FCFC part 1

Fake chips from China (FCFC): fake Z80 chips

This blog post is about testing faulty retro chips from China revealing the faults.
Back in 2016 when I started the reZet80 project, ebay and Aliexpress sellers from China offered thousands of retro chips. Being interested in chips from the 70s and 80s I ordered a few Z80 CPUs and many 74LS series chips.
I never bought from the cheapest sellers because I never trusted them (interesting how they try to outbid each other by 1 cent). A wide range of chips is still offered today but prices went up a lot after Corona and the war in the Ukraine.
Most of the chips are OK, at least they do the job they are supposed to do according to the datasheet. For the price they might be Chinese clones of branded chips but I don't care as long as the functionality is not faulty or limited.

Let's get started with two Z84C0020PEC chips, 20 MHz CMOS Z80 CPUs.

The scratches on the back side of the chips were caused by myself.

Inserting the CPUs in a working reZet80 system does nothing. An oscilloscope reveals that both Z80 are dead.

An oscilloscope capture of a "good" Z80 CPU looks like this:

Channel 0 (white) is /RESET, Channel 1 (brown) /MREQ, Channel 2 (red) /RD, Channel 3 (orange) /WR and Channel 7 (purple) /IORQ. Channels 4, 5 and 6 (yellow, green and blue, respectively) are a mix of address and data lines. My test program repeatedly reads some data from ROM and outputs it to I/O, so note that /IORQ and /WR are both low at the same time with /MREQ and /RD high.

Firstly the visual inspection, I examined the logo and production date: Nothing conspicuous but the marking on the bottom is missing. This might be a clue.
The pins look brand new.
Second step: Use acetone to expose very cheap fakes. No, still the same top markings:

Using nail polish remover in the third step makes the markings look paler but still OK:

Time for destructive behavior: Remove the top finish with a sharp tool (well, too much destruction):

BUT revealing some part numbers on the back side:

After some investigation on the net I found that for earlier Z80A CPUs 12- and 13-digit codes were used. So one chip might be a resurfaced NMOS Z80A. 14-digit codes were used for Z8000 processors and peripherals, so the second chip could be a Z85xx. Or maybe a newer NMOS Z80A.
With these clues I tried to measure the internal resistance between +5V (Vcc) and GND for all 40-pin Z80 and Z8000 processors and peripherals. Nothing, absolutely dead. So RIP, whatever you were!
But wait, one last step that comes to my mind, decapping with hammer and pliers:

This is how far I get with brute force:

IMO these don't look like pieces of a Z80 die...