Custom Sally Replacement

[+mksmith]

Hi everyone,

 

Whilst I don't unfortunately have an XL machine myself I often watch various Youtube channels and videos covering these machines.  I came across this channel recently called the Byte Attic where the chap (I can't find his name unfortunately) has been designing a Sally replacement which plugs directly into the socket (replacing it with a PCB containing a W65C02S and some logic and transceiver chips).  He has just completed the series and has made the board available on Github here.

 

The design starts in Part 3 of his 5 part series:

 

Hopefully some of you might be able to resurrect a machine or two using it ?

[+DrVenkman]

Clever but it's a solution in search of a problem. A new WDC65C02S runs about seven bucks at Mouser. A new old-stock SALLY is eight bucks from Best Electronics.

[+mksmith]

14 minutes ago, DrVenkman said:

Clever but it's a solution in search of a problem. A new WDC65C02S runs about seven bucks at Mouser. A new old-stock SALLY is eight bucks from Best Electronics.

Oh - I didn't realise these were still reasonably readily available.  I guess for those without easy access to Bests it could be an option but an original chip is certainly the preferred of course.

[+mytek]

Even with the original Sally still available from Best, you have to factor in a pretty steep shipping and handling fee + minimum order requirements. I could also see an advantage in the Sally replacement having a much lower power requirement due to being CMOS (Sally gets very hot - meaning it's drawing quite a bit of power), and a few more efficient OP-Codes. Cons would be no support for illegal OP-Codes, although that's rarely a problem.

 

I just went to the GitHub link, but nothing appears to have been uploaded.

 

 

Never mind - it was my browser's javascript blocker causing the problem.

 

Edited August 21, 2020 by mytek
commented about problem being on my end

[+mksmith]

21 minutes ago, mytek said:

Even with the original Sally still available from Best, you have to factor in a pretty steep shipping and handling fee + minimum order requirements. I could also see an advantage in the Sally replacement having a much lower power requirement due to being CMOS (Sally gets very hot - meaning it's drawing quite a bit of power), and a few more efficient OP-Codes. Cons would be no support for illegal OP-Codes, although that's rarely a problem.

 

I just went to the GitHub link, but nothing appears to have been uploaded.

 

 

Never mind - it was my browser's javascript blocker causing the problem.

 

For sure - I live in Australia and ordered some joystick parts from Brad a few years ago and it cost me a fair amount on top of the $ conversion.  No doubt it works better for some than others.

 

This guy does some really interesting videos around replacing these old and getting harder to find components - he did a similar 6502 replacement for a Vic-20 in a previous episode. 

[Kyle22]

I'd like to see the schematic of this. I'd also like to know if a PAL/GAL was used, and I'd like to know if the code in the PAL/GAL is the SAME as I wrote for the FTe Sweet-16.

 

 

Edited August 21, 2020 by Kyle22
Bold the I.

[+mksmith]

7 minutes ago, Kyle22 said:

I'd like to see the schematic of this. I'd also like to know if a PAL/GAL was used, and I'd like to know if the code in the PAL/GAL is the SAME as I wrote for the FTe Sweet-16.

 

 

The 3 videos explain his entire dev process and the issues he encountered. These are just off the shelf parts from my understanding so no code involved. Well worth a watch!

[+mksmith]

The schematic is here:

https://github.com/TheByteAttic/Sally-to-W65C02S-adapter/blob/master/Schematics W65C02S to Atari Sally Adapter 20 August 2020.pdf

 

[+mytek]

I like the electronic design not requiring programming of a PLD. But due to the through-hole parts being used it's a rather large board that may interfere with other upgrades depending on the model of Atari this gets used in.

 

 

I think down the road what I would like to do is layout a new version of the PCB using SMD components, which should allow it to be shrunk down to a footprint barely larger than just the CPU itself. If I do this (I'm not saying that I will), I would of course release the gerbers into the public domain same as the original creator and credit him on the silkscreen for the original design.

 

Here's an example of what I am thinking the size could be reduced down to...

 

 

Unlike this example board that utilized a straight thru piggyback, it would still require an offset header similar to what was done in the original layout for signals that don't pass all the way thru, and it would need to have SMD components top & bottom in order to stay inside the CPU footprint. Overall I think it would end up with a footprint the same as my example photo.

 

[re-atari]

A pal of mine, Guus Assmann, designed a similar upgrade already back in '86-'87. He also included a hardware realtime clock, could be either a OKI 5832 or a Motorola 146818, and a 6520/6821 PIA based printer port. Schematics and PCB layout were published in Atari Magazine (NL). The PCB was really big.

 

I wonder if The Byte Attic was inspired by his design. If so, it would be nice of him to mention it and give credit where credit is due.

 

re-atari

[Rybags]

I agree a slimline form factor would be better - space for a given upgrade usually steals from adjacent areas.

 

Off topic somewhat but in the modern day why don't we yet have a plugin emulated replacement?

The thought is a PIC, AVR, something, that runs a 6502 core and can behave exactly like the OEM part.  Then add things like 65C02 mode, maybe 65816, and obviously why not have a 6809 mode as well?

[Mathy]

Hello guys

 

On 8/21/2020 at 10:15 AM, re-atari said:

A pal of mine, Guus Assmann, designed a similar upgrade already back in '86-'87.

 

Here's some information (with schematic and pictures) of Guus' upgrade.

 

Sincerely

 

Mathy

[+mytek]

On 8/21/2020 at 1:15 AM, re-atari said:

A pal of mine, Guus Assmann, designed a similar upgrade already back in '86-'87. He also included a hardware realtime clock, could be either a OKI 5832 or a Motorola 146818, and a 6520/6821 PIA based printer port. Schematics and PCB layout were published in Atari Magazine (NL). The PCB was really big.

 

I wonder if The Byte Attic was inspired by his design. If so, it would be nice of him to mention it and give credit where credit is due.

It's not the same circuit. Guss's circuit was very similar to the original 800 (in fact it might be identical - my memory isn't sure), where both the data bus and the address bus required tri-state buffers to float the CPU. Whereas this new design leaves the address lines connected and uses the W65C02S to float the address bus internally, thus its more compact with one less tri-state buffer required.

 

[Gunstar]

On 8/21/2020 at 3:28 AM, Rybags said:

 

Off topic somewhat but in the modern day why don't we yet have a plugin emulated replacement?

The thought is a PIC, AVR, something, that runs a 6502 core and can behave exactly like the OEM part.  Then add things like 65C02 mode, maybe 65816, and obviously why not have a 6809 mode as well?

I wanted to like this idea more than once!

Edited August 23, 2020 by Gunstar

[+mytek]

On 8/21/2020 at 1:28 AM, Rybags said:

Off topic somewhat but in the modern day why don't we yet have a plugin emulated replacement?

The thought is a PIC, AVR, something, that runs a 6502 core and can behave exactly like the OEM part.  Then add things like 65C02 mode, maybe 65816, and obviously why not have a 6809 mode as well?

Yes it would be the best of all worlds.

 

Unlike using an FPGA, where there are proven cores to work from, coding a microcontroller of any kind to do this would be quite the challenge. Although at least we live in a time when they are certainly getting fast enough to do so.

 

We won't see this happening anytime soon, although I'll be happy to be proven wrong .

 

[ivop]

16 hours ago, mytek said:

We won't see this happening anytime soon, although I'll be happy to be proven wrong .

I acquired a development board with the same CPU as the UnoCart specifically for this reason. But that was years ago

 

[ijor]

On 8/21/2020 at 5:28 AM, Rybags said:

Off topic somewhat but in the modern day why don't we yet have a plugin emulated replacement?

The thought is a PIC, AVR, something, that runs a 6502 core and can behave exactly like the OEM part.  Then add things like 65C02 mode, maybe 65816, and obviously why not have a 6809 mode as well?

 

It might be feasible to emulate Sally at real time using a fast modern MCU. May be with something like an Arm Cortex perhaps. But I think it is obvious that this task is much more suitable for an FPGA and I think there are already such 6502 chips replacements. A FPGA could easily do much more, not only "clone" Sally at 1.79 MHz, but could also do, i.e., a 65816 at much higher frequencies.

Edited August 25, 2020 by ijor

[+slx]

I totally lack the skills to design my idea:

 

A plug-in board with a 65C816 (real or emulated) that accesses system RAM through the socket for data bank register 0 and "on-board" RAM for everything above and includes circuitry to make it Sally-compatible. As 65C816s are available in SMD it might be possible to design this so small that it can be swapped for a real Sally. (Mostly for 800s, XL/XEs can have the functionality with Antonia.)

 

There's a guy in Bulgaria who builds 65C816s on a board that makes them pin compatible with normal 6502s (effectively a 65c802 replacement).

 

 

[+mytek]

I was looking at the schematic for this adapter board today and it dawned on me that it could be simpler in theory.

 

Byte Attic Version

 

My Version with 74HCT245 Transceiver chips removed

W65C02S-SALLY-Adapter_schema.pdf

 

The W65C02S already has the ability to float both the Data and Address bus via the 'BE' pin, and it also has better drive capabilities than either SALLY or a stock 6502. So the inclusion of the 74HCT245 chips in the original version seemed redundant to me.

 

Any thoughts?

 

[reifsnyderb]

6 hours ago, mytek said:

I was looking at the schematic for this adapter board today and it dawned on me that it could be simpler in theory.

 

Byte Attic Version

 

My Version with 74HCT245 Transceiver chips removed

W65C02S-SALLY-Adapter_schema.pdf 36.3 kB · 4 downloads

 

The W65C02S already has the ability to float both the Data and Address bus via the 'BE' pin, and it also has better drive capabilities than either SALLY or a stock 6502. So the inclusion of the 74HCT245 chips in the original version seemed redundant to me.

 

Any thoughts?

 

The Byte Attic guy has an explanation on youtube as to why he used used the 74HCT245 chips.  But if it works without them that would be fantastic.  I've got a couple of the W65C02S PLCC chips around and may give this a try.

 

 

Edited August 5, 2022 by reifsnyderb

[+mytek]

2 minutes ago, reifsnyderb said:

The Byte Attic guy has an explanation on youtube as to why he used used the 74HCT245 chips.  But if it works without them that would be fantastic.  I've got a couple of the W65C02S PLCC chips around and may give this a try.

Yeah he also said that the other glue logic had to be LS and not HCT, which is starting to sound like possible timing problems that might be correctable by simply going to ACT or an F chip instead. I remember that the 1088XEL didn't like using a 74HCT08, but worked fine with a 74LS08, and of course a 74F08.

 

[+selgus]

51 minutes ago, mytek said:

Yeah he also said that the other glue logic had to be LS and not HCT, which is starting to sound like possible timing problems that might be correctable by simply going to ACT or an F chip instead. I remember that the 1088XEL didn't like using a 74HCT08, but worked fine with a 74LS08, and of course a 74F08.

 

It was used to strengthen the signals. He states that it probably would still work without the bus transceivers, but according to the data sheets, it might not work consistently. I've used the bus transceivers in my SALLY replacement boards, because with add the additional components on the address bus, it seems logical to want to make sure these signals were strong. 

[+mytek]

12 minutes ago, selgus said:

It was used to strengthen the address signals. He states that it probably would still work without the bus transceivers, but according to the data sheets, it might not work consistently. I've used the bus transceivers in my SALLY replacement boards, because with add the additional components on the address bus, it seems logical to want to make sure these signals were strong. 

I think you meant to say the data bus, which was where one of the transceivers was placed on his circuit. On the other hand the address lines went straight into the W65C02S same as what I show. The other Transceiver was used as a simple buffer for some of the other signals. Funny thing was he also had the SOB line buffered from the SO on the SALLY, however on the Atari that signal is grounded, so really no need to buffer it.

 

"...according to the data sheets, it might not work consistently."

 

Yeah I'll stand corrected on my earlier statement about the W65C02S having more drive capability than a standard 6502 or possibly SALLY, because it's hard to get a proper comparison between the datasheets as to exactly what is the difference in regards to fan-out capability. Although I would think that all the chips would be similar in this regard. And since SALLY works in an Atari without buffering , I would assume the same for the W65C02S, but that has yet to be determined. I'll have to test to be sure

 

[+selgus]

2 hours ago, mytek said:

I think you meant to say the data bus, which was where one of the transceivers was placed on his circuit.

Yes, sorry, misspoke there.. meant the data bus.