CDX S-Video Help

[Segasonicfan]

Okay so far the only known S-Video fix for the CDX (due to the Fijitsu chroma encoder and not the better sony one) is the mod posted here:
http://www.neogeousa.com/mod-index.html
The Neo Geo uses the same chip so you take the same steps to achieve a proper S-Video signal.  There's only one small problem...future-active.com (the site it suggests) no longer sells the ECG-85 chip required to do the mod.   I cannot find this chip on any other website to purchase either.  Does anyone know how I can find this chip or is there a similar circuit I can use to get the same effect?  I know D-Lite successfully did the mod and I'm hoping he'll let me know.  Many thanks to anyone who can help me out with this.

[Segasonicfan]

ahhh, alas i found the solution!:

http://lukesandel.com/videogames/SNK/NeoGe...eo/default.html

I've been looking for that for a long time.  The NTE-85 is a breeze to find making this mod entirely possible now.  CDX S-Video awaits!

-Segasonicfan  

[Segasonicfan]

Success!  After many hours of tinkering away I have finally managed to get a beautiful S-Video out of my CDX.  Unfortunately, the digrams and info on the neogeousa and other sites aren't entirely helpful with the mod.  Maybe they worked for the Sony CXA1145, but they sure as hell aren't the same as the MB3514....even though they are identical chips.  For example, I had to remove the 100uf and .1 pf caps because the took the image away.  Also, the recommended 33ohms for chroma (or luma, whichever line they said) is simply wrong.  Both lines should have 75ohms as the datasheet recommends.  So yeah!  Those are the differences you need to know in modding the CDX for S-Video.

Now if only I didn't feel like I'm talking to myself...

-Segasonicfan

[Drewman21]

Good job on the work you did.  I have been following what you were doing and checking the sources you had mentioned.  My question is(not being great at the technical apect) on your little board that you put together in stead of soldering right to the chips.   Could you post a diagram on how you put them together and how you put in the pointometer for the one?  I like the tight design and I want use it on some of my machines.  Please post it up!
Thanks
Drewman21

[Segasonicfan]

nice to know somebody is reading my posts =)  Unfortunately I think you have me confused with someone else that did another mod because I didn't use a board or a potentiometer.  I did solder right to the chips (i had to cause there's no space to work with inside the CDX).  Using a potentiometer is a good idea but you would need one with very low resistance, from 0 to 100 ohms I'd say.  I probably won't eb doing that though because I get a perfect picture with 75ohms.  I just can't believe the change in quality of the picture.

[viletim!]

Segasonicfan,
The MB3514 is a 100% compatible (as far as I'm aware) drop-in replacement for the CXA1654. See the datasheet on how to connect luminance and chrominance. It requires just a coupling cap + resistor for each signal.



While I'm at it here's a critique of that neogeousa.com mod thing:
First NTE/ECG transistors are (expensive!) repacement transistors... not a great choice, the bc558 or 2n3904 are better, easier to obtain and cheaper. Putting a little tiny cap in series with a huge electrolytic is really silly when coupling signals. Instead, better to use a 10� electro cap for luma and a 0.1� ceramic for chroma (chroma cap can be tiny because the frequency is very high). I couldn't decypher the transistor bit (where's the circuit diagram?). It looks like an emitter follower config though, pretty standard... Asside from that it should work ok on anything containing a CXA1145.

[Segasonicfan]

I think you mean drop in replacement for CXA1145 not the 1645...  well you are correct about it only needing the caps and resistors to output S-Video.  In fact, it doesn't even need that.  However, if you want the best S-Video possible from the chip, you have to use an additional circuit.  Check the bottom of the MB3514 schematic (You can get it here: http://katalog.elektroda.net/indx400-BA.html)  And I'm not sure why you think the NTE-85 is expensive.  www.jameco.com sells them for 85 cents.  I actually live 2 miles from them so I picked up 4 for less than a dollar =)  And I know the CXA1145 and MB3514 are supposed to be the same, that's why I didn't understand the caps not working.  I spent hours working on it, even putting in diff caps, resistors etc.  But I didn't make any mistakes (obviously, cause it works without them) and it would be ard to with such a simple circuit.  It just doesn't work with caps+the transistor amp.  I hardly care anymore though, since I'm getting really great S-video! =)

[viletim!]

Thanks for the datasheet. You're right, it looks like an enhansed version of the 1145. See The data for the ES71145 URL=http://members.optusnet.com.au/eviltim/gamescart/es71145.pdf]here[/URL]  (it's compatible with the CXA version, the CXA data is only available in japanese). The main difference seems to be that the Fujitsu chip's Y/C outputs "Can be directly driven at 75 ohms". The CXA version definately can't do that so requires the transistor amplifiers. The bottom right of the test circuit shows how the Fujitsu's Y/C output should be connected.

As for the NTE-85, damn expensive! I just compared the Jameco prices - at a quantity of 10 the NTE-85 costs 79 cents while the 2N3904 costs just 6.9 cents.

[D-Lite]

Thanks for the datasheet. You're right, it looks like an enhansed version of the 1145. See The data for the ES71145 URL=http://members.optusnet.com.au/eviltim/gamescart/es71145.pdf]here[/URL]  (it's compatible with the CXA version, the CXA data is only available in japanese). The main difference seems to be that the Fujitsu chip's Y/C outputs "Can be directly driven at 75 ohms". The CXA version definately can't do that so requires the transistor amplifiers. The bottom right of the test circuit shows how the Fujitsu's Y/C output should be connected.

As for the NTE-85, damn expensive! I just compared the Jameco prices - at a quantity of 10 the NTE-85 costs 79 cents while the 2N3904 costs just 6.9 cents.

But you'll find that the MB3514 does not look correct without the transistor modification.  Same with the CXA1145.  If you look at the results with and without the full on transistor installation you will see exactly what I mean.

[D-Lite]

And I'm not sure why you think the NTE-85 is expensive.  www.jameco.com sells them for 85 cents.  I actually live 2 miles from them so I picked up 4 for less than a dollar =)

You live in/near Belmont??  

[D-Lite]

Yes, I live in Foster City.  Oh, and the transistor mod gave me a much better picture than just the passive component mod, I recommend u try it.

-Segasonicfan

Holy crap we're neighbors!  I live in Belmont, about 3 miles up Ralston from Jameco!  

Oh, and I know the transistor mod works awesome.  That's what I put in the CDX I did too.  And in Genesis systems, in the Neo Geo, etc....

Small small world.  We'll have to meet up sometime!  

[Segasonicfan]

how crazy.  yea we hsould meet up...I've never actually had the chance to play an AES system before  And I could show you my portable CDX mod.

[viletim!]

But you'll find that the MB3514 does not look correct without the transistor modification. Same with the CXA1145. If you look at the results with and without the full on transistor installation you will see exactly what I mean.

I've got a MB3514 in my Megadrive. As I previously mistook it for a CXA1645, I connected  pins 15 + 16 to my monitor through series resistor (75) and capacitor (220�). The video output was at the correct level and it worked flawlessly. The sony CXA1145 certainly does require a 75ohm driver circuit.

[Segasonicfan]

That's strange Viltim.  I foudnt he result of just using passives was more like described here:
http://lukesandel.com/videogames/SNK/NeoGe...eo/default.html
"Incidentally, if you try to simply tap S-Video from the CXA1145 without buffering it, you'll get a deceptively decent signal (assuming you filter it first, and stick a 75ohm luminance/33ohm chrominance resistor on it). The only problem is, during high-luminance transients the screen will become unstable. It looks good, but isn't optimal. Also, the colors may bleed excessively."

I found only after sing the amp mod (and my small changes to it) was the picture truly perfect S-Video.

-Segasonicfan

[viletim!]

Segasonicfan,
It's not strange at all...The part (MB3514) can drive a 75 ohm load (TV/monitor) from it's Y/C outputs without an external amplifier as advertised in the data sheet (with the aid of a pair of 470� caps and 150 resistors). Are you claiming that the data sheet is wrong?

Perhaps the neo-geo's video circuit is loading the output in your case???

[Segasonicfan]

I'm not saying the datasheet is wrong.  Read page 12:

"Note: Although there is no filter between Y-out and C-out, and between Y-in and C-in, in the test circuit, in actual use, the band pass filter (B)
and delay line (D) must be provided as shown in the above figure."

-Segasonicfan

[viletim!]

ok...I've played around inside my Mega Drive and found:

The data sheet IS wrong! Connecting the Y/C signals as shown in the test circuit will not work (well to be fair it is a test circuit, not an application circuit). It should be connected up CXA1645 style. That is, with a 75 ohm resistor and a 220� capacitor in series with each output. This resaults in video suitable for any TV/Monitor (Y=1Vpp, C=0.5Vpp). If that doesn't work for you then I can only conclude that either the neo-geo isn't using the chip correctly (ie feeding it with too high/low amplitude RGB) or there are multiple 'versions' of the chip.

BTW, the band pass filter and delay line are only only required for ther generation of composite video.

[Segasonicfan]

no, the band past filter and delay line are for Y/C output.  The circuit outputs composite video without them.  They are used (or substituted with a transistor amp) to output the proper Y/C signals.  The datasheet is correct.

-Segasonicfan

[viletim!]

no, the band past filter and delay line are for Y/C output. The circuit outputs composite video without them. They are used (or substituted with a transistor amp) to output the proper Y/C signals.

Bullshit!

If the band pass filter and delay line are not required in the process of generating composite video then why are these circuits (however crude) present in two video game systems that were designed to output only ever output RGB and composite video???

The band pass filter is required to limit the chrominance bandwith (to about 1Mhz, I think...) so the TV can decode it properly. The delay line is there to compensate for the delay introduced by the band pass filter.

The MB3514 WILL directly output Y/C. It does NOT require a BPF/delay line (such a thing will only limit the bandwith of the Y/C output, making it look not much better than composite) or any transistors to do so. The Y/C outputs should be connected as described in the CXA1645 application circuit.

The MB3514 datasheet isn't incorrect, but it is misleading and provides no application circuit.