Just a quick query for those who know better than me - if I decided to use the '265 for my next project rather than the '816, then it might have advantages:

Full 24-bit address bus brought out, so no latch & buffer.
Internal VIAs (or VIA type things with parallel ports and timers)
Internal UARTs
Other internal stuff I'm not that fussed about - e.g. tone generators.

But it's a big PLCC - so not having used one in any of my own designs so-far, can I assume that 84-pin PLCC through-hole sockets are fairly generic and would "just work"? (and this might be completely newb/dumb question, but just want to make sure!) e.g.

https://uk.farnell.com/multicomp/mc-84p ... dp/2097218

Basically I'm thinking it would save me some board space (the upper 8 address latch and databus buffer, and a VIA or 2) - however the only down-side I can see right now is that it's only rated to 8Mhz and not the 14Mhz of the '816. (anyone overclocked a '265 ?)

Anything else I ought to consider? (I'm looking at the same setup as I have so-far, essentially treating it as a RAM only system, so would disable the internal RAM and ROM with an ATmega doing the bootstrap and some IO stuff) but one of these, a can oscillator, a Garth memory module or 2, and ... what could possibly go wrong ...

Cheers,

-Gordon

_________________
--
Gordon Henderson.
See my Ruby 6502 and 65816 SBC projects here: https://projects.drogon.net/ruby/

Hi Gordon,

using that socket should work perfectly. You only need to double check if your CAD has a correct definition of the socket and its pin numbers (their counting is somewhat special).

In order to get it easy up and running you only need two crystals (32768 Hz and 3.6844 MHz) and a TTL<->USB Abdapter (ok, plus some C's). Then it should boot into the Mensch Monitor with 9600,8,N,1 IIRC.

Once you have your own monitor/bootloader/etc. in say a flash EPROM you can use faster clocks. I think I briefly tested 18.432 MHz (RAM only) with no issues. With a 55ns EPROM it runs hours @ 14.7456 MHz.

Some additional information you may find at New WDC W65C268QBX Board (pg.3++).

For additional I/O there are two /CS built in: /CS0 (32 byte) and /CS1 (64 byte). Both have a timing that corresponds to 65xx I/O (i.o.w /CS wents low during PHI1 and is released past end of PHI2). So it should be no issue to attach 2 VIAs with minimum efforts.

The /CS3..CS7 are gated with PHI2. This eases generating /WE and /RD for RAM/(E)EPROM.


Have Fun.
Arne

My CAD is ... Fritzing, which is/was OK for the basic 6502 board I've just done, however somewhat sub-optimal for this and it's probably time I learned KiCAD anyway.




OK, thanks. My plan for my '816 project is to copy my existing 6502 project and have an ATmega share 256 bytes of RAM with the CPU (at $00FFxx), so with the '265, if I keep the '265 in reset, BE & Rdy low, poke the bootloader in, the shared RAM, then raise Rdy, BE and Reset in that order - which if I've interpreted the datasheet right puts the '265 core into what they call "emulation mode" which means no internal ROM/RAM decoding and full 24-bit external address bus then it ought to boot directly from RAM with no ROM needed.




Not sure I need any additional IO for my system - it "uses" 4 x 8-bit ports for the 24-bit address bus and 8-bit data bus in "emulation" mode, that's leaves 4 ports free - and as I only need 2 ports for my expansion plans, it leaves me with plenty in-hand.

It's making me wonder why more people aren't using this/talking about it and I feel I'm missing something, but maybe not and it's the fear of the unknown!

Thanks,

-Gordon

_________________
--
Gordon Henderson.
See my Ruby 6502 and 65816 SBC projects here: https://projects.drogon.net/ruby/

Well, there is more about the '265. Have a look at =W65C265SXB.

I don't know why the '265 isn't more prominent. But there are a few reasons: big case (and only SMD, not easy to use on a breadboard, only perf board), more expensive (but when you add one 65C51 (if that beast would work) there is no much difference), and most likely: only rated for 8 MHz.

If you boot from your ATmega you should release first RESB then BE/RDY. If BE/RDY is tied hi it boots into internal ROM. If BE/RDY is low when RESB is deasserted nothing happens because the bus isn't available. The CPU has to wait until BE gets true.

It should start up as 65C02 (emulation mode) using the provided RESET vector running at CLK speed.


Arne

I can think of a few possible reasons that the '265 is overlooked.
It's got no user-programmable ROM, and almost no on-chip RAM.
It has no hardware SPI or I2C, and the four onboard UARTs are either all on or all off. EDIT: This is wrong. I misread the datasheet. The UARTs have individual enable bits and configuration registers.
The processor bus being present at all is nice, but it isn't particularly flexible, in that you can't turn off the bank address, so those eat 8 IO pins whether you want them or not.

Here begins my opinion:
In comparison to, say, an ATMega328, I think it's a fairly bad microcontroller, even including the ATMega's fewer IO pins, just on the basis of what you need to do to make it nice to use. It only eliminates the UART chip and address decoder(and not always that). You still need some user-programmable rom and you'll want more ram than what it has on chip. If you want more IO, that's a VIA you need to include, and if you want hardware SPI, that's yet another IC. The '816 can get a system with a VIA and hardware SPI with 7 chips, where the '265 results in a system with 5. It is an improvement, but not as much as a more ideal MCU that could get three. As far as I know, such a thing doesn't exist in our realm in ASIC form.

Mind you, I'm not going to jump ship to the AVR, despite my comparison making it look better. I'm just planning to use the '816 instead.

I would just suggest getting a PLCC remover tool also, to be able to get the part out of the socket easily and without damage.


I would be very interested to find out what you find. When I was talking with Bill Mensch one time, he asked me why forumites aren't using the '265. I mentioned the speed rating. He said something like, "It might go a lot faster. We just never tried it any faster than 8MHz" (or however fast they try it in order to mark it for 8MHz). Wha??? This and a few other things he has said put me of the opinion that he's really shortchanging his own product in various ways. [Edit, a few minutes later: GaBuZoMeu's 18.432MHz is good news. Now I'd like to know what an '816 can really do!]


I keep seeing this kind of reluctance on this forum. Hopefully your CADs aren't that hard to make your own packages or edit what was supplied in the libraries. It's very quick to make up new ones in my 25-year-old CAD, and in fact I have hundreds of packages I myself made, partly because of all the new proprietary parts and partly just to get higher density than what the standard parts in the libraries that came with it gave.

Thankyou for the additional comments, DerTrueForce.

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

I think this is wrong. There are four registers ACSRx ($DF70, $DF72, $DF74, $DF76) and each of them have a Tx enable bit and a Rx enable bit.

Well, that's embarrasing. I misread the datasheet. You're right, each UART does have individual enables. I'd better edit that.

Oh no, never mind!

Reading WDCs datasheets without getting puzzled is likely impossible

I got confused by the order of BE & Reset signals earlier too....

-Gordon

_________________
--
Gordon Henderson.
See my Ruby 6502 and 65816 SBC projects here: https://projects.drogon.net/ruby/

If you extend your memory above 64K then you lose a lot of the I/O port pins.

On my 65C265SXB I have 1MB of extended RAM using /CS7, four Uarts (but no RTS/CTS) and a bit-bashed SPI (4 pins). That only leaves one or two spare pins and the two PWM outputs spare.

That said I think the 265 is a nice all round package. The only problem with the SXB board is its low clock speed and too little RAM.

_________________
Andrew Jacobs
6502 & PIC Stuff - http://www.obelisk.me.uk/
Cross-Platform 6502/65C02/65816 Macro Assembler - http://www.obelisk.me.uk/dev65/
Open Source Projects - https://github.com/andrew-jacobs

For me, the real show-stopper is the non-reprogrammable internal ROM. Every other MCU on the market (whether 8-bit or RISC) now has internal Flash that you can reprogram not only in-circuit, but in many cases with user code; by comparison, not being able to change the internal code at all is absurdly limiting. Working around it requires nearly as much external hardware as a 'C02 or '816 system - so we just use the latter.

To be fair, it's a very different process to have a programmable ROM. Probably a more expensive process. Perhaps they could have made it easier to hook up external ROM (and RAM).

Somehow I start smiling and shake my head when people have difficulties simply to ignore the internal ROM of the '265 (and '134). If I understand the initial behaviour of RESB and BE/RDY correctly, a simple RC combination is enough to delay BE/RDY=1 from RESB=1 and in turn getting rid of the internal ROM at all (it may re-enabled if required). Getting rid of the ROM you have all degrees of freedom with your code. During startup there is a /CS for the upper half of the first bank where you can place your monitor/os-boot/etc. And again, you can get rid of that section by disabling that /CS. This appears to me not being very restrictive.

Of course Flash-ROM would be nice. But back in that days (I assume late 80's) only ATMEL (and perhaps Cypress) have the capabilities (and patents) to mix HCMOS and Flash on one chip. In that days that was very very advanced stuff (and expensive too). This may simply prohibit its implementation. (And later there was probably not enough money and customers to do this.)


my 2 cents
Arne

Back in the late 80s is one thing, but now I'm looking at a Mouser page chock-full of SOIC and LQFP Cortex-M0+ MCUs for about 1 Euro each. Every single one of them has Flash and SRAM on board, as well as a good variety of I/O peripherals. People are buying those by the shovelful.

The one thing they don't do is bring the raw address and data buses out to the pins. But if that's what you need, the plain '816 works with less hassle and in a P-DIP form factor.