Does anyone know of a SMD Z80 core that can run on 3.3V?

[Richard Lewis]

I know the Z180 - eZ80 come in 3V3 versions but I looking specifically for a fully z80 compatible 3.3v part (I want it to pass Zexall). Not sure if Zilog ever made one but perhaps Toshiba or NEC etc did? I'm building an Z80 expansion card for a CPLD board that is max 3V3 IO and don't want to have to drop in 4 8-bit level converters just so I can play nicely with 5V. I don't want to use a Z180 because of space constraints and I'm not sure how compatible an eZ80 is with a vanilla Z80 core. I believe the eZ80 has a register that sets Z80 compatibility mode but never had a chance to play with one. 

I'm going to preemptively answer one of your questions: "Couldn't you run a Z80 core on the CPLD?", Answer: "Nope, the chip isn't large enough..." :-)

-Richard

[Alan Cox]

On Saturday, 20 July 2019 20:57:48 UTC+1, Richard Lewis wrote:

I know the Z180 - eZ80 come in 3V3 versions but I looking specifically for a fully z80 compatible 3.3v part (I want it to pass Zexall). Not sure if Zilog ever made one but perhaps Toshiba or NEC etc did?

eZ80 is almost Z80 compatible. It borrows nonsense instructions like ld l,l for prefixes. To all intents and purposes its Z80 compatible in that it runs CP/M etc fine. Z180 can be made to be compatible but you have to write a handler for the illegals (they trap to address 0 in a detectable way and your JMP 0 code splits CP/M warm start from a trap and fixes it up).

NEC btw are not quite Z80 compatible. It's really tricky but you can tell them apart in software. ZEXALL isn't good enough to pick it up (in fact ZEXALL is actually missing some standard corner case tests and doesn't even notice a 16bit SBC bug in libz80 or an obscure LD A,R/LD A,I bug in the T80 core).

Alan

[Greg Holdren]

Richard,

Edit: Opppssss nevermind...... Ignore below. Didn't read the I know the Z180 - eZ80 come in 3V3 versions close enough. :)

Check out the Z80S180/Z80L180 CPUs or the Z80182/Z80L182 (uses S180 core). The S180 core can run at 3.3V

S180:

https://docs-emea.rs-online.com/webdocs/0c83/0900766b80c8370a.pdf

182:

https://www.mouser.com/datasheet/2/450/z80182-35610.pdf

Why does it need to pass zexall?

Greg

[Richard Lewis]

Doesn't necessarily need to pass zexall but I am aware of z80 specific CP/M applications (from distant memory) that made use of undocumented instructions. I'm also keen to emulate a TRS-80 and I know for sure there were plenty of games that used undocumented op-codes. 

-Richard

[Tadeusz Pycio]

I want it to pass Zexall

The ZEXALL test does not pass Z180 and some CMOS processors of second sources (NEC D70008A, Toshiba TMPZ84C00).

[Greg Holdren]

So zexall test undocumented instructions as well as the documented instructions?

[Bill Shen]

I believe the CMOS version of Z80, Z84C00XX, can work down to 3.6V at reduced speed. Don't know how reliable but it is worth a try.
Bill

[Bill Shen]

Just for kick, I lower the voltage on a Simple80 rev1 to 3.3V (verify the voltage with a meter) and booted it up.  It passed the monitor's built in memory diagnostic.  It booted CP/M2.2, PIP one entire disk (3 meg of data) to another disk with verify OK, ran the mbasic80 asciiart benchmark normally, compiled a simple C program with Hi-Tech C, downloaded big file with xmodem, ran zork OK and started running zexall and passed the first few tests (it would take 106 minutes to run the full test, I figure 10 minutes of running OK is good enough).

The simple80 rev1 has the following parts:
Z84C0020PEC
Z8442B1 (SIO-2)
KM681000ALP-8L
W27C512-45Z
74HC00
74LS32
7.37MHz oscillator
CF disk

This is unexpected.  I've read that CMOS Z80 can run down to 3.6V, but didn't expect the entire board to run at 3.3V at 7.37MHz.

  Bill

[Richard Lewis]

Heard rumors online but this is the first time anyone has confirmed it. This is awesome. I wanted to use a vanilla Z80 without resorting to shifting 5V <-> 3.3V. 

[Bill Shen]

I built up two more Simple80 rev 1 boards and populated them with new set of ICs and CF disks.  One board works at 3.3V and passed the complete zexall tests.  The other works at 3.4V but failed at 3.3V.  Both are running at 7.37MHz.  Pretty amazing results.
  Bill

[Richard Lewis]

I'm building a Z80 expansion card for my CPLD board. I'm using an Intel/Altera MAX V as my CPLD and have no trouble driving 5V TTL but would need a level shifter or external diode clamp to go the other way, which means anything on the Z80 that is an output or a bidirectional pin. Looks like the I/O pins have an absolute maximum of 3.6V so maybe running the Z80 at 3.5-3.6 would be more stable? 

[Karl Albert Brokstad]

Great news Bill

My initial thought about this is that your system is quite minimalistic and draw very little power.

So there is sufficient power to drive whole the system.

What if you tested a more complicated system with more devices and system logic, would it still run?

Karl

[Greg Holdren]

Richard,

Any reason why you are using the Max V then? I remember back you talking about it but don't remember the reason.

I was told that the Xilinx XC9500XL series of of devices work just fine in Z80 based circuits. So my guess VIH is being meet by whatever the system this person had designed with 5V parts. The XC9500XL runs on a single 3.3V supply but are fully 5V tolerant. Yeah, different tool. meh Ford/Chevy. :)

Greg

On Saturday, July 20, 2019 at 12:57:48 PM UTC-7, Richard Lewis wrote:

[Bill Shen]

Richard,
Z80 should work reliably at 3.5V
Bill

[Bill Shen]

Karl,
The Z80 in my minimalistic SBC is driving 4 loads: CF disk, RAM, EPROM, and SIO. Some address lines are also driving additional TTL logics. 3.3V parts are far more complex than the 5V parts, so I think 4 loads drive are enough for most 3.3V systems. BTW, the current draw of Simple80 is around 15...@3.3V
Bill

[Richard Lewis]

One reason is because I bought 2 off of Ebay to play around with and by mistake they sent me 20. So I want to use them in something. But mostly because I was looking for the largest CPLD/FPGA I could find that is reasonably recent, I could still solder by hand (BGA is still too scary to attempt) and was supported by the latest version of Quartus. Interesting however is that Quartus 18.1 is no more or less stable than 13.0.1 (the last version that supports the Max7000). I crashed Quartus 13.1 doing a JTAG boundary scan and thought: "I bet they fixed this in 18.1 since it is 6 year newer... nope... still crashes". 

I used to play around exclusively with Xilinx stuff like the Spartan 6, Zync etc but got bored and thought to give Intel/Altera a try. I'm agnostic though. Anyway jumping from Vivado to Quartus was like jumping into a different car, once I learned how to turn on the radio I was fine. 

-Richard