Genesis model 1 s-video problem

I recently modded my genesis for s-video, but I didn't have a 27ohm resistor, so i used a 43ohm and it looked great. After I got the 27ohm resistor I swapped it and rewired my system. This is what my image looks like now.

http://imgur.com/Hm6X8jT

Any idea what could be causing that color distortion? Is it the 27ohm resistor? Because the 43ohm worked well before, long-term would that cause a system failure?

The composite a/v works well.

Thanks in advance,
- John

 

I don't think the encoder's Y/C outputs are actually meant to drive a TV so it's unlikely the mod you're following is competently thought out. The value of 27 ohms itself makes no sense on the engineering side, so it would have been chosen by the author because it "worked" for their them and they naively would assume it should work for others. The conventional and correct way to install a S-video mod would be to amplify Y and C with a video operational amplifier, then connect it via a 75 ohm resistor which serves to match the cable's impedance and terminate reflections at the source. The video amplifier with a fixed gain of 2 is necessary in this configuration to deliver the proper signal level (voltage amplitude) to the TV because the series resistance attenuates the signal seen by the TV.

43 ohms would work better than 27 ohms because the encoder's pin probably isn't capable of supplying the current to drive a low impedance load (a TV). Typically video encoders provide Y/C outputs so the signals may be externally filtered and looped back into the encoder (which has a high input impedance) to be further joined into composite video. So the higher impedance of 43 ohms + 75 at the TV is easier for it than 27 + 75. Neither are correct however and both would put a strain on the 1145 because the outputs are forced to supply many times the expected current from the pins.

 

Wow, thanks for the in-depth response. I was using this tutorial:

http://retrotimegames.com/page13.html

Would this not work?

 

I used this same tutorial to mod my model 1 Genesis. While I am happy with the signal I'm interested in knowing if this tutorial is dated and if there may be better methods to use. This was my first video mod on a console so seeing anything better than composite may have been enough for me.

 

The circuit may "work" for some (eyes are very bad judges of signal integrity), but the signal delivered to the TV will always be abhorrently out of spec irrespective of the resistor chosen. TVs are robust and often try to compensate for such bad signals, but in many cases it's impossible to. Even if the TV recovers from amplitude issues, it can never recover from distortion from the mod or added noise from the recovery process.

Such a mod is just bad form. Notice how even the cheapest DVD player doesn't have compatibility issues with any TV? That's because DVD players and TVs follow signal standards and the vast majority of video mods do not--they're typically authored by people who are probably oblivious to their existence and only have a basic grasp of the circuits they publish. The mod you linked is a perfect example. I don't want to put people down, but it simply doesn't make any engineering sense.

Just about every video mod requires an operational amplifier at the least to implement correctly. Today video op amps are a standard IC component (an example being the popular THS7314) --single transistor amplifiers as seen in many amateur mods haven't cut it since TVs started accepting baseband video inputs in the 70s, and even then those amplifiers were still competently designed to interface with subsequent circuitry.

 

Calpis what would you recommend as the proper way to do S-Video on the Genesis Model 1 and while we're at it the Model 2? You say the net is filled with improper circuits so let's try to get the proper ones out there.

 

A proper circuit is basically the application note from any video amp datasheet. There are plenty of examples of RGB mods for the THS7314, it's the exact same theory for S-video.

 

So something like this:



should there be some caps in series with the 75ohm like the typical 220uf?

 

I dont think you are supposed to amplify the chroma. All the ones I've seen only have the luma going through the transistor.

 

Almost. The CXA's Y/C outputs will probably have a significant DC offset which must be removed. Y can use the THS' video clamp mode (so just a coupling capacitor), but C cannot as it doesn't have a sync component, so you must AC couple the input (with capacitor and pull-up resistor).

It's up to you. If there is a significant dc offset burning current it would be advised, if not it's better to leave them off for lower distortion.

Both signals need amplification because the series-parallel termination divides the amplitude in half, so unless the signal out of the DAC is already 2x the desired level (and can source the necessary current), an amplifier is required to deliver the correct amplitude to the TV. The single transistor "amplifier" configuration seen most video mods doesn't actually even amplify the signal (voltage), it simply buffers the signal so it may drive a heavy load (such as a TV's terminated input)--the actual amplitude delivered to the TV is anywhere from 90% to <50% of the desired signal level, and will almost certainly be distorted from improper biasing. To get 100% amplitude with low distortion and perfect impedance matching, you need an op amp on every signal driving a coax cable.

 

I believe this should be correct now based off the Datasheet and what you have told me.

 

Yes, exactly. I'm not sure about the pull-up though, I think 5.1 or 5.6 M is recommended for a 5V VDD. Since chroma has a smaller amplitude than luma you can actually use an even larger value to further lower the dc offset, but you must be careful that the signal doesn't clip. I would adjust it until chroma's color burst is offset at least 100 mV or match it to luma's offset to be safe.

 

You think 5.1 to 5.6M as a solid value for most TV's? or will that be something that will have to be tweaked per TV set?

 

It has nothing to do with the TV, the resistor biases the signal's dc offset at the amp's input and thus the dc offset at the output. The higher the dc offset on the output the more current the amp must waste driving the signal to the TV since the extra dc isn't part of the signal. A little dc offset is good because it isn't possible to perfectly bias the signal due to component tolerances and signal tolerances. (The clamp function you're using on luma attempts to minimize the dc offset, but it only works for signals with embedded sync.) If the output signal swings too close to or past the negative rail the signal will clip (distortion) since transistors aren't perfect or magical.

I would start at 5.1 M and measure the offset. If the offset is significant (like 300 mV+), I'd increase the resistor to 5.6, 6.8, 7.5 M etc to get the offset down to or just higher than whatever the offset is on luma because we know the amp can handle that offset with little distortion and component tolerances must be accounted for.

This is probably the best solution available now since it's low-distortion, power-friendly and low-cost, and it conforms to industry standards (assuming the levels from the encoder aren't far out of spec).

 

Thanks for the info, hopefully some people try the circuit out and this will be the defacto circuit to use. While we are on video discussion I've seen some IC that uses an amp similar to the THS that feeds RGB to another amp that spits out Component Video, what are your thoughts doing this with the THS or similar components? Here are the components I'm referring to but one of them is hard to come by in most distributors and there likely must be alternatives.


http://www.linear.com/product/LT1395

http://www.linear.com/product/LT6559

Combo of ICs to produce Component Video from RGB.
http://www.linear.com/solutions/1312

 

Those ICs are various op amps (THS are fixed gain op amps with input clamp/level shifters and high performance filters too). You can use a fast op amp to directly transcode YPbPr, but that particular circuit is more of a clever hack to do it with absolutely minimal parts and it's only meant to be PART of a transcoding system. You still have to do your own sync insertion and ensure that the input signals are biased correctly, plus it expects you to have a negative supply.

To make a good transcoder for all purposes you probably want it to operate from a single supply, and you want to be able to give it ac coupled RGB signals so you need a clamp circuit triggered by sync, and you need a robust and fast sync slicing circuit (probably better than LM1881) because sync needs to be added back to Y at just the right time and amplitude. It's a tall order. For the best results it'd take an well-researched original design, that plenty of people will be happy to copy.

 

Can that circuit be easily adaptable to a positive supply and what circuit or IC is better than the LM1881? That seems to be the one most used by most people and I'm not familiar with any alternative.

 

Definitely not, without the negative supply you'd have to make the whole transcoder ac coupled which will need to be re-biased which requires the circuit and all resistors to be changed. Plus it'd have to be slightly modified regardless because there are no provisions for sync. Sync must be processed (since video games can't be relied upon to have the precise amplitude), then arithmetically added to Y, which will change the resistor balancing act that is that circuit.

There are lots of IC choices by different vendors with better performance but I don't have any experience with them. Personally I'd try to go with a discrete circuit since the better sync separator ICs are rather expensive and you can buy a lot of discrete parts for $6 or $7.

That that doesn't mean much when people don't understand what they're using it for and why; people just throw them into their project regardless of whether they're necessary or not, and always inappropriately connect them directly to their TVs...

 

Bringing a thread back from the dead

Listening to this advice I decided to try this circuit. I built it exactly like Helder schematic using the THS7414 and the 5.1Moh pull-up resistor. If I tested the DC offset correctly this is the proper resistor to keep it close to zero.

I will admit that the colors seem more accurate when compared to the mod by the OP... the problem is that there is some major ghosting/smearing/ringing in the luma channel. The jailbars were about equal to the the OP mod.

I tried changing a few resisters and rewiring the lines further away from each other to prevent ringing. No avail. It could be that I bread boarded it instead of soldering it to a protoboard or could also be my svid to HDMI converter box... but I have none of these problems with my other s-vid devices.

I would be afraid to admit that this is what properly amplified s-video looks like for the MD1, but I am skeptical to make this assumption because the OP svid mod has no ghosting whatsoever on my television using my MD1 and svid to hdmi converter.

Read the data sheet for the THS7414 for some tips to deal with oscillation but I couldn't find anything that would make a difference. Any Tips?

 

That's a pull up resistor, it's connected to 5V and not GND. Maybe this could be the issue you're having and of course it could be your converter doing the ghosting even though it works fine with your other setups.