Z80 discontinued after 48 years

This link was posted to a Discord I frequent. A sad day.


1 Like

Is there a chance they might license the Z80 to other manufacturers? Are there other manufacturers currently producing licensed Z80 versions or derivatives?

I’m afraid I don’t have the foggiest idea. I’m confident that such components will be available for purchase for a long time, though.

It would be lovely if they open-sourced the design. We could contact them.

FYI, there’s high quality die shot here (19MB JPEG): http://visual6502.org/images/Z84C00/Z84C00_die_shot_20x_1b_7000w.jpg
This is of course a complete redesign of the original nMOS Z-80.

1 Like

The Z80 was second sourced by almost every body in the 80’s.
I can see the slower 2 and 4 mhz z80’s gone, but Zilog may have
faster versions around. The Japanese had some nice Z80 like chips.


The Sharp MZ-series used Sharp’s own Z80 clone called LH0080A.

Who has z80 projects on the go?

I have to ask. So if one of my TRS-80’s or Kaypros needs a new CPU, would I be able to put in a Z80 made recently? That’s opposed to project systems like the RC2014 that use Z80s, but are built with modern parts.

Try to stick to the same speed of a Z80, timing delays do matter in most cases,
Ebay has z80’s for about $10 or less.

Worth mentioning, the eZ80 will be still available.


And that’s what I was wondering. Unless you have a vintage system, that uses a Z80, and needs a new one, you’ll probably just go with the eZ80.

So I can see kit systems like the RC2014 just changing over to the eZ80.

You would need to do the analysis on the schematic but generally you can stick a modern CMOS Z80 into a historic NMOS Z80 system. Unlike the WD65C02 they didn’t change all the timings or mess with some pin usages.

The gotcha is more normally with CMOS v NMOS peripherals. NMOS devices don’t get angry with bus contention. They use a bit more power but meh so what you can calculate it. CMOS gets very upset and will die early or smoke.

There are quite a few machines where peripheral chips are partially decoded such that you can get more than one responding at a time. Those parts you should not replace with CMOS.

For a CPU though - it’s usually fine. There are two differences instruction wise and they never really show up (ok one game on the ZX spectrum goes a bit odd with a CMOS Z80 out of zillions).

  1. The NMOS part has a bug in IRQ handling that the CMOS one does not. The NMOS work around works fine on CMOS so it’s all good
  2. The undocumented out(c),255 on NMOS is out (c),0 on CMOS.

ez80 is a different beast altogether. It’s not 100% Z80 compatible, it’s not available in soldering friendly formats and it has very fast signal edges and all the usual stuff that goes with higher speed design.

I can still get 6809s easily and they are a better CPU that’s been discontinued since forever. I can still get 6303s, 1802s and so on.

And they still make 8051s (What was the Iron Maiden lyric “why do the good die young only evil seems to live forever”) 8)


What are the major incompatibilities of the eZ80 with the Z80?

To clarify, I know the eZ80 and the Z80 have significant hardware differences but I was wondering about software incompatibilities.

I don’t know about incompatibilities - undocumented opcodes and behaviour would probably qualify, and so replacing the Z80 in a CPC or ZX Spectrum and expecting every game and demo to run flawlessly is probably not going to work out. That’s a guess. But I would rather expect running a Z80 CP/M or BBC Basic or output of a C compiler would be fine.

Here’s a post “Zilog Z80 End Of Life Notification” from the rc2014 blog:

What the announcement does not cover is the eZ80, which is likely to continue to be manufactured for quite some time to come. Whilst the eZ80 is somewhat backwards compatible with the Z80 and Z180, it is by no means a drop-in replacement. Making changes to existing hardware and software to take this 100 pin LQFP package is non-trivial, but it isn’t out of the question.

But in the same post:

As a business owner of z80kits.com, a site that as the name suggests, is dedicated to the sale of kits based around the Z80, I find this news quite devastating. Over the past 8 years I have sold almost 5000 kits with a Z80 in them, plus a lot of extra modules that make use of the Z80. The EOL of the Z80 is signalling an EOL of z80kits

1 Like

I guess, it may depend on the form factor, as well: while I can imagine some kind of tiny daughter board for a compatible eZ80 package, using this as a drop-in replacement for densely packed PCBs or in cases where there are several cards tightly installed side-by-side (like with the RC2014) may pose some problems, which may not be overcome but with a complete redesign.

On the other hand, there are so many Z80s floating around that there shouldn’t be any shortage in the next several years. There are so many devices from which they can be salvaged… and there will be a market for this.

I saw an adafruit video indicating that there are also some Z84 chips not being discontinued. I’m not clear how they differ from the Z84C chips listed though.

There’s now an open source Verilog definition of the Z80:

And it’s going to be implemented in hardware:

The first fabrication of FOSS Z80 is scheduled for June of 2024!

Zilog Z80 modern free and open source silicon clone

On the path to become a silicon proven, pin compatible, open-source replacement for classic Zilog Z80.

FOSS Z80 leverages OpenROAD flow and FOSS 130 nm Skywater PDK to synthesize production ready silicon. Tiny Tapeout infrastructure is used to test and pool design with many others to reduce the cost of physical chip fabrication at Skywater Foundries.

The first iteration of FOSS Z80 silicon

The first iteration is developed with Tiny Tapeout 07 using 130 nm process and fits on a 0.064 mm2 die area. The first fabrication is scheduled for June of 2024 as a part of CI 2406 Shuttle.

Mind though that – with a look at the to-do list – undocumented instructions are not implemented, yet.


…when companies no logner derive value from a design, it should be publicly avalible for free and for all time. As it may be of value to iterate on in ways the original creator was not able to envision.


I suspect the cost is the manufacture, not the actual design - after all, we have the design or at least FPGA models and soon a real model by the looks of it, but manufacture is still going to the the costly part.


1 Like