MegaDrive 1 CXA1645 mod

I checked another datasheet and it had pin 6 as "xtal in" so yep, that's right. I just lifted pin 6 on one of my MDs (model 1 obviously) and... it seems to work! There's definitely an improvement there on the Sonic 2 intro screen. I'll have to give it a proper play to say for sure whether the jailbars are gone completely but I'm pretty optimistic.

Thanks Calpis!

 

What motherboard/case revision are they.

I have a VA4 motherboard. The case has the High Definition Graphics and sound text around the ring as well as the serial connector on the rear. The system I sent to Masterturk was identical except the white piece which houses the LED on his was paint whereas on mine it's a clip out piece.

There seem to be so many different variations.

I have an issue with my RGB Scart cable. It requires some adjusting before it displays the picture correctly however this has it's benefits. In one position it shows jailbars but the image is softened like composite. When in correctly it's pixel perfect with no jailbars and very crisp pixels. The only downside is that if I knock the console I often lose the picture all together.

Anyway sorry to go off-topic. I'm just really into my MD1 at the moment and I'm fascinated with the things people are doing to them. TmEE went nuts with the MD2 but nobody does much with the MD1.

 

Sorry I know the one I use now is a JAP VA5/VA6 (it says that on the motherboard). I can't recall what the others were :/

Alchy> LOL imagine if your simple fix does what I've been spending so long trying... Please let us know how it compares to my Sonic screenshots

 

I've taken it for a spin the last hour or so and as far as I can tell this pretty much fixes it entirely, it's certainly much improved and if there are still jailbars they're at the point where I can't really notice them. I swapped a second unfixed Megadrive back in to double check and the difference is pronounced. I'd do comparison shots but I don't have a suitable camera or the skills to photograph a CRT screen.

Calpis is a genius I'm really happy, this has fixed the final image quality issue I had with the Megadrive.

 

Haha goddammit that's awesome! I'll try it later and will take a few shots myself

What is lost by doing this? Composite and S-video entirely? Or would you still be able to use composite video as sync if needed?

 

Can someone be a little bit more specific on what is being done here? Might give it a go. Pretty pleased with my picture as mentioned but would love to have MOAR power.

 

There's a Sony chip inside the Megadrive that converts the raw RGBS signal from the VDP chip into composite (and potentially s-video). This chip also buffers the RGB signal and accidentally degrades it (blue specifically, it seems). We can get rid of the Sony chip altogether and build a new amp to get the RGB signal up to spec for a TV as Bearking did (kudos on that, btw), or thanks to Calpis' advice we can just remove the offending signal from the Sony chip.

Locate the Sony chip marked "Sony CXA1145". It's under the heatsink attached to the voltage regulators, so you will need to completely disassemble the Megadrive and get access to the underside of the PCB. Once you've got the usual case/RF shield screws out the way, unscrew the voltage regulators (do this before unscrewing the heat sink as it's reasonably heavy and can put some strain on them), then take out the rest of the screws by the cart slot, corner, underside etc.

Now, once you can see the chip, compare to this:



It's pin 6 we're after, note the indentation on the bottom of the chip in that diagram and compare with the chip itself for orientation. I clipped the leg with a pair of small pliers and then lifted it away from to chip with a tiny screwdriver to ensure no contact was being made.

Screw everything back together, job done.

 

Awesome thanks. I'm not unfamiliar with taking the system apart so i'll have a go at this later when I've finished work.

Actually sod it I might do it now. Hard concentrating on work when the builders are destroying the kitchen.

 

Is this definitely caused by internal chip interference? Or could it be caused by an abnormal color subcarrier signal?

Could we fit a new crystal oscillator (and any necessary components) to the lifted chip legs in order to restore Composite and S-Video and see if the problem still persists? Then we should know for sure if its the CXA1145 or the external signals that are to blame.

 

Ok I just fired up Sonic 1 to put an idea of "before" into my head before I carry out the mod.

I noticed that I couldn't see any jailbars. Also noticed that the Digital Noise reduction on my TV was at the highest so I put it to off and reduced colour contrast and brightness all to default.

Dear god! Looks like a prison cell. Bars in every colour. Definitely going to do the mod now and see what happens.

 

I didn't want to be rude but I did wonder if your set was in some way responsible. I'm pretty certain any sufficiently sharp RGB image that's come from the Sony CXA chip in a model 1 Megadrive will show this defect. That's what people who are smarter and more knowledgeable than me have said, anyway

 

No your absolutely right. I should have checked my TV settings before I said anything.

I figured though that any contrast or brightness change would exacerbate such a problem.

One thing I have learned though is that the DNR of my TV is very good.

 

Ok just finished it, my pliers were a little too wide to cut the pin so I used a pair of nose hair scissors (McGuyver eat your heart out!) Then I bent it out straight to ensure it wasn't making contact. I made sure that it wasn't touching the heat sink or anything else when I put it back together.

Absolutely flawless. The first noticeable difference was around the edge of the picture was actually stripy light blue on dark blue. That is now all one solid colour.

Again the blue sky in Sonic 1 was stripy to the extreme and the other colours suffered some jail barring too.

Now...I cannot find a single fault with the picture.

Thanks everyone in this thread who dicked around and put their MD at risk to get to the bottom of this!

Very happy

 

x

 

I've highlighted the chip and pin in the picture (this is not my motherboard by the way...which it was, the region jumpers are in a much nicer place!)

As far as I can see the traces for that pin run under the chip so clipping it is probably the best option unless there is something on the underside of the board you can do, or you can follow the trace back far enough to it's origin but as you can see there's a lot going on in that corner.

On the plus side the legs are pretty hefty and could be fixed by even the worst solderer.

It disables colour through composite and S-video yes.

 

I just cut pin 6 and I still have jailbars :/ No improvement at all. What's in your cables? From each color I have a 220uF > 75ohm resistor.

EDIT: What do you use as sync? I'm using pin 11 > 220uF > 75ohm

EDIT: Pin 10 or composite video doesn't make a difference

EDIT3: I think I've tried everything. I have removed everything in the cable and whatnot. For some reason this doesn't work on mine. What a shame... I'll try that simplified amp suggested on the previous page next

 

Doesn't work :/ All black screen.

 

If your cable has resistors and caps then remove them from the amp's output.

For sync use composite sync coming out of the '1145 which is already routed to the DIN connector, or if using a SCART input use composite video.

I don't think cutting the PCB trace is a more humane way to do it, it's harder to reverse when there's any cutting involved. I'd try to find whether the interference is just from the traces being routed too close or in the chip itself. If the traces just need to be moved apart there isn't too much harm in cutting the composite trace and jumpering it to the DIN but I'm sure it can be done less destructively. If it turns out to be internal to the '1145 I'd desolder it, lift the pin and solder the '1145 back down and maybe put a switch to re-enable the clock. Building an amp like Bearking would allow you to leave everything intact pretty much, and in theory it can be a cleaner signal.

link83, I don't think there's anything you can do externally to fix the carrier wave on the '1145, it's a digital input unlike the '1645 which accepts a true sine wave. There has to be a filter inside the chip to shape it into a sine just as there is probably a filter to generate the quadrature wave instead of using a PLL. I don't think it's the problem since the encoder works pretty well, too well for modulated square waves.

One final and important thing to consider is that the interference might not be coming from the composite output but maybe from pin 5, the oscillator-inverter output. If the output must have enough current to drive the crystal, and if it doesn't have any load it could act as an antenna or the switching noise alone could affect the B input which is then amplified. To test this a few K resistor to GND should work.

 

I only tried the simple amp without anything in the cable

So I should cut pin 5 as well?

 

Hmmm I don't have an RGB display to test anything myself, just simulation. I think this means it's necessary to use a PNP version like this which sinks current: or use the unideal capacitive coupling and bias resistors as you already are...

That's the circuit used in the AV Famicom (better R values would be 330 and 100) but it's probably not worth bothering with..

That's how it should be.

No you shouldn't cut it, if you disconnect pin 6 pin 5 is no longer an issue. Even cutting it wouldn't necessarily stop the noise since I don't think there is a trace in parallel with blue, as an antenna it'd need a path to ground or if it's the switching noise you can't do much here but put more bypass capacitors across the VCC and GND rails of the chip.