shmups.system11.org

I recently stumbled on this video where there is a comparison between a ""stock"" SNES (read the description for details) and a modded one that try to sharpen the video output.

The result of this mod is looking really good on the video. It is designed by a japanese guy named Torapu (とらぷ).
This is the page (in japanese) where the mod is presented whith schematics for it.
The soldering looks horrible but maybe it must be looked into so we can improve upon his idea?

So what do you think?
Is this finally a good (wip) solution for 2-CHIP consoles?

Interesting mod, I like that he shows comparison pictures and the quality difference is clearly visible.
I thought I would be interested in testing this on my 2chip but then I saw this...



...wow, Can one of our resident experts interpret that into something that won’t catch fire?

That's definitely a prototype. One or more of the retro community's more talented members could probably turn that into a simple QSB.

Also, Voultar teased on the Retro Roundtable that he's working on a 2-chip video improvement.

I think this is the same guy who figured out how to match the imbalanced audio levels of the AV Famicom with the original Famicom, with a similar arrangement of jungle gym madness.

That fire hazard hurts to look at.

The comparison pics are easier to look at though. Not quite to a Jr. level of clarity, but a major improvement. Maybe I should sell my 1 chips before this becomes a thing and devalues them!

Why bother if the Reverse-LPF on the OSSC does exactly the same?
On second thought, if a board is produced I would definitely buy it.

Because not everyone who wants this effect has or wants an OSSC?

notice the date on the linked page is several years old, IIRC this has been discussed before and was ultimately abandoned by the creator because of problems with it

It's still somewhat possible to do, just complicated and error prone.

If we can figure out some ways to make it better, I think it can be a viable mod

Edit:
Just an idea for a start:
The sharpening method tends to pick up more high frequency (switching) noise, similar to what other contrast enhancements tend to do.
It's probably a good idea to use a THS7314 for the output stage, and benefit from its low pass filter right at the source.

Got any USA 1chip? I’m in for one

The reverse LPF on the OSSC doesn't work without introducing artifacts in other areas of the picture. I tried it extensively and realized I was trading a bad blurry picture for a bad sharp one.

At any rate, I had a look at it and I simply don't understand why these people don't learn PCB design software. There's no excuse for that mess, not even as a prototype. Also, I think he could have made his efforts even easier by lifting the subcarrier signal so it doesn't reach the encoder. I noticed his starting point had that classic subcarrier interference going on.

@fbx
With this analog design from his scope images I can see overshoot, just like with the digital reverse-lpf on the OSS, but to a lesser degree.
According to google translate it has some sort of overshoot limiting.
Since the snes has different amounts of smear on darker and brighter shades any reverse-lpf will work only on certain shades, and overshoot/undershoot for others.

In which case it's not worth it to me, although his results look quite a bit better than on the OSSC.

I made some test quite a whila agon (begin of 2016). But it was without success - I observed a lot of background noise in composie video and in RGB the results were not as good as promised by Torapus post. So I put it away...



Maybe it was because I only had a PAL model at hand back in the days.

Well, if he can sort it out, that would be fantastic. The RGB coming out of the 2 Chip looks like composite. It's depressing.

OK, i have an SHVC-001 like in the mod described, and want to give it a go. If i actually get around to laying out a proper PCB for the job, and it works, i'll report back. Might take a few months. The LMH6683 isn't very cheap and i can only find it in lots of 5. And does anyone know what transistors were used for Q3/5/7 and Q4/6/8?

EDIT: and omfg there are no good schematics out there for the SHVC-001! I'm going to try to trace the video circuit properly first to confirm that Japanese modder is right. He's clearly way smarter than i am but i find his methodology and style hard to follow.

And does anyone know what will happen if i sub in 1N4148 diodes rather than the 1S2076A items? Forward voltage of 1V rather than 0.8V... I guess i can always bridge them.

While there are some differences with early SHVC boards, the available schematics pdf applies to everything you need:
https://wiki.superfamicom.org/schematic ... nd-pinouts

If you look at borti4938's PCB, you'll find what to use for the transistors.

The LMH6683 was probably chosen for its high speed.
Using another chip may be the tweak point to address some of the issues with this mod.
(Ie: Using one of the TI video amps should work and reduce complexity while increasing quality )

Good man, be sure to report back

Cheers for the info. Sadly borti's board simply says to use the transistors from the SNES mainboard, which i'd prefer not to do. They're probably just A1037's tho, so i'll try some of those. And re the THS7374 i would have thought the precise filtering was the point of this, but given i'm on CRT i'm going to have to break the signals out to an oscillscope anyway to test. I'm hardly any good at this though, so i wouldn't be holding my breath if i were you

EDIT: i've done up a schematic - https://imgur.com/a/hFyyEeZ, layout comes next. I'm going to have to wait a while for parts to arrive as some are coming from China, so i won't be able to test anything for a few weeks at best.

EDIT: slight change to schematic above, to reflect the original SNES R7, R12 and R17 remaining in circuit. You have to not only look at Torapu's schematic but also at what he's actually soldered in. Hopefully the SFC i bought for testing will arrive soon.

PCB is ready but not ordered or tested - I'll order some tonight. I put it up as part of this post on AussieArcade - https://www.aussiearcade.com/showthread ... -video-fix

Hi all. I am the one who made the video comparison. Because I couldn't find any good quality captures of the mod in action, I made my own board for it (the board design was terrible, so I won't publish it). My particular implementation bypassed everything on the stock video driver. The inputs tapped the PPU2 RGB traces and the outputs were soldered directly to the RGB multi-out pins. All the stock video driver components on the motherboard were removed. Overall, the mod shows that it's possible to do something about the 2-chip blur in hardware, but I don't know if a "good enough for me" solution exists yet. The filtering is a bit on the weaker side and I think there was some nonlinear distortion/clipping on the lower end of the signal. However, its low noise was one of the major advantages over using the OSSC's Reverse LPF and is one of the motivators to look further into it.

The hope of the video comparison was to get people interested in investigating an acceptable hardware solution for the blur. I have since been investigating it myself and made over half a dozen prototypes, but it's been tough because I'm a newbie at PCB design and analog electronics. Trial-and-error methods are a LOT harder to do when you have to spin up a new board for every attempt! There are also virtually no technical discussions about addressing this problem, so I've been groping in the dark to find something that works. Maybe we can get the ball rolling on this.

Right now, the biggest hurdle to overcome is jailbars. There are some striking bars that occur at a frequency of about 670 kHz (every 8 pixels or so) and appear strongest in the red channel. You may notice them in the comparison video. I'm want to lift the video output pins on the PPU2 chip and solder to them directly to see if it helps, but the thought of trying that scares me.

So i got around to installing the mod PCB, and did a little testing, and i notice a nice improvement. I'm using RGB into an Ikegami TM10-17RA 10" shadow mask pro CRT monitor, and the difference is immediate. I did use new 2SA1037A transistors rather than recycling the SNES ones, but that shouldn't make any difference. I also left off the 270pF filter caps by mistake, but the image isn't suffering. And i did the 10uf caps mods first, as Torapu describes, and also the 7805 filter cap mod on console5 (https://console5.com/wiki/SNES#Vertical_Line)

The modded image is crisper, but also brighter and more vivid and colourful without seeming to crush blacks or highlights. I have another unmodded SHVC-CPU-001 SFC to compare, and the unmodded picture now seems a little... muddy. The new picture probably isn't perfect, I'm not able to scope anything let alone get decent photos of the CRT, but it is a definite improvement. This is based on running images from 240p test suite alone, so i really should try a few other carts too for further testing - i want to triple check that highlights are not being crushed at all.

My own experience would indicate Torapu's pictured improvements are borne out. Why others have not had equal success i cannot guess.

PCB files, install guide and pics are in the link in my post next above.

Still no idea if this mod will work for a PAL SNES though

This looks quite promising! Unfortunately, there was only one image I could see that didn't require registration to the aussiearcade forum. Would you mind reposting pics here to make it easier for folks to see them?
Also would be useful if someone could test this on an OSSC + capture, as things such as noise/vertical bars are more noticeable that way.

Yep, could you please post your pictures here as well? Imgur.com also works nicely for this.

As yoshiyukiblade said, the mod is viable. The community needs to come together and improve it a little, but it can be done.

Sorry, i didn't realise you'd have to register to see everything. Yes i know what imgur is, but I was hoping not to post everything twice. If you cannot see the parts list or download the PCB, let me know and i will upload those here too.

There is only one pic of the end results, i tried to get more but i'm no photographer. Someone else will have to snap some. Or simply capture the video output, or use a scope. The below pic isn't even very good, but if you try it at a few different scales and keep squinting at the right hand edges of the black lines, you can see the blur is noticeably reduced. This pic is full size, and if you can see the dots of the CRT shadowmask you're zoomed in far too close. So download it, shrink it, and squint away


The other shots were just install pics of the filter amp PCB and showing some electrical tape over the RFI shielding which will stop it making contact just in case.



And then a couple of install pics stolen from Torapu's page, so you can see those at the link in the OP on here.

Having looked at a few more screens, and a few games, i think the mod isn't crushing anything but is skewing tones very slightly toward the brighter end. Black is still black, but a 1% tone might become 2%, up to 50% becoming 55%, and 90% becoming about 92%, up to the point where only 100% is actually fully 100% still. So it's not an even skew. And if it's not just me imagining things to begin with, it can be corrected with brightness and contrast adjustment anyway. I should probably install those 270pF bypass caps and see if they make any difference.

Hmm, thinking about this, I wonder if the brighter image is because the transistors are now able to switch cleanly? Probably not, but one of the things i played with before the mod was swapping them for 2SAR512R's which only smeared things far worse. Another clue in the puzzle, perhaps.

Well, i'm sure those of us who care have been following Opatus's SNES digital video thread - viewtopic.php?f=6&t=66597 (EDIT: by which i meant it's awesome, and potentially a far better solution if you don't need analog video)

But i thought i'd report that jeremybh1 on the AussieArcade forums tried the Torapu mod on a 2-chip PAL SNES with equal success to my own results. He then realised he'd left off the 270pF filter caps on the transistors, and added them, only to have the sharpness reversed again. I left them off by mistake, myself, but never bothered adding them as things were so much better already.

https://www.aussiearcade.com/forum/game ... -video-fix

Update: New gallery of photos https://imgur.com/gallery/e1QtvVj
2-chip RGB Bypass
Resident forum member, Buttersoft, had previously shown his Super Famicom that was modified with an RGB bypass to enhance sharpness. It was thought that only 1-chip Super Nintendo consoles were an option if one wanted truly sharp video output. 1-chip refers to Nintendo's later revisions of the SNES that use only 1 picture processing unit chip. Personally I didn't want to seek out a 1-chip as I'd already come so far with this existing console.

Today I have completed my install of this RGB bypass and the results are excellent! I knew from the second I turned on the console that edges of text were improved, and I could see every pixel defined, including some used for dithering effects that I hadn't quite noticed before. My complaints about the softness are eliminated. I also believe the saturation of colours is that bit better too. For example purple tones are stronger.

Games tested thus far with are Super Mario Kart, Aladdin, 240p Test Suite, Super Mario World and Castlevania.

Please note my findings include leaving out a step - there is no benefit to installing 270pf capacitors, rather it reversed my sharpness!

I followed Buttersoft's instructions which worked out as follows on my SNSP-CPU-02:
Populate the PCB with SMD 1206 components of resistors and caps. Then diodes and amp chip. Tin pads and clean any flux thus far.
Open up the SNES to remove the main board, leaving the heat sink on as we are only using the rear side of the mainboard.
Desolder Q3, Q5, Q7 from the mainboard and solder them to the bypass PCB.
Prep the pads on the SNES board by using solder wick, clean with alcohol, add fresh solder, clean with alcohol and add a tiny bit of rosin flux to each pad.
Prepare my wires by laying them over each point and marking where to cut and trim. Once trimmed I twisted the exposed wire a bit, tinned it with solder, then used snips to trim the exposed wire right down. The transistor pads are small and I wanted a neat fit, but left a longer bit of wire for the PCB pads.
The PAL SNES has +5v at different locations to the SFC (Super Famicom), and I settled on the right side of R21 which traces back to the output pin of the 7805 regulator.
For ground I used a corner of a ground plane.

Tested the console, worked first go using that custom cable that has C-SYNC on pin 3 and 75-ohm resistors to ground on R, G,B.
I mention this as I was unsure if the bypass board requires a straight through cable or still needs the resistors to ground (like all PAL SNES consoles need). It does need the resistors.
This RGB Bypass absolutely benefits S-Video output also, and I tested it with my PAL S-Video cable to great effect.

In addition to my above installation steps, here are the details for a Japanese SNES when using the PCB by Buttersoft https://oshpark.com/shared_projects/Ha2tBzZa

Parts:
The diodes are DO-35 package 1S2076A. Each set is two diodes, one forwards one backwards.
U1 is the amplifier chip, LMH6683MAX package SOIC14. Buy from Mouser
Re-use the three transistors from the SNES mainboard, or buy new ones which are Bourns 2SA1037A, package SOT-346
All other packages are SMD 1206 package size:
2kR - R2, R3, R12, R13, R19, R20
300R – R1, R14, R21
200R – R4, R5, R10, R11, R17, R18
47R – R6, R7, R8, R9, R15, R16
All Capacitors are 1nF
After Q3, Q5, Q7 have been relocated to the PCB, observe below
Collector
__|__
[____]
| |
base emitter

Solder the respective wires as described above, including +5v to the PCB pad and a single ground wire to the capacitor shown.
Affix the board with electrical tape for shielding and your glue of choice. I still swear by hot glue as it's got an antidote - alcohol! Strain relief and fixation is never a bad thing.

https://imgur.com/6kJMZ1m
https://imgur.com/Rk3Qbzt
https://imgur.com/HxRTe0F
https://imgur.com/Iwahpee

That's amazing news! Thanks for the report captaineos

I'm eagerly anticipating someone making a vid tutorial, on an NTSC-U console would be best (my ADHD really makes it hard for me to use text as directives )

Damn fine work, so 2-chip softness has been fully defeated? I would love to see someone do a 1-chip/Jnr bypass comparison.