SNES-Mini RGB Measurements

You would have an impedance mismatch. Keep the output at 75 ohms, read the thread, and add the resistors to the proper points on the mini.

 

I did the mod without any divider at the input pins and using std 75Ohm resistors at the out, I have to say that connected to the XRGB Mini works reasonably well.

The impedance mismatch we seem all to talk that much about cannot be obviously at the input of the 7314, at the output (unless there's clipping) even if I have to put a 100 Ohm or a 150 Ohm if needed wouldn't make so much of a difference ... instead of 50% at the TV with the std 75 Ohm I'll get 33% at the TV with a 150 Ohm if that's what it takes.

All the 75Ohm matching talk is wrt coax carrying RF signals to minimize reflections
https://en.wikipedia.org/wiki/Coaxial_cable#Choice_of_impedance
here we are in baseband with non shielded cables and using 10% components (I know I am) ... so I doubt that if I have to "shrink" a little the signal at the out it would be such a disaster ... again unless the 7314 is clipping or working in a non linear region.
I'm fine that you guys are making the circuit the best it can be conditioning the signals before the amp, but it should be possible to tweak the out stage as it is without too much of an issue, certainly not one of impedance mismatch that is any serious.

Moreover in my case I have to account for the XRGB Mini that would process the signal and recondition it and send it via HDMI to my TV which will have to adapt it to the panel native colorspace (which if I recall correctly is around 12 bits these days) ... I am convinced that the voltage divider at the input may not matter that much as long as the original circuit is not making the 7314 work in a completely wacky part of the spectrum.

Anyway as I stated in the opening sentence, the default proposal worked well enough for me for now, and given its simplicity if I have to raise the out resistor to get a better picture I will.

EDIT:
The input voltage divider proposed uses a 10.7K and a 82.5K = total of 93.2K
82.5/93.2 = 0.885 -> it means we reduced the input signal a mere 12% -> given the 7314 has 6dB = 2x gain at the out we have a 24% difference with or without the divider. Given we have a source at the 0.7-0.8V level the amplifier is plenty in linear region [it can work all the way to 2.2V just fine] -> a normal 75 Ohm at the output may give you a signal that is 12% higher than expected at the TV (remember that the TV has another 75 Ohm so the final signal is divided by 2) -> raise that to 100 Ohm and the signal at the TV is now 75/175 = 0.42 = 42% of the total -> so we have an initial signal (we would like it to be 100 which multiplied by 2 by the amp is 200) that is 12% too high (so really 112) multiplied by 2 by the amp = 224 in out but with a 90 Ohm we have at the TV 75/(75+90) = 0.45 = only 45% of it (instead of the 50% with a 75 Ohm) -> 224 * o.45 = 100.8 ~ 100 -> and we're back and have absorbed completely that initial 12% skew, if we use 100 Ohm instead we have 75/(75+100) = 0.43 -> 224 * 0.43 = 96.3 so again very close to get the signal as we want it.

Again please understand I appreciate you guys making the circuit the best it can be it's just that if the whole difference is 12% I believe it can be absorbed while keeping the circuit simple.

 

I understand what you are saying, but if you plan on adding resistors to make a divider on the output, why not add the proper resistors to fix the output of the PPU? Then the amp would buffer the correct signal level, and you don't have to worry about correcting it at the output? It's very simple, just 3 resistors tied to GND.

I am also using a xRGB-Mini BTW.

 

The input dividers require 6 new resistors (2 for each input), the change I am proposing replaces the 3 75Ohm with say 90/100 Ohm and that's it, no 6 new components.
That's what I meant by keeping it simple.
If I had to purchase a premade PCBs like the ones floating around I'd go for the one that does it best obviously, but as I am soldering myself the least number of components the lest likely to screw up etc....etc....

[I reedited my previous post with the correct math with a 90 Ohm and a 100 Ohm to exemplify the transfer function of 0.45 and 0.43 respectively, 90.9 Ohm and even more 100 Ohm are common enough to find]

 

Oh ok. Well, good luck then. We were just looking for the proper design, and it was important to keep the output with a 75 ohm resistor since that is what the TV/upscaler expects to see. I found that the output from the PPU was not as expected.

I would suggest using 1% tolerance resistors if you can. We are talking about very small signals, and a slight change between resistors can cause a difference in color balance.

 

I'm not 100% sure what that means.
Take the SNES Mini PAL, in the Scart cable Nintendo added a 75 Ohm in parallel with the TV (from R-G-B to GND), thus halving the total value to 37.5 which would require the same at the source to halve the original signal (if it's 2x to start with that is), so saying the the TV "expects" 75 Ohm is inexact.
All that the TVs want is correct signals at their end and they tend to be quite forgiving depending on the amount of conditioning they perform (the less of it the less forgiving of course), whatever you do to minimize reflections, noise, spurious harmonics etc... whether that means losing 50% of the signal on the console side with an extra 75 Ohm resistor (or whatever) after having amplified 2x the original signal or anything else is fine by the TV.
The TV itself does not expect anything from the source with the exception of a signal that when applied to its 75 Ohm behaves correctly, the rest is left for exercise to the transmission line implementers ;-)

I recently had to "shortcut" at the console a 75 Ohm resistor over a composite signal out of a CXA1145 on a US SMS because for some strange reason it was way too dark (this after having made sure the RF modulator was isolated from the composite out at the chip, there was nothing else on the line) ... I didn't shortcut it just added a 10 Ohm in parallel ... ideally the CXA 1145 is supposed to drive a 150 Ohm (75 + 75) load and yet it did not for reasons I have yet to understand (faulty CXA1145 or else), in all of this the TV was fine (actually now it's better, I tried 3 different TVs all with the same results before/after the mod).
On the same vein I found out that there exists both 50 ohm and 75 ohm optimized connectors for the cabling (at least for coax/shielded pairs), again hard to judge their impact in baseband.

Again please don't take this as negative feedback, I'm just stating that it should be possible with negligible effects to have a simple circuit. I use LCD TVs (no CRT) so my experience may be different than others. Also I noticed that when you go digital it's really heavily dependent on how much conditioning the TV applies, I have a LCD TV 1080p whose SVideo input is amazing (it's hard to judge how much Scart RGB + XRGB Mini really improves), but another whose SVideo input is hardly better than composite (and here XRGB Mini really shines) ... the point being that what you see on your particular TV/Monitor may not work for others the same way you intended however it is hard to debate that if you adhere to strict standards it would have issues (even when the standard tells you to double a signal just to halve it again the very next stage).

Alright, I deserve a TL;DR

 

If you are designing a mod or anything, you should stick to standards - not rely on the TV being forgiving.

That was the whole point of the thread. If you want an "easy" and incorrect mod - just go back to using 100ohm resistors on the output that everyone was doing before this thread.

 

Impedance mismatching can cause ringing and ghosting on the video image. But, hey if it's acceptable to you, by all means go ahead. I don't see how our circuit is difficult, it works, and it is designed correctly.

 

I completely missed that part of the posts.
And it was right there in the OP, now I feel like an arse .... oh well, sorry!!!

 

As I was at it I educated myself about matching etc ...
http://www3.eng.cam.ac.uk/DesignOffice/mdp/electric_web/AC/AC_14.html

I found interesting the remarks about "short" and "long" lines.
Only for "long" lines the (mis)matching seems to make a difference wrt to wave reflections and power transfer.

For a 6Mhz signal (that is the high frequency of our video signal on each channel) the wavelength is 50 meters, 1/4 of that = 12.5 meters according to http://www.csgnetwork.com/freqwavelengthcalc.html,
my cable is not even 2m so relatively speaking short for the baseband signal hence effects of mismatch should be almost negligible.
Somewhere else I was reading that anything < 1/10 of the wavelength classifies as "short" line, that's up to 5 mt of cable in the RGB scart connection case (as well as composite and SVideo).

Anyway just wanted to add the article that explains pretty well matching, line lengths etc.....

 

I will probably make a fool of myself by posting this but here goes...

I've been wanting to mod my old rev. snes to bypass the internal rgb circuit because of the bad output. So after reading this thread I got curious about doing the same measurements on it.
I should stress that I do not really know what I'm doing so I may possibly have screwed up, but I happened to have an old oscilloscope lying around so...

This is measuring the red signal with a Tektronix 475A on a PAL SNSP-CPU-01 (running in 60hz mode) using the 240p test suite

From PPU2 pin 95:
0.5 V/Div (roughly 2.5 Vpp)
0 DC offset

From the multi out:
0.5 V/Div (roughly 1.3-1.4 Vpp)
DC offset: 2.3V


To confuse matters I have used different horizontal settings for these two measures so the top picture shows a 15.5KHz signal and the bottom one a 60Hz signal. I guess what matters is the peak-to-peak voltage anyway...

Seems like this version has a different signal coming from the ppu or I may have made some error. How will this affect the rgb circuit if I was to mod this thing?

 

I have this problem to, i tried to bridge C48 but i still have interference, specially noticable in magical quest intro.
Im using an rgb modded (internal amp) snes mini (USA) official power supply and csync rgb cable.

 

I know this possibly doesn't qualify as a SNES, but those mini SNES look alike clones like the Yobo FC Twin (with the cloned CPU and PPU chips) usually come with either the CXA1145 or CXA1645 video encoders and they do include a 1K pull down resistor to GND for each RGB line coupled with a 0.1uf ceramic cap into the encoder.
I recently did a RGB mod to a similar clone and I'm getting the same video noise as the others no matter if I use Csync, Composite Sync or a Sync separator like the LM1881. I wonder if I should try amplifying the signal with a transistor instead.



These are examples of such clones:

 

For the 2 posts who are having interference, how well made is your RGB cable? It is important that the cable is shielded, the shield is connected to GND, and it definitely helps if each R,G,B and Sync cable is also shielded separately. As a crude test, you can try wrapping your cable in some aluminum foil, and make sure the foil is connected to a ground.

 

I mod today my SNES Jr (NTSC-J) with a RGB amp from borti with this cable: http://www.ebay.de/itm/201443007751?_trksid=p2060353.m2749.l2649&ssPageName=STRK:MEBIDX:IT

I use no resistors on the RGB signal. On a LCD LED Samsung TV. Looks really nice, tested with Star Ocean. Result:

 

In this guide he added a interference filter, I want to try this but its hard to see how it shall be done.
http://www.gamesx.com/rgbadd/snes2rgb.htm

 

anyone?

 

To clarify, aside from the cleaner sync signal, would this design provide better video quality on an SNES Mini than using something like the RetroRGB amp, so long as you use 1.1k Ohm resistors to tie the signal to ground?

 

The THS7374 is slightly different than the THS7314 used in RetroRGB amp. The only difference would be the extra input to buffer the CSYNC signal. Video quality would be the same. If you want to try and improve video quality, use well-shielded RGB cables, and keep the length of the cables used for the mod as short as possible, and away from power sources.