A fix for SNES jailbars

The place for all discussion on gaming hardware
syboxez
Posts: 94
Joined: Sun Jul 24, 2016 4:33 am

A fix for SNES jailbars

Post by syboxez »

I have finally discovered a fix for SNES jailbars (example: viewtopic.php?f=6&t=61573). This fix applies to all models including all 2CHIPs, all 1CHIP revisions, and the Mini (was tested on an SHVC-CPU-01 and a Mini). This is not aliasing as several people have mentioned, as this issue occurs on CRTs as well as digital displays (and scalers). A low pass filter may hide this issue (although it did not for me when using a CRT), but it will not eliminate the cause of it.

For those familiar with PC Engine jailbars, this is a very similar issue with basically the exact same fix as described here by Tim Worthington: https://etim.net.au/av-driver/pcebars/

To fix the issue, I replaced all the SMD ceramic decoupling capacitors (stock: 0.1uF) with 10uF X7R ceramic capacitors on CPU, PPU1, PPU2, WRAM, and both VRAM chips, although you probably just need to replace the one on PPU2 (or S-CPUN for 1CHIPs) since that is what produces the final analog RGB output.

Note that this is separate from the white vertical line fix, which is to replace the VRM with a 78S05 (or equivalent 2A linear VRM) and to add an extra decoupling capacitor to the output.

I would highly appreciate any criticism and verification of this method, as I do not own a good oscilloscope.

This mod may also improve the results of the reverse LPF in the OSSC for 2CHIP consoles, although I cannot test this seeing as I do not own an OSSC. nor do I play my games on modern displays.

For those that would like to test, the issue seems to be most noticeable on the intro to Final Fantasy VI or in Chrono Trigger's indoor void (generally the same places that the white vertical line is visible).

Note that this will not fix other types of interference, such as checkerboarding. Checkerboarding on the SNES is caused by interference from the Chroma line with other video lines, generally by either a bad PCB layout and/or a video cable without proper shielding.
User avatar
FBX
Posts: 2347
Joined: Wed Feb 18, 2015 10:18 am
Location: DFW area, Texas
Contact:

Re: A fix for SNES jailbars

Post by FBX »

Interesting find! Although I've not seen so much as a HINT of jail bars on any of my 3 SNES consoles, which are an APU, 1CHIP-03, and the junior (which is Voultar bypassed).
syboxez
Posts: 94
Joined: Sun Jul 24, 2016 4:33 am

Re: A fix for SNES jailbars

Post by syboxez »

FBX wrote:Interesting find! Although I've not seen so much as a HINT of jail bars on any of my 3 SNES consoles, which are an APU, 1CHIP-03, and the junior (which is Voultar bypassed).
Do you have a Low Pass Filter anywhere in your chain? If so, try turning it off. Also try the console on a CRT if possible (high quality CRTs such as a PVM preferred).

Also try other inputs such as composite or S-Video.

Some consoles are affected more than others since the health of these capacitors can greatly vary. Same deal as how ghosting varies greatly with 1CHIPs.
User avatar
FBX
Posts: 2347
Joined: Wed Feb 18, 2015 10:18 am
Location: DFW area, Texas
Contact:

Re: A fix for SNES jailbars

Post by FBX »

syboxez wrote:
FBX wrote:Interesting find! Although I've not seen so much as a HINT of jail bars on any of my 3 SNES consoles, which are an APU, 1CHIP-03, and the junior (which is Voultar bypassed).
Do you have a Low Pass Filter anywhere in your chain? If so, try turning it off. Also try the console on a CRT if possible (high quality CRTs such as a PVM preferred).

Also try other inputs such as composite or S-Video.

Some consoles are affected more than others since the health of these capacitors can greatly vary. Same deal as how ghosting varies greatly with 1CHIPs.

I use the OSSC's LPF with Voular's board, but that's because I have the board jumpered to no filter. My 1CHIP-03 is stock (with 750 Ohm brightness resistors) and looks fine without LPF. When I get some time, I'll check out all three consoles on my PVM. What I need to do is get an original launch edition SNES for testing.
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: A fix for SNES jailbars

Post by rama »

It's easy to see these old console's minor jailbars on a GBS-8200 scaler. This just has an LPF at 30Mhz, so stock systems will always show some jailbars.
If you turn on edge enhancement and contrast enhancing features, it becomes easily visible.
For example, I can see a 1Chip SNES calculating / shovelling around data :p

And yea, my go-to solution for these things is adding SMD caps of pretty high values (10 or 22uF work best) to the already existing caps.
SNES definitely needs one on the work RAM, and should have one per PPU.
This brings down the jailbars significantly.

The only console where this doesn't work or at least doesn't do much: Sega Genesis. The 5V and ground lanes are so contaminated on these, nothing seems to help. The source of much of it is definitely the VRAM bus and chips, but it'll be hard to avoid these since we need to get the RGB out of the region somehow.
syboxez
Posts: 94
Joined: Sun Jul 24, 2016 4:33 am

Re: A fix for SNES jailbars

Post by syboxez »

rama wrote:It's easy to see these old console's minor jailbars on a GBS-8200 scaler. This just has an LPF at 30Mhz, so stock systems will always show some jailbars.
If you turn on edge enhancement and contrast enhancing features, it becomes easily visible.
For example, I can see a 1Chip SNES calculating / shovelling around data :p

And yea, my go-to solution for these things is adding SMD caps of pretty high values (10 or 22uF work best) to the already existing caps.
SNES definitely needs one on the work RAM, and should have one per PPU.
This brings down the jailbars significantly.

The only console where this doesn't work or at least doesn't do much: Sega Genesis. The 5V and ground lanes are so contaminated on these, nothing seems to help. The source of much of it is definitely the VRAM bus and chips, but it'll be hard to avoid these since we need to get the RGB out of the region somehow.
Interestingly enough, my Sega Genesis (Model 1 VA6 w/ TMSS) doesn't have any noticeable jailbars on my CRT. Due to the PCB routing being a major issue, I wonder if copper tape would help like it does with the GBS-8200 (I don't have one, but I've heard reports that it does help). Copper tape would basically be the same as adding a third layer to the PCB that is only a ground plane. The Sega Genesis jailbar issue is a seperate issue caused by bad PCB layout, so it makes sense why adding more capacitance to the IC decoupling caps wouldn't fix anything.

Also I did more experimentation. I replaced the THS7314 in my Mini with a THS7374 with disabled LPF, and I do notice very mild jailbars (separate from the ones caused by low value decoupling caps), so a mix between high capacity low ESR ceramic capacitors for decoupling and an LPF should result in basically perfect video. I do still notice these "other" jailbars on my SHVC-CPU-01 more than on my Mini, so I may try to RGB bypass that with borti's PCB and enable the LPF when I get the chance. Note that all of my testing is on a consumer CRT, with nothing digital in the chain (unless an Extron Crosspoint and generic RGB to YPbPr converter counts).

On a separate note, I've been meaning to get a GBS-8200 to use GBS-Control on, but the only problem is that it doesn't support composite or S-Video, which I need for my C64, and I can't find anything on converting CVBS or Y/C (specifically with a PAL color carrier in my case) to RGB or YPbPr.

And on an even more separate note, does anyone know of any universal RGB to YPbPr converter that is actually good? I was thinking of picking up some Genesis HD Retrovision cables and attempting to adapt that for my Crosspoint output, but I have doubts that it will work correctly. Right now, I'm using a CSY-2100 clone, and the colors are terrible (even after I spent countless hours calibrating it).
User avatar
maxtherabbit
Posts: 1763
Joined: Mon Mar 05, 2018 4:03 pm

Re: A fix for SNES jailbars

Post by maxtherabbit »

syboxez wrote:
rama wrote:It's easy to see these old console's minor jailbars on a GBS-8200 scaler. This just has an LPF at 30Mhz, so stock systems will always show some jailbars.
If you turn on edge enhancement and contrast enhancing features, it becomes easily visible.
For example, I can see a 1Chip SNES calculating / shovelling around data :p

And yea, my go-to solution for these things is adding SMD caps of pretty high values (10 or 22uF work best) to the already existing caps.
SNES definitely needs one on the work RAM, and should have one per PPU.
This brings down the jailbars significantly.

The only console where this doesn't work or at least doesn't do much: Sega Genesis. The 5V and ground lanes are so contaminated on these, nothing seems to help. The source of much of it is definitely the VRAM bus and chips, but it'll be hard to avoid these since we need to get the RGB out of the region somehow.
Interestingly enough, my Sega Genesis (Model 1 VA6 w/ TMSS) doesn't have any noticeable jailbars on my CRT. Due to the PCB routing being a major issue, I wonder if copper tape would help like it does with the GBS-8200 (I don't have one, but I've heard reports that it does help). Copper tape would basically be the same as adding a third layer to the PCB that is only a ground plane. The Sega Genesis jailbar issue is a seperate issue caused by bad PCB layout, so it makes sense why adding more capacitance to the IC decoupling caps wouldn't fix anything.

Also I did more experimentation. I replaced the THS7314 in my Mini with a THS7374 with disabled LPF, and I do notice very mild jailbars (separate from the ones caused by low value decoupling caps), so a mix between high capacity low ESR ceramic capacitors for decoupling and an LPF should result in basically perfect video. I do still notice these "other" jailbars on my SHVC-CPU-01 more than on my Mini, so I may try to RGB bypass that with borti's PCB and enable the LPF when I get the chance. Note that all of my testing is on a consumer CRT, with nothing digital in the chain (unless an Extron Crosspoint and generic RGB to YPbPr converter counts).

On a separate note, I've been meaning to get a GBS-8200 to use GBS-Control on, but the only problem is that it doesn't support composite or S-Video, which I need for my C64, and I can't find anything on converting CVBS or Y/C (specifically with a PAL color carrier in my case) to RGB or YPbPr.

And on an even more separate note, does anyone know of any universal RGB to YPbPr converter that is actually good? I was thinking of picking up some Genesis HD Retrovision cables and attempting to adapt that for my Crosspoint output, but I have doubts that it will work correctly. Right now, I'm using a CSY-2100 clone, and the colors are terrible (even after I spent countless hours calibrating it).
I think the KD-VA5 is supposed to be pretty damn good

ETA: according to some posts I found on neo-geo it does in fact support 480i and 240p even though the only 15kHz resolution listed on the box is 576i
Last edited by maxtherabbit on Thu Jun 07, 2018 11:36 pm, edited 1 time in total.
User avatar
Syntax
Posts: 1774
Joined: Wed Aug 09, 2017 12:10 am
Location: Australia

Re: A fix for SNES jailbars

Post by Syntax »

syboxez wrote:
rama wrote:It's easy to see these old console's minor jailbars on a GBS-8200 scaler. This just has an LPF at 30Mhz, so stock systems will always show some jailbars.
If you turn on edge enhancement and contrast enhancing features, it becomes easily visible.
For example, I can see a 1Chip SNES calculating / shovelling around data :p

And yea, my go-to solution for these things is adding SMD caps of pretty high values (10 or 22uF work best) to the already existing caps.
SNES definitely needs one on the work RAM, and should have one per PPU.
This brings down the jailbars significantly.

The only console where this doesn't work or at least doesn't do much: Sega Genesis. The 5V and ground lanes are so contaminated on these, nothing seems to help. The source of much of it is definitely the VRAM bus and chips, but it'll be hard to avoid these since we need to get the RGB out of the region somehow.
Interestingly enough, my Sega Genesis (Model 1 VA6 w/ TMSS) doesn't have any noticeable jailbars on my CRT. Due to the PCB routing being a major issue, I wonder if copper tape would help like it does with the GBS-8200 (I don't have one, but I've heard reports that it does help). Copper tape would basically be the same as adding a third layer to the PCB that is only a ground plane. The Sega Genesis jailbar issue is a seperate issue caused by bad PCB layout, so it makes sense why adding more capacitance to the IC decoupling caps wouldn't fix anything.

Also I did more experimentation. I replaced the THS7314 in my Mini with a THS7374 with disabled LPF, and I do notice very mild jailbars (separate from the ones caused by low value decoupling caps), so a mix between high capacity low ESR ceramic capacitors for decoupling and an LPF should result in basically perfect video. I do still notice these "other" jailbars on my SHVC-CPU-01 more than on my Mini, so I may try to RGB bypass that with borti's PCB and enable the LPF when I get the chance. Note that all of my testing is on a consumer CRT, with nothing digital in the chain (unless an Extron Crosspoint and generic RGB to YPbPr converter counts).

On a separate note, I've been meaning to get a GBS-8200 to use GBS-Control on, but the only problem is that it doesn't support composite or S-Video, which I need for my C64, and I can't find anything on converting CVBS or Y/C (specifically with a PAL color carrier in my case) to RGB or YPbPr.

And on an even more separate note, does anyone know of any universal RGB to YPbPr converter that is actually good? I was thinking of picking up some Genesis HD Retrovision cables and attempting to adapt that for my Crosspoint output, but I have doubts that it will work correctly. Right now, I'm using a CSY-2100 clone, and the colors are terrible (even after I spent countless hours calibrating it).
Got jailbars on my SMD 2 and it has a 7374 RGB bypass with RGB coming from lifted pins on the gpu via fully sheilded RCA cable. Copper tape on the RGB bypass board. New 100N caps on the mobo for decoupling. Orginal CXA encoder completly removed. Seperate LDO used for the RGB bypass board. Fully recapped.

Ill revisit it soon but im working on this bloody SSDS3 diagonal noise at the moment.
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: A fix for SNES jailbars

Post by rama »

Yeah, I tried copper shielding (tape) on my Genesis. It does work .. a little, I guess.
Definitely still looking for that one fix though ;p

The gbs8200 memory bus is under software control. Copper shielding will do something, but it's far easier to just adjust memory timings a little for the same effect :)
User avatar
maxtherabbit
Posts: 1763
Joined: Mon Mar 05, 2018 4:03 pm

Re: A fix for SNES jailbars

Post by maxtherabbit »

rama wrote:It's easy to see these old console's minor jailbars on a GBS-8200 scaler. This just has an LPF at 30Mhz, so stock systems will always show some jailbars.
If you turn on edge enhancement and contrast enhancing features, it becomes easily visible.
For example, I can see a 1Chip SNES calculating / shovelling around data :p

And yea, my go-to solution for these things is adding SMD caps of pretty high values (10 or 22uF work best) to the already existing caps.
SNES definitely needs one on the work RAM, and should have one per PPU.
This brings down the jailbars significantly.

The only console where this doesn't work or at least doesn't do much: Sega Genesis. The 5V and ground lanes are so contaminated on these, nothing seems to help. The source of much of it is definitely the VRAM bus and chips, but it'll be hard to avoid these since we need to get the RGB out of the region somehow.
I'm going to give this a shot (stacking them on the existing bypass caps)

What's the correct size for the full size SNES, 0805?
User avatar
FBX
Posts: 2347
Joined: Wed Feb 18, 2015 10:18 am
Location: DFW area, Texas
Contact:

Re: A fix for SNES jailbars

Post by FBX »

maxtherabbit wrote: What's the correct size for the full size SNES, 0805?
For surface mount ceramics, yes. The SNES Junior uses 0603.
fernan1234
Posts: 2175
Joined: Mon Aug 14, 2017 8:34 pm

Re: A fix for SNES jailbars

Post by fernan1234 »

syboxez wrote:
For those that would like to test, the issue seems to be most noticeable on the intro to Final Fantasy VI or in Chrono Trigger's indoor void (generally the same places that the white vertical line is visible).
It looks like running these games from an SD2SNES also eliminates or at least greatly diminishes the appearance of both jailbars and the white lines in the middle of the screen in scenes like these. I forget the technical reasons behind this, but it's worth considering as an alternative or complement to this mod.
User avatar
maxtherabbit
Posts: 1763
Joined: Mon Mar 05, 2018 4:03 pm

Re: A fix for SNES jailbars

Post by maxtherabbit »

alright - got 10uf X7R ceramics stacked upon C90-95

this covers work RAM, both VRAMs, both PPUs, and the CPU

super easy to install, let's hope it does something
User avatar
maxtherabbit
Posts: 1763
Joined: Mon Mar 05, 2018 4:03 pm

Re: A fix for SNES jailbars

Post by maxtherabbit »

annd we're good :D
diagonal noise GONE
jailbars in SMW night screens GONE

when the OSSC samples the SNES in generic 4:3 mode there are still some faint "jailbars" but they are 100% not present when optimal sampling is enabled - I think these are just sampling artifacts from the sample rate being wrong for the SNES and not "jailbars" at all
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: A fix for SNES jailbars

Post by rama »

Yep, I can see similar artefacts on my scaler, when the sampling rate is off or some other ADC parameters don't match well.
The extra capacitors decrease noise on Vcc and this shows when scoping the RGB video lines :)
DJ Kevgeez
Posts: 41
Joined: Tue Mar 19, 2019 6:42 am

Re: A fix for SNES jailbars

Post by DJ Kevgeez »

Is anyone dealing with Horizontal Jail bars?
I notice it most when the screen is black.
I guess its always been there but once you notice them you cant unsee them.
I'm pissed off because i've spent all this money on cables etc.

I'm using a scart cable from the UK and a auto bandridge switcher until my Gscartw arrives in the mail.
RGB0b
Posts: 543
Joined: Wed Dec 05, 2012 1:52 pm

Re: A fix for SNES jailbars

Post by RGB0b »

DJ Kevgeez wrote:Is anyone dealing with Horizontal Jail bars?
I notice it most when the screen is black.
I guess its always been there but once you notice them you cant unsee them.
I'm pissed off because i've spent all this money on cables etc.

I'm using a scart cable from the UK and a auto bandridge switcher until my Gscartw arrives in the mail.
My guess would be the PSU, or the Bandridge switch.
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: A fix for SNES jailbars

Post by rama »

Or maybe even the TV?

A horizontal bar is unusual. Is it just with this SNES?
User avatar
Kez
Posts: 818
Joined: Thu Jul 20, 2017 7:09 am

Re: A fix for SNES jailbars

Post by Kez »

Do you have a picture of the issue?
DJ Kevgeez
Posts: 41
Joined: Tue Mar 19, 2019 6:42 am

Re: A fix for SNES jailbars

Post by DJ Kevgeez »

rama wrote:Or maybe even the TV?

A horizontal bar is unusual. Is it just with this SNES?
Yes its only with SNES gaming.
Its not one bar. Its Jailbars but horizontal.
The PS2, NES, Genesis, Dreamcast, XBOX and Gamecube show up crystal clear.
All are scart except the Gamecube's Component cables.
Displayed on a Trinitron 36".

I just ordered a 1-chip SNES and a new SCART cable from RGBs american seller on ebay.
I'm hoping its just the SNES.
copy
Posts: 232
Joined: Fri Apr 29, 2016 10:38 pm

Re: A fix for SNES jailbars

Post by copy »

Thanks syboxez for this thread. I did a Voultar RGB bypass on my 1CHIP a while ago, and immediately saw these jailbars. Glad to know there's a fix.

I also have a Mini that I'd like to try the fix on, but I'm having trouble finding X7R 0603 caps. Would X6S be suitable?
copy
Posts: 232
Joined: Fri Apr 29, 2016 10:38 pm

Re: A fix for SNES jailbars

Post by copy »

I decided to order these X5R caps for my Mini. I hope they are okay:

https://www.digikey.com/product-detail/ ... ND/5251372
User avatar
RIP-Felix
Posts: 140
Joined: Sun Mar 31, 2019 7:54 pm

Re: A fix for SNES jailbars

Post by RIP-Felix »

I assume C90-95 are on the SNES Jr. board revision? Which cap numbers I should stack or replace on an SNS-CPU-1chip-01?

Also, does anyone know of a source for hi resolution pics of the various board revisions (Front and reverse)? I can't find anything that allows me to read the silk screening, and reverse images where these caps are located are even harder to find pics for.
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: A fix for SNES jailbars

Post by rama »

copy
Posts: 232
Joined: Fri Apr 29, 2016 10:38 pm

Re: A fix for SNES jailbars

Post by copy »

RIP-Felix wrote:I assume C90-95 are on the SNES Jr. board revision? Which cap numbers I should stack or replace on an SNS-CPU-1chip-01?
Judging by this schematic, I believe C90-95 are the cap numbers for pre-1CHIP boards.

As for the 1CHIP and Mini, I would also be eternally grateful if someone could list the target cap numbers for us amateurs. :)
copy
Posts: 232
Joined: Fri Apr 29, 2016 10:38 pm

Re: A fix for SNES jailbars

Post by copy »

Okay, on my 1CHIP-02, I believe these are the decoupling caps that need to be replaced:

Code: Select all

 C5 -- U1: S-CPUN A
C57 -- U3: WRAM
C63 -- U5: VRAM
C64 -- U4: VRAM
Correct?
User avatar
maxtherabbit
Posts: 1763
Joined: Mon Mar 05, 2018 4:03 pm

Re: A fix for SNES jailbars

Post by maxtherabbit »

copy wrote:Okay, on my 1CHIP-02, I believe these are the decoupling caps that need to be replaced:

Code: Select all

 C5 -- U1: S-CPUN A
C57 -- U3: WRAM
C63 -- U5: VRAM
C64 -- U4: VRAM
Correct?
Don't replace them just stack the higher value MLCCs on top
copy
Posts: 232
Joined: Fri Apr 29, 2016 10:38 pm

Re: A fix for SNES jailbars

Post by copy »

Did I find the right 1CHIP caps? They all seem obvious except for the CPU, which has several caps nearby.
maxtherabbit wrote:Don't replace them just stack the higher value MLCCs on top
Too late. I have already replaced them on my SHVC-CPU-01, and have started on my 1CHIP-02.

Syboxez did say to replace them. Is there any problem caused by replacing instead of stacking?
User avatar
maxtherabbit
Posts: 1763
Joined: Mon Mar 05, 2018 4:03 pm

Re: A fix for SNES jailbars

Post by maxtherabbit »

copy wrote:Did I find the right 1CHIP caps? They all seem obvious except for the CPU, which has several caps nearby.
maxtherabbit wrote:Don't replace them just stack the higher value MLCCs on top
Too late. I have already replaced them on my SHVC-CPU-01, and have started on my 1CHIP-02.

Syboxez did say to replace them. Is there any problem caused by replacing instead of stacking?
not especially, although in theory having both caps in parallel offers superior decoupling performance
paulb_nl
Posts: 340
Joined: Sat Feb 20, 2016 5:05 pm

Re: A fix for SNES jailbars

Post by paulb_nl »

maxtherabbit wrote:annd we're good :D
diagonal noise GONE
jailbars in SMW night screens GONE
Which board revision is this? Under which circumstances did you see the diagonal noise?
Post Reply