XRGB-mini Framemeister

[Fudoh]

well, one mode is just a click from the other - so everbody should just try. Not exactly something on which you have to rely on recommendations.

[TheShadowRunner]

I'm taking Movie over Natural for 480i as it seems Natural has some kind of greener tint Movie doesn't have. Case in point: in the PS2 Browser..

[Joelepain]

For me Movie just looks like all the other mode in terms of color/saturation.
Paradoxically, Natural is the only mode which is not natural. Natural should have been called "Vivid" or something like that. (maybe it's a translation mistake from japanese ?!)
And so the mini is lacking a true Natural mode, which should have been similar to Movie mode (in terms of color, deinterlacing and LPF) but without the dithering/grain effect.

Anime is for me too saturated to be useful in any situation.

[Fudoh]

And so the mini is lacking a true Natural mode, which should have been similar to Movie mode (in terms of color, deinterlacing and LPF) but without the dithering/grain effect.

I don't get the presets anyway. Since the Marvell has all the options available seperately, we should just get access to set them individually as well. First thing we requested, never happened.

[MojoBox]

Paper Mario TTYD is the only 480i game I've noticed any real comb artifacts in. Simply using the hammer attack will reliably produce them every time. I suspect this is due to the high frame rate and how brief the animations are, maybe only a few frames.

I was able to freeze frame it: http://i.imgur.com/GyUVphn.jpg

Not a big deal at all, but the only mode that doesn't present it is Game1 and Game2.

[Marcel_Ger]

Hey guys, I´m the new guy in here!

First of all I wanna say...great forum and from what I saw by now...cool community

I´m kinda new to the Framemeister (and the whole rgb and modding stuff) . I used it with my Neo Geo AES (RGB, Unibios Mod) and it worked quit good without any big changes (Image_mode: picture, Sync mode: auto, sync level: around 9-10). No flicker or anything like that.

I use an european SCART to Framemeister adapter (passive) to hock up my modded systems
Now I got a PC Engine Duo and did the "Otaku´s-store.net" RGB Mod with the internal AMP (I also replaced the composite sync signal by pure sync and installed the cap parallel to the C909) and the RGB SCART EURO cable (I live in germany btw.) but now I have to turn the sync mode off and crank the sync level up to 18. Otherwise I get a total mishmash of a picture (with HuCards) or an extremely high rate of flicker and picture lost (with CD-ROM Games) and even then the picture is not that stable but playable.

When I plug the Duo directly at my Sony LED TV it works fine but it looks kinda bad (obviously)

So now to my question(s)

1) There´s a second cap that comes with the otakus kit...for all I understand, its a replacement cap (100uF) for the C 961 and I read in an other forum that this improves the quality of the picture even more but could it stop the flicker and stuff?

2) I read about some sync strippers like these or these. Could one of these cables solve my sync problems with the Duo?

Sry for the long text and maybe question 1 is more like a "jailbars on the pc engine" kinda question, so sry if off topic

Regards
Marcel

[Joelepain]

Paper Mario TTYD is the only 480i game I've noticed any real comb artifacts in. Simply using the hammer attack will reliably produce them every time. I suspect this is due to the high frame rate and how brief the animations are, maybe only a few frames.

I was able to freeze frame it: http://i.imgur.com/GyUVphn.jpg

Not a big deal at all, but the only mode that doesn't present it is Game1 and Game2.

I don't know if we're talking about the same king of artefacts, but running F-zero GX at 480i is a good example. You have to look at the white flashing light at the back of the of the car engine.
In movie mode you have some kind of combing artefact, whereas in game 1 and 2 it looks correct (as if the game is running in 480p, but with the shakyness of bob deinterlacing).
As Fudoh already said, you can get rid of these artifacts by setting v_scaler to 8.

Another example could be Metroid prime in 480i, if you hold the charge beam and look around, you can see some artifact on the bulb of the charge.

It looks like the problem is present for real 60hz interlaced content, so when every single field come from a different frame. But I suppose it's the hardest kind of content to deinterlace, and it must be really hard to do better.

[blizzz]

Even Yadif has problems with F-Zero GX in 480i, you need QTGMC to deinterlace it at 60fps without artifacts and that algorithm is too slow for real time application. The better solution would be to play in 480p without the Framemeister. The game looks simply gorgeous.

Marcel, sync strippers work on composite video and create composite sync (which is the same as pure sync btw). I doubt that it helps in your case, since you're already using c-sync.

[Marcel_Ger]

Hmm I thought so...I´m gonna open it up and check everything again...maybe I missed something and install the C 961.

[FBX]

I use an european SCART to Framemeister adapter (passive) to hock up my modded systems
Now I got a PC Engine Duo and did the "Otaku´s-store.net" RGB Mod with the internal AMP (I also replaced the composite sync signal by pure sync and installed the cap parallel to the C909) and the RGB SCART EURO cable (I live in germany btw.) but now I have to turn the sync mode off and crank the sync level up to 18. Otherwise I get a total mishmash of a picture (with HuCards) or an extremely high rate of flicker and picture lost (with CD-ROM Games) and even then the picture is not that stable but playable.

With regard to the SCART cable for your PC Engine Duo, did you make sure it's actually wired for c-sync? I think with these cables you have to specify how you want it wired. I know I did for my JP-21 SNES cable.

[Marcel_Ger]

I know I have to open it up to check that but how should it look? Does anybody have a circuit diagram for that? (sry I´m really lost in the rgb, sync, jungle)

[TheShadowRunner]

I know I have to open it up to check that but how should it look? Does anybody have a circuit diagram for that? (sry I´m really lost in the rgb, sync, jungle)

You can most likely compare with this:
http://www.gamesx.com/rgbadd/duorgb.php


Another great source:
http://www.mmmonkey.co.uk/category/nec/

[FBX]

After exhaustive SNES picture analysis, I found an optimal setting that gives "B+" results on the RGB input flicker tests, and that's as follows:

Brightness: 23
Gamma: 9
Black: 1
A/D: 131

Mega Man X looks pretty good, Zelda looks superb, and it also passes the shrub test in Super Mario World (i.e. no crushed green shades).

BTW, I was doing some screen size, scaling, and output res tests, and I found that the absolute sharpest possible SNES picture with no artifacts for 1080p displays is in fact my 4x custom scale setting and the TV screen size set to "Full Pixel". I was able to take a magnifying glass and actually see that the vertical resolution on each pixel was EXACTLY 4 rows of LCD elements. Here again is the quick setting breakdown:

HDMI Output: 1080_60p
Auto_Scale: Off
H_Scaler: 5
V_Scaler: 7
Zoom: On
Zoom_Size: 97
Zoom_Overscan: 86
TV display screen size: Full Pixel

Similarly, you can perfectly scale the SNES image to 5x by using the same settings above, only change the Zoom_Size to 90 and the Zoom_Overscan to 100. The only drawback here is you lose 4 rows of pixels on the top as well as on the bottom. This confirms the math of 224x5 = 1120, which cuts off 4 pixels above and below on 1080 res. This would be a good setting for games that are letterboxed like Final Fight, or for games that don't have structural graphics in those areas of the screen (Space Megaforce for example).

[micro]

Hey guys! Most of the time I'm just lurking but today I want to show you something.

Owning a Framemeister since 3 months I've experienced problems with consoles that are not running on their original refresh rates. For example PAL Megadrive on 60 Hz or jap. Saturn on 50 Hz. (I don't even want to start talking about the PSX )
I also observed that using these configurations somehow mess up the black level settings of the Framemeister. Running the PAL MD2 on 60 Hz makes black appear as grey. It's really annoying...

So naturally I wanted to replace the PAL crystal inside the MD2 with an NTSC crystal:


Unfortunately, 53.693175 MHz oscillators are hard to get. Eventually I've found them available on aliexpress and ordered 2 pieces. I've bought them 5 weeks ago and I'm still waiting for them to arrive...

So imeanwhile I've come up with another solution:


It's a dual frequency oscillator designed to be a drop in replacement for the original metal can oscillators. The main part is a CDCE925 clock synthesizer which can generate almost any frequency with its PLL's (phase locked loop). The oscillator got a CLK1/!CLK2 pad to select between two different frequencies. When this pad is connected to your 50/60 Hz switch the required clock signal will be selected automatically:


So the question is: Does it solve the problems with the Framemeister? Yes, it does!


No more stuttering and messed up black levels. Head over to http://nfggames.com/forum2/index.php?topic=5744.0 if you want to find out more about it.

[BuckoA51]

Good solution! Could this be adapted to other machines? I'd really love something like this for the Amiga.

[micro]

So I guess running a PAL Amiga on 60 Hz will introduce the same problems?

But yes, this oscillator should also work inside the Amiga. You just need to:

1) find out the frequencies of the original PAL and NTSC metal can oscillators.
2) for automatical clock switching you need to have some sort of PAL/NTSC signal that toggles between high and low when switching from 50 to 60 Hz and vice versa.

It should also work in other consoles (Saturn for example) and I will try to install it into the PSX. (I need to fix my PAL PSX consoles, got a lot of us and jap. games I can't play on the Framemeister atm)

[BuckoA51]

Yeah, the Amiga is way out in 60hz and of course it's quite rare to find an NTSC Amiga since the machine was way more popular here in PAL land.

This is an exciting development, I hope you will consider selling the boards to the wider community.

[RGB32E]

It's a dual frequency oscillator designed to be a drop in replacement for the original metal can oscillators. The main part is a CDCE925 clock synthesizer which can generate almost any frequency with its PLL's (phase locked loop). The oscillator got a CLK1/!CLK2 pad to select between two different frequencies. When this pad is connected to your 50/60 Hz switch the required clock signal will be selected automatically:

Pretty cool!

Could you describe the audio and RGB circuits I see in your MD2?

[micro]

Pretty cool!

Could you describe the audio and RGB circuits I see in your MD2?

Of course.

The big audio circuit is the CCAM I've built according to this diagram: http://www.sega-16.com/forum/showthread ... post511814

The RGB circuit is just a THS7314 bypassing the video encoder with input voltage dividers to bring the RGB signals down to the proper level.

Yeah, the Amiga is way out in 60hz and of course it's quite rare to find an NTSC Amiga since the machine was way more popular here in PAL land.

This is an exciting development, I hope you will consider selling the boards to the wider community.

I don't intent on selling these oscillators but I will release all the informations and software. In fact you could order your PCB's right now and rebuild the oscillators but without a proper programmer & programming software the oscillator won't work as intended.

But if you're interested then I could sell you one of my three oscillators. But we'd need to figure out the AMIGA frequencies first. Which Amiga model are we talking about, A1200?

[Marcel_Ger]

I know I have to open it up to check that but how should it look? Does anybody have a circuit diagram for that? (sry I´m really lost in the rgb, sync, jungle)

You can most likely compare with this:
http://www.gamesx.com/rgbadd/duorgb.php


Another great source:
http://www.mmmonkey.co.uk/category/nec/

Thanks man, that helped me checking my cables (all sync btw)
I opened up my duo and found a cold soldering joint at the c 909 and replaced the c 941 with a 100uF cap and....everything turned out quite good. A little shaking at bright animations but all in all quit perfec. I'm 98 percent satisfied

Thanks for your help quys

[TheShadowRunner]

Thanks for your help quys

You're welcome

I also observed that using these configurations somehow mess up the black level settings of the Framemeister. Running the PAL MD2 on 60 Hz makes black appear as grey. It's really annoying...

Super nice job micro, this is very curious though, do you have any hint as to why different frequencies would fuck around with the black level on the Mini?
Also do you plan to make some more SNES pad wireless PCBs?

[Fudoh]

do you have any hint as to why different frequencies would fuck around with the black level on the Mini?

different IRE offsets in different regions. The Marvell might have a compensation for that and for some reason it might be kicking in when some non-standard NTSC signals are detected.

[Ed Oscuro]

different IRE offsets in different regions.

Those are voltage differences, not frequency related, right?

[Unseen]

It's a dual frequency oscillator designed to be a drop in replacement for the original metal can oscillators.

Cool! (one more thing I can strike from the list of things I wanted to look into ;) )

If your circuit can be adapted for a 3.3V system, it could probably be used for the PSX too: The console uses a PLL to generate three internal clock frequencies from a single crystal and the GPU clock happens to be the same 53.693MHz / 53.203MHz for NTSC/PAL. The CPU clock (67.7376MHz) appears to be the same for PAL and NTSC systems and the third clock is just the color carrier which RGB users don't care about anyway. "Live" switching based on an internal NTSC/PAL signal of the console will probably need a glitch-free switching circuit though.

[Edit: Proofread carefully to see if you any word out.]

[micro]

Thanks!

Yes, I've already had a look at the Playstation's clock generation circuit. As you said it also uses PLL's to generate the 3 different clocks (CPU clock, GPU clock and PAL/NTSC subcarrier).

Today I've installed my dual frequency oscillator into the Playstation. I did a stupid mistake (bridging two pins) which I didn't realize in the beginnning. Thanks to that mistake I've wasted several hours. Anyways, in the end the oscillator works as intented.

That means:

* perfect black levels (50 + 60 Hz)
* no more stuttering or other image problems (50 + 60 Hz)
* bonus: Konami's anti-PAL protection found in the Bemani games is now useless.

I'm really happy right now!
Detailed post with PSX installation pics is coming to the gameSX thread soon.

Also do you plan to make some more SNES pad wireless PCBs?

Yes, sometime this year.

[BuckoA51]

But if you're interested then I could sell you one of my three oscillators. But we'd need to figure out the AMIGA frequencies first. Which Amiga model are we talking about, A1200?

I'll drop you a PM and we can work something out. I'm interested in this for several consoles.

[micro]

Damn, I'm sorry but just an hour ago I've installed the last of my three oscillators.

Well, today I fiddled with my Neogeo MVS. It's a consolized MV-1C. Although the MVS works with XRGB-Mini fairly well (no stuttering), the black levels still get messed up. And of course it's recognized as 241p and you have to adjust V_WIDTH from 32 to 33.

The MV-1C got a 24.00 MHz crystal and the resulting refresh rate is 59.185606 Hz. So it's an easy calculation: To get a refresh rate of 59.94 Hz the MVS needs a 24.30590978 MHz clock signal. I've programmed the dual frequency oscillator to produce these two clocks (24 MHz and 24.30... MHz). Here's the result:



The video signal is now detected as 240p, so no more adjusting V_WIDTH. Also black is black and not gray anymore. The MVS now runs 1.27% faster. I only played for about 15 minutes but I couldn't notice any issues. (I still need to play more to make sure there are no drawbacks.)

[FBX]

So I got my RGB cable in for the model 1 Genesis today, and did some testing with the mini. Picture quality was superb, but the real test was Castlevania: Bloodlines, since I already knew from emulator use that this game switches from 320x224 to 256x224 back and forth at certain moments like the map screen for example. The mini reacted to this res change by blanking the screen for about 1 second during the switch-over. Just an FYI if this of course hasn't already been discussed.

[austin532]

It has no problem switching from 240 to another 240 signal. It's when it switches from 240 to 480 that it throws a hissy fit.

[FBX]

It has no problem switching from 240 to another 240 signal. It's when it switches from 240 to 480 that it throws a hissy fit.

No, it does in fact seem to have a problem with the H-res changing at least on my end. However, I had also noticed jail bars on blue shades of my Genny's RGB picture, so I did some Googling and learned this was encoder interference in the CXA chip. So I broke the trace line and this cleared up the jail bars completely. Upon doing some more testing after this fix, I found the mini will now retain the picture during the res change from 320 to 256, but the sound will still drop for about a half-second to a second.

Edit: Booted the game up again and it's back to dropping the picture for a second as well as the sound. It appears the mini just is very finicky about those res-changes in Bloodlines. It would be nice if someone else could test this independently. Make sure it's an NTSC original model Genesis, using RGB boosted CSYNC. Could it possibly be the boosted SYNC causing the issue?