Megadrive/genesis model 1 dual overclock + electronics questions

[Linux-SWAT]

Hi,

as pre-pre-pre-orderers are so cocky these days, I modded again my EU Megadrive (a version this a text before the usual SEGA logo) today. There was already JP/US and 50/60Hz switches.

I bought these two oscillators:

Mouser No:	520-2200BX-100
Mfr. No:	ECS-2200BX-100
Desc.:	Standard Clock Oscillators Standard Clock Oscillators ECS-2200X,OSC XO 10.000MHZ TTL PC PIN

and

Mouser No:	520-TCH1200-X
Mfr. No:	ECS-2100AX-12.0MHZ
Desc.:	Standard Clock Oscillators Standard Clock Oscillators DIP-8 5V 12MHz

Used an on/off switch to halt the 68000 CPU.

And an on/on/on switch to select clocks 7.6MHz/10MHz/12MHz.
Here's how it works:


My experiments so far, with only a few tries with a few games:
- game cartridges and flashed Everdrive can boot directly from 10MHz
- flashing the Everdrive at 10MHz works but the flashed data is corrupted, e.g. games have color problems.
- 10MHz really enhance some games e.g. I saw no Thunderforce IV 1st stage slowdowns.
- pausing the CPU seems not required is advisable when switching between clocks.
- I wasn't able to play at 12MHz, colors got weird instantly and the game crash, even with games reported as working@12MHz.

My questions about 12MHz:
- should I try to shorten the length of the clock cable ?
- should I try to use a grounded/isolated cable ?
- should I try a better/more precise oscillator ?
- should I ground the switches ?

Some pics (of course !).
I got +5V and GND at the region jumpers (random cable colors, I have to resupply):

Oscillators, switches, and the Almighty 68000 on steroids:

 

[levi]

I'm not sure what you mean by using an isolated and grounded cable. Do you mean coax? I've never seen a clock crystal connected using coax, although usually they're mounted on PCBs so the use of any cable is limited. Shortening the cable could well help, since a clock signal is effectively an AC waveform. I'd have thought that a crystal is a crystal, either it works or it's busted, but maybe I've just been lucky this far. Without an oscope debugging these kind of things is difficult. As to grounding the switch, do you mean the casing? I see no electrial benefit in doing that, especially in a double insulated plastic case.

 

[Linux-SWAT]

Like an audio cable.
Yup, the casing.

@hns answer:

- should I try to shorten the length of the clock cable ?

Yes - the longer they are the worse the signal quality and the more noise they can collect

- should I try to use a grounded/isolated cable ?

maybe but can have other negative effects if cable impedance is not right

- should I try a better/more precise oscillator ?

those are already good.

- should I ground the switches ?

Could help a little but may not.

Generally you could use digital switches (i.e. AND + OR combinations or some Multiplexer chip) so that you only switch the MUX control.

And switching those processors while they are operating may leave them without clock for a while.
And as far as I remember the 68000 it was quite at the limits of the technology back then. So overclocking 12 MHz may be beyond limits.

Just my 2 ct and have a happy new year!

 

[Linux-SWAT]

Not pausing the CPU when switching frequencies often lead to crashes, corrected 1st post.

 

[levi]

Yes, I guess you're running off three different clocks, switching between the randomly will at least 50% of the time lead to a single duty cycle far less than what it's meant to be, which is implementation defined on a CPU but generally bad news.

 

[Linux-SWAT]

Yeah, I was skeptical when I read here that the pause switch was optional:

how to overclock your sega genesis model 1 and NOTHING ELSE

Good news, I learned how to overclock the model 1 genesis! :D Honestly I got tired of seeing people not just talking about overclocking the Sega Genesis model 1 but seeing them including things and mixing the mod with other mods using these little boards in there guides that just makes the...

www.sega-16.com

 

[Linux-SWAT]

Super Hang-On greatly benefits from the 10MHz overclock.
I can't tell if the game speed increased, but the controls are much more responsive.
Normally the Megadrive version feels like maneuvering an elephant, now it's closer to the arcade.

 

[Linux-SWAT]

I shortened the oscs<->switch and switch<->CPU cables, and soldered the clock switch casing to the gnd.
It helped as it takes more time to crash ! But I didn't found a 12MHz playable game. Most of the time, the colors go weird and at best the game resets.
I could try to solder the clock pin of the oscillator directly on the switch, but I'm good with 10MHz. Everything work well so far, and the games which are enhanced have no more slowdowns.
Also could try a 11MHz one day...
I just remember that Herzog Zwei can be a good candidate to push the machine as we can build tons of units in that RTS, tremendously slowing down the game.

 

[spud42]

as levi said without a CRO it would be hard to figure out what is wrong. however wire length and capacitance could most definitely affect the square wave leading and trailing edges. rounding them off so edge triggering could be dodgy. 7.6 to 10 MHz is quite the overclock so if it works great at 10MHz thats a 30% OC

 

[Linux-SWAT]

Yup, it clearly makes a gap.
Herzog Zwei crashed @10MHz at around 40 units.
After that, the switch ground solder didn't hold so I saw that it changed basically nothing.