which upgrade is going where at the moment (one of my IntSDX boards looks like it has a bad GAL flash, so that's scuppered some temporary plans), but naturally if and when I get around to this mod, I'll write it up for others to share.

[sloopy]

fjc,

if you leave the /CE in place, will this allow a 16k rom to be used instead of BASIC? or didnt you look into that deep?

 

sloopy.