Atari 800 256Kb upgrade

[ClausB]

I could see in your pictures where the black wire goes. It carries the /REF signal and should go from pin 2 of the LS10 to pin 1 of the LS158.

 

I have studied the schematic again an see no fault. We must find out why Q3 does not work. Could you solder a wire to Q3 and leave the other end loose so that it reaches above the board when the board is plugged in to the computer? Then you can measure the voltage on the Q3 signal with a volt meter. You must also solder another loose wire to GND of course. Also solder another one to pin 6 of LS158 Z503 pair (signal 2B).

 

In BASIC type POKE 53247,8 and check that the voltage on Q3 is above 3 V and the voltage on 2B is below 1 V. Then type POKE 53247,247 and check that the voltage on Q3 is below 1 V and the voltage on 2B is above 3 V.

[1050]

My bad there, I was looking at the previous schematic. Apparently I don't have enough confusion in my life as is and have to go this extra step to find enough. At best, by my feeble understanding, only one of the additional 158 (z503 pair) should be timed to the originals and the one used for refresh creation (z504 pair) should be as Claus has it being timed with the /REF signal to get that job done properly. Please revert to Claus' approved schematic and adhere to his suggestions/setup and tests. He is the master here, I'm the noob with my own questions hoping to get a few incidental answers resolved while you do all the work. And sorry for the interruption, please proceed.

[ebiguy]

My bad there, I was looking at the previous schematic. Apparently I don't have enough confusion in my life as is and have to go this extra step to find enough. At best, by my feeble understanding, only one of the additional 158 (z503 pair) should be timed to the originals and the one used for refresh creation (z504 pair) should be as Claus has it being timed with the /REF signal to get that job done properly. Please revert to Claus' approved schematic and adhere to his suggestions/setup and tests. He is the master here, I'm the noob with my own questions hoping to get a few incidental answers resolved while you do all the work. And sorry for the interruption, please proceed.

 

Please, do not apologize. You tried to help me and that's what I keep in mind. No problemo !

 

I could see in your pictures where the black wire goes. It carries the /REF signal and should go from pin 2 of the LS10 to pin 1 of the LS158.

 

I have studied the schematic again an see no fault. We must find out why Q3 does not work. Could you solder a wire to Q3 and leave the other end loose so that it reaches above the board when the board is plugged in to the computer? Then you can measure the voltage on the Q3 signal with a volt meter. You must also solder another loose wire to GND of course. Also solder another one to pin 6 of LS158 Z503 pair (signal 2B).

 

In BASIC type POKE 53247,8 and check that the voltage on Q3 is above 3 V and the voltage on 2B is below 1 V. Then type POKE 53247,247 and check that the voltage on Q3 is below 1 V and the voltage on 2B is above 3 V.

 

Here are the results:

POKE 53247,8 gives 0V on Q3 and 0,2V on 2B

POKE 53247,247 gives 4,1V on Q3 and 0,2V on 2B

 

1) It seems that the voltage of Q3 is the inverse of what is expected

2) 2B is always 0V whatever we have at Q3.

 

Here the layout of the board (maybe you will detect something wrong in my layout)

 

 

U1 = 74LS02

U2 = 74LS175

U3 = 74LS158

U4 = 74LS158 Z503 pair

U5 = original 74LS158 removed from Atari board and plugged in my PCB.

U6 = 74LS393

 

Without the yellow text so the path are more visible on the PCB

 

 

I did not check voltage on pin 14 but really on pin 15 of 74LS175. I add this information because pin 14 is /Q3 (which would explain the inverted results if I would have used the wrong pin)

Edited September 28, 2015 by ebiguy

[ClausB]

There is a mistake in the schematic: pin 5 of Z502 is really D3, so move your blue wire to pin 5. That should solve the inverted Q3. (Z502 pin 14 is really D4.)

 

The black wire should go from pin 2 of the LS10 at Z501 to pin 1 of the LS158 at U3.

[ebiguy]

There is a mistake in the schematic: pin 5 of Z502 is really D3, so move your blue wire to pin 5. That should solve the inverted Q3. (Z502 pin 14 is really D4.)

 

The black wire should go from pin 2 of the LS10 at Z501 to pin 1 of the LS158 at U3.

 

YES ! it works !!!!!!!!!!!!!!!!!

 

You are he best !

I would never think about a mistake in the Atari field manual (I checked that they also inverted pin 5 and pin 14 of Z502).

 

Thank you very much. I am really happy with my upgrade !!!

 

But, please, let me ask you a last advice.

All the mods have been done on one PCB since the start of this thread but I built 2 of them as I said in the first post.

I did all the changes and all the checks unsing the volt meter on the first PCB.

You last modification about Q3 was successful on the first PCB as I could check the volt with the 2 wires still in place.

But I only have 8 banks and pin 6 of Z503 pair (2B) is allways at 0,2V

 

I tried on the second PCB and discovered that the Q3 change fixed everything (I have the 16 banks)

 

So there is something which is broken on the first PCB.

Of course, after trying the mod suggested by 1050, I reverted to connecting pin 2 of Z501 with the black wire so both cards are wired the same way to conform to your diagram.

I was wondering if I broke something (maybe with too much heat while soldering - I am not an expert).

 

Do you have a check I could do to understand what is the problem with the pin 6 of Z503 pair (check the volt at a given pin for example) ?

If you don't have any clue, I will start a second upgrade from scratch

 

Again, thank you very much for your time and your expertise.

[ClausB]

That's good news. De rien!

 

Yes, perhaps too much soldering damaged the LS158 at U3.

 

You have done a very nice job on the PCBs. Congratulations.

 

My aunt lives in Paris (Sucy-en-Brie) and I visited her and her daughter from Lyon last year. Great town, Paris!

Edited September 29, 2015 by ClausB

[1050]

Only thing I would suggest is a new LS158 for your U3 which I have been calling Z504 pair? I assume this is the first thing you tried though. And I too fail to understand how my suggestion could have ruined it, I'm as baffled as you are that one works and the other doesn't.

 

My original concerns are no more since you posted the board diagrams where I see that the timing of the original 158s is driving the Z503 pair (your U4) which was my only concern due to the time delay injected by L501 - R501 pair driving the originals. It sounds as if I wasn't clear enough in expressing my concerns first time around since even I wouldn't have paid any attention to my own advice if I were you, but only because I would have known that the issue of concern was already taken care of. I never meant to suggest that the refresh creation circuit be driven from any other source than the /REF signal.

 

Further thought leads me to suggest that you do a thorough cleaning of the test board with the two added output wires for monitoring. Sometimes flux left on the board can influence the data state of nearby chips and introduce errors where there should be none. With soldering heat, the flux can break down into acidic compounds and combined with different metals in the pins soldered VS the copper of the board for example, we have in essence created a micro sized battery feeding the wrong signals into the circuit and this battery is not in the schematic. By cleaning the electrolyte away (burnt flux, rosin, crud, etc.), the battery ceases to function and/or influence the circuit. For cleaning, one can use your sister's toothbrush combined with automotive brake cleaner in a spray can alternating with automotive carburetor cleaner in a spray can until visible signs of flux are all gone with a scrubbing rinse done for good measure. Since these cleaners can be flammable and/or often have noxious fumes this should ONLY be done outside and let the cleaner dry completely. Compressed air works great to speed this part up but is hard to come by without a dedicated air compressor for the purpose, in a pinch, you don't really need it, it's just really nice to have.

[ebiguy]

That's good news. De rien!

 

Yes, perhaps too much soldering damaged the LS158 at U3.

 

You have done a very nice job on the PCBs. Congratulations.

 

My aunt lives in Paris (Sucy-en-Brie) and I visited her and her daughter from Lyon last year. Great town, Paris!

 

Thank you !

I really happy with the result because electronics is not my "cup of tea" !

Maybe I will begin another thread with all the files updated and the PCB files so that other guys can order both PCB from a manufacturer and build the upgrade.

With a PCB it's quite easy to do. Without it's a nightmare to wire all pins !!!

 

I live in Paris. I would love to meet people from the Atari community here in France.

If you ever come back in France, please PM me first. That would be great. You are the man who make this upgrade possible !

 

I thought you were german at first because of your name.

Are your parents German ? Did you live in Germany at one time ?

 

Only thing I would suggest is a new LS158 for your U3 which I have been calling Z504 pair? I assume this is the first thing you tried though. And I too fail to understand how my suggestion could have ruined it, I'm as baffled as you are that one works and the other doesn't.

 

My original concerns are no more since you posted the board diagrams where I see that the timing of the original 158s is driving the Z503 pair (your U4) which was my only concern due to the time delay injected by L501 - R501 pair driving the originals. It sounds as if I wasn't clear enough in expressing my concerns first time around since even I wouldn't have paid any attention to my own advice if I were you, but only because I would have known that the issue of concern was already taken care of. I never meant to suggest that the refresh creation circuit be driven from any other source than the /REF signal.

 

Further thought leads me to suggest that you do a thorough cleaning of the test board with the two added output wires for monitoring. Sometimes flux left on the board can influence the data state of nearby chips and introduce errors where there should be none. With soldering heat, the flux can break down into acidic compounds and combined with different metals in the pins soldered VS the copper of the board for example, we have in essence created a micro sized battery feeding the wrong signals into the circuit and this battery is not in the schematic. By cleaning the electrolyte away (burnt flux, rosin, crud, etc.), the battery ceases to function and/or influence the circuit. For cleaning, one can use your sister's toothbrush combined with automotive brake cleaner in a spray can alternating with automotive carburetor cleaner in a spray can until visible signs of flux are all gone with a scrubbing rinse done for good measure. Since these cleaners can be flammable and/or often have noxious fumes this should ONLY be done outside and let the cleaner dry completely. Compressed air works great to speed this part up but is hard to come by without a dedicated air compressor for the purpose, in a pinch, you don't really need it, it's just really nice to have.

 

I tried to replace U3 with a new one but it does not work either. So I guess there is something else.

I don't have all the material you're talking about to clean the PCB. I think it will be much easier for me to start a new one.

Thank you also for your advices. The important is the fact that you wanted to help.

That's really great to ask people in this forum because you always find someone to help you.

[ClausB]

Perhaps U4 is damaged?

 

Yes, my parents are from Germany and my relatives are there, except for the aunt who married a Parisian. Last summer my children and I visited most of them. We drove a tiny Peugeot 4000 km in two weeks! Approaching Paris, we laughed at this sign. We went straight and our car did not explode.

 

Edited September 29, 2015 by ClausB

[ebiguy]

Perhaps U4 is damaged?

 

Yes, my parents are from Germany and my relatives are there, except for the aunt who married a Parisian. Last summer my children and I visited most of them. We drove a tiny Peugeot 4000 km in two weeks! Approaching Paris, we laughed at this sign. We went straight and our car did not explode.

 

Maybe you won't believe it but I laughted for 3 minutes (and my wife also) when I saw your picture !!!

[Magic Knight]

Im away to make make gerbers and produce boards for making this - and see if it works (hopefully within a few weeks) .

 

Anyone tried this with larger memory than 256k or know if it would work?

[Gunstar]

This upgrade doesn't give you 4K of extra main ram does it? I've sort of fallen in love with the idea of an extra 4K main ram, until I can get an Incognito. I'd be satisfied with 128K ram with the extra main ram, 256K or 512K would be better, If I can get them with 52K main memory.

[ClausB]

No but this one gives 52K.

http://atariage.com/forums/topic/185500-using-the-atari-400-48k-board-in-the-800/page-2?do=findComment&comment=2727204

 

Though it's not compatible with this 256K mod.

[Gunstar]

Thanks! ClauseB, I'm looking into doing the 62K mod you pointed to in the other thread. I've got a couple other expanded memory computers with XE and Rambo compatibility, so with my 800 I'm more concerned in getting as close as I can to 64K XL compatibility rather than expanded Axlon memory, at least until an Incognito board finds it's way into my hands.

Edited November 15, 2018 by Gunstar

[Magic Knight]

Would Axlon clones in general allow me to load Dropezone (64k) using either tape loading, side 2, the!cart etc? Or is this flash memory for internal pre dos handling only?

 

What I mean is can I boot a 64k game xex or Atr from a sidecart using this on an 800?

Edited November 15, 2018 by Magic Knight

[Gunstar]

No. Axlon is adding extended banks of memory only, beyond the normal 48K limit of main memory and swaps them in 8K(?) banks with part of the 48K main memory. 130XE and XL memory upgrades do the banking in 16K banks, and, of course, have the extra main memory of 64K too. You have to have main memory on the 800 beyond 48K mapped the same way the extra 16K memory for 64K XL/XE's are mapped for it to be compatible. I think the 800 62K upgrade gets it close, and more software will work that is "64K" but not all? Apparently, the maximum main memory that XL/XE's have access to is really 62K, because 2K ram absolutely has to be used for I/O. I don't know if you would have to take that into account also, with the 800 62K upgrade, or if that is already accounting for the I/O too.

Edited November 15, 2018 by Gunstar

[Magic Knight]

No. Axlon is adding extended banks of memory only, beyond the normal 48K limit of main memory and swaps them in 8K(?) banks with part of the 48K main memory. 130XE and XL memory upgrades do the banking in 16K banks, and, of course, have the extra main memory of 64K too. You have to have main memory on the 800 beyond 48K mapped the same way the extra 16K memory for 64K XL/XE's are mapped for it to be compatible. I think the 800 62K upgrade gets it close, and more software will work that is "64K" but not all? Apparently, the maximum main memory that XL/XE's have access to is really 62K, because 2K ram absolutely has to be used for I/O. I don't know if you would have to take that into account also, with the 800 62K upgrade, or if that is already accounting for the I/O too.

Thanks for clarifying. I'm looking for the capacity of an 800XL but with feel of the nice 800 :0)

 

Guess going for 62k might be best option for me.

[ClausB]

Yes, 62K RAM is the most you can get without banking, because of the 2K I/O space, even in an XL or XE. It's always been called 64K mode but it's only 62K really.

 

Axlon uses 16K banks too, although the banking register is not PORTB. Axlon upgrades won't help in playing 64K games.

[_The Doctor__]

Yes, 62K RAM is the most you can get without banking, because of the 2K I/O space, even in an XL or XE. It's always been called 64K mode but it's only 62K really.

 

Axlon uses 16K banks too, although the banking register is not PORTB. Axlon upgrades won't help in playing 64K games.

not without changing the code to follow the banking scheme, not that difficult if you think about it....

[+Nezgar]

Axlon banks are 16KB, in the same address range as 130XE/RAMBO/Compy Shop ($4000-$7FFF), but use a different control register.

 

Mosaic banks are 4KB, mapped in the C000-CFFF range "normally unused" in the 800. Without banking, you achieve the so-called 52K of main memory with this, with the last 4K bank switchable.

[Gunstar]

Yes, 62K RAM is the most you can get without banking, because of the 2K I/O space, even in an XL or XE. It's always been called 64K mode but it's only 62K really.

 

Axlon uses 16K banks too, although the banking register is not PORTB. Axlon upgrades won't help in playing 64K games.

That's right, it's PortB...I don't know why I always remember it as an 8K vs. 16K issue. I mean, I know the PortB verses Axlon mapping where ever, but for some reason I've got it stuck in my head that some extended memory upgrades use 8K banks (maybe it's a carry-over from rom banking) instead of 16K and that Axlon standard was one of them...thanks for setting me, and the record, straight.

Edited November 15, 2018 by Gunstar

[Gunstar]

Thanks for clarifying. I'm looking for the capacity of an 800XL but with feel of the nice 800 :0)

 

Guess going for 62k might be best option for me.

A lot of the commercial productivity software like Visicalc and Syncalc and other Syn-series software, and many word processors have/had Axlon compatible versions and then XE/Rambo conversions of them came later with the 130XE's release. So if that and ram-disks is all you are after, otherwise compatibility up to 48K, then Axlon is the way for you to go. But if you are just looking for better compatibility with 64k Software then I guess the 800 62K upgrade will help some. But without the Incognito board, you will never get full 64K base and XE/Rambo compatible extended memory. Apparently. Unless, of course, some one here builds an Incognito Lite or something without all the extras of U1MB memory and SDX, CF card, etc. I'd be satisfied with 64K base XL compatible and a switch, manual or software, to go into Axlon mode for extended memory. I was thinking I might be able to put some kind of kludge together in this regard for my own use, a one-off prototype, using a combination of upgrade schematics or something.

Edited November 15, 2018 by Gunstar

[ClausB]

Again, my 62K upgrade should be compatible with the "64K base XL" mode. It even uses PortB (you have to run a wire to the joystick port). That said, I never built and tested it but I'm willing to help you test and debug it, as I did for ebiguy.

[Gunstar]

Again, my 62K upgrade should be compatible with the "64K base XL" mode. It even uses PortB (you have to run a wire to the joystick port). That said, I never built and tested it but I'm willing to help you test and debug it, as I did for ebiguy.

OK, so it get it done some how, using my 3 16K boards, and then maybe expand it via an 800 card version of your original 256K XL upgrade! Or, I create a switch to turn off PortB (that is back to controlling controller ports 3 and 4) out of "64K mode" back to 48K base and extended Axlon memory. The Axlon memory will not be coming from the same board(s) used for the 62K upgrade, that's not what I'm saying. I'm saying a board or two used for the 62K upgrade and then a separate Axlon board that can be activated when the 62k Board(s) go into standard 48K base mode. Unless of course someone here builds an Incognito Lite for it all to be on one board-and no Claus, I'm not hinting at you or anyone else in particular to do this, just a general hint to the community, maybe someone will decide to pick up the torch.

 

It's funny how things change so quickly in life. A week ago anything like this was the furthest thing from my mind. If I ever happened across an Incognito at the right time for the right price, I would then get an 800. Then I'm gifted with an 800 out of the blue, and looking for a new XL compatibility board. Of course I have to hack it now, that's half the hobby for me!

Edited November 15, 2018 by Gunstar

[ClausB]

I have been thinking about picking up that torch actually. I plan to do an experiment this winter to see whether my idea would work.