YPbPr ouput mod for SNES consoles

[Pasky]

[mikejmoffitt]

Ok, I replaced the 10uF going from the S-Enc to the encoder with a 220uF.

The result is the yellow stays great for much longer, but just drains away much more slowly. Instead of draining within 8 seconds like the 10uF it now takes about 30 seconds to drain. I'm not certain what's going on and why the yellows are being drained away. Anyone have any idea?

Or maybe people are creating broken "amps" due to minimal understanding of video signals, termination or transistor amplifier topologies.

I'm not claiming to be any kind of electrical engineer. I'm simply following a schematic and reporting my findings. You may very well be correct, I'm open to suggestions from anyone with knowledge on the subject.

I am not versed in either amplifiers nor buffers, but this situation I have seen before remedied with a simple buffer made with an NPN transistor.

[Pasky]

I am not versed in either amplifiers nor buffers, but this situation I have seen before remedied with a simple buffer made with an NPN transistor.

Details please?

[kyuusaku]

The NJM is designed specifically for composite/luma, it doesn't seem to have a way to disable the clamp function so it can't be used as a general purpose 6 dB amp. Maybe this is a problem, maybe not..

Your 220 uF input capacitors are way too big. The bigger the capacitor the longer it takes to charge and discharge for the clamping function. The datasheet specifies 4.7 and later 10 uF, which is still very large for a clamp cap, but it's a BJT chip... So I dunno which one I'd choose. Probably 4.7 uF.

The NJM chip also has a sag-correction feature. You don't want to use that--it compensates for a condition affecting only sync carrying signals. To bypass it tie pins 3&4 and 5&6 together and get rid of the 22 uF caps. It's also better to put the 75 ohm resistors first, and 100 uF capacitors second, which you can do now after getting rid of sag-correction.

[Pasky]

Hrmmm.

I kinda want to build a completely new circuit. I ordered a breadboard a couple days ago so I can play around with some of the settings more quickly. I've seen people use this for the n64 rgb mod:

http://www.ti.com/product/ths7314

http://mmmonkey.co.uk/console/nintendo/ ... gb_new.htm

But I'm unsure what circuit I'd need to build for this with the SNES. Would this be a better video amp to use? If so, could you help me by providing a circuit to use? Thanks.

Maybe there is a better amp to use for this than that one?

[NightWolve]

Ok, I replaced the 10uF going from the S-Enc to the encoder with a 220uF.
The result is the yellow stays great for much longer, but just drains away much more slowly. Instead of draining within 8 seconds like the 10uF it now takes about 30 seconds to drain. I'm not certain what's going on and why the yellows are being drained away. Anyone have any idea?

Pasky, very quickly, before I respond to other posts, you made a mistake here in not paying attention to polarity and why those 2 capacitors were wired the way that they were. You've got +3V DC from the pins and you're facing the -/negatve end of a polarized capacitor towards it! Not good! The original circuit here called for a 100 or 220 uF cap with the +/positive leg connected to the S-ENC pin. This would filter out the +3V DC, and the 2nd capacitor, if you look closely, had the -/negative leg connected to the previous cap's -/negative leg.

What you just did in your picture shows that you connected the negative end of a polarized 220 uF cap to a +3V pin in changing the original design because of what your engineer friend said. You can damage or blow a cap if you don't pay attention to polarity and when you have a source with voltage, etc. Anyway, that was the idea behind why the two were connected that way. 1st cap, +positive leg to the pin, filters DC, which makes it safer to connect the 2nd cap in the opposite direction. If you could deal in 10 uF range, you can get ceramic caps and not worry about this, but yeah, gotta watch that with polarized aluminum caps!

I don't like circuits where the direction flow is ambiguous, but I have seen them where caps have the negative pin facing something generating a signal, so that's not necessarily wrong. I don't know what that accomplishes, but yeah, I've seen it.

Err, did you just not change the image where the + sign is ?? Probably should get changed lest somebody goes and tries it that way.

[Pasky]

Ok, I replaced the 10uF going from the S-Enc to the encoder with a 220uF.
The result is the yellow stays great for much longer, but just drains away much more slowly. Instead of draining within 8 seconds like the 10uF it now takes about 30 seconds to drain. I'm not certain what's going on and why the yellows are being drained away. Anyone have any idea?

Pasky, very quickly, before I respond to other posts, you made a mistake here in not paying attention to polarity and why those 2 capacitors were wired the way that they were. You've got +3V DC from the pins and you're facing the -/negatve end of a polarized capacitor towards it! Not good! The original circuit here called for a 100 or 220 uF cap with the +/positive leg connected to the S-ENC pin. This would filter out the +3V DC, and the 2nd capacitor, if you look closely, had the -/negative leg connected to the previous cap's -/negative leg.

What you just did in your picture shows that you connected the negative end of a polarized 220 uF cap to a +3V pin in changing the original design because of what your engineer friend said. You can damage or blow a cap if you don't pay attention to polarity and when you have a source with voltage, etc. Anyway, that was the idea behind why the two were connected that way. 1st cap, +positive leg to the pin, filters DC, which makes it safer to connect the 2nd cap in the opposite direction. If you could deal in 10 uF range, you can get ceramic caps and not worry about this, but yeah, gotta watch that with polarized aluminum caps!

I don't like circuits where the direction flow is ambiguous, but I have seen them where caps have the negative pin facing something generating a signal, so that's not necessarily wrong. I don't know what that accomplishes, but yeah, I've seen it.

Thanks for the heads up, I'm just gonna completely scrap that and try something else, the Luma and the yellow being incorrect is really bothering me.

[kyuusaku]

It's not so wrong to face the AC coupling cap towards the NJM, like basically every amplifier the input is biased so it's sort of standard practice when you're expecting an AC coupled signal, or low-offset DC coupled.

Actually here it really doesn't even matter. Consider:

-It takes over -1V across the leads to damage the cap.
-220 uF @ *1 Hz* has a reactance of 723 ohms. Neglecting the NJM's bias voltage even a 5V DC offset wouldn't damage the cap: (5 / (723 + 1000000)) * 723 == 0.003 V
OK the input impedance could be lower, but you get the picture. You wouldn't want to short it to a high voltage however.

Anyways this is irrelevant because the component choices are totally inappropriate.

[Ziggy587]

Hrmmm.

I kinda want to build a completely new circuit. I ordered a breadboard a couple days ago so I can play around with some of the settings more quickly. I've seen people use this for the n64 rgb mod:

http://www.ti.com/product/ths7314

http://mmmonkey.co.uk/console/nintendo/ ... gb_new.htm

But I'm unsure what circuit I'd need to build for this with the SNES. Would this be a better video amp to use? If so, could you help me by providing a circuit to use? Thanks.

Maybe there is a better amp to use for this than that one?

I've already tried that amp, look at my results on the top of page 4.

viewtopic.php?p=107496#p107496

[Pasky]

Hrmmm.

I kinda want to build a completely new circuit. I ordered a breadboard a couple days ago so I can play around with some of the settings more quickly. I've seen people use this for the n64 rgb mod:

http://www.ti.com/product/ths7314

http://mmmonkey.co.uk/console/nintendo/ ... gb_new.htm

But I'm unsure what circuit I'd need to build for this with the SNES. Would this be a better video amp to use? If so, could you help me by providing a circuit to use? Thanks.

Maybe there is a better amp to use for this than that one?

I've already tried that amp, look at my results on the top of page 4.

viewtopic.php?p=107496#p107496

Ah I missed that....in that case, I'll wait for my breadboard to arrive and see if I can tweak this amp unless someone has a better one in mind.

[NightWolve]

Here go some quicki screenshots with my USB YPbPr capture device using the latest circuit here with 33 Ohms on the Blue-Y output. Think I'm just gonna have to recommend a 0-100 Ohm pot on the outputs because determining the best value is tricky and given what Mike said about the imbalances which was true for me, my case of not having to amp Blue-Y, the results for what's "best" is gonna be all over the friggin' place, even for the exact same chip and board revision, etc. That's really too bad. It's still an interesting mod and all, I've enjoyed the time spent fiddling around with it.





In the SSFII case below, I took a screenshot with ZSNES and resized it to match the YPbPr shot. That green water there actually is a good test to see if your TV's circuitry is properly decoding green. My CRT doesn't handle it well; it'll come out bluish while it'll be perfect on a modern Samsung LCD as mentioned previously. Another thing, I turned up the sharpness a bit on my capture device's settings, so that's why it's mismatched compared to the ZSNES shot. Anyhow, pretty good on that one! Red seems alright, green is decoded proper by the device... Yellow though, yeah, like Pasky is mentioning, there's an issue with that elsewhere... But eh, not bad for a poor man's generic transistor amp and only having to do it on one line.



Below is what that green looks like on my old school Panasonic 32" CRT 1999 era. It's "cool," doesn't have that fluorescent appearance that my Samsung 19" LCD has, nor the USB device or ZSNES. However, if you switch to S-Video/Composite, it works, so this is strictly a matter of the circuitry that processes the YPbPr signal. Note: the circuit used at the time of these CRT shots was my previous one.



I wish I hadn't done the +5V mod to my system though. Some time back, my +5V regulator burned out and I decided that since I had a good, well-regulated +5V 3A power supply, I should try to mod the system to work with it, then I could get rid of that big heat sink metal that covers the relevant work area with the chip, plus it'd be better for the system... Well, I succeeded with some hiccups at first, but yeah, I didn't have to order a new regulator which I didn't wanna have to do... Anyhow, I wonder if that change affects my results in terms of signal strength levels and so forth, so I kinda wish I hadn't done it now... So yeah, it's a +5V modded SNES: I had to remove a few caps in the power area near the power switch, added clips for a 1A glass fuse (burned a few), had to bridge two traces and connect the input to output holes/solder-points where the regulator was together, obviously. Poor thing is still working.

[Pasky]

Anyone know if this would be a direct replacement for the SNES video encoder?

http://www.utsource.net/ic-datasheet/BA ... 57098.html

[NightWolve]

I've been chatting with Ace9921 on/off concerning these chips actually. He likes the BA7230LS (you can buy that at the same site you linked) over the BA6592F for custom mod jobs he's been doing and he has a circuit for it. Anyway, I looked up a quote with him from youtube:

As for the BA6592F (which is not the same thing as the S-ENC, that's a BA6594AF), I recommend you switch to a different encoder, the BA7230LS. With minimal tweaks, this thing is capable of delivering superior video to the BA6592F. It's just as sharp, but the colors are more vibrant than with the BA6592F. Better yet: it's A LOT easier to solder to as it's not a surface-mounted chip. Either way, take you pick and head to UTSource.net. That's where I get my chips from. Ask for a quote ($10 minimum).

I believe the pinouts are the same, but they're somehow different, so... Whatcha planning now ??

[Ziggy587]

I've already tried that amp, look at my results on the top of page 4.

viewtopic.php?p=107496#p107496

Ah I missed that....in that case, I'll wait for my breadboard to arrive and see if I can tweak this amp unless someone has a better one in mind.

Sorry, I posted that really quick before work, I didn't mean to say that you should avoid the THS7314 amp. I just wanted to give you a heads up that I've already tried it and show you how I wired it. My results were... bad, but I think it's because I was doing something wrong. It's a 6dB amp, and even specified in the datasheet YPbPr, so I'd assume you'd be able to use it. When I tried it, the colors were all wacky. I should have taken a screenshot because it's kinda hard to describe what it looked like. Not that I ever have, but I'd imagine if you ate mushrooms or dropped acid then played SNES, that's how it'd look.

Give the THS7314 a try if you have the spare time, it might be an good amp to use. I don't know.

Your 220 uF input capacitors are way too big. The bigger the capacitor the longer it takes to charge and discharge for the clamping function. The datasheet specifies 4.7 and later 10 uF, which is still very large for a clamp cap, but it's a BJT chip... So I dunno which one I'd choose. Probably 4.7 uF.

You'll have to excuse my ignorance because I basically no nothing...

The reason why I used the 220uF caps was to drop the voltage down before inputting Pr and Pb into the amp, it wouldn't work otherwise. I don't know of any other way to kill the voltage, which is why I used that RC circuit or the 220uF caps. Without anything, none of the amps I've tried would work.

After reading the latest replies in this thread, with the NJM amp I've tried removing the 220uF caps and running Pr and Pb directly from the chip to the 10uF cap on the NJM's input. The amp wouldn't work.

The NJM chip also has a sag-correction feature. You don't want to use that--it compensates for a condition affecting only sync carrying signals.

Awesome! Good to know, thanks. Not to go off topic, but should I assume the same for the RGB mod for the N64? That's were that diagram came from, and since the RGB signal is outputted along with a dedicated sync, then I'd assume the RGB signals contain no sync and the same logic applies to them for the SAG pins?

The NJM is designed specifically for composite/luma, it doesn't seem to have a way to disable the clamp function so it can't be used as a general purpose 6 dB amp. Maybe this is a problem, maybe not..

Hmm... Well like I said, I basically know nothing. I've tried the OPA360, the THS7314, and the NJM2267 amp. The OPA360 was very hard to use. It requires a 3v power supply, it's a pathetically small package to hand solder, and you need two of them. So I abandoned it for those reasons. The THS amp seemed like a good choice, but I had bad results with it. Perhaps we should revisit the THS?

Assuming the NJM amp is a bad choice, what amp would YOU use?

With whatever amp we use, what do you recommend we do to kill off the voltage before feeding the signals into the amp? Like I said, I was unable to get any of the amps working unless I did something to kill off the voltage first (like the 220uF cap).

[Pasky]

I ordered a THS7314 so I'll still have one to play around with. While I wait on parts, I'm trying to repair this broken snes I got today.