Z80-512K with 3V3 mods howto?

[7alken]

hi, as I finally settled on some RCBus parts it seems, wanting to abuse it by 3V3, I plan to use Sergey Z80-512K at 3V3 too and here is my proposal what to do here with it, as I anyway need some unified adapters for testing RAM chips I have here, so this system may be working fine, I hope - pls, if you have any comments to this craziness, feel free to do :-)

Sergey Z80-512K 3V3-mods, also as SRAM tester (DIP32 universal testing/ZIF adapter even for 7xmem SRAM-TEST !!!)

 Steve: (generally for testing of 7xmod or anything else also at 3V3 against these too)
 - RED    SC702 passive backplane (+ my (LOW-PROFILE CARD or PROTOTYPE now) providing external 3V3 to entire system + 3V3 RXTX BT UART)
 - GREEN  SC726 clock oscillator
 - BLACK  SC729 CF card
 - WHITE  SC727 RTC
 - BLUE   SC705 ACIA (with Hitachi R63B50P CMOS)

 Sergey:
 - BLUE original Z80-512K
 - XOSC in socket, replaceable for 3V3 version (or ext clk from bus/somewhere/sc726)
 - CPLD swappable for 3V3 version (Atmel ATF or Altera epm3064) - there probably isnt dual 3V3/5V0 capable chip, ya !!!
 - dip32 flash replaced by soldered plcc32 adapter (is here) - so plcc32 flash only, 5V0 or 3V3 (no dual voltage)

 - dip32 SRAM replaced by precision socket - ready for DIP32 UNI-ADAPTERS (useable for FAST 7xmem SRAM-TEST too):
 - using "BGA FlexZIF" + "SOP BackZIF" + SOJ/DIP sockets (basic testing max 512k/8bits data even on 16/32 chips)
     (2M)    bga90 13x8, tsop48_ATYP (3V3)
     (1M)    bga90 13x8, tsop44 (3V3)
     (512K)  bga48 8x6, bga48 10x8, tsop44, soj36, sop32 (3V3, 5V0)
     (256K)  
     (128K)  bga48 7x7, tspo44_ATYP, soj32, sop32, dip32 (3V3 + mosel-retro 5V0)
     (64K)   pdip32 (narrow/cache, umc 5V0)
     (32K)   sop28, dip28 (5V0)

Petr

[7alken]

umm, as I want to test also big SRAMs in 7xmem, such hybrid DIP40+20 module (16bit data + more addr) can fit into Z80-512K (moving the R-network to bottom right-angled)
can this be useful as compact memory expansion thing generically (some discrete 74138 possible there with more smaller smd chips)? its weirdo but, maybe?? ))

[Bill Shen]

Just try it.  As long as the Zilog parts are CMOS, they generally can run down to 3.3V.  Compact flash, SD are good at 3.3v, AS6C4008 is rated at 3.3V, flash is not rated at 3.3V, but should work; TTL logic actually works OK at 3.3V.  7.37mhz is very conservative frequency, parts will slow down, but should be fine at 3.3V.  I know CMOS Z80 will work around 16Mhz at 3.3V.  So, as long as you don’t have voltage supervisor reset, you can just lower voltage until it doesn’t work.  You’ll be surprised how low it can go.

Bill

[Bill Shen]

This is my experiment with Z80 at 3.3V.  It runs RomWBW with Z80 at 16.7Mhz. 

https://www.retrobrewcomputers.org/doku.php?id=builderpages:plasmo:epm240dev:3vz80:3vz80rev1_1

[7alken]

ya, thanks Bill, I finally (after few hours of reconsidering the full RCBus kits vs PCBs only as I have here lot of things and chips will be required different) pulled the trigger on Stephen tindie and also got the PCB and CPLD (although also will need 3V3) of Sergeys Z80-512K and I plan to place it into quite nice box SP-7771 (front+back panels are just flat plastics, replaceable by custom PCB-like design, if somebody wants - 248x75x2 mm) which I have here for almost 2yrs planned as "constraint" for my system (even that RCBus then developed into miniaturized 7xsys in progress ... no doubt) so here is draft of it all ... (and fingers X'ed for chips selection, probably (a)LVC will be okay ... TI docs are good for selections)

Petr



  

the box is available here for cca $10 ... mfg has nonsense price $27 for 1pc which goes down below $10 after 10+ pcs;
I have this small box here really for 2years almost, planned for RCBus or something ... so these are my "constrains" ))
(so everything rather low-profile - will see if CF will be exposed outside ... probably yes ...)
https://www.gme.cz/v/1508539/sp7771-abs-grey-krabicka-plastova

[7alken]

just now realized that SC726 will need to be laid upside-down, no big issue here, sure I need female RCBus header also on left side of SC702 ... ))
... original plan was to have "some" way to provide different clocks for different CPU modules, or simply just possibility to use different clocks

as for me here is crucial 12.288MHz (or half of it) instead of default RCBus clock (or experiments with base at 2x too - in case chips on bus will comply,

those old mostly for sure not ... but thats okay)

P.

[7alken]

now I found also nice "dinoboards" resources about CPLD, quite updated ...
https://github.com/dinoboards/cpld-toolchain
https://github.com/peterzieba/5Vpld

and also, noting again nice "digital" logic designer and simulator (simple schematic experiments end exports) written in java:
https://github.com/hneemann/Digital
(from my point of view, best to try as first steps ...)

[7alken]

I tried to peek at some existing memory modules with DIP32 SRAM (usually 512k) and it seems that on all cards where this is used can DIP40+20 fit by shifting -2 pins left, so in fact "centering" it horizontally around DIP32 and by ignoring on RCBus the right-most pins (or omiting them alltogether ...) by maintaining DIP32 SRAM pinout compatible, and in some crazy scenarios, extra pins can be wired, or new custom modules defined - this is only quick draft, but I will use this, it seems ... will try to do some extra pinout, but if anybody will have some notes, they are welcomed - its ad-hoc try to standardize some tiny modules with SMD memory parts also for something bigger/faster than 8bits, mostly ... (sure always best to have anything fast near cpu...)
P.

[7alken]

hi, awawiting delivery from Steve,
checked at mouser that DIP logic discretes 688 and 393 are available as "hc" and 02 04 32 74 125 138 are available also as "ahc" ... = 2V - 6V mostly
in smaller smd packages as narrow soic/sop (1.27mm pitch - also weird availability though) are available also "lv" "lvc"(ti,nxp) or "lvx" "lcx" (onsemi, toshiba)  ... = 1V5 - 3V6 mostly
(these "lxx" have at least mostly 5V tolerant inputs, but cant go over 5V CMOS HI (4V?) without pullups there)

so, in fact most interesting are still CPLD, even smaller ones as max3000 epm3064 (or those Atmel ATF still) - 3V3 powered with 5V tolerant inputs
(! MAX II as epm240 epm570 ... has NOT 5V tolerant inputs, beware !)

usually, entire decoding logic and in some cases entire simple boards can be replaced by single such small CPLD
I considered for a while experimenting with redrawing the address decoders/glues into CPLD piggyback wired by flying into sockets ... )) 
(having all cards linked with JTAG by flying on top TDx/GND and passing TCK/TMS on RCBus user pins .... will see)

Another option can be tiny PCB adapters for (T)SSOP  (0.635mm pitch) "Lxx" discrete logic chips, but those like 393 and 688 aren't available anyway ))

DIP (A)HC will be okay probably in entire range of experiments, even with 5V CMOS processors, though - okay for now;
(nice to have any peripherals capable of even lover voltages or work at 5V -- single supply, but more power domains can be used too)

so, another option are JTAG-chained new PCBs with epm3064 in tqfp44, no problem to solder these at all ...
(and in such setup, you can even omit jumpers for address selection and bake everything by each rewrite of CPLD, no problem)

that reversed SC726 scared only few minutes, as a) its only power and clocks b) I will just mount the jumpers on other side, done ))))
(there will be anyway custom notes, as my clock will start higher - may be at 24.576MHz to allow my loved 12.288 MHz
for some audio chip I have here and it needs this ...) so RCBus can have easily this too or half, so 6.144MHz
(will see if I use clock from here or local for ACIA - Si5351a will allow independent clocks for various audio PSGs, here its important)
... btw, this entire board may be later replaced by custom extension, moving here I2C master and Si5351a clkgen and I2C control of LDOs also
(Si5351a exists as ready I2C module, having 3 outputs ... cant wait for experiments with it ASAP)

in fact, I "wildminded" some future here below around SC702 repurposing ... but that's far away )) generally, I think that can be nice challenge to
try to bring *any* regular RCBus systems below 5V to 3V3 using CMOS parts and playing with even lower voltages and frequencies which it sustain ...
(nothing can stop you to use your own logic at say 2.56V and torture the CMOS, the space/moon technology, even more ....... )))

design proposal below can also use any wasted small CPLDs jtag-chained on "cpld base" (it's really NOT necessary to burn old chips in weapons ... enough!)
-7alken

[Mark T]

If you are looking for a faster 8 bit comparator there is also the 74ac(t)521, with the same pinouts as the 74hc(t)688.

You might find some of the 74lvc in dip format, but these have a Vcc less than 5v.

The 74lvc1g series can operate at 5v

[7alken]

thanks a lot Mark! will check it too... :-)
umm, in fact here now working on DIP32 extended module for SRAM, post was twice eaten by some digital weirdo )) ....kindly pls, is this too weird or can it be usefull this way ???
(checked it physically fits on most existing boards, by luck ...)
Petr

[Bill Shen]

As you’ve speculated, CPLD is the key.  A modest CPLD like EPM570 or even EPM240 can do everything for you except processor and memory.  Here is a design blog building a 3.3V RomWBW SBC with just EPM570, Z80, and RAM. I’ve forgotten that Z80 can actually run at 25Mhz at 3.3V, but that’s probably only achievable by few selected Z80.  https://www.retrobrewcomputers.org/forum/index.php?t=msg&goto=10742&&srch=Epm570#msg_10742

Bill

[7alken]

thanks guys, here also for Mark, I almost overlooked digikey has some ac521 ... but here we with Lyra consulted all options (its quite compatc overview of different logic features) too ...
I even initially went to discrete transistors idea, but was also bring down by Lyra quickly .... ))
https://chatgpt.com/share/68c66564-0990-8000-a209-7b9065e34f43
P.

[7alken]

ya, guys, excuse me, I realized I wildened again about 3V3 (here!) and will be better to discuss this all in retro-comp; still working on DIP32+16, now having this pinout but extending the PCB always to DIP40+20 size for more space and holding also my sodimm-200/80 (DDR2 confirmed as requiremen, only this one fits with wider part under 51mm, it is precisely 50.85)
this was result of linkage to DIP 2.54 pitch ... so anything related more at retro-comp, ya
Petr

[Sergey Kiselev]

On Saturday, September 13, 2025 at 10:26:39 PM UTC-7 7alken wrote:

umm, in fact here now working on DIP32 extended module for SRAM, post was twice eaten by some digital weirdo )) ....kindly pls, is this too weird or can it be usefull this way ???
(checked it physically fits on most existing boards, by luck ...)
Petr

FYI, Dallas/Maxim/now Analog makes a 2 MiB NVRAM module in DIP package. While the module itself is crazily expensive (probably because it is a niche product with low demand?!), you can reuse its pinout:

https://www.analog.com/media/en/technical-documentation/data-sheets/DS1270AB-DS1270Y.pdf

Pins 1 and 34 are not used in this case, but they can be used as A22 and A21 respectively, allowing up to 8 MiB capacity with this pinout.

--Sergey

[7alken]

Thanks Sergey, I want to have also 16bit data possible, so this is DIP32+16 as-is now, it was really quick here, after checking physical possibility to insert this into lot of existing memory boards (sadly few of Bill Shen are already very tight, but he uses mostly fast SOJ anyway) - these you mention are really crazy expensive - and in case anybody want try them, will be possible to wire extra address into that "16" 2nd bottom precision line, or have it extended on board in parallel - or put in on that piggyback (slightly higher, off vertical center), but thanks for this tip ... umm fact is those battery backed chips are quite high also, not sure if they fit with socket anyway ... but, most important difference is that 16bit data width

it was quite wild how this was born ... randomly, watching on your Z80-512K )) ... as primary reason for my entire setup and use Z80-512K as "flat" inserted into passive SC702 is ability to TEST chips laid momentarily on that special module, that was original idea...

(sure, now its about some initial rational pinouts of 7xmem (draft done see "7xsys" here - but while DDR2 sodimm allows nice GNDs on bottom, it still will be better to have spread some GND on top, in case somebody would want to use rather 2x40 (no 80+bottom) 1.27mm instead of sodimm, its still possible, the more that 1mm thick PCB nicely ZIF-fits to such headers, tested here) ... and of course 7xmod to CPLD - but I will start with those memory modules first, it seems - this is obviously quite hard to do it right - if you have any ideas, I will be glad to consider everything - clocks will go throigh CPLD from Si5351 and at 2 power lines from base too, using probably TPS7A7001DDA - having feedback steered by I2C pots MCP4442-104E)

Petr