"New" 6510 from 6502

[rupy]

https://monotech.fwscart.com/MOS_CPU_Re ... 91072.aspx

[GARTHWILSON]

Thanks for the information.  I put it in the head post of the "65xx parts sources, genuine and fake" sticky topic.

[rupy]

No problem, I don't even know if it works, just ordered one... hopium!

[BigDumbDinosaur]

rupy wrote:

No problem, I don't even know if it works, just ordered one... hopium!

Well, the site does mention some problems, e.g.:

Quote:

A local enthusiast has discovered that using this adapter in 8501 config in their 64K-modded C16, results in Slipstream (https://psytronik.itch.io/slipstream) hanging...

I’d guess one’s definition of “it works” would depend to some extent on how much such problems interfere with the intended application.

One of the things that could be a “gotcha” with these chip replacements is how the 6502 used as the core responds to “illegal” (undocumented) instructions. I seem to recall that someone somewhere mentioned that Rockwell 6502s don’t behave exactly like the genuine MOS Technology parts when confronted with some illegal opcodes. If this is true, it would surely trip up any program relying on them. Of course, the onus is on the software developer who chooses to use the illegal instructions, not the producer of the hardware.

Caveat emptor, as always.

[GARTHWILSON]

It was good of them to tell about the known problems on the front page; so I don't have anything against giving the link to someone who might be interested in trying it.

[rupy]

I was waiting for this chip, it means as long as you heatsink (and/or undervolt) the VIC2 and SID properly you should be able to run your C64 hardware a long time into the future. I'm not a fan of emulation/FPGA, but this is different.

For example I recommend the SL100 version of the PLA.

Also this exposes the differences and the complexity of those in a compact yet explicit way!

Now the only part we need a modern (non FPGA) version of is the CIA:

https://retrocomputing.stackexchange.co ... e-6526-cia

[rupy]

Hm, would this work with the WDC 6502?

https://www.mouser.se/ProductDetail/Wes ... KH5w%3D%3D

[BigDumbDinosaur]

rupy wrote:

Hm, would this work with the WDC 6502?

https://www.mouser.se/ProductDetail/Wes ... KH5w%3D%3D

Not sure I understand your question. The part to which you linked is the WDC 65C02. WDC never produced an NMOS 6502.

BTW, you should update your profile to indicate where you are located. That link was a non-English page.

[BigEd]

(That comes across as rather hostile, BDD. If rupy, or anyone, would like to note their location in their profile, they are free to do so, and it's a friendly thing to do. But it's not at all required. Of course, location and language are only loosely correlated, and any of us can apply some common sense or machine translation if faced by a language we don't understand. A link to a Swedish-language site which has a chooser for English as an option is not a great inconvenience. It doesn't even merit a reaction, in my view.)

[rupy]

So I received the PCB and it's mixed:

In the reloaded v1 motherboard this CPU replacement does not work: http://move.rupy.se/file/20221228_212659.jpg

But in an old breadbin it "works", sort of; many things don't, so far 50/50:

The "bomb jack" I used fails after the intro.

The "rescuing orc" I have also fails mid readme.

That said "power glove" and "duck hunt" worked.

Seems workable and I hope Jacob can keep working on the PCB:

The W65C02 might work (without undocumented opcodes) but you need to cut a trace, I ordered one but I would like the PCB to have pins instead so you can switch without cutting/soldering!

This (together with the new keycaps, which also have their ups and downs, the stem holder is too thin) finalizes my deep dive back into C64 and the 6502. I managed to make a small "demo": https://csdb.dk/release/?id=188442

But I probably wont prioritize the C64 action MMO I was planning. Got 2 MMO engines on modern hardware and they take enough time.

That said: the whole point for which I created an account here is now solved!

I will make my own stack VM! Believe it or not but Java, C#, 6502, Risc-V, ARM everything is just opcodes/instructions in bytes; we don't need more hardware we need better software = VM that runs alot of bytecode without GC = static heap allocation!

[Dr Jefyll]

rupy wrote:

The W65C02 might work [...] but you need to cut a trace [...] I would like the PCB to have pins instead so you can switch without cutting/soldering!

I suppose it would be nice if the PCB had a jumper setting for this. But instead perhaps you should consider taking the 40-pin W65C02 and bending pin 1 (VPB) sideways slightly so it doesn't engage with the socket. This will have the same effect as cutting the trace... and the effect disappears when a different CPU is inserted.

-- Jeff

[rupy]

Dr Jefyll wrote:

But instead perhaps you should consider taking the 40-pin W65C02 and bending pin 1 (VPB) sideways slightly so it doesn't engage with the socket.

Ok, that's neat... I don't want to break the 65C02 either... but now I know how to fix it.

[BigDumbDinosaur]

Something else that could give you some trouble is the WDC 65C02’s outputs have aggressive edges, unlike those of NMOS parts. Aggressive edges often provoke ringing and ground bounce, which can be “deadly” in a system with 74LS logic and its poor noise immunity.

[BigDumbDinosaur]

Dr Jefyll wrote:

rupy wrote:

The W65C02 might work [...] but you need to cut a trace [...] I would like the PCB to have pins instead so you can switch without cutting/soldering!

I suppose it would be nice if the PCB had a jumper setting for this. But instead perhaps you should consider taking the 40-pin W65C02 and bending pin 1 (VPB) sideways slightly so it doesn't engage with the socket. This will have the same effect as cutting the trace... and the effect disappears when a different CPU is inserted.

-- Jeff

I should mention some other pin differences between the WDC 65C02, the Rockwell renditions and the NMOS part. All referenced parts are in DIP packages.

• Pin 2 (RDY) on all versions of the WDC 65C02 is bi-directional and will be driven low if a WAI instruction is executed (intentionally or otherwise). RDY should be pulled up to VCC with a 3.3k resistor—do not directly connect it to VCC. If RDY is allowed to float, the 65C02 might not run.
• Pin 5 on all static-core WDC parts and some versions of the Rockwell 65C02 is ML (memory lock), which is an output that is normally high, but will be driven low during portions of read-modify-write instructions. Pin 5 is a no-connect on the NMOS part. You should verify that nothing is indeed connected to it. Otherwise, the MPU could end up DOA due to this pin being inadvertently connected to VCC or ground.
• Pin 35 on the NMOS and WDC parts is a no-connect. It is also a no-connect on some versions of the Rockwell 65C02. As a fairly general rule, no-connects should not be connected to anything.
• Pin 36 on the NMOS part, earlier versions of the WDC 65C02 and some versions of the Rockwell 65C02 is a no-connect. On static-core WDC parts and other versions of the Rockwell 65C02, this pin is BE (bus enable), which must be pulled up to VCC during normal operation. If pulled to ground, A0-A15, D0-D7 and RWB will be tri-stated and cease to be outputs. If BE is left floating, there’s no telling what might happen. The usual practice is to tie BE to VCC with a 3.3k resistor.

[barnacle]

BigDumbDinosaur wrote:

• Pin 36 on the NMOS part, earlier versions of the WDC 65C02 and some versions of the Rockwell 65C02 is a no-connect. On static-core WDC parts and other versions of the Rockwell 65C02, this pin is BE (bus enable), which must be pulled up to VCC during normal operation. If pulled to ground, A0-A15, D0-D7 and RWB will be tri-stated and cease to be outputs. If BE is left floating, there’s no telling what might happen. The usual practice is to tie BE to VCC with a 3.3k resistor.

I recently discovered this the hard way: on a w65c02 part (purchased about ten years ago but not used until last week!) if BE is left unconnected per the NMOS datasheet, the output pins are all apparently low - certainly there is no action on them. At least it's an easy one to catch.

Neil