MicroMac Speedy for IIfx

[petteri]

I managed to get stable operation with 100MHz crystal oscillator, random wires, breadboard etc as seen here: https://68kmla.org/bb/index.php?threads/macintosh-iifx-with-pronitron-80-21-v1-5.42667/#post-467388

So I guess there is some tolerance for stupidity but with limits

I think the clock speed is halved before the CPU somewhere on the logic board. Any info what kind of circuitry that is?

 

[Phipli]

I managed to get stable operation with 100MHz crystal oscillator, random wires, breadboard etc as seen here: https://68kmla.org/bb/index.php?threads/macintosh-iifx-with-pronitron-80-21-v1-5.42667/#post-467388

So I guess there is some tolerance for stupidity but with limits

I think the clock speed is halved before the CPU somewhere on the logic board. Any info what kind of circuitry that is?

I.... wouldn't recommend breadboard for 100MHz.

I push my luck with 20MHz, 40 might be possible with luck... 100MHz is... significantly out of manufacturer specification.

I'd suggest rebuilding the circuit on some proto board.

 

[Phipli]

I think the clock speed is halved before the CPU somewhere on the logic board. Any info what kind of circuitry that is?

Probably a flip flop or a counter inside a PAL or custom silicon.

 

[Bolle]

custom silicon.

Memory controller.

 

[cheesestraws]

I managed to get stable operation with 100MHz crystal oscillator, random wires, breadboard etc as seen here

Wow. I'm surprised that worked. Nice one.

 

[Phipli]

I managed to get stable operation with 100MHz crystal oscillator, random wires, breadboard etc as seen here: https://68kmla.org/bb/index.php?threads/macintosh-iifx-with-pronitron-80-21-v1-5.42667/#post-467388

So I guess there is some tolerance for stupidity but with limits

I think the clock speed is halved before the CPU somewhere on the logic board. Any info what kind of circuitry that is?

Another thought, have you grounded the inputs on the other gates on the 74125? They could be ringing. When you're making your protoboard version it's worth doing that.

 

[petteri]

Alright, some progress finally. I used 5V from SCSI power connector to trigger an external interrupt on Arduino (https://www.arduino.cc/reference/en/language/functions/external-interrupts/attachinterrupt/) and that then starts the clock signal at the desired frequency. Still using breadboard and mess of dupont cables.

I was able to create the slowest ever IIfx at 20MHz. It also was able to boot at 40MHz and rather strange 42.43MHz. But above that it would not give me a happy chime.

So this proves this combo is actually providing somewhat ok CLK signal. I have left other outputs of 74125 open so far, have to try that as well. Next step will be to create a protoboard version and also to find someone with proper scope to inspect the output.

 

[petteri]

And to shock those who have some idea of electronics:

 

[petteri]

55MHz and seems to be stable, at least to run performance tests. I managed to make it boot at at 57.4Mhz but that was not stable enough to run tests. Maybe some additional cooling for CPU, FPU and memory controller would help?

I replaced the breadboard with a perforated protoboard. I also discovered that is important to disable other clocks of the Si5351 and to have short signal wire without other wires going nearby.






I wonder what was the reason for Apple to settle with 40MHz...

 

[Phipli]

I also discovered that is important to disable other clocks of the Si5351 and to have short signal wire without other wires going nearby.

If you solder a female header onto your protoboard, wire it in and plug the Si5351 into it you'll get an even cleaner signal. Wires at those kind of frequencies are acting as transmitters and recievers.

I wonder what was the reason for Apple to settle with 40MHz...

Motorola defined the maximum speed of the chips, based on testing, and it will include a margin for degradation, apply to a significant temperature range and be on a batch, not chip, basis. Most chips will happily overclock 20%, some more. Technically it shortens their life and increases energy consumption and heat generation. I'm not worried about the life thing with my machines given they don't see that much use compared to the full days/weeks/months/years of work they did when new.

 

[Phipli]

Also, if you get yourself an ATTiny45 chip, you can program it like it is an Arduino, but it just an 8 pin chip instead of a whole board you could add that to your board as well!

Awesome project

I make similar things, but havent implemented the Si5351 in anything yet. This is my latest board, it solders to the underside of the board directly to the pins on the clock (this is when I was bench testing it - pretend there is a IIfx sandwiched between the tin can and the blue board!) the wires are just for the testing, the only wires you need are for the switch to pick the overclock or stock frequency normally :


View attachment 20230222_141644_1.mp4

This was a test with a 66.666MHz clock and the blue board is 50MHz.

 

[petteri]

I went through some junk computers I had and placed heat sinks on MCU and CPU. I also placed a random 12V fan to blow some air over them and memory.



First I reached 57.4MHZ (115MHz on the Si5351) and that seemed to be stable. So I tried 60MHz but that barely booted on the happy mac icon. Stepping down the highest I was able to run tests with was 58MHz (115.9MHz on the Si5351) but I had to try several times to get test results. Also TattleTech 2.59 crashed so I used 2.17 for CPU info. I think I would need to clean up the signal path from my board and the modified aliexpress hook to connect the original oscillator is not optimal either. RAM chips are all 70ns and that might become bottle neck as well.

 

[Phipli]

Its worth keeping an eye that your serial and floppy drive are still working, the CPU isn't normally the first thing to stop.

 

[olePigeon]

Wow. Nearly 60Mhz.

 

[petteri]

Its worth keeping an eye that your serial and floppy drive are still working, the CPU isn't normally the first thing to stop.

Good point! I reverted back to original and tested floppy drive. Everything fine. I don't have anything using serial so cannot test that.

I think I'll let this be for a while. Unless there is really cold winter weather coming up and I could do a overclocking test in freezing conditions :-D

 

[joshc]

Nice work. However...excuse my ignorance but how is this different/better than simply swapping the oscillator for a faster one?

See https://lowendmac.com/2014/overclocking-the-mac-ii-series/ and http://www.applefool.com/clockchipping/iifx.html

I just heard from someone who replaced the 40MHz 68030 and 68882 with the 50MHz versions, and he is now running at 60MHz without any problems.

 

[Phipli]

Nice work. However...excuse my ignorance but how is this different/better than simply swapping the oscillator for a faster one?

See https://lowendmac.com/2014/overclocking-the-mac-ii-series/ and http://www.applefool.com/clockchipping/iifx.html

You don't have to buy loads of incremental clocks to find the limit, and you can't revert to stock speeds at the push of a button?

 

[joshc]

Fair enough. I thought it was somewhat accepted that 50MHz is safe on IIfx, I had always planned to just go with that but it's nice to see 58Mhz is stable too.

 

[Phipli]

Fair enough. I thought it was somewhat accepted that 50MHz is safe on IIfx, I had always planned to just go with that but it's nice to see 58Mhz is stable too.

It varies quite a lot machine to machine. I my dad and I both have the same 601 upgrade, both with 66MHz grade chips. Mine will run at 80MHz all day, my dad's won't even boot at 80MHz.

I usually aim for 20% as a "almost always fine" number, but my Acorn A3010 can take a /100%/ overclock!

I love that machine

 

[petteri]

Fair enough. I thought it was somewhat accepted that 50MHz is safe on IIfx, I had always planned to just go with that but it's nice to see 58Mhz is stable too.

Stable enough to run tests, I wouldn't use that for production use. Another important reason for me not to swap the crystal oscillator was to keep this IIfx as original as possible (collectable, driven only on Sundays...) and I didn't want to risk with soldering oscillator several times.