MegaDrive 1 CXA1645 mod

[Bearking]

So that really only leaves you with PS1 games played on a PS2 with component? You should consider an XRGB then, there is also the cheaper alternative the HDBOX Pro:
http://www.hdboxpro.com/eng/shop.htm

I don't have any personal experience with that one, and the results will definitely not be as nice as the XRGB-3's, but it's a hell of a lot cheaper.

[Alchy]

I've got a HD Box Pro (a variant on the name actually but same chipset/hardware inside). It's junk, avoid - it's only slightly better than the scaling hardware in your set is likely to be. If you've only got one set and it only has VGA/DVI then I guess it'd do for PS2 but it's really not a good solution, even with endless tweaking I couldn't get a result that I thought was acceptable.

[link83]

4) composite -> luminance is a side-effect of disabling the oscillator and also not the point, the point was just to stop the oscillator which is interfering with the signal one of the following ways:

-the power rail (the inverter switching current dropping the input bias voltage or something)
-via EMI from the inverter output -> blue input
-or through parasitic capacitive coupling of the composite output -> blue output

So couldn't someone test this 'scientifically' by going through the variables until we narrow down the actual culprit? Just seems a bit silly to cut Pin 6 and completely disable Composite and S-Video if its unnecessary, especially if the problem could be solved another way? :shrug:

[Lum]

So that really only leaves you with PS1 games played on a PS2 with component? You should consider an XRGB then, there is also the cheaper alternative the HDBOX Pro:
http://www.hdboxpro.com/eng/shop.htm

I don't have any personal experience with that one, and the results will definitely not be as nice as the XRGB-3's, but it's a hell of a lot cheaper.

Pretty close. That I'm aware, Mega Man X Collection was the sole commercially released Gamecube title in history to output 240p (unless parts of Mega Man Anniversary Collection did?). And no Xbox. Unless I've been horrible mistaken.

[Bearking]

MegaMan Anniversary runs in 480i on the GameCube :)

[Calpis]

So couldn't someone test this 'scientifically' by going through the variables until we narrow down the actual culprit? Just seems a bit silly to cut Pin 6 and completely disable Composite and S-Video if its unnecessary, especially if the problem could be solved another way? :shrug:

Absolutely, it just seems nobody wants to do further testing. If I were doing this myself I'd test for all possibilities before making any irreversible changes.

[Bearking]

What are we supposed to do? I tried removing the amp, restoring to the original RGB lines and cutting pin 6. I also tried building your simpler amp and it didn't work. If there is a better/simpler solution I'm all for it.

[Druid II]

Alright, I finally finished soldering the RGB cable for my MD1. (I had to order a 262° din8 connector from Poland of all places, and the plastic that holds the pins together started melting at the last few connectors, almost disfiguring the whole thing.)

Someone explain this wizardry.

With the RGB cable, I get jailbars on the otherwise astonishingly perfect picture (way better picture quality than what I managed to get on my Saturn, in fact). I then disconnected C28 on the motherboard, thereby removing the oscillator signal right before it goes into pin 6 of the cxa1145. The jailbars are gone, but now the picture crawled to the left so much that a little part is cut off, and a black bar is on the right.

This is with using composite signal on the scart composite pin. If I use the c-sync pin, I get picture on bootup that immediately goes out of sync and fades away, before the "Produced by or licensed..." screen could finish. My multimeter says that the c-sync pin outputs 4.47v, which I suppose is the cause. Are there any physics professors here who could tell me what sort of resistor would I need to reduce this to whatever is the nominal voltage of pure composite sync?

What are we supposed to do? I tried removing the amp, restoring to the original RGB lines and cutting pin 6. I also tried building your simpler amp and it didn't work. If there is a better/simpler solution I'm all for it.

Install a simple on/off switch connected to C28 on the bottom of the motherboard, and call it "Composite / RGB". If you have a model without the EXT port, you even have a perfect place for it on the back side.

[Calpis]

What are we supposed to do? I tried removing the amp, restoring to the original RGB lines and cutting pin 6. I also tried building your simpler amp and it didn't work. If there is a better/simpler solution I'm all for it.

I can't say, I don't have the setup for this. If you're satisfied with what you have now I'd leave it at that. I know for sure that a suitable RGB amp can be built with a transistor, two resistors and no input capacitors, but what I came up with didn't work.

The jailbars are gone, but now the picture crawled to the left so much that a little part is cut off, and a black bar is on the right.

Strange... It's something with the TV, sync is sync is sync so the picture should be in the right place. Is it a HDTV? If so each resolution will have different parameters to center the image and maybe since composite is now irregular (it's luminance) the TV is confused.

My multimeter says that the c-sync pin outputs 4.47v, which I suppose is the cause.

Did you measure the pin with a 75 ohm load?

[Druid II]

Panasonic p42g10e.

According to my multimeter, the voltage is the same with a 75o resistor (I just touched the ground pin on the scart output without disassembling the cable, and had the resistor pushed into the scart socket, then touched its other side with the other end of the multimeter).

5v dc line was also measured as 5v with the resistor.

Not sure whats going on there. The multimeter can measure resistance as well, and it says the resistor is 75ohm.

[link83]

Would this schematic help at all?
http://emudocs.org/Genesis/mega2.png
I dont know if its official or reverse engineered, but it does show the entire video circuit for the CXA1145.

[Calpis]

That couldn't be properly loaded. From the schematic I guess sync has an open-collector output: (5V / (2200 + 75)) * 75 = 0.165V but that's too low, it should be 0.287V. Like usual there's probably more going on inside the chip...

[TmEE]

Csync is an open-drain output and needs a pullup resistor. Have a 1K or less instead of 2.2K and lot of problems should get solved.

[Bearking]

Install a simple on/off switch connected to C28 on the bottom of the motherboard, and call it "Composite / RGB". If you have a model without the EXT port, you even have a perfect place for it on the back side.

Yes but cutting pin 6 didn't help anything on my MegaDrive. Removing C28 is the same thing so that wouldn't make a difference :)

[Druid II]

Csync is an open-drain output and needs a pullup resistor. Have a 1K or less instead of 2.2K and lot of problems should get solved.

So, disconnect R16 on the back of the pcb and install a 1k resistor there?

edit: wait a sec, doesn't the sync line need a 220uf cap too? That may be the problem. But, I have no more space in this scart to install it.


edit2: I just noticed, c-sync is not coming from the cxa1145 to begin with, but from the vdp. I'll try replacing that signal with the c-sync out from the cxa, with a 220u cap and 75o resistor.

[Calpis]

Csync is an open-drain output and needs a pullup resistor. Have a 1K or less instead of 2.2K and lot of problems should get solved.

Are you sure? Seems kinda hard to think they could make that big of a mistake; 40 IRE for sync is widely universal.

So, disconnect R16 on the back of the pcb and install a 1k resistor there?

(4.5Vpp / (1100R + 75R)) * 75R = 0.287Vpp

4.5Vpp because the output can't swing all the way to GND, really it's 4.3 to 4.8 depending on load and process.

So I'd use a 1.1K.

edit: wait a sec, doesn't the sync line need a 220uf cap too?

edit2: I just noticed, c-sync is not coming from the cxa1145 to begin with, but from the vdp. I'll try replacing that signal with the c-sync out from the cxa, with a 220u cap and 75o resistor.

Under some circumstances sync doesn't need to be capacitively coupled, but here it should be because the output can't swing down to 0V w/o it. The MD must be assuming the signal is AC coupled at the display input. It also normally doesn't need impedance matching (Zout = 75R) since it's very low frequency; even the 9th harmonic is <200 kHz and that makes a very adequate pulse.

It's a good idea to use the CXA's sync out though because it's probably buffered to the right level and impedance matching which will help with finicky equipment. When you use the proper 75 Rout then you also need the 220uF Cout since they make a suitable RC time-constant.

[Druid II]

CXA1145 voltage measures to 2.68v.

Using it with a 220uf cap and 75o resistor, I get rock solid picture... shifted to the left. I'm gonna try disconnecting pin6 again, see what happens then.

edit: still shifted to the left. What the fuck.

[Calpis]

It's not surprising the CXA output will have a high level like that, the level should be 0.287 * 2, and it probably is with a 2V offset which is removed by the cap. You can't really trust what a multimeter tells you, it's a relatively fast AC signal and the meter will just give you the RMS voltage, not the true Vpp without respect to duty cycle. For that you need a scope.

[link83]

The recommended component layout for XO IN and XO OUT shown in the CXA1145 datasheet does not seem to match the layout Sega used, could this be the cause of the problem? The datasheet shows a 15K resistor to ground on XO OUT but Sega left it unconnected :confused:

[Druid II]

Okay, tested with a regionswitch (by using composite or c-sync, I refer to what I connected to the scart comp out pin):
- in PAL mode with c-sync, screen is shifted to left, and is jumping around occasionally
- in PAL mode with composite, screen is stable and almost centered (only shifted to the left by like 4 pixels)
- in NTSC mode with c-sync, screen is shifted to left, but is stable
- in NTSC mode with composite, screen is shifted to left, but occasionally jumps
Additionally, in NTSC mode, the red color is bleeding in a dither pattern, while in PAL it does not.

This is all with pin 6 lifted.

But it looks as if this whole ordeal is a "warmup" problem. After testing for 10-20 minutes (using composite), I noticed that the jumping occured less and less, there was no color bleeding anymore, and the screen was near perfectly centered almost all the time. Occasionally, in NTSC mode, the screen still started up shifted to left AND had bleeding reds, but I could always fix both by switching to PAL and then back to NTSC while still on the "produced by" screen. Additionally, sometimes I noticed that the game started shifted to left, then immediately jumped to the center while still on the "produced by" screen.

So yeah, the problem goes away by itself after generally using the console.

Fucking ghosts in the machine...

edit: yet again in another test, with Ranger X, in NTSC mode it jumped left and right nonstop, but after leaving it running past the intro and another demo, the screen stabilized. What the heck is going on here?

edit2: by the way, composite still works, and it has jailbars (pin6 still removed). Not sure if it had them before this, I'm not disassembling this thing again to check. Colors etc. are all there. RF however, is fucked.