Hi...

I accidentally found a Z84C0020FEC which is a Z80, has the wrong package but runs at a stunning 20 MHZ... anyone experience with using one of these beasts in an MSX ? The package problem is easily solved with a adapter print

Wonder what timing issues you might run into when going beyond 7 MHz :-)

Jos

the VDP May explode! The PSG will burn...

You could probably connect it and run it off the VDP clock. The problem will be the other hardware. PingPong may actually be right that the PSG would burn if it was also running at 21MHz. So that would probably have to be clocked at a lower speed. Same for communication with the cartridge slots. As the OCM mentions that running at 21MHz gives problems with external hardware, you would have to account for that as well. Maybe RAM and other components would be in trouble as well. It would be an interesting project to get it to all work together. I have little experience with this stuff, so maybe someone else can give a more complete view on this.

Of course you have to switch back to access the VDP and PSG, just like with normal 7 MHz machines. I was thinking of a 3.58 / 14.32 MHz switch or something :-)

@foske: the ram may be the main problem also switching at 3.5mhz. As you probably know, the old z80 @ 3.5 mhz on msx receive a wait state on every m1 cycle, so you have an idea of how much slow is the msx ram.

There is no speed gain if you have a z80 at 20mhz stopping at every:

fetch cycle
vdp access
psg access
ppi access.

what the z80 could do at full speed? only instruction decoding and part of instruction execution?

Quote:

Hi... I accidentally found a Z84C0020FEC which is a Z80, has the wrong package but runs at a stunning 20 MHZ... anyone experience with using one of these beasts in an MSX ? The package problem is easily solved with a adapter print Wonder what timing issues you might run into when going beyond 7 MHz :-)

Ofcourse many experiments have been performed in the past on MSX 1 and 2 machines, even turboR, to speed up the lazy thing.
The findings are that double speed can be achieved reliably, and 10 MHz is the top without replacing most of the ICs and rewrite the BIOS.

The quality of the components typically found in MSX computers do not push far beyond their guaranteed clock speed of around 4 MHz. Going to 7 MHz (doubling the clock speed) already pushed early MSX 2 machines in to trouble, mostly the ROM (EPROM) had to be replaced by a faster type. RAM ICs also are limited.
Expansions like slot and memory expanders are even more harder to design for speeds beyond the standard speed. Only later generations of the MK/Gouda/HPN designs and combinations allow 7 MHz clocks.

Another source of obstacles is the way most software is designed. Especially when dealing with hardware timing is often important, and when the cpu is faster than the original programmer anticipated the hardware is called upon too fast.
The vdp is known best for this, software delays are required, done by dummy instructions and that time wasted is ofcourse clock dependent.

My experience with faster clocks on MSX machines is that it is very valuable for the more boring things like editing, compiling, business applications (most CP/M software works fine on MSX-DOS in screen 0) and self written programs tested on higher speed.
But start a program requiring graphics and especially sound and you have to switch back to the standard speed.
This includes all games and demos!

You can use this fast Z80 ofcourse at 7 MHz clock speed!
I hope you know Alwin Henseler turbo design, nice way to experiment.