PlayStation PAL Series Colour/Color Modification (PAL60)

[TriMesh]

The PIC in the original mod is used as driver ( I guess. I can't imagine what else it should be good for. )
In any case, you obviously need to have a 4.43Mhz solution for PAL.

Yeah, that was fairly common back in the day - the PIC was set up with the oscillator in XT mode and the code for the delays adjusted so that the timing was correct with the 4.43MHz clock. Went out of fashion when Sony started putting anti mod code in, since it was non-stealth.

[rama3]

It's clever use of resources, I guess. So you get color on CXA1645 using boards, but did anyone ever check the quality of it :p

[TriMesh]

It's clever use of resources, I guess. So you get color on CXA1645 using boards, but did anyone ever check the quality of it :p

Really no worse than any of those other "stick some unsynchronized subcarrier signal into the video encoder" mods. Very noticeable dot crawl in NTSC mode, more or less noticeable in PAL depending on the exact frequency of the xtal. Of course, there is still plenty of room to mess things up (long wires and no decoupling caps, for example), and quite a few of the installs I saw did exactly that.

[rama3]

It is some tricky electronics for sure, very unusual stuff for people doing digital stuff like modchips.
Say, you don't happen to have some tips to improve an exisiting design, maybe?
For example, is it better to use a sine wave than square? Or is triangle any good?
Is there a way to shape the signal? I already know an inductor of the right size can be used to make a sine from a square.
My aim is to reduce dot crawl in NTSC on both, the CXA1645 designs, and the newer encoders.

[TriMesh]

It is some tricky electronics for sure, very unusual stuff for people doing digital stuff like modchips.
Say, you don't happen to have some tips to improve an exisiting design, maybe?
For example, is it better to use a sine wave than square? Or is triangle any good?
Is there a way to shape the signal? I already know an inductor of the right size can be used to make a sine from a square.
My aim is to reduce dot crawl in NTSC on both, the CXA1645 designs, and the newer encoders.

Sine is probably the ideal signal to use, if you can generate it. Both square and triangle waves are similar in that they only contain odd harmonics, although the higher harmonics on the triangle wave drop off more rapidly (square law rather than linear). I'm not sure the harmonics are a real problem, though - the 3rd harmonic will be around 12MHz in PAL and 10.7MHz in NTSC, and they are both well outside the video passband.

If you lock the subcarrier to the GPU clock (say using the PAL clock and a fixed /12 in both video modes) then that should improve the dot crawl in PAL mode, since it will restore the fixed phase relationship between the chroma signal and the rest of the picture, but I'm not sure it will help much in NTSC mode because the phase relationship will still be constantly changing. It will probably be more consistent, but not really any better.

[nocash]

The clock shape (sine/square/other) isn't the issue. The first problem is how to synchronize the color clock to the GPU clock (that will give you a perfectly stable picture with "perfectly stable" color artifacts). And the second problem is how to de-synchronize it again (by adding some small timing offsets in each Nth frame, to "hide" the artifacts; by moving them to different locations in each frame).

If you have a PAL board, and want it make it using PAL colors in both PAL and NTSC mode, then it might be simple: Just take the onboard 53.20MHz signal and divide it by 12 to get the 4.43Mhz clock. I haven't tried that, but it should be properly synced to the GPU, and hopefully desynced enough for not giving you artifacts.
The later models like PSone are actually using a very similar method with a similar external clock divider.

If you have an imported NTSC board and want to make it using PAL colors then you won't get around using an extra oscillator. I don't know of an "analog" method for forcing that oscillator to run in sync with the GPU, but you could use an oscillator that is, say, 8 times faster than 4.43MHz, and then divide it with a 3bit timer to get 4.43MHz.
For the syncing you could simply reset the divider at VSYNC. And for desyncing you could use a 2bit timer incremented at VSYNC, and then tweak the circuit to output new clock pulses when the 2bit timer gets equal to the MSBs of the 3bit timer.
That, in theory. In practice, many people might prefer a less complicated method with slightly dirty composite output, or a RGB cable without composite at all.

[rama3]

Those are some very nice explanations and tips, thanks!
Regarding signal shape: I found that designs using a sinus exhibit no "jialbars" when hooked up to an upscaler.
This is only true for well made designs (ie: made so in the factory, not tacked on) though.
Example of this: GH-006 PS2s look great through an upscaler.
For other designs, it really depends on your luck. PU-18 PSXs often shows jailbars and PU-8 not so much, for example.

nocash:
I know that I'm just looking for a synchronized artifact look. Your reset on VSYNC suggestion is great. It would probably work well!
Offsetting artifacts a la Nintendo looks good on regular 50/60Hz television. I have a nice 100Hz Trinitron though, which doubles the field rate probably in AABB fashion. The "Nintendo offset" doesn't help the picture there, it just ruins the otherwise nice SNES output. But I'm fine with it there, it's the look I expect of it
For PSX, I only use it in NTSC mode and with correct NTSC timings (53.69Mhz GPU oscillator). The picture can be really good. I'm just looking to make it even better

Where an NTSC PSX output still has a slight artifact problem is when tiny areas of different color are used. Some color combinations produce a shimmering dot crawl then. 90% of games don't do this though and then the picture is great.
Example of an offender (and even this is mostly fine on PU-8!):
Side note: This picture is also an example on where I really prefer Composite. Composite makes this look very nice

[Shadow Moon]

Here's a picture that explains how to do this mod on an SCPH-102 console:

[K4DoS]

You can do that on the underside of the board too.

[x7502x]

Ok, I have most retail versions of the ps1, 1002 - 102.

On the newer versions in my case (not sure ive tested 7002 for BW or the 102 for color) 7502+ they play NTSC games in full color with no modification, I know my tv supports NTSC but older version are still black and white, I remember seeing an explanation for this but it slips my mind at the minute.

So my question is why is this modification needed on newer systems? for tv's that don't support NTSC? as I get full color on my 7502.

[TriMesh]

It's dependent on your TV, and which video formats it supports.

All the PAL consoles starting from the PU-20 (SCPH-7002) will always output a signal with a 4.43MHz PAL subcarrier irrespective of the video mode. Without this mod, the effective video mode when running a 60Hz game will be NTSC4.43 and with the mod it will be PAL-60.

If you get an image with correct colors without this mod then your TV supports NTSC4.43 and you don't need to do the mod. If you get a stable image that has messed up colors then you have a TV that only supports PAL60 and you should do this mod.

[K4DoS]

I've been doing some research lately now that I have a PU-8 machine on my hands (chipped before I got it and came with a US supply despite being a PAL machine, beats me why did that happen) and found this on Youtube.

Anyone can explain if this would work? Might be a way to do the mod without a special programmed PIC or 4 pin XTAL. I'd assume it works since there's the transistor to drive the circuit, but I might be wrong. Also, can I use a BC547 (or any NPN transistor for that matter) to drive it?

[TriMesh]

I've been doing some research lately now that I have a PU-8 machine on my hands (chipped before I got it and came with a US supply despite being a PAL machine, beats me why did that happen) and found this on Youtube.

Anyone can explain if this would work? Might be a way to do the mod without a special programmed PIC or 4 pin XTAL. I'd assume it works since there's the transistor to drive the circuit, but I might be wrong. Also, can I use a BC547 (or any NPN transistor for that matter) to drive it?

It looks reasonable - that seems to be a Colpitts oscillator circuit using an NPN transistor running in emitter follower mode. A BC547 would probably work, although I would probably go for the BC547C version to get more gain margin. The plain BC547 only has a gain bandwidth product of about 150MHz, so at 4.43MHz it would only have a gain of about 33 which might cause startup problems. The C version would have a gain of > 100 at the same frequency and should oscillate very easily.

Having said that, if you already have the transistor to hand then try it.

[K4DoS]

Tried it and no dice. It will boot in B/W in both PAL and NTSC mode. I checked the wiring and it's done correctly.

Maybe it's because I have a early PU-8? I know the S/N on the board ends in -467-11, and at least I couldn't find the 0 ohm links for the main 53MHz XTAL.

[TriMesh]

Tried it and no dice. It will boot in B/W in both PAL and NTSC mode. I checked the wiring and it's done correctly.

Maybe it's because I have a early PU-8? I know the S/N on the board ends in -467-11, and at least I couldn't find the 0 ohm links for the main 53MHz XTAL.

I don't think so - that part of the circuit is pretty much the same all the way to the SCPH-55xx models. It's possible that you just need a much higher gain transistor - that circuit seems to be using parasitic capacitance for feedback. It looks like the pinout should be right for a BC547 (CBE), with the collector connected to 5V and the load on the emitter.

To be honest, that looks like a pretty iffy circuit - for reference, here is the textbook version (on the left) and that one (on the right):



There are several components missing compared to the canonical version - C3 is an output coupling cap, and you can probably get away with that being missing. R2 (in conjunction with R1) sets the bias point of the transistor, so with it missing the output voltage is going to be close to the positive rail. Finally, C1 is missing - and this is the cap that provides the feedback between the emitter and the base without which the circuit won't oscillate at all. I assume that the original circuit was relying on parasitic capacitance to do this.

The other thing is that in an oscillator like this the port the xtal is connected to has to effectively exhibit negative resistance - and this has to be larger than the ESR of the xtal otherwise the circuit simply won't oscillate.

[wanderer]

Can someone clarify what PAL60 is ? I get that that it is 60 hz. What I don't get is whether it is PAL color and PAL resolution or just PAL color with the lesser NTSC resolution

[rama3]

525 lines, 60Hz, PAL color encoding at 4.43Mhz

[Jack Solidus]

hello how could I modify the ntsc 7501 console, to show colors in games pal. Will they have a diagram?

[rama3]

As far as I know, you can't easily modify an NTSC console for PAL output.
You need the 4.43Mhz carrier clock that NTSC consoles cannot produce.

[likeabaus]

You can force it with software. If you download and burn ImportPlayer (any version), you can boot that using you modchip or boot disc/swap trick and then it comes up to a menu, that lets you force the video mode. Then you can swap in your game and it "should" display in PAL. The only downfall is you'll have to do this everytime....

https://consolecopyworld.com/psx/psx_utils_bootcd.shtml