Thanks, Jan! Great review! I did encounter a lot of that flicker during development of the board, even in an 800XL. But eventually I changed things around in the design and got rid of it entirely, both in the 800XL and 130XE. When I saw the flicker you were getting, I even thought I might have sent you an earlier prototype board, instead of the final design! But checking in my inventory here, I think I did send you the correct one. It is curious that you got the flicker. It has to do with the clock lines of the Atari's two-phase clock; those are both very timing-sensitive and very load/signal-level-sensitive, which work against each other: to resolve the load/signal-level issue, you want to boost everything through an HCT buffer, but that screws up the timing, as HCT parts have 18ns propagation delay. Standard LS parts have only 8ns delay, but less drive. So the end design is a balance. The only thing I could think of was that those extra pin-headers you used first didn't make good contact, but then you changed those for proper pin-headers, so it can't be that. Anyway, this is a good cautionary tale, so I updated the documentation to inform people that you had this problem with a 65XE. On a different note: the pin-headers I sent you do fit in precision sockets, but you have to install them with the thick side up, and the thin side down. Cheers!
20:41-"yeah, and we're going to have to take the whole circuit board out to get the shield off ............which i put back in there. " The moment of realization of the act responsible for this was priceless! lol
So much variety in these old machines, as always this hobby has its quarks to work out, you never can predict what will happen either, you would think that the adapter would replace somethin' specifically and should in theory then replace that chip or component without any change or harm to anythin' else but clearly Freddie doesn't like a few new chips in this case, oh well at least it does work. Danke Jan!
Well when we were kids, my bros and I used to play mini basketball in the room while waiting for a game to load...So the rust on my 800xl shield is sweat....lol.
Rockwell manufactured chips on contract for several companies. The 6502s they made for Apple were not the Atari SALLY version of the 6502. After all, Apple didn't need the halt line for use with Atari's ANTIC or MARIA graphics chips.
The rust can just came from moisture, where the device was stored. Maybe a cellar - that's enough getting some rust on that part. It's not real liquid damage...
My guess as to the rust is that there was a humidifier in the room it lived in and the moisture settled through the vents and condensed on the colder (usually) metal.
I am amazed to see 6502's still being manufactured today. Perhaps this upgrade may be a great idea, for the new model should be far more efficient, thus even cooler than the original processor.
Unfortunately the modern ones are not drop-in replacements for the original 6502 either, as some of the "illegal" op-codes are not supported afaik that many old programs (games/demos etc) used.
The 65xe was manufactured in different variants so not all of them are easily upgradable to 130XE. If you see a 65xe that does not have ECI slot next to the cartridge slot, it's a sign that it uses a dedicated board. Anyway, if you are aiming for an upgrade, you not only need the extra RAM chips but also the EMMU that comes only in the original 130. You can replace the EMMU with some programmable logic chips but when you do, it is more reasonable to do an upgrade to something like 1meg of SRAM (kits available from hobbyists around the world).
U34 was called Enhanced Memory Management Unit, or EMMU for short. It can be substituted with a GAL and some generic 74xx chip (as U35). Real EMMU didn't need U35.
I learned to code on a 65 XE. Loved the angular design. Loved older ThinkPads because of those design elements too. The rusty shield with no internal component corrosion may be due to use in a coastal area. Mine had that too when I opened it.
I have done a similar adapter for 65C816, and experienced similar problems. It kind of works, but is unstable. I've managed to run YOOMP! on it. It also worked better on XL. My adapter used only 2 TTL chips (I've used SMD), which I've fitted under the CPU socket.
Judging by the flicker etc. That came up with the adapter and replacement chip.! It seems to be more of an issue with grounding.! All you need to do is find a good ground point on the motherboard and patch a wire directly to the ground point to the replacement chip.!! I have discovered that this is a common fault upgrading older CPU.!
Might be worth trying again once you upgrade the ram. The addition of the multiplexer logic and the extra chips onto the bus might have the effect of synching things up, perhaps by adding more delay to things. I'd source similar (ideally identical) chips and see what happens.
This is plausible indeed. The flicker has to do with both timing of the PHI2 clock line, and the load on that line, which may exceed the drive capacity of the two-phase clock generation logic.
@@thebyteattic Hmm, so it's not just a timing issue then: the added logic for the extra ram bumps the current up enough to run the memory subsystem properly with the modern CMOS (it's CMOS, right?) 6502, correct?
Hi Jan! Probably stupid question to ask before I watch video till the end but what is the reason for swapping 65c02 instead Sally? Just for a compatibility test?
They are still in production and run more efficiently. They have some incompatibilities though, especially with illegal opcodes that are used by some software.
@@JanBeta "and run more efficiently" as opposed to what? is this a power/thermal thing or a code thing? does the replacement run at a faster clock or is this more a case if you have a dead CPU this could be used as a replacement? Ive never had stability issues with the original SALLY and they seem pretty reliable
@@DanafoxyVixen the power consumption of CMOS processor is 10 to 20 times lower than nMOS based (wikipedia). But I wouldn't consider swapping Sally with 65c02 just because of this if Sally is working well.
You will most likely run into issues with many modern demos that use the so-called illegal instructions, which do not work with the 65C02 (I mean even with the XL machines).
i seems that there is some timing investigation necessary. do you have a picoscope? it would also show how clean the signals are. maybe changing the logic chips to faster variants could help. ok, i read somewhere that the sally chip had a halt-pin that halted the cpu and decoupled it from the bus. because you have glitches on the graphic, i assume there is a collision on the bus. the graphic chip needs to access the bus but the cpu is not fully decoupled with the bus transceiver chip. so, somehow the halt signal does not do it's job fast enough. this could also be problematic when halt goes away and the cpu access is to late for a read or write. maybe you take a look on the 74xxx variants that are available. there are endless variants of ls and hc and hct and f and what not. i know that ti has some fast ones. i don't know if Microchip also makes some (which i should probably know). if you need some samples from Microchip (Atmel, Micrel) , please let me know. ok, i found some schematic from 1987: atariage.com/forums/uploads/monthly_04_2018/post-14026-0-22693400-1523129848.jpg i don't know if this is the same that you have. but as i saw, there is a 74ls02 and 74f244, 74f245 on it. these are old chips from the 80s. please look for faster variants and check the signal delay from the halt to the phi2 signal. and the "glitches" on the output of the 74x244 as i never owned an atari, i am no expert on that but if you have some questions regarding that stuff (or samples from Microchip), contact me on facebook.
@@JanBeta Ah!! ... check that the bass guitar back there at 23:04..... "@Jan Beta The bassist"?? ;-) .. with "Pedal Mom" @ Fran Blanche (another credit from the good old Germany) we have half of "The Retro Tech rock band " :-DD >> ruclips.net/video/kNrXpn_YwbM/видео.html
I'm a Commodore guy, but looking at the older Freddie-less boards, they don't have too many extra logic ICs. My guess would be to make a fully compatible Freddie-replacement board instead? ...Just an idea.
_I'm guessing that the rust resulted from window cleaner mist that settled after the housekeeper sprayed the CRT to which that computer was connected._ 📺🔫 _As for the adapter module, I bet that Bernardo needs to slow one or more signal timings to make it compatible with the old hardware._ ⏳📉
Was gonna say the same. The rust pattern is too uniform and abrupt. Either storage or it came from a humid environment where humidity was allowed to build up just through/below the grille.
Ah!! ... check that the bass guitar back there at 23:04..... "@Jan Beta The bassist"?? ;-) .. with "Pedal Mom" @ Fran Blanche (another credit from the good old Germany) we have half of "The Retro Tech rock band " :-DD >> ruclips.net/video/kNrXpn_YwbM/видео.html
Because the later Atari's (beginning during the production of the 400/800) use a custom variant of the standard 6502, called the 6502C: The 6502C (Sally) has a HALT signal on pin 35 and a second R/W on pin 36 (these pins are not connected (N/C) on standard 6502s). Pulling HALT low latches the clock, pausing the processor. This was used to sync the CPU with the video circuitry (ANTIC)., enabling DMA for video. Earlier 400/800 computers use a standard 6502 with additional circuitry on the processor board.
How exactly does the Rockwell 6502 differ, if any, from the standard MOS chip which apple and the VIC-20 used? And therefore how does it differ from the 65C02? Isnt the C02 pin compatible with the original? I understand some of the undocumented op-codes are not supported in the C02. Would that cause problems with some copy protection software (and likely some games that pushed the envelope).
The 6502C Sally has an extra HALT pin that is used by the ANTIC to make the CPU skip a cycle. It is kind of similar to the AEC pin of the 6510, but with the difference that the Sally samples the HALT pin at the start of the cycle and the CPU skips a full cycle, while the AEC of the 6510 just connects the CPU from the bus while the CPU continues to work normally. The ANTIC has the bus available for the full cycle that is skipped, while the VIC-II in the C64 can only access the bus during the PHI1 phase when the 6510 doesn't use it. The 6510/AEC approach allows the VIC-II to work invisibly to the programmer, there is no slow-down when the VIC-II is fetching data. The advantage of the Sally approach is that you can achieve higher clock speeds, but the CPU is continuously being interrupted by the ANTIC, especially if you use the higher resolutions, the slowdown becomes significant. The 6520C Sally uses an NMOS 6502 core therefore is as different from the 65C02 as the normal 6502.
Hmmm... wie ich in einem früheren Video erwähnt habe, arbeite ich auch an Prozessoradapters und habe auch einen Sally-Prototypen. Während mein Adapter ziemlich ähnlich aussieht, habe ich keine Kompatibilitätsprobleme mit Freddie festgestellt. Ich könnte dir meinen Adapter schicken, vielleicht willst du sie vergleichen?
The big downside is that software that uses undocumented opcodes won't work. The 65C02 has a bunch of nice new opcodes using up some of the undocumented opcodes, and turns all the rest into NOPs. I wish it added indirect JSR, which would be useful for OS entry points, but ultimately the new features aren't useful when the primary purpose is to run legacy code. The compatibility issues are probably similar to those with the 65816 upgrades, but I haven't looked into that.
12:33 I think the problem there was that you appear to have soldered the pin headers in upside-down. Looking at the photos in the documentation: github.com/TheByteAttic/Sally-to-W65C02S-adapter/blob/master/DOCUMENTATION%20Sally%20Adapter.pdf ... it appears that the side with the smooth pins are meant to pass through the board and be soldered, and the side with the "stepped" profile pins are meant to protrude from the bottom of the adaptor.
yes, under magnification i can see that the pin thickness is different from top to bottom side, so needs to go the right way to plug into ic sockets, and if Jan had held them in ic sockets instead of holes in pcb he would have seen that. and later he used different ones that may have been machined differently, and perhaps look more unidirectional.
Ah, there's a memory. I made a 65c02 adapter for my XL years ago via Eagle and the press, peel, etch, PCB method. Indeed a good idea because: -Sallys are hard to find -Sallys drive their address lines quite weakly which is a cause of instability, especially with heavily upgraded systems -Mexico Sallys (most of them) have a raft of other reliability issues So it's indeed best to get a 'proper' 6502 but a little extra logic is needed. Also adding buffers to the address lines while you're at is is a good idea. ...sorry, I'll shut up now and watch the video.
That's pretty common, Big Clive live streams are great "radio" in the background and occasionally you might pick up something particularly interesting on the side.
@ always something new to learn about MOS/CBM after all these years.
3 года назад+1
@@Alphadec after Commodore purchased MOS, MOS no longer manufactured chips for Atari. So the Atari had to cope somehow, and "Sally" (aka 6502C) was produced under license by various manufacturers, mentioned above.
@ that wasn't the reason. Atari specifically paid for the license for the 6502 and its support chips from MOS so it could work with any chipfab company they wanted to. It was more practical for them since they were in Sunnyvale CA and MOS was on the East Coast. Atari preferred to work with Synertek. And that was before Commodore took MOS over. There were several MOS staff who asked Atari to acquire them instead but Warner decided it wasn't sound because Atari was too engrossed in getting the 2600 to market at the time... back in early 1977. Commodore never cut Atari Inc off. In fact, several Commodore people have claimed they set up a dummy company to sell ROM chips to Atari Inc because it was Atari who didn't want to buy from Commodore since they were a competitor. Commodore even shopped around the MOS VIC-II to several video game companies hoping it would be used in consoles and arcade machines but were rebuffed by every company they approached. They then decided to put it into what became the C64.
Indeed, in the limited footprint available, I couldn't place them close. But this shouldn't be a problem: a breadboard prototype without the caps worked perfectly well.
Is it possible? Yes. But is it wise? Well now, with a 020 or 030 you lose to much compatibility. I would rather look for a Mega STE (68000 at 16 mhz) or a TT or Falcon.
Yes. There's several boards from back in the day to modern versions. And there's a lot of patched software. You'll want to move up to TOS 2.06 though. And get the patched software versions.
@@TheJeremyHolloway that is a bit of the point I was trying to say. TOS 2.06 is from the Mega STE which has improved graphics (larger palette, blitter) and sound. So for a regular ST I don't know if it is a wise decision, because you're not making it STE compatible and like you said you need patched software. Sure some software (or games) might run better/faster, but still I believe it is more useful to upgrade a STE.
@@xXTheoLinuxXx there are boards that allow non STe machines to use TOS 2.06. The reason to do it is because that version of TOS is supported more with newer products than TOS 1.06. There are also hacked versions in the AtariAge Forums.
Sorry if this is off topic... but is there any chance you'll be taking a look at the "6502 Plus 4" board for the Commdore C16 / Plus 4 that replaces its 7501/8501 CPU in much the same way as you did in the Atari? Just saw this was available the other day, and I like to wait to see a review from a name I know and trust before taking the plunge. :) amigaonthelake.com/6502-plus-4/ though I'm guessing it's available in other places
Great stuff. Does anybody know what I could check as my atari 520ST can't find channel on my old philips TV i'm using RF cable. I have test the RF with c64 on my TV whci works fine. Thanks
Oh, I got a deal on two 520STf untested but they don't show video or one is garbled. They are completely different beast than XL/XE so I haven't got info to troubleshoot them.
@JanBeta for science, try the board in a 7800. It also uses a SALLY. I'd love to see what it would do in a 5200 but there's so few of them in Europe...
4:18 Another bit of MOS trivia: MOS was originally part of Allen-Bradley. (If you have old high-quality carbon resistors, they may have been made by Allen-Bradley.) Allen-Bradley later became wholly owned by Rockwell Automation. en.wikipedia.org/wiki/MOS_Technology#History In fact, Allen-Bradley's Milwaukee headquarters is now Rockwell's headquarters. The headquarters building is famous for having one of the world's largest 4-faced clocks. en.wikipedia.org/wiki/Rockwell_Automation (My father, uncle, and aunt each worked at Allen-Bradley for about 40 years.)
I think you have timing problems. Most modern chips timing is more precise, the older systems have different capacitance/inductance on the mobo which could mess with timings being on the edge
Commodore's 6510 variant of the 6502 (the 6510 and 8502) also added a HALT pin (as well as a Chip Select line). This was used for similar purposes as in the Sally, but also allowed the addition of a Z80 processor (in a cartridge for the C64) that allowed the use of CP/M. (CP/M was pretty much dead by the 1986 launch of the C128, but it did work. The CP/M cartridge for the C64 mostly didn't, as it was fairly early on that the power supply was made cheaper and no longer had the oomph to power it.)
@@neophytealpha There are, and have been for a long time. All the way back to when they were still being made, in fact. You could get accelerator boards for C64 and C128's at least as far back as the late 80's, based on the 65C816, which runs at up to 14 MHz (IIRC). These work well, and include additional circuitry to operate the built-in I/O ports of the 6510/8502. The main problem you get with these is when people decided to be clever and use the undocumented opcodes, which were disabled in all the CMOS descendants of the 6502 (which was NMOS, being a mid-70's design). This raises possible problems with all the true 6502-based devices as well (like Commodore's 8-bit disk drives), which might use undocumented opcodes to do some things more efficiently. I know of no such problems with Commodore's ROM code in their computers, but who knows what they might have done to make their disk drives more efficient?
@@JeffSmith03 The part that I'm sorry nobody ever used is that initial bootup of the machine is managed by the Z80 (they had to do this to resolve certain problems relating to making cartridges work correctly, IIRC, check the relevant videos on Bill Herd's RUclips channel for more specific details), but so far as I know nobody ever took advantage of that to boot it into a ROM cartridge-based Z80 program.
@@evensgrey that's right, I would love to write Z80 code for it if I can get one. I also saved an old Sinclair ZX81 and TRS-80 Model 4 for the sake of coding with Z80
There is a chance that this is a system timing/clock issue with the 74LS08. I don't see how Freddie would cause the issue you are seeing but I could be wrong. You might try installing a 74F08N as a better and more modern replacement. "However the 74F series is definitely noticeably faster than any of these, and has proven itself to often times fix issues with phase2 clock latency which appears to be the real issue we are seeing in the A8." this is a quote from Mytek on the AtariAge forums. Also I don't see the need to replace a fully working CPU with something that could still sometimes cause compatibility/stability issues, etc. Sally is a great chip but yes later in the XE era Atari used the absolutely cheapest sources for their IC's. A primary fail point was the OS chip in the late years due to the extreme Tramiel cost cutting. Maybe it was a good thing. They at least kept the platform alive while Warner could not.
Maybe try 74ALS245 instead of HCT, it looks like their delays are a bit shorter - actually in the old BBC Micro, when using Solidisk sideways RAM, the LS245 often needed to be replaced with a faster one, either a selected brand ALS, or the 74F245 (not cheap, but looks like you can still get) - they are even faster than the best ALS, and may hold it together if the timings are being pushed past the edge by the HCT
You need CMOS logic levels for the 65C02. Most electronics is pretty tolerant for (A)LS, but the 65C02 is not. This is why he needs a HCT245, to get the logic level up to 5V. You might try a 74AHCT245.
If you want to understand why you were having counter-intuitive issues with your board vs. breadboard: ruclips.net/video/ySuUZEjARPY/видео.html It is the most insightful electronics video I've EVER watched. It's absolutely mind-blowing. Everything I knew about power, signals, routing, and how power even moves was wrong.
Atari didn't buy 6502s from MOS which Commodore allegedly profited from. There were several MOS employees who asked Al Alcorn at Atari Inc to go to Manny Gerard at parent company Warner Communications and ask for Atari to purchase MOS right before Commodore bought them. At the time, MOS was near bankruptcy because they kept shipping chips to Commodore who didn't pay for them [Commodore later bought the company and cancelled their debts with it in the process]. Manny didn't like the idea and Al agreed with him. Instead, Atari Inc signed a licensing deal with MOS for the 6502 - and its support chips - so Atari could modify the design freely as they wanted to and also go to other manufacturers - Synertek especially - and work directly with them. In exchange, Atari bought 50,000 6507s - and RIOT chips - for use in the early Atari VCS [2600] models in 1977. It's a myth Atari Inc - or Atari Corp - purchased 6502s after that from Commodore owned MOS after that. Others at MOS were so against being acquired by Commodore that they left. Bill Mensch had his own 6502 license granted to him and he left MOS and set up Western Design Center (WDC) to do his own development and improvements to his beloved chips. Atari primarily worked with Synertek as they had divulged to MOS. Atari and Synertek were even working on a 6502 chip with 16-bit extension instructions called the "6510" - not to be confused with the Commodore MOS chip of the same name later used in the C64 - in 1979 but it didn't apparently work and wasn't released. It's still debated whether those instructions were later incorporated into WDC's 65816 but one of the co-designers of that chip previously had worked for Atari. After Synertek collapsed following the North American video game industry crash, the successor Atari Corp had companies like Rockwell manufacture their SALLY CPUs for them. It's not even clear how many 6502s - or its support chips - Apple bought from MOS since Apple also worked with Synertek and Rockwell so the assertion Commodore profited from selling 6502s to other companies at inflated prices while cutting the C64's prices and running them out of business is highly in doubt. It made for a good story though at the time. They ran TI out of the home computer business but TI used their own CPU, not the 6502. Sinclair used Z80s. The Colecovision and the ADAM also used the Z80. Mattel used a GI CPU in the Intellivision. Tandy and Vectrex used Motorola 6809s in their computers and consoles...
Wow, thanks for all the background. I added a separate comment: MOS was originally an Allen-Bradley company (1969). Later Rockwell acquired Allen-Bradley and uses it as its headquarters (Milwaukee) today. I enjoyed learning this, given that my dad worked for Allen-Bradley in 1969 (and about 40 years, all told). In 1983, he bought me a Commodore 64 with a MOS processor. (My guess is dad never knew the connection, but he might have!)
BTW, the Bill Mensch interviews here are interesting: ruclips.net/video/Ne1ApyqSvm0/видео.html (I watched after the Chuck Peddle one. Mensch comes off as a bit more... pleasantly humble. :) )
I think it flickered because you didn't stack the board high enough so you should definitely try and stack it to a height 30.48 cm from the main board. I'm only kidding of course but that would be funny to see!
1. You can't solder for shit 2. As a desperate measure, Try variatng one or both of those small caps on the board. They are obviously emant for generating delay of some signal edge and can be finicky. 3. Use good sockets. DOn't put those shitty long adapters between board and motherboard. 4. Real reason why your board doesn't fit could be that pins aren+t 100% aligned. You should have stuck both of those 20-pin rows into DIP40 package and then solder onboard. PCB holes of the chip are far too wide.
Thanks, Jan! Great review! I did encounter a lot of that flicker during development of the board, even in an 800XL. But eventually I changed things around in the design and got rid of it entirely, both in the 800XL and 130XE. When I saw the flicker you were getting, I even thought I might have sent you an earlier prototype board, instead of the final design! But checking in my inventory here, I think I did send you the correct one. It is curious that you got the flicker. It has to do with the clock lines of the Atari's two-phase clock; those are both very timing-sensitive and very load/signal-level-sensitive, which work against each other: to resolve the load/signal-level issue, you want to boost everything through an HCT buffer, but that screws up the timing, as HCT parts have 18ns propagation delay. Standard LS parts have only 8ns delay, but less drive. So the end design is a balance. The only thing I could think of was that those extra pin-headers you used first didn't make good contact, but then you changed those for proper pin-headers, so it can't be that. Anyway, this is a good cautionary tale, so I updated the documentation to inform people that you had this problem with a 65XE. On a different note: the pin-headers I sent you do fit in precision sockets, but you have to install them with the thick side up, and the thin side down. Cheers!
I caught that Perifratic nod lol. I have to say it that way myself.
Great video!!
20:41-"yeah, and we're going to have to take the whole circuit board out to get the shield off ............which i put back in there. "
The moment of realization of the act responsible for this was priceless! lol
Right on time, fixing my 65xe at the moment so yesss!
So much variety in these old machines, as always this hobby has its quarks to work out, you never can predict what will happen either, you would think that the adapter would replace somethin' specifically and should in theory then replace that chip or component without any change or harm to anythin' else but clearly Freddie doesn't like a few new chips in this case, oh well at least it does work. Danke Jan!
So happy that Focus 10 is on Spotify! Great music!
They should call this mod the "Harry", as in, "When Harry met Sally"... :P
You win the Internet today. That was awesome
and Freddie meshed things up...lol
I never had an Atari 8 bit machine. I had a 2600 like everyone did, but I had Commodore commputers (VIC 20, 64, 128).
Hey Jan, thanks for sharing!
Love the channel by the way.
that rust looks like someone had it in a room with high humidity, rather than it getting rained or spilled on.
I thought the same too
this and probably also due to very rust prone sheet metal
Well when we were kids, my bros and I used to play mini basketball in the room while waiting for a game to load...So the rust on my 800xl shield is sweat....lol.
Strange how i missed this video, an Atari one too! Great seeing you do some Atari stuff ;)
Just seen Adrian's digital basement use a similar adapter in a C64, sure it's the same guy!! Cool!
Yes, that's Rockwell Semiconductor. Made a huge number of 6502s. They were used by Apple as well, probably depending on availability.
Rockwell manufactured chips on contract for several companies. The 6502s they made for Apple were not the Atari SALLY version of the 6502. After all, Apple didn't need the halt line for use with Atari's ANTIC or MARIA graphics chips.
danke jan für ein weiteres tolles video grüße aus den Niederlanden
The rust can just came from moisture, where the device was stored. Maybe a cellar - that's enough getting some rust on that part. It's not real liquid damage...
One of my favourite computers
That rust pattern is indicative of being in a humid/salt water mist environment, near the ocean.
My guess as to the rust is that there was a humidifier in the room it lived in and the moisture settled through the vents and condensed on the colder (usually) metal.
I am amazed to see 6502's still being manufactured today. Perhaps this upgrade may be a great idea, for the new model should be far more efficient, thus even cooler than the original processor.
Unfortunately the modern ones are not drop-in replacements for the original 6502 either, as some of the "illegal" op-codes are not supported afaik that many old programs (games/demos etc) used.
The 65xe was manufactured in different variants so not all of them are easily upgradable to 130XE.
If you see a 65xe that does not have ECI slot next to the cartridge slot, it's a sign that it uses a dedicated board.
Anyway, if you are aiming for an upgrade, you not only need the extra RAM chips but also the EMMU that comes only in the original 130. You can replace the EMMU with some programmable logic chips but when you do, it is more reasonable to do an upgrade to something like 1meg of SRAM (kits available from hobbyists around the world).
U34 was called Enhanced Memory Management Unit, or EMMU for short. It can be substituted with a GAL and some generic 74xx chip (as U35). Real EMMU didn't need U35.
What IC's did you use? Looking at the github the 245s have to be HCT, the 74 and the 02 have to be LS.
I've just stepped frame-by-frame through the appropriate part of the video, and that's exactly what the chips are.
I learned to code on a 65 XE. Loved the angular design. Loved older ThinkPads because of those design elements too. The rusty shield with no internal component corrosion may be due to use in a coastal area. Mine had that too when I opened it.
Jay Miner = Total Genius
I have done a similar adapter for 65C816, and experienced similar problems. It kind of works, but is unstable. I've managed to run YOOMP! on it. It also worked better on XL. My adapter used only 2 TTL chips (I've used SMD), which I've fitted under the CPU socket.
I can't help but wonder if the extra chips in the 130XE is where the difference lies.
Something about bus capacitance and rise/fall times perhaps.
Judging by the flicker etc. That came up with the adapter and replacement chip.!
It seems to be more of an issue with grounding.!
All you need to do is find a good ground point on the motherboard and patch a wire directly to the ground point to the replacement chip.!!
I have discovered that this is a common fault upgrading older CPU.!
Amazing English Jan. Have you lived in the UK? How did you learn?
Salt water moisture in the air from living near the ocean rusts everything pretty badly.
Yep, and maybe even lived it's life near an often open window.
That rust on the shielding is caused by dust falling on the shield. The dust holds moisture and rusts the shielding.
Might be worth trying again once you upgrade the ram. The addition of the multiplexer logic and the extra chips onto the bus might have the effect of synching things up, perhaps by adding more delay to things. I'd source similar (ideally identical) chips and see what happens.
This is plausible indeed. The flicker has to do with both timing of the PHI2 clock line, and the load on that line, which may exceed the drive capacity of the two-phase clock generation logic.
@@thebyteattic Hmm, so it's not just a timing issue then: the added logic for the extra ram bumps the current up enough to run the memory subsystem properly with the modern CMOS (it's CMOS, right?) 6502, correct?
Hi Jan! Probably stupid question to ask before I watch video till the end but what is the reason for swapping 65c02 instead Sally? Just for a compatibility test?
They are still in production and run more efficiently. They have some incompatibilities though, especially with illegal opcodes that are used by some software.
@@JanBeta "and run more efficiently" as opposed to what? is this a power/thermal thing or a code thing? does the replacement run at a faster clock or is this more a case if you have a dead CPU this could be used as a replacement? Ive never had stability issues with the original SALLY and they seem pretty reliable
@@DanafoxyVixen the power consumption of CMOS processor is 10 to 20 times lower than nMOS based (wikipedia). But I wouldn't consider swapping Sally with 65c02 just because of this if Sally is working well.
never before seen footage.... Jan tidies up a bit!!!! lol
just kidding I'm hardly a tidy soul myself
PCBWaaaaayyy??? Channeling Perefractic now?
You will most likely run into issues with many modern demos that use the so-called illegal instructions, which do not work with the 65C02 (I mean even with the XL machines).
Don't confuse the 6502C with the custom Tri-State 65C02. Interestingly, the Atari 400 and 800 use the Rockwell 6502.
@@theannoyedmrfloyd3998 That was my point.
i seems that there is some timing investigation necessary. do you have a picoscope? it would also show how clean the signals are. maybe changing the logic chips to faster variants could help.
ok, i read somewhere that the sally chip had a halt-pin that halted the cpu and decoupled it from the bus. because you have glitches on the graphic, i assume there is a collision on the bus. the graphic chip needs to access the bus but the cpu is not fully decoupled with the bus transceiver chip. so, somehow the halt signal does not do it's job fast enough. this could also be problematic when halt goes away and the cpu access is to late for a read or write.
maybe you take a look on the 74xxx variants that are available. there are endless variants of ls and hc and hct and f and what not. i know that ti has some fast ones.
i don't know if Microchip also makes some (which i should probably know). if you need some samples from Microchip (Atmel, Micrel) , please let me know.
ok, i found some schematic from 1987:
atariage.com/forums/uploads/monthly_04_2018/post-14026-0-22693400-1523129848.jpg
i don't know if this is the same that you have. but as i saw, there is a 74ls02 and 74f244, 74f245 on it. these are old chips from the 80s. please look for faster variants and check the signal delay from the halt to the phi2 signal. and the "glitches" on the output of the 74x244
as i never owned an atari, i am no expert on that but if you have some questions regarding that stuff (or samples from Microchip), contact me on facebook.
That's a long tall sally.
Lol, indeed!
@@JanBeta Ah!! ... check that the bass guitar back there at 23:04..... "@Jan Beta The bassist"?? ;-) ..
with "Pedal Mom" @
Fran Blanche (another credit from the good old Germany) we have half of "The Retro Tech rock band " :-DD >> ruclips.net/video/kNrXpn_YwbM/видео.html
I'm a Commodore guy, but looking at the older Freddie-less boards, they don't have too many extra logic ICs. My guess would be to make a fully compatible Freddie-replacement board instead? ...Just an idea.
The problem must be in this adapter, my own 6502 adapter has zero problems inside Freddie boards.
yo jan your English as come a long way,i guess making these videos helped
_I'm guessing that the rust resulted from window cleaner mist that settled after the housekeeper sprayed the CRT to which that computer was connected._ 📺🔫
_As for the adapter module, I bet that Bernardo needs to slow one or more signal timings to make it compatible with the old hardware._ ⏳📉
I don't think this is liquid damage... Looks like it was stored in a humid environment
Was gonna say the same. The rust pattern is too uniform and abrupt. Either storage or it came from a humid environment where humidity was allowed to build up just through/below the grille.
I think so, it looks like some condensation happened there.
Ah!! ... check that the bass guitar back there at 23:04..... "@Jan Beta The bassist"?? ;-) ..
with "Pedal Mom" @
Fran Blanche (another credit from the good old Germany) we have half of "The Retro Tech rock band " :-DD >> ruclips.net/video/kNrXpn_YwbM/видео.html
Why the elaborate adapter, the BBC model B as far as I know just ads a capacitor to alter the timing on a pin.
Because the later Atari's (beginning during the production of the 400/800) use a custom variant of the standard 6502, called the 6502C:
The 6502C (Sally) has a HALT signal on pin 35 and a second R/W on pin 36 (these pins are not connected (N/C) on standard 6502s). Pulling HALT low latches the clock, pausing the processor. This was used to sync the CPU with the video circuitry (ANTIC)., enabling DMA for video.
Earlier 400/800 computers use a standard 6502 with additional circuitry on the processor board.
@@almerian not just for the ANTIC. It was also used in the 7800 with its MARIA graphics chip for the same reason.
Some of the later 800xl also used the Freddie chip, so the adaptor may not work with them too.
I don't think it's the Freddie chip, as it did work fine in an 130XE. But hey, who knows!
How exactly does the Rockwell 6502 differ, if any, from the standard MOS chip which apple and the VIC-20 used? And therefore how does it differ from the 65C02? Isnt the C02 pin compatible with the original? I understand some of the undocumented op-codes are not supported in the C02. Would that cause problems with some copy protection software (and likely some games that pushed the envelope).
The 6502C Sally has an extra HALT pin that is used by the ANTIC to make the CPU skip a cycle. It is kind of similar to the AEC pin of the 6510, but with the difference that the Sally samples the HALT pin at the start of the cycle and the CPU skips a full cycle, while the AEC of the 6510 just connects the CPU from the bus while the CPU continues to work normally. The ANTIC has the bus available for the full cycle that is skipped, while the VIC-II in the C64 can only access the bus during the PHI1 phase when the 6510 doesn't use it.
The 6510/AEC approach allows the VIC-II to work invisibly to the programmer, there is no slow-down when the VIC-II is fetching data. The advantage of the Sally approach is that you can achieve higher clock speeds, but the CPU is continuously being interrupted by the ANTIC, especially if you use the higher resolutions, the slowdown becomes significant.
The 6520C Sally uses an NMOS 6502 core therefore is as different from the 65C02 as the normal 6502.
Hmmm... wie ich in einem früheren Video erwähnt habe, arbeite ich auch an Prozessoradapters und habe auch einen Sally-Prototypen. Während mein Adapter ziemlich ähnlich aussieht, habe ich keine Kompatibilitätsprobleme mit Freddie festgestellt. Ich könnte dir meinen Adapter schicken, vielleicht willst du sie vergleichen?
The big downside is that software that uses undocumented opcodes won't work. The 65C02 has a bunch of nice new opcodes using up some of the undocumented opcodes, and turns all the rest into NOPs. I wish it added indirect JSR, which would be useful for OS entry points, but ultimately the new features aren't useful when the primary purpose is to run legacy code. The compatibility issues are probably similar to those with the 65816 upgrades, but I haven't looked into that.
12:33 I think the problem there was that you appear to have soldered the pin headers in upside-down. Looking at the photos in the documentation:
github.com/TheByteAttic/Sally-to-W65C02S-adapter/blob/master/DOCUMENTATION%20Sally%20Adapter.pdf
... it appears that the side with the smooth pins are meant to pass through the board and be soldered, and the side with the "stepped" profile pins are meant to protrude from the bottom of the adaptor.
yes, under magnification i can see that the pin thickness is different from top to bottom side, so needs to go the right way to plug into ic sockets, and if Jan had held them in ic sockets instead of holes in pcb he would have seen that. and later he used different ones that may have been machined differently, and perhaps look more unidirectional.
Yes, maybe I should have warned of it in the documentation... This whole thing is serving to show me how poor my documentation is...
Does this mod only work for PAL Ataris?
It should work for both PAL and NTSC models, as far as I know. But obviously I don't have an NTSC Atari to test it in.
Hey Jan, could you try to frame the workbench differently as to remove the soldering iron from the shot? Its movement is rather jarring.
Ah, there's a memory.
I made a 65c02 adapter for my XL years ago via Eagle and the press, peel, etch, PCB method. Indeed a good idea because:
-Sallys are hard to find
-Sallys drive their address lines quite weakly which is a cause of instability, especially with heavily upgraded systems
-Mexico Sallys (most of them) have a raft of other reliability issues
So it's indeed best to get a 'proper' 6502 but a little extra logic is needed. Also adding buffers to the address lines while you're at is is a good idea.
...sorry, I'll shut up now and watch the video.
rapidus? new ula?
Is it weird that sometimes I just put Jan's videos on in the background just because his voice is somehow soothing when I'm working?
That's pretty common, Big Clive live streams are great "radio" in the background and occasionally you might pick up something particularly interesting on the side.
You're not the only one...
Can't win them all huh.
But still an amazing video.
Did not know about Rockwell, intresting fact. Thought MOS was the only producer of 6502.
"Several manufacturers produced the SALLY 6502 for Atari, including MOS Technology, Synertek, Rockwell, NCR, and United Microelectronics (UMC). "
@ always something new to learn about MOS/CBM after all these years.
@@Alphadec after Commodore purchased MOS, MOS no longer manufactured chips for Atari. So the Atari had to cope somehow, and "Sally" (aka 6502C) was produced under license by various manufacturers, mentioned above.
I'm not an Atari expert, I'm a commodore guy. how does a standard 6502 differ from the 'Sally' chip?
@ that wasn't the reason. Atari specifically paid for the license for the 6502 and its support chips from MOS so it could work with any chipfab company they wanted to. It was more practical for them since they were in Sunnyvale CA and MOS was on the East Coast. Atari preferred to work with Synertek. And that was before Commodore took MOS over. There were several MOS staff who asked Atari to acquire them instead but Warner decided it wasn't sound because Atari was too engrossed in getting the 2600 to market at the time... back in early 1977. Commodore never cut Atari Inc off. In fact, several Commodore people have claimed they set up a dummy company to sell ROM chips to Atari Inc because it was Atari who didn't want to buy from Commodore since they were a competitor. Commodore even shopped around the MOS VIC-II to several video game companies hoping it would be used in consoles and arcade machines but were rebuffed by every company they approached. They then decided to put it into what became the C64.
Probably not the problem but the decoupling caps on the adapter board look like they are too far away from their chips.
Indeed, in the limited footprint available, I couldn't place them close. But this shouldn't be a problem: a breadboard prototype without the caps worked perfectly well.
Is it possible to upgrade the ST to a faster processor, like a 68030?
Is it possible? Yes. But is it wise? Well now, with a 020 or 030 you lose to much compatibility. I would rather look for a Mega STE (68000 at 16 mhz) or a TT or Falcon.
Yes. There's several boards from back in the day to modern versions. And there's a lot of patched software. You'll want to move up to TOS 2.06 though. And get the patched software versions.
@@TheJeremyHolloway that is a bit of the point I was trying to say. TOS 2.06 is from the Mega STE which has improved graphics (larger palette, blitter) and sound. So for a regular ST I don't know if it is a wise decision, because you're not making it STE compatible and like you said you need patched software. Sure some software (or games) might run better/faster, but still I believe it is more useful to upgrade a STE.
@@xXTheoLinuxXx there are boards that allow non STe machines to use TOS 2.06. The reason to do it is because that version of TOS is supported more with newer products than TOS 1.06. There are also hacked versions in the AtariAge Forums.
Sorry if this is off topic... but is there any chance you'll be taking a look at the "6502 Plus 4" board for the Commdore C16 / Plus 4 that replaces its 7501/8501 CPU in much the same way as you did in the Atari? Just saw this was available the other day, and I like to wait to see a review from a name I know and trust before taking the plunge. :)
amigaonthelake.com/6502-plus-4/ though I'm guessing it's available in other places
Try , try again ……. “As always thank you for sharing”. Greetings from Karachi, Pakistan.
Great stuff. Does anybody know what I could check as my atari 520ST can't find channel on my old philips TV i'm using RF cable. I have test the RF with c64 on my TV whci works fine. Thanks
Oh, I got a deal on two 520STf untested but they don't show video or one is garbled. They are completely different beast than XL/XE so I haven't got info to troubleshoot them.
@JanBeta for science, try the board in a 7800. It also uses a SALLY. I'd love to see what it would do in a 5200 but there's so few of them in Europe...
Actually, the 65C02 in Atari 8-bit computers, the C stands for Custom as those processors are Tri State.
6502C [SALLY] is Atari's version of the 6502 regardless of the contract manufacturer [Rockwell, Synertek, etc]. The 65C02 is a WDC chip.
4:18 Another bit of MOS trivia: MOS was originally part of Allen-Bradley. (If you have old high-quality carbon resistors, they may have been made by Allen-Bradley.) Allen-Bradley later became wholly owned by Rockwell Automation.
en.wikipedia.org/wiki/MOS_Technology#History
In fact, Allen-Bradley's Milwaukee headquarters is now Rockwell's headquarters.
The headquarters building is famous for having one of the world's largest 4-faced clocks.
en.wikipedia.org/wiki/Rockwell_Automation
(My father, uncle, and aunt each worked at Allen-Bradley for about 40 years.)
I think you have timing problems. Most modern chips timing is more precise, the older systems have different capacitance/inductance on the mobo which could mess with timings being on the edge
Right on. It's both a timing and signal strength issue. It's a fine balance to strike.
🤘😁🤘
another adapter by Piotr D. Kaczorowski: ruclips.net/user/shortsZMjztv8HGWg
Commodore's 6510 variant of the 6502 (the 6510 and 8502) also added a HALT pin (as well as a Chip Select line). This was used for similar purposes as in the Sally, but also allowed the addition of a Z80 processor (in a cartridge for the C64) that allowed the use of CP/M. (CP/M was pretty much dead by the 1986 launch of the C128, but it did work. The CP/M cartridge for the C64 mostly didn't, as it was fairly early on that the power supply was made cheaper and no longer had the oomph to power it.)
They need a modern replacement for the 6510. One that could be used in a 64 or 128.
@@neophytealpha There are, and have been for a long time. All the way back to when they were still being made, in fact. You could get accelerator boards for C64 and C128's at least as far back as the late 80's, based on the 65C816, which runs at up to 14 MHz (IIRC). These work well, and include additional circuitry to operate the built-in I/O ports of the 6510/8502.
The main problem you get with these is when people decided to be clever and use the undocumented opcodes, which were disabled in all the CMOS descendants of the 6502 (which was NMOS, being a mid-70's design). This raises possible problems with all the true 6502-based devices as well (like Commodore's 8-bit disk drives), which might use undocumented opcodes to do some things more efficiently. I know of no such problems with Commodore's ROM code in their computers, but who knows what they might have done to make their disk drives more efficient?
I wanted a C128 more when I found out the Z80 was built into it just for CP/M. I collect things that were overkill
@@JeffSmith03 The part that I'm sorry nobody ever used is that initial bootup of the machine is managed by the Z80 (they had to do this to resolve certain problems relating to making cartridges work correctly, IIRC, check the relevant videos on Bill Herd's RUclips channel for more specific details), but so far as I know nobody ever took advantage of that to boot it into a ROM cartridge-based Z80 program.
@@evensgrey that's right, I would love to write Z80 code for it if I can get one. I also saved an old Sinclair ZX81 and TRS-80 Model 4 for the sake of coding with Z80
Okay, this is a silly Sally question, but have you tried replacing the 65C02 with a 6502 (no C)?
There is a chance that this is a system timing/clock issue with the 74LS08. I don't see how Freddie would cause the issue you are seeing but I could be wrong. You might try installing a 74F08N as a better and more modern replacement. "However the 74F series is definitely noticeably faster than any of these, and has proven itself to often times fix issues with phase2 clock latency which appears to be the real issue we are seeing in the A8." this is a quote from Mytek on the AtariAge forums. Also I don't see the need to replace a fully working CPU with something that could still sometimes cause compatibility/stability issues, etc. Sally is a great chip but yes later in the XE era Atari used the absolutely cheapest sources for their IC's. A primary fail point was the OS chip in the late years due to the extreme Tramiel cost cutting. Maybe it was a good thing. They at least kept the platform alive while Warner could not.
I am willing to bet dust trapped moisture and that's what caused the rust.
Maybe try 74ALS245 instead of HCT, it looks like their delays are a bit shorter - actually in the old BBC Micro, when using Solidisk sideways RAM, the LS245 often needed to be replaced with a faster one, either a selected brand ALS, or the 74F245 (not cheap, but looks like you can still get) - they are even faster than the best ALS, and may hold it together if the timings are being pushed past the edge by the HCT
You need CMOS logic levels for the 65C02. Most electronics is pretty tolerant for (A)LS, but the 65C02 is not. This is why he needs a HCT245, to get the logic level up to 5V. You might try a 74AHCT245.
The difference between theory and practice is that in theory, there is no difference.
If you want to understand why you were having counter-intuitive issues with your board vs. breadboard: ruclips.net/video/ySuUZEjARPY/видео.html
It is the most insightful electronics video I've EVER watched. It's absolutely mind-blowing. Everything I knew about power, signals, routing, and how power even moves was wrong.
No point replacing the cpu unless It didn't work!
Atari didn't buy 6502s from MOS which Commodore allegedly profited from. There were several MOS employees who asked Al Alcorn at Atari Inc to go to Manny Gerard at parent company Warner Communications and ask for Atari to purchase MOS right before Commodore bought them. At the time, MOS was near bankruptcy because they kept shipping chips to Commodore who didn't pay for them [Commodore later bought the company and cancelled their debts with it in the process]. Manny didn't like the idea and Al agreed with him. Instead, Atari Inc signed a licensing deal with MOS for the 6502 - and its support chips - so Atari could modify the design freely as they wanted to and also go to other manufacturers - Synertek especially - and work directly with them. In exchange, Atari bought 50,000 6507s - and RIOT chips - for use in the early Atari VCS [2600] models in 1977. It's a myth Atari Inc - or Atari Corp - purchased 6502s after that from Commodore owned MOS after that. Others at MOS were so against being acquired by Commodore that they left. Bill Mensch had his own 6502 license granted to him and he left MOS and set up Western Design Center (WDC) to do his own development and improvements to his beloved chips. Atari primarily worked with Synertek as they had divulged to MOS. Atari and Synertek were even working on a 6502 chip with 16-bit extension instructions called the "6510" - not to be confused with the Commodore MOS chip of the same name later used in the C64 - in 1979 but it didn't apparently work and wasn't released. It's still debated whether those instructions were later incorporated into WDC's 65816 but one of the co-designers of that chip previously had worked for Atari. After Synertek collapsed following the North American video game industry crash, the successor Atari Corp had companies like Rockwell manufacture their SALLY CPUs for them. It's not even clear how many 6502s - or its support chips - Apple bought from MOS since Apple also worked with Synertek and Rockwell so the assertion Commodore profited from selling 6502s to other companies at inflated prices while cutting the C64's prices and running them out of business is highly in doubt. It made for a good story though at the time. They ran TI out of the home computer business but TI used their own CPU, not the 6502. Sinclair used Z80s. The Colecovision and the ADAM also used the Z80. Mattel used a GI CPU in the Intellivision. Tandy and Vectrex used Motorola 6809s in their computers and consoles...
Wow, thanks for all the background. I added a separate comment: MOS was originally an Allen-Bradley company (1969). Later Rockwell acquired Allen-Bradley and uses it as its headquarters (Milwaukee) today. I enjoyed learning this, given that my dad worked for Allen-Bradley in 1969 (and about 40 years, all told). In 1983, he bought me a Commodore 64 with a MOS processor. (My guess is dad never knew the connection, but he might have!)
BTW, the Bill Mensch interviews here are interesting: ruclips.net/video/Ne1ApyqSvm0/видео.html
(I watched after the Chuck Peddle one. Mensch comes off as a bit more... pleasantly humble. :) )
Atari....
Wir haben ihn an die dunkle Seite verloren....
😁
I think it flickered because you didn't stack the board high enough so you should definitely try and stack it to a height 30.48 cm from the main board.
I'm only kidding of course but that would be funny to see!
Lol, that would be fun to try... :D
1. You can't solder for shit
2. As a desperate measure, Try variatng one or both of those small caps on the board. They are obviously emant for generating delay of some signal edge and can be finicky.
3. Use good sockets. DOn't put those shitty long adapters between board and motherboard.
4. Real reason why your board doesn't fit could be that pins aren+t 100% aligned. You should have stuck both of those 20-pin rows into DIP40 package and then solder onboard. PCB holes of the chip are far too wide.