Overclocking a PS2

[grokr989]

Lechkovitz,
I checked and in your model X102 drives IC106(The clock generator for both the both the EE(through IC101) and the GS)  so simply replacing X102 with a higher frequency part won't work. It works in my model because X101 drives only IC101(The clock geerator IC for the  EE).  The GS uses a separate crystal(X202).

[panzeroceania]

Some games rely on the refresh rate for timing - if you double the refresh rate on those games it will double the speed.

And yes, the XRGB2 does suck at line doubling interlaced video, that's why you buy an adaptive deinterlacer that only deinterlaces the area of the image that suffer from "toothing". 

Hmm. that sound interesting. Can you recommend one ?

I also would like to know good alternatives, even if they are expensive.

[RobIvy64]

:golf clap:

Good to hear you accomplished it! I never tried this with a fat PS2. Sadly my soldering iron has long been retired for such awesome tomfoolery. 

[jdtheretro]

Yea, if you isolated the clock signal, and since the cpu bus multiplier is 2x, if you want it at 343, then have the clock coming from a 171.5 crystal osc.

[RobIvy64]

Buuuuuuuuuuuuuuuuuuump!

Reading grokr989 information, I realized that I had the exact same version of PS2 sitting in my closet!

I cracked it open and got to work. The stock crystal oscillator is 18 MHz.

First, I tried 27 MHz (27 x 16 = 432 MHz..... LOL). Nothing. Green light, followed by red.

Next, 24 MHz (384 MHz). Powers on, but nothing. Blank screen.

Unfortunately, I don't currently have an oscillator between 18 and 24 MHz, so I decided to UNDERCLOCK in the meantime!

12 MHz (192 MHz CPU).

See results below:
https://www.youtube.com/watch?v=JBhYElftOSg

[RobIvy64]

The audio pitch changes because I suspect the sound CPU (PS1 CPU) shares the same clock speed (x2).

If I can isolate that clock input, set the sound CPU input back to 18 MHz, this will resolve the sound pitch changes.

[RobIvy64]

I cut the clock to the R3000A, and I still had sound but no CD/DVD drive. Apparently, the PS1 CPU controls the DVD-ROM.

I'll have to probe around the other two ICs and find the clock input.

[maxdub432]

UPDATE:
Now that summer's over I finally had some more time to throw at this project. I grabbed a fresh PS2 (30001R, mainboard's more like a 39k than a 31k) and tested out my theory about MSCLK needing to be 36.864 rather than 36. System still wouldn't boot with my 36.864MHz oscillator signal connected to the IOP. Switching from the EE signal to my signal after bootup caused an unrecoverable freeze.

Probed the signal from the IOP -> SPU2, also 36.864. Substituting my signal here also failed to boot. Switching from the IOP signal to my signal after bootup caused sound to glitch or drop out. Trying to load any programs led to a soft hang until I switched back to the stock signal. Switching back resulted in normal sound resuming maybe 20% of the time. Unless someone else bumps this topic with new ideas I'm giving up on fixing the audio desync.

I was able to reproduce my overclock on this PS2, so I know firsthand that 30001, 30001R, and 39001 can all be overclocked; I would assume that this applies to all fat models.

Overclocking results in:
noticeable framerate improvement in Odin Sphere (only game I've tested)
audio/video desync, irritating in cutscenes
loss of network & internal HDD capability
usb still works, so you can still load games with OPL

I originally assumed that networking still worked because my oc'd ps2 grabbed an IP from my router. Actually trying to load games with OPL over SMB/ETH from a known working server, however, always times out

Because of the audio desync, I think the best application of the overclock is to install a switch between the stock and overclocked crystals and only use it for games that you know have a lot of lag. This comes with two additional caveats: one, put the switch as close as you can to X101/X102, because if your wires are too long bootup is unreliable; two: you can't switch the OC on in the middle of a game like you can on the PS1, here it causes a hard freeze. I'm out, hope this helps at least one person out there looking for the info

I also used grokr's success as a jumping-off point to overclock a PS2 last fall, but rather than sourcing a 30001 I successfully OC'd a 39001. I upgraded the unified CPU/GPU clock crystal (X102), then interrupted the signal from the unified clock generator (IC106) to the GPU at resistor R159 and substituted my own non-OC'd signal from a 54MHz oscillator.

I successfully booted with a ~24MHz X102 but thought it was too fast and eventually settled on ~21MHz. I was bothered not only by the audio desync but also by the inability to use an internal hard drive. I tried to correct both problems at once by interrupting the signal from the CPU to the IOP/IC303 (which is, IIRC, the PS1 R3000A) and substituting my own.

I found that the PS2 wouldn't boot at all with the CPU -> IOP signal missing, presumably because the IOP is needed to pull the BOOTROM into memory, but would continue to run at the OSD if the signal was cut after bootup. The PS2 failed to boot at all when I substituted my own 36MHz signal to the IOP even when using the stock ~18MHz X102.

My multimeter indicated the stock CPU -> IOP signal is not actually 36MHz, but 36.864MHz, exactly twice the value of X102, which is something I could tackle if I could source the proper oscillator, but for now I haven't had the time