The Z80's secret feature discovered after 40 years!

[RoGeorge]

You are right again, a copper rod works even better.

First thing I do after fitting a new folded wire is to snip it in half.  Then twist the two parts together.  A proper diameter copper rod might be hard to find, while thick solid copper is easy to get.  I use leftovers from mains wiring.

For example, this Z80 system I designed in the late 80s was soldered using twisted copper wire as a soldering tip, and a very powerful soldering gun of (100-200W and no temperature control):






https://hackaday.io/project/1411-xor-hobby-a-vintage-z80-computer-prototype

The wrapping wire was soldered instead of wrapped, and that type of wire was covered in something that couldn't stand the hit, so I've start soldering "through" the plastic, without stripping or even cutting the wire.  Not exactly the best soldering practices.  Though, last time I've powered it, it was still working.  That was about 30+ years later after soldering through the cover of unstripped wrapping wire.

It was stupid, but it works. 

[DavidAlfa]

Definitley, you don't need $1K equimement, it's a lot more the skill than the machine.
When I was younger and poorer I only had the typical 30W unregulated pen-style JBC soldering iron. Yet  I soldered a ton of tqfp and all kind of small stuff.
This is an old picture on the OpenWRt forum, made by me.
Wanted USB, the router didn't have, but the soc did. So I had to become creative.
Why the scratching? Well, the AR9331 uses a weird package (Dual-row LPCC whatever) where one row will be hidden under the package.
Pins on each row are separated by 0.5mm, then both rows are routed together to the top layer, so those wires are 0.25mm apart from each other!

[RoGeorge]

That's very hardcore! 

[wraper]

Watch this pron. In Russian but video shows it all. A 8 minutes you can see spiderweb.

[RoGeorge]

Wow! 

[DavidAlfa]

Someone had a lot of free time!
What the hell happened to that card? Got ran-over by a train?

[RoGeorge]

I guess a short circuit in the original voltage stabilizer heated to hot red and carbonized the PCB.
Or maybe the board was shot by a bullet. 

[wraper]

Faulty buck converter burned the PCB. Someone removed PCB material before he received this board. He joked that it was given to beavers for fixing.

[DavidAlfa]

He was lucky to have the pcb files, otherwise it'd be unfixable.

[wraper]

He was lucky to have the pcb files, otherwise it'd be unfixable.

He had a boardview for a similar board but not this one. So needed to figure out where traces by going over components/pads with copper braid to find continuity.

[radiolistener]

Z80 really has some secrets which are discovered after 40 years, this is how some instructions affecting undocumented flags. Z80 has internal register which is updated on some instructions and undocumented flags are set depending on that hidden register.

[iMo]

Those "hidden features/instructions" in the chips from 70/80ties are remnants of the design/testing/verification practices from that era. The focus was on the intended functionality only, their tools were simple (pencil and paper), they worked under time stress, etc., and when something toggled somewhere on the chip without an impact on the given specification they did not care much about it.
A complete test of all the stuff on the chip is not possible even today..

[radiolistener]

A complete test of all the stuff on the chip is not possible even today..

Do you know some feature of Z80 which is not implemented in modern emulators (which implements latest finding on MEMPTR register and all other undocumented features)?

As I know Z80 die were scanned layer-by-layer on electronic microscope, it's schematic was reverse engineered and all its undocumented features were explained.

There are several clones of Z80 with a little different behavior, for example OUT (C),0 can lead to write #FF on some clones.

Here is picture of Z80 die: https://s.zeptobars.com/z80-Z0840004PSC-HD.jpg

Here is some details on how its registers are implemented: https://www.righto.com/2014/10/how-z80s-registers-are-implemented-down.html

What secret is still not reversed is FD1793. As I know some attempts to reverse engineering were made, it has some firmware inside. But there is still no full emulators for that chip. Here is a die photo of Russian clone 1818ВГ93: https://upload.wikimedia.org/wikipedia/commons/thumb/5/5a/Kr1818vg93-3-HD.jpg/800px-Kr1818vg93-3-HD.jpg
You can identify internal ROM with firmware on the left top side. But I never seen that firmware content and it's description on internet. If someone have it, please let me know.

[MarkT]

I thought the fairly well known obscure undocumented Z80 feature was that the two bytes of the IX and IY registers can be addressed individually like HL by adding DD or FD prefix before H and L referencing opcodes, or something like that - basically the DD and FD prefixes just alter the meaning of "HL", "H" and "L".  Never thought to try DD FD as a prefix though, suspect the last one wins.

Whoops, still able to remember many Z80 opcodes decades after programming it.  I blame the Sargon chess book that had a machine-code listing you could enter into your Z80 system of choice!!  Still have the MOSTEK Z80 manual somewhere...