MegaDrive 1 CXA1645 mod

[Bearking]

If I could've made digital captures of the signal, they probably would've looked the same as the ones Bearking posted above.

Awsum! Alles klapt in the end...

This is probably as good as it gets. I'm very happy with it :)

[Emperor Udan]

The same for me, my picture is near perfect. So near that for me it don't need to be better, and i'm a perfectionnist.

[Druid II]

Yeah, at this point I'm more interested in what exactly causes the interference, than how to fix it.

[l_oliveira]

Yeah, at this point I'm more interested in what exactly causes the interference, than how to fix it.

I think I mentioned it already ...

Two things:
VRAM is too close to the encoder

And I was not sure about the chroma carrier but it seems like the chroma carrier clock on the MD is too strong. On other machines using the CXA1145 the image is perfect or near perfect.

On my own personal designs using the CXA1145 (SUPERGUN), I have dot crawl problem but that's caused by thermal fluctuations on the quartz oscillator crystal.

If I change the circuit for an square wave generator (just like the square wave signal the MD sends to the CXA chip) I'll get rid of the dot crawl but I'll get JAGGIES and crappy RGB output on the buffer ....

Nice, huh ? :shrug:

[Druid II]

And I was not sure about the chroma carrier but it seems like the chroma carrier clock on the MD is too strong. On other machines using the CXA1145 the image is perfect or near perfect.

So how would you go about reducing that? Add a resistor somewhere?

[l_oliveira]

So how would you go about reducing that? Add a resistor somewhere?

Maybe tinkering with the values of the existing parts on the circuitry already on the MD... Maybe reducing the value of the capacitor which sends the Chroma clock to the encoder IC is a good start ...

[TmEE]

You need a RC filter to cut down the corners of the square wave. There actually should be one in MD, but the capacitor is on the weak side.
VRAM too close to video is one thing to consider, but then again, MDs in same board version have it and don't have it so there's more to it...

[SGGG2]

Without pin 50 connected:


It's not PERFECT as you can see. There is still a hint of jailbars to see. But add scanlines to that and it's pretty great.

Hey, this worked for me, too! Thanks link83!!! :clap::clap::clap:

There's still a hint of jailbars, but they're invisible when I set noise reduction on the Optoma HD3000 at maximum.

(Scaler chain is XRGB-3 > Gefen VGA to DVI Scaler plus > Optoma)

[l_oliveira]

It's impossible to completely get rid of the jailbars because all Yamaha VDPs are based on the Texas Instruments TMS9918 and that chip do suffer of the jailbars effect too ... :lol:

The MSX2/2+ (V9938/V9958) is also plagged with the jailbars stuff. Even the V9990 chip (does not have anything from the TMS9918 in it) has some issues with these jailbars.

Oh well. Glad to know you guys could figure out means of making it acceptable. :thumbsup:

[Druid II]

Checking the schematics: Pin 50 on the VDP is the same line that C28 sits on, so removing that resistor is the same thing as disconnecting pin 50 on the VDP. I think it's safer to do that then trying to pull the chips leg off...

[SGGG2]

Yeah, that's far better. I slightly warped pin 49, and I was very careful. I've been looking for a way to get rid of the jailbars for a year now, and nothing was going to stop me.

[Lum]

It's impossible to completely get rid of the jailbars because all Yamaha VDPs are based on the Texas Instruments TMS9918 and that chip do suffer of the jailbars effect too ... :lol:

The MSX2/2+ (V9938/V9958) is also plagged with the jailbars stuff. Even the V9990 chip (does not have anything from the TMS9918 in it) has some issues with these jailbars.

Oh well. Glad to know you guys could figure out means of making it acceptable. :thumbsup:

Then that striking revelation is true after all? I thought I was just coming off as a naysayer when I'd suggested the possibility... Such a MD graphics architecture flaw gone largely unexplained so long after release, is big news! Too bad any involved chip designers have likely moved on. I doubt we'll ever get confirmation from the source.:-(

What you're saying would suggest all standard MD based devices have identical RGB artifacts under equal conditions. (proper cables, no outside interference like TV settings, etc). No matter its model and revision. Whether or not 32X, Sega CD, SMS mode, or other upgrades are involved. Given those still must use the console's VDP. I don't recall any significant VDP improvements being done in MD's life. Perhaps it may have messed with game compatibility too much?

Which leads me to another question I never even considered to ask. Do the Pico, MegaPlay, or MegaTech also experience these bars? They're just different enough it might be fun to look into.

[bart_simpson]

Ok with my offical rgb cable for my model one it reduces the lines aka jail bars.

I can post a nice picture of the board of my rgb cable and what chips are one there.

[l_oliveira]

This awful image is output from a Yamaha YM2302C chip (V9938) composite output pin.

Board layout is pretty poor. The computer in question is an NMS-8020 that was converted (by myself) into an MSX2. I used leftover parts and the VDP was salvaged from an 80 columns expansion cartridge which was meant to be used on MSX1 computers (what a waste of a great VDP chip lol :lol:).

There's a LOT of room for improvement but I would need to use RGB instead.

Edit: If you look at the picture closely you will notice that along with the usual jailbars from the VDP DRAM refresh cycle you see fuzzy noise. That's noise coming from the Z80 bus. ^_^



(oh and also there's a bit of fuzziness from JPEG compression too) ;)

[sdekaar]

Without pin 50 connected no more jailbars on my plasma.
thanks for the tips guys :)


(japanese megadrive (haa-2510 ic bd m5 VA1 837-6832)
rgb cable
original ac adaptator 9v/1.2a +step-down converter 220/110
rgb cable
tv 60pk250)

[keropi]

can the same workaround be applied on the SMS2? I added an rgb-out on my console and I get those lines, IIRC the sms2 uses the same sony chip... has anyone looked at this by any chance?

[bart_simpson]

Try a crt tv and there are no jailbars on a crt tv with scart rgb.

[l_oliveira]

Try a crt tv and there are no jailbars on a crt tv with scart rgb.

Actually the original complaint from Bearking (the OP) was how the blue had jailbars on RGB. Turns out we discovered that the CHROMA carrier frequency was being irradiated to the bue input pin of the encoder and was being aplified by the RGB buffer. That was the whole problem here.

Pin 4 of the encoder chip is the blue input ... Guess what, pins 5 and 6 are the CHROMA carrier input pins. :)

[Lum]

Hardware engineers missed that back in the 80s and we're only now discovering the problem... Geez.

[Druid II]

What you're saying would suggest all standard MD based devices have identical RGB artifacts under equal conditions. (proper cables, no outside interference like TV settings, etc). No matter its model and revision. Whether or not 32X, Sega CD, SMS mode, or other upgrades are involved. Given those still must use the console's VDP. I don't recall any significant VDP improvements being done in MD's life. Perhaps it may have messed with game compatibility too much?

Which leads me to another question I never even considered to ask. Do the Pico, MegaPlay, or MegaTech also experience these bars? They're just different enough it might be fun to look into.

There were some 7-8 different chips that had the Megadrive VDP on it through the years, it's possible that the problem was fixed eventually - or just fixed by circumstances of being manufactured by more modern ways. My Megadrive had one of the early type standalone VDP chips and it had jailbars. Later systems were more and more integrated and presumably built on a smaller node, so they may have been fixed. Does the MD2 have this problem?

I wish we had documents, notes, diaries of the design process of early Sega hardware. Reading that kind of stuff is very, very interesting, there are many of them from the old Commodore staff (the C128 one is insane).