THS7374 vs THS7314 + 1 chip brightness fixing

The place for all discussion on gaming hardware
borti4938

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by borti4938 »

With the log disabled you these 'jailbars' might be not jailbars; they might be aliasing effects occurring.

EDIT: "log" = LPF -> auto correction of my "smartphone"
Last edited by borti4938 on Mon Jul 03, 2017 5:56 am, edited 1 time in total.
User avatar
FBX
Posts: 2347
Joined: Wed Feb 18, 2015 10:18 am
Location: DFW area, Texas
Contact:

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by FBX »

borti4938 wrote:With the log disabled you these 'jailbars' might be not jailbars; they might be aliasing effects occurring.
"Log" ?

All I know is this:

LPF on = blurry pixels.

LPF off = Sharp pixels with jail bars and/or aliasing effects.

What's the solution?
User avatar
FBX
Posts: 2347
Joined: Wed Feb 18, 2015 10:18 am
Location: DFW area, Texas
Contact:

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by FBX »

Update:

The 7374 vertical lines appear to be caused by the Framemeister itself, just as in the case of the Saturn Jail bars. In both instances, the vertical lines shift left and right independent of the pixels as you adjust SYNC_LEVEL. A compromising solution is to lower H_SCALER to a value of 1 and it will reduce the visibility of the lines, but as a slight cost of sharpness.

.
leonk
Posts: 1080
Joined: Sun Mar 13, 2011 9:29 pm
Location: Toronto, Canada

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by leonk »

I gave a friend the 7374 mod chip (borti design) to try it.

He too noticed that with LPF on, the image is more "blurry" but with LPF off, feeding into XRGB with latest firmware, FBX 5X profile, picture mode. He has no alias noise on his brand new high end samsung 4K TV. In fact, he's sending me his gscartsw to disable the lpf because he can see the blurring the switch is doing!

Very strange. You can actually can see the results of using the 7374 with no lpf in his youtube video:

https://youtu.be/oMEjQh-nHG0
User avatar
FBX
Posts: 2347
Joined: Wed Feb 18, 2015 10:18 am
Location: DFW area, Texas
Contact:

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by FBX »

leonk wrote:I gave a friend the 7374 mod chip (borti design) to try it.

He too noticed that with LPF on, the image is more "blurry" but with LPF off, feeding into XRGB with latest firmware, FBX 5X profile, picture mode. He has no alias noise on his brand new high end samsung 4K TV. In fact, he's sending me his gscartsw to disable the lpf because he can see the blurring the switch is doing!

Very strange. You can actually can see the results of using the 7374 with no lpf in his youtube video:

https://youtu.be/oMEjQh-nHG0
Must be the difference in manufacturing of each Framemeister then? It can't be the display, because both my Sony display and capture card reveal the same vertical lines.

And to be clear, I don't believe this issue is 'aliasing' because the lines can be shifted with Sync_Level. They seem to be independent of the graphics themselves.

.
leonk
Posts: 1080
Joined: Sun Mar 13, 2011 9:29 pm
Location: Toronto, Canada

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by leonk »

I'm not sure.

I tested with 2 XRGB's here on my TV (1080p Samsung) and I can see the "jailbars". Turning the LPF on amp or switching to GAME1 mode on XRGB gets rid of the noise (hence why I always believed it to be alias noise)

The only thing I can think it could be is the TV!! Its own 1080p to 4K scalar is removing the jailbars somehow??
User avatar
FBX
Posts: 2347
Joined: Wed Feb 18, 2015 10:18 am
Location: DFW area, Texas
Contact:

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by FBX »

leonk wrote:I'm not sure.

Turning the LPF on amp or switching to GAME1 mode on XRGB gets rid of the noise (hence why I always believed it to be alias noise)
On mine, NONE of the image modes remove the lines. In fact, "Picture" seems to have the least visible, though still visible from the pic I posted. The only way I could make the lines less visible was turning H_Scaler down to 1.
leonk
Posts: 1080
Joined: Sun Mar 13, 2011 9:29 pm
Location: Toronto, Canada

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by leonk »

hmmm. maybe you have a different problem. SCART cable choice or not using csync?

I had a customer who had jailbars using csync rgb cables from that lady on ebay. He complained to her. She sent him a new cable. All jailbars gone.
User avatar
FBX
Posts: 2347
Joined: Wed Feb 18, 2015 10:18 am
Location: DFW area, Texas
Contact:

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by FBX »

leonk wrote:hmmm. maybe you have a different problem. SCART cable choice or not using csync?

I had a customer who had jailbars using csync rgb cables from that lady on ebay. He complained to her. She sent him a new cable. All jailbars gone.
I've got a few different cables to try, I will check them and report back.
User avatar
FBX
Posts: 2347
Joined: Wed Feb 18, 2015 10:18 am
Location: DFW area, Texas
Contact:

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by FBX »

Okay so I opened one of the SNES SCART cables I got from retro_console_accessories and removed all of the caps from the RGB line (I don't believe these are needed with Borti's board?), but I also noticed some strange wiring maybe you guys can explain to me what's going on with:

There's a mysterious blue wire connected to pin #8 that is then 'bridged' to pin #16 with some sort of shrink-wrapped resistor (the shape reminds me a of a resistor anyway). Pin #18 is paired with this particular bundle as the grounding line. Pin #20 is of course the csync line, with pin #17 as the grounding line. Online Googling showed me that pin #8 is the voltage switch line, and pin #16 is 'RGB Blanking'. I'm guessing the wiring to these was some sort of failsafe in case someone hooked the cable into a JP21 adapter?

I desoldered the blue wire, the resistor 'bridge', and their grounding line from pin #18. I then hooked the cable back up to the Framemeister, and I want to say the vertical black lines look fainter, but I can still make them out if I stare hard enough or walk up to my TV screen and look closely.
borti4938

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by borti4938 »

FBX wrote:And to be clear, I don't believe this issue is 'aliasing' because the lines can be shifted with Sync_Level. They seem to be independent of the graphics themselves.
With changing the sync level you automatically shift the sampling point, which means a phase shift in the frequency domain. This automatically changes the aliasing overlapping area in its phase and also the position of the vertical lines. This is also indepent of the graphics. Still it can be both.

However, your last post point out that the effect are jailbars with quite high probability.
FBX wrote:Okay so I opened one of the SNES SCART cables I got from retro_console_accessories and removed all of the caps from the RGB line (I don't believe these are needed with Borti's board?)
Optional.
FBX wrote:There's a mysterious blue wire connected to pin #8 that is then 'bridged' to pin #16 with some sort of shrink-wrapped resistor (the shape reminds me a of a resistor anyway). Pin #18 is paired with this particular bundle as the grounding line. Pin #20 is of course the csync line, with pin #17 as the grounding line. Online Googling showed me that pin #8 is the voltage switch line, and pin #16 is 'RGB Blanking'. I'm guessing the wiring to these was some sort of failsafe in case someone hooked the cable into a JP21 adapter?
Pin #8 is initially used to tell the receiving device to switch over to AV (level above 2V) and whether the picture is 16:9 (level between 4.5V and 7V) or a 4:3 picture (9.5V to 12V). I don't have a framemeister but my guess is that you need the connection there. I don't see any reason why the Framemeister should use that.
Spoiler
Also in the popular cable schematics the connection to pin 8 is not correct in my point of view for both - NTSC and PAL cable. For NTSC cable its connected to +5V and for PAL to +12V which is always on if the PSU is active (also with console switched off). Personally I suggest to use a step-up-converter from 5V to 12V if you really need that.
Pin 16 is the blanking level. It's initially used to tell a TV with voltage levels between 1V to 3V that the SCART connector transmits RGB and not composite video. Because you have 5V in the console you need the resistor here to drop it down to an appropriate level.

Pin 18 is the reference ground for pin 8 and 16.



Just to get some more ideas:

Is you cable full shielded, i.e. every single line? If yes, is there a wire used for a ground connection or only the shields?
Normally one should use a wire for ground connection. I've read somewhere that the shield should only be connected to a single end - either console- or SCART-end. (Although I also have my shields connected to both ends)

Is pin 21 of your SCART connector connected to a ground wire?
User avatar
FBX
Posts: 2347
Joined: Wed Feb 18, 2015 10:18 am
Location: DFW area, Texas
Contact:

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by FBX »

borti4938 wrote:
Pin #8 is initially used to tell the receiving device to switch over to AV (level above 2V) and whether the picture is 16:9 (level between 4.5V and 7V) or a 4:3 picture (9.5V to 12V). I don't have a framemeister but my guess is that you need the connection there. I don't see any reason why the Framemeister should use that.

Pin 16 is the blanking level. It's initially used to tell a TV with voltage levels between 1V to 3V that the SCART connector transmits RGB and not composite video. Because you have 5V in the console you need the resistor here to drop it down to an appropriate level.

Pin 18 is the reference ground for pin 8 and 16.
With that all being the case, it seems these pins are effective useless being hooked up to the Framemeister from my NTSC Super NES, which explains why it worked perfectly fine without them when I desoldered them from the pins. You only need R,G,B, Csync, AL, AR, and grounding lines, as that's all I had feeding into the framemeister at that point.


Just to get some more ideas:

Is you cable full shielded, i.e. every single line? If yes, is there a wire used for a ground connection or only the shields?
Normally one should use a wire for ground connection. I've read somewhere that the shield should only be connected to a single end - either console- or SCART-end. (Although I also have my shields connected to both ends)

Is pin 21 of your SCART connector connected to a ground wire?
All cables are fully shielded, there's a ground wire paired for each bundle (itself also shielded), and there's a shielded ground wire hooked to Pin #21. When I hook a different SCART cable to the same SNES, the vertical lines are shifted to a different position on the screen.

At any rate, none of it looks as clean as the unmodded 1CHIP-03 I have, so I'm just going to leave the RGB unmodded in that one and add in your TOSLINK digital audio mod. I've already ordered it and it's on the way. I've got one of those installed in my SNES mini, and it sounds fantastic!
copy
Posts: 232
Joined: Fri Apr 29, 2016 10:38 pm

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by copy »

FBX, your overall experience with the THS7374 mod sounds similar to mine, as I posted a couple pages back. I had a totally clean picture with a 1CHIP-02, but immediately saw large dark bars once I installed the THS7374 bypass.

Perhaps the THS7374 just doesn't pair well with the Framemeister (or at least some units)? I've gotta say I was pretty discouraged at the result, and regretted doing the mod.
User avatar
FBX
Posts: 2347
Joined: Wed Feb 18, 2015 10:18 am
Location: DFW area, Texas
Contact:

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by FBX »

copy wrote:FBX, your overall experience with the THS7374 mod sounds similar to mine, as I posted a couple pages back. I had a totally clean picture with a 1CHIP-02, but immediately saw large dark bars once I installed the THS7374 bypass.

Perhaps the THS7374 just doesn't pair well with the Framemeister (or at least some units)? I've gotta say I was pretty discouraged at the result, and regretted doing the mod.
Yeah it's unfortunate. However, with the recent revelations about the RGB cables from ebay being way out of spec for these consoles, I'm optimistic that might help. I've got my resistors and cap kits coming in this evening from Amazon, and I will be testing each and every cable I own to make sure all signals are within industry standards (which right now, I know a couple of them are not based on Voultar's video).
leonk
Posts: 1080
Joined: Sun Mar 13, 2011 9:29 pm
Location: Toronto, Canada

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by leonk »

my experience with THS7374 vs THS7314 is a tad different.

Using Borti's design and LPF on, technically the picture quality should be the same. but it's not!! I used to do the simple DIY 7314 mod but on snes mini it always generated a faint shadow to the right of big color changes. On xrgb, it made it unplayable.

With 7374 and LPF, the shadow is gone!
Ikaruga11
Posts: 1454
Joined: Thu Apr 07, 2016 1:32 pm

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by Ikaruga11 »

wait

Is the THS7374 the same thing that's in the N64 RGB mods?
User avatar
FBX
Posts: 2347
Joined: Wed Feb 18, 2015 10:18 am
Location: DFW area, Texas
Contact:

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by FBX »

FBX wrote:
copy wrote:FBX, your overall experience with the THS7374 mod sounds similar to mine, as I posted a couple pages back. I had a totally clean picture with a 1CHIP-02, but immediately saw large dark bars once I installed the THS7374 bypass.

Perhaps the THS7374 just doesn't pair well with the Framemeister (or at least some units)? I've gotta say I was pretty discouraged at the result, and regretted doing the mod.
Yeah it's unfortunate. However, with the recent revelations about the RGB cables from ebay being way out of spec for these consoles, I'm optimistic that might help. I've got my resistors and cap kits coming in this evening from Amazon, and I will be testing each and every cable I own to make sure all signals are within industry standards (which right now, I know a couple of them are not based on Voultar's video).

Just got done attenuating the Csync line in my RGB cables, and unfortunately it made no difference on the dark vertical stripes. Switching back and forth between my unmodded 1CHIP-03 and the Mini with the 7374 Borti board, the black lines are non existent on the 1CHIP-03, whilst plainly visible on the mini. This is a big sad face. :-(

So the next step would be to compare with Voultar's board. If the black lines show up there too, then there's something the Framemeister simply does not like about the 7374 with LPF disabled.


Edit: Reviewing Voultar's earlier posts & pics, one would think the issue is the Framemeister is picking up on the unfiltered frequencies. But why then does the unmodded 1CHIP-03 look crystal clear on the Framemeister? Does the SNES 1CHIP-03 have some sort of LFP onboard for RGB output?


.
RGB0b
Posts: 543
Joined: Wed Dec 05, 2012 1:52 pm

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by RGB0b »

I'm really starting to think that many of the issues we're seeing are 100% Framemeister-related. I've tested and screenshotted more SNES' then I ever thought I'd see in a lifetime and the same few things always stand true:

- Different processors definitely processes the images differently from others...and the FM is far from perfect.

- SNES 1CHIP's do not have a LFP in them. If you take the "raw" signal into a capture card...or through an OSSC with the LPF turned off into a capture card, you'll get aliasing noise. Same with Mini's that have RGB enabled through their own S-RGB chip.

- People with Framemeister's, or people who go direct into capture cards should either turn their 7374 LFP on, or make sure there's another one in the chain. Everyone else should leave it off and use the LFP on another device. As an FYI, the LFP on the OSSC seems to work a bit better then the one built into the 7374.

FBX - The 7374 chip in your mini in one that came from a hand-assembled prototype batch, months before Borti's final design was complete. For gaming and making your profiles, it's perfect. Truthfully, I don't know if it's good enough to be doing zoomed-in comparison shots. As as FYI, that is NOT a statement against Borti's board, it's me doubting my assembly and installation of the board. I'll email you and get you a new one to use.
leonk
Posts: 1080
Joined: Sun Mar 13, 2011 9:29 pm
Location: Toronto, Canada

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by leonk »

Unlike OSSC, the XRGB mini performance varies greatly depending on what firmware you are on.

I know many people that use xrgb with 0 issues (including myself). If anything, for the 2 weeks I had the OSSC, I found the XRGB mini to be a more mature product, better compatibility and overall more configurable. I personally like the frame buffer that allows me to fill the screen vertically. I paid extra for a 55" TV darn it, why can't I use all of it. ;)

I also found the OSSC LPF to be too temperamental. Either not strong enough or too strong. Using the LPF of a TI THS series RGB amp resulted in more pleasing output on my 7300 series Samsung 1080p TV.

Your experience will vary based on your TV.
User avatar
FBX
Posts: 2347
Joined: Wed Feb 18, 2015 10:18 am
Location: DFW area, Texas
Contact:

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by FBX »

retrorgb wrote:
- SNES 1CHIP's do not have a LFP in them. If you take the "raw" signal into a capture card...or through an OSSC with the LPF turned off into a capture card, you'll get aliasing noise. Same with Mini's that have RGB enabled through their own S-RGB chip.
Begs the question: Why does my 1CHIP-03 look crystal clear on my Framemeister, when it should show aliasing like the modded mini is doing?
leonk
Posts: 1080
Joined: Sun Mar 13, 2011 9:29 pm
Location: Toronto, Canada

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by leonk »

FBX wrote:
retrorgb wrote:
- SNES 1CHIP's do not have a LFP in them. If you take the "raw" signal into a capture card...or through an OSSC with the LPF turned off into a capture card, you'll get aliasing noise. Same with Mini's that have RGB enabled through their own S-RGB chip.
Begs the question: Why does my 1CHIP-03 look crystal clear on my Framemeister, when it should show aliasing like the modded mini is doing?
VERY GOOD QUESTION!!

A friend with 1CHIP-03 has LPF off and no alias noise directly into xrgb mini

My SNES mini has horrible alias noise with LPF off!! So we are experiencing the exact same results!
borti4938

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by borti4938 »

I doubt that there is no LPF between in- and outputs of R, G and B implemented in the S-RGB.
But I've never seen a measurement so far.
User avatar
Voultar
Posts: 550
Joined: Fri Jan 10, 2014 8:29 pm
Location: USA

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by Voultar »

borti4938 wrote:I doubt that there is no LPF between in- and outputs of R, G and B implemented in the S-RGB.
But I've never seen a measurement so far.
Video Encoders from that era typically never employ any low order filtering. Other than a band-pass for the chroma. Analog devices don't discriminate against that stuff, for the most part. The S-RGB doesn't do have that, either.

Just a guess. But I'm betting that throughout all of the Firmware revisions, Micomsoft has simply manipulated the low order filter in different ways. Some F/W allowing more high-frequency content to fold back in. Others, to a much lesser degree. That would certainly explain the radical differences between firmware revisions.

I really don't like that thing..

For clarity, here's a bone stock 1CHIP-03 being piped into a device which lacks any kind of 9.5MHz LPF suitable for SD content.

Image
User avatar
FBX
Posts: 2347
Joined: Wed Feb 18, 2015 10:18 am
Location: DFW area, Texas
Contact:

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by FBX »

Voultar wrote: For clarity, here's a bone stock 1CHIP-03 being piped into a device which lacks any kind of 9.5MHz LPF suitable for SD content.
What doesn't make sense to me is why my 1CHIP-03 looks crystal clear on my Framemeister. It's supposed to have aliasing lines, but it doesn't. The only modification I have done this far on the 1CHIP-03 is the csync jumper wire fix to restore csync.

Then I plug in the modded mini and it does have aliasing lines (they appear as thin black vertical stripes and are sort of phased in and out across the screen). It's not like I'm changing firmwares either. It's the current firmware for the Framemeister on both tests.
leonk
Posts: 1080
Joined: Sun Mar 13, 2011 9:29 pm
Location: Toronto, Canada

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by leonk »

the csync 1 wire restore on the -03 is not a good idea. The csync will run out of spec resulting in the xrgb loosing sync as often as every few minutes.

Compare to 1CHIP -01/-02 to see what components are missing or do proper 7374 bypass amp install that will correctly amplify csync to the correct level.
User avatar
FBX
Posts: 2347
Joined: Wed Feb 18, 2015 10:18 am
Location: DFW area, Texas
Contact:

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by FBX »

leonk wrote:the csync 1 wire restore on the -03 is not a good idea. The csync will run out of spec resulting in the xrgb loosing sync as often as every few minutes.
I attenuated the csync line from the cable end. Besides that, it never dropped out even once in all my use of it.
borti4938

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by borti4938 »

What I don't understand is that most people wants to have CSYNC as TTL compatible and design / modify their cables (in the SCART plug) such that it is compatible to 75ohm terminated sync. Why not outputting 75ohm compatible CSYNC from the console and using simply a wire without any components in the wire as it should be. There are two advantages:

- You can achive line matching to the cable and have NO reflections due to the impulses other than with TTL sync.
- Your console does not transmit any current through the wire while being off as the impulses by specifications are -300mV impulses.

@FBX: have you ever tried to uses a CSYNC signal as described above.

I borrow me a Framemeister from a friend of mine and I corious to see if my consoles also have the vertical lines on it (SNESes modded with THS7374 and outputting standard CSYNC (0V downto -300mV impulses if 75ohm terminated)).
I'm also willing to send aborad one of my consoles to FBX although it might be somehow expensive to send the SNES from Germany to USA).

@Voultar: Thank you for the picture :) So I have to invert my assumption. However, I will measure the spectrums of a random picture content quite soon to get clarity ;) I guess it might be interesting though.
paulb_nl
Posts: 340
Joined: Sat Feb 20, 2016 5:05 pm

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by paulb_nl »

borti4938 wrote:What I don't understand is that most people wants to have CSYNC as TTL compatible and design / modify their cables (in the SCART plug) such that it is compatible to 75ohm terminated sync. Why not outputting 75ohm compatible CSYNC from the console and using simply a wire without any components in the wire as it should be.
Because stock consoles output TTL sync and not everyone wants to mod their console. Also you don't want to (accidentally) connect a CSYNC cable without components in it to a stock console.
borti4938

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by borti4938 »

The questions goes to people who perform mods on their consoles and or modify their cables or give everything away for have it modificated anyway.
If you use a stock console and a 75ohm sync termination one should use composite video or luma in my point of view. The sync information in there is the same and with a good cable composite video / luma part of the signal has very very minor influence on the R, G, B lines.
RGB0b
Posts: 543
Joined: Wed Dec 05, 2012 1:52 pm

Re: THS7374 vs THS7314 + 1 chip brightness fixing

Post by RGB0b »

paulb_nl wrote:
borti4938 wrote:What I don't understand is that most people wants to have CSYNC as TTL compatible and design / modify their cables (in the SCART plug) such that it is compatible to 75ohm terminated sync. Why not outputting 75ohm compatible CSYNC from the console and using simply a wire without any components in the wire as it should be.
Because stock consoles output TTL sync and not everyone wants to mod their console. Also you don't want to (accidentally) connect a CSYNC cable without components in it to a stock console.
Right, exactly. If I were going to use a custom cable I built (DSUB, BNC, etc), I'd absolutely set the board to output 75 ohm sync from the console. If I'm using a SNES SCART cable, I'm going to assume that at some point, someone will plug it into a different (or unmodified) SNES, causing issues. It's more of a better safe then sorry" attitude then anything else.
Post Reply