[TUTORIAL] Overclock for Gradius III original game board

Joined
Jan 8, 2019
Messages
985
Reaction score
1,013
Location
Spain
Hi. Just wondered if the game could be improved in speed-up for avoiding all the huge slow-downs.

Maybe trying to change both Motorolas 68000 for ones with more MHz speed and also the crystal oscillators...

Anyone could highlight some info about?
 
Last edited:
I've played around with this on other boards, ultimately it depends on how the clocks are managed the board
some boards have a single clock source and everything from the video output to the main and sound CPUs are derived from that clock, other times they're independent and the video output or audio timer triggers an IRQ in the main CPU, in which case you could probably up the clock speed without causing the graphics or sound to speed up.

sometimes they're independent but just use a single clock on the PCB that gets divided and split around to reduce part count.

I don't own a Gradius III but I'd recommend you look at the board identify the oscillators on board and what they're attached to, and look at the machine state in the mame driver as well.
 
I've played around with this on other boards, ultimately it depends on how the clocks are managed the board
some boards have a single clock source and everything from the video output to the main and sound CPUs are derived from that clock, other times they're independent and the video output or audio timer triggers an IRQ in the main CPU, in which case you could probably up the clock speed without causing the graphics or sound to speed up.

sometimes they're independent but just use a single clock on the PCB that gets divided and split around to reduce part count.

I don't own a Gradius III but I'd recommend you look at the board identify the oscillators on board and what they're attached to, and look at the machine state in the mame driver as well..
I think both motorolas are controlled by one 10mhz osc. Maybe upgrading to 12mhz the OSC and motorolas should make a significant improvement.
 
You want to make it harder!?
I want to make it playable, xD.

@twistedsymphony More info

PXL_20260223_160814786.jpg

PXL_20260223_160822365.jpg

PXL_20260223_162224670.jpg


68000 CPU and 68000 collisions are interchanged from the 10MHz osc with a LS244 input-output play.

68000 collisions has DTAC (PIN 10) tied to +5V through a tiny resistor.
 
so that 3.579545 is likely for Audio CPU. and the 10MHz is likely for the main CPU. mame driver implies this uses an IRQ to trigger the main CPU.

what clock is running the video output/screen refresh?

looking at the MAME driver there should be a 24MHz Xtal for the screen timer:
https://github.com/mamedev/mame/blob/master/src/mame/konami/gradius3.cpp

so I'd just try swapping that 10MHz out for a 12MHz and see what happens.
 
so that 3.579545 is likely for Audio CPU. and the 10MHz is likely for the main CPU. mame driver implies this uses an IRQ to trigger the main CPU.

what clock is running the video output/screen refresh?

looking at the MAME driver there should be a 24MHz Xtal for the screen timer:
https://github.com/mamedev/mame/blob/master/src/mame/konami/gradius3.cpp

so I'd just try swapping that 10MHz out for a 12MHz and see what happens.
But changing 10mhz to 12mhz would OC both Motorolas, right? Isn't that dangerous?

Yes, there is also a 24mhz osc
 
So people have swapped in a 12MHz osc on a 10MHz 68K on CPS1 to make it a "dash" board. I honestly don't know how stressful or not stressful that is for the CPU, probably perfectly fine for a test, maybe perfectly fine for long term home use (maybe not? genuinely don't know), might still want to upgrade the CPUs if you're worried about it long term.

but IMO it's not likely to cause any damage to test just swapping the osc and a whole lot easier than swapping out the whole CPU only to find out it doesn't work.
 
PXL_20260223_220955755.jpg


Changed the OSC, and I don't know if it's placebo, but I note the gameplay even better. At least on second level, where the bubbles are, the slow-downs are faster while occuring and dissapear in less time.

I have to test it further / record videos for anyone to note differences.
 
yeah I think if you record a video with both speeds and do a side-by-side comparison you may pickup on differences that you didn't notice, or confirm that it's just a placebo :D
 
so looking up the "SCN68000CAN64" it seems the "A" in the suffix denotes that this is a 10MHz 68K
the "CBN64" and "CBN64" variant is a 12.5MHz model.
https://www.cpu-world.com/CPUs/68000/Signetics-SCN68000CAN64.html
Motorola also makes some 16MHz and 16.67MHz 68Ks if you really want to crank up the juice :D
I have 16MHz oscillator from Taito F3, and If I need those 16MHz Motorolas, I have plenty spares.

I'll make a test today trying a 16MHz oscillator and seeing if REALLY the gameplay changes.
 
will be interesting to see how a 10MHz 68K responds to such a big increase :D
 
Well, tests done. I have recorded them, and yes, there is a huge difference in CPU working only increasing from 10mhz to 12mhz (changing J11 oscillator).

My 2 gameplays have the same ship distribution A and the same order in power-up sequence. 4 important scenarios where the improvement can be seen:

- Just after getting laser on first level. When firing laser and missiles, in 10mhz a lot of slow-downs can be seen. Hardly a slow-downs on 12mhz
- Spider boss. A lot of slow-downs on 10mhz when firing laser and missiles constantly when it fires its sand bomb. On 12mhz much less slow-downs
- Just after entering second level and after getting 3rd option. A lot of slow-downs on 10mhz. Much less slow-downs on 12mhz
- In the second level, on 10mhz video at 4:40 aprox., there is a massive slow-down for all the objects in screen. On 12mhz video at 4:20 aprox, at the same point, the slow-downs are not that massive.

 
Last edited:
Back
Top