RIP Z80

[SiliconWizard]

Frankly, I still don't see much point, if you want to do that, compared to a soft core on a small FPGA.

It's just for fun, to remind myself of the bus timings, etc.  Any CMOS SMT version would work just as well.  I don't have any actual use for it.

Then you shouldn't have any problem buying a bunch of real Z80's, which can still be bought from various places. And the Z180, again, which is in stock at Digikey.

You probably misunderstood my point: I didn't say there was no point in working with real hardware, if you're interested in that.
I just said that for this, actual parts are still available, so what I don't see is the point of going as far as designing brand new silicon for that. Doesn't make sense.

Just buy a couple Z80s (probably on Ali or eBay), or Z180s (Digikey, I think Mouser has some as well), and start hacking away!

[bson]

Then you shouldn't have any problem buying a bunch of real Z80's, which can still be bought from various places. And the Z180, again, which is in stock at Digikey.

This is true, but being part of a community effort still tickles something positively, psychologically speaking, that a commercial part can't.  Since it's purely for fun this is a non-negligible aspect.

[RoGeorge]

Many microcontrollers today can emulate a Z80 almost flawless.  Only a few days ago I've seen an entire Sinclair ZX Spectrum home computer emulated with an ESP32:  https://hackaday.com/2024/06/16/esp32-powers-single-pcb-zx-spectrum-emulator/

[Sal Ammoniac]

Digi International makes a series of MCUs called the Rabbit which include a Z80 clone.

https://www.digi.com/products/embedded-systems/digi-connectcore/system-on-modules/rcm3000

[SiliconWizard]

Then you shouldn't have any problem buying a bunch of real Z80's, which can still be bought from various places. And the Z180, again, which is in stock at Digikey.

This is true, but being part of a community effort still tickles something positively, psychologically speaking, that a commercial part can't.  Since it's purely for fun this is a non-negligible aspect.

Well, certainly, I was speaking from a purely pragmatic POV, but if people want to design a brand new Z80 on silicon, I'm nobody to say they shouldn't!

But since this all has a significant cost, I think it's in their best interest to make sure they know all their options before investing time and money in that. Also a bit wary of the "hype" effect triggered by the EOL announcement, that may fade off in a few months from now, before the project actually gets completed. We'll see.

[zapta]

1k resistor outlast the z80. What a shame.

[SiliconWizard]

Just thought I forgot to mention the eZ80, which is still in production, Littelfuse has no intention to stop producing it, and should be binary-compatible with the Z80 while adding tons of features and being much, much faster. Still used in some TI calculators AFAIK.

So if one wants to build a Z80 board for various purposes including running CP/M on real Zilog hardware, then that would be it IMO. Still easily available everywhere.

[woofy]

For me the only reason to use a Z80 is to scratch a nostalgic itch, and that's what I did a couple of years ago. An FPGA was used to mop up all the random logic, memory mapping, interrupt logic, vga, that sort of thing. I could have put the Z80 into the fpga, but no. That would not scratch the itch, it had to be a real Z80. A look-a-like chip eZ80, 64180, fpga or whatever just would not do.
I must be getting old, nostalgia isn't what it used to be. 

[MT]

[westfw]

Don't some vendors other than Zilog make "Z80 compatible" chips?Are they all also going to stop producing them?
Ebay listing: "Z80 compatible CPU.  New, from Russia." (like Nixies!)

[SiliconWizard]

I know this has made a lot of "RIP Z80" videos, and the "hype" will pass, and soon enough no one will be talking about the Z80 anymore, at least no more than just a few months ago.

Littlefuse, who owns Zilog, has made a clear statement about the why: there is not much market for original Z80 parts anymore, which is a dog by today's standards, and they had to maintain a 4µm process node just for this chip. Crazy. They should have pulled the plug earlier.

As I already said, there are newer, "compatible" Z80 parts, that Littelfuse has been clear about keeping production lines for. On more modern process nodes, and not 5V anymore, but who needs that anymore except hobbyists keen on vintage stuff? Why would companies bother making 5V clones for a market that would never repay the investment?

To me it all really looks like yet again one of these hype waves, that triggers a frenzy about something that people otherwise mostly didn't care about, and that will pass as quickly as it started, probably as soon as Youtubers will be tired of making retrospective Z80 videos. Just my 2 cents anyway.

[brucehoult]

Littlefuse, who owns Zilog, has made a clear statement about the why: there is not much market for original Z80 parts anymore, which is a dog by today's standards, and they had to maintain a 4µm process node just for this chip. Crazy. They should have pulled the plug earlier.

Do they only have the original hand-drawn masks and not a net list?

TSMC's 130nm and 180nm nodes (for example) are both 5V compatible. Fully-automated layout from a net list would probably get you a 200 MHz part.

Personally I'm just glad I don't have to do anything serious with either the 6502 or z80 any more. Both are just absolutely horrid. If you need 8 bit then AVR is infinitely nicer, MSP430 for 16 bit (which is pretty pointless these days), or RV32EC or Cotex-M0 for tiny 32 bit cores.

CH32V003 (2k RAM, 16k flash, 48 MHz) is 1732x1172 µm. I found numbers for MSP430FR5738 (1k RAM, 16k FRAM, 24 MHz) of 2280x2080 µm, 2.3 times bigger. z80 is 3545×3350 μm. 6502 is 3890x4270 μm and would be called 5μm today (the very first ones were I think 8μm).

[coppice]

TSMC's 130nm and 180nm nodes (for example) are both 5V compatible. Fully-automated layout from a net list would probably get you a 200 MHz part.

TSMC's 130nm and 180nm nodes are 5V I/O compatible, but doing the core in the big 5V I/O transistors would made the part large and slow. You won't get to anything like 200MHz unless you go to the complexity of split supplies.