PC Engine RGB jailbars, let us find a working solution

[rtw]

I recently did an RGB mod to my IFU-30A using otaku's NEC RGB amplifier.

Overview. Only need to remove the three screws at the back



Cabling and placement. GND and 5V is taken from the 7805, audio taken from rightmost connector



Before closing



Side view



The output is fed into an XRGB-mini Framemeister.

To test everything I used Artemio Urbina's brilliant 240p test suite: http://junkerhq.net/xrgb/index.php/240p_test_suite.

This suite also has a sound test which will verify the left and right audio channel.

Now, please tell me what is the best way to check for banding ?

[Artemio]

I didn't include a test in the suite for banding, since I am not 100% sure on how to check for it across all mods and systems. I do have a white PCE with IFU-30, and the mod inside the PCE, a Turbo Grafx-16 and a PCE Duo. The TG-16 doesn't show banding for me. The other two do at varying degrees, all of them use the THS7414 amp right now.

I did make a small rom before making the suite that has the patterns that show banding in my setup: https://dl.dropboxusercontent.com/u/1403041/banding.pce

I didn't include these on the suite since I have no idea if these are consistent or the best way to test for it. For teh record, I use a PVM and they show there, I'd need to check on the Framemeister.

I have a low bandwidth scope, If I can help just let me know what to do from either software or measuring, my background is software so I am clueless as of what to look for in terms of noise.

[rtw]

I didn't include a test in the suite for banding, since I am not 100% sure on how to check for it across all mods and systems. I do have a white PCE with IFU-30, and the mod inside the PCE, a Turbo Grafx-16 and a PCE Duo. The TG-16 doesn't show banding for me. The other two do at varying degrees, all of them use the THS7414 amp right now.

I did make a small rom before making the suite that has the patterns that show banding in my setup: https://dl.dropboxusercontent.com/u/1403041/banding.pce

I didn't include these on the suite since I have no idea if these are consistent or the best way to test for it. For teh record, I use a PVM and they show there, I'd need to check on the Framemeister.

Thank you Artemio, I tested the banding.pce and I cannot see anything on my Framemeister.

I re-read this entire thread, are you saying that the IFU-30/IFU-30A amplifies the video signal from the EXT connector ?

[Artemio]

Thank you Artemio, I tested the banding.pce and I cannot see anything on my Framemeister.

Just connected my PCE+IFU-30 system to the framemeister, and it is indeed not visible in my setup. I also tested by PCE Duo, and the signal jumps a lot. I cannot verify if the banding is visible due to synch issues with my Duo and XRGB Mini. I can build a sync stripper if needed. I also tested both units again in my PVM, and the PCE+IFU is almost unnoticeable when compared to the Duo. This at least with my test cases.

I re-read this entire thread, are you saying that the IFU-30/IFU-30A amplifies the video signal from the EXT connector ?

What I found out is the following. When I modded my white PCE system I initially took the RGB lines from the EXT port, and it worked fine. However when I plugged it into the IFU-30, everything got way brighter. I unplugged it and tested again, and it had lower levels again. So I went ahead and tapped the RGB lines directly from the Hu6260, tested it with and without the IFU-30 and levels didn't vary.

Please let me know if there is something I can measure with the scope, I only have a 40 Mhz one.

[rtw]

What I found out is the following. When I modded my white PCE system I initially took the RGB lines from the EXT port, and it worked fine. However when I plugged it into the IFU-30, everything got way brighter. I unplugged it and tested again, and it had lower levels again. So I went ahead and tapped the RGB lines directly from the Hu6260, tested it with and without the IFU-30 and levels didn't vary.

Do you mean that you took the signal from the EXT port on the IFU-30 (same as I did) ?

Please let me know if there is something I can measure with the scope, I only have a 40 Mhz one.

To be honest I do not know what to look for so I can only suggest to look for differences. Use the first image in banding.pce as the visible test.

Make a small breakout SCART (male/female) cable where you can tap the RGB and Sync signal consistently.

Find two units, one which exhibits the problem and one without and start looking at the signals.

The problem is that you need accurate probes for these measurements.

[Fudoh]

I tested the banding.pce and I cannot see anything on my Framemeister.

what kind of PCE are you using ? A classic white one ?

I have two interface units (2nd revision), one with a white PCE, one with a Core Grafx II. The combo with the PCE gives me banding on the Framemeister, the CGII combo doesn't (or hardly). The two interface units have the same RGB mod and of course I exchanged the PCE systems as well...


@Bancho: I would never expect banding from a standalone CG or CGII.

[Artemio]

Do you mean that you took the signal from the EXT port on the IFU-30 (same as I did) ?

No, sorry. I took the signal from the EXT port inside the pc engine. I wanted the mod to be stand alone. When using the white pce unit alone vs with the IFU 30, there was a huge visual difference in video levels.

I'll compare the signals and post my results here. I only have the Tektroniks 1/10x probes that came with my TBS1042 scope.

[rtw]

what kind of PCE are you using ? A classic white one ?

Classic white one. My IFU is a IFU-30A. What game/program are you using to show the banding ?

Do you mean that you took the signal from the EXT port on the IFU-30 (same as I did) ?

No, sorry. I took the signal from the EXT port inside the pc engine. I wanted the mod to be stand alone. When using the white pce unit alone vs with the IFU 30, there was a huge visual difference in video levels.

Strange, according to the TG-16 schematics, the RGB & Sync signals go directly from the Hu6260 to the EXT connector. So the only difference would be a shorter signal path by a few centimeters...

I'll compare the signals and post my results here. I only have the Tektroniks 1/10x probes that came with my TBS1042 scope.

Brilliant, please put the probes at 10x to increase bandwidth.

[Fudoh]

Classic white one. My IFU is a IFU-30A.

mine too - both.

What game/program are you using to show the banding ?

always the very first level of New Adventure Island. Just start running and you'll immediately notice the banding on the light blue sky.

Any other fixes to your PCE ? RF unit removed/disabled ? Capacitor mod to the graphics chip to reduce banding ?

[rtw]

always the very first level of New Adventure Island. Just start running and you'll immediately notice the banding on the light blue sky.

Any other fixes to your PCE ? RF unit removed/disabled ? Capacitor mod to the graphics chip to reduce banding ?

Tested New Adventure Island. I do not see any banding or I do not see any obvious banding.

Tomorrow I will take my XCAPTURE-MINI and record the start of New Adventure Island and post it for your scrutiny.

No mod done to the PCE. Unit is PI-TG001

[Fudoh]

Amazing - really! I've been fighting this RGB mod for a decade now. With the white PCE I also get a lot of (video-) noise on the Framemeister.

When you say GND and 5V is taken from the 7805, what's that 7805 exactly ? On my mod GND and 5V are taken from the EXT port as well and I would certainly love to change that to your configuration.

[Artemio]

The 7805 is a voltage regulator: http://potentiallabs.com/cart/image/cac ... 00x416.jpg

That image shows how it looks like and the pins for 5v and ground

[Fudoh]

excellent! rtw, just waiting for your video and I'll give it a try after that

[ApolloBoy]

When you say GND and 5V is taken from the 7805, what's that 7805 exactly ?

All that technical knowledge and you still didn't know what a 7805 is? Wow.

[Pasky]

LoL.

The 7805 is the linear voltage regulator Fudoh. You can feed it a higher voltage and it will output 5V and burn off the excess as heat.

[Artemio]

Here are basic measurements with the scope, please orient me in order to get better information. All images are available in full size if you click teh link and then the image at the site.

================
PC Engine Duo images:
================

Banding:



Composite video as sync @ line 80

Notice that the signal is slightly curved, was about the only difference I noticed.

Close up of the same line:


FFT of the full line:

Of course this is still composite video, a 3.6 Mhz dominant signal

Composite video in yellow vs Green channel in blue:

Closeup of the same line green vs composite video:


=============
TG-16 photos
=============

Composite video as sync:


TG-16 Green vs composite:



I expected less noise in the TG-16, and it seems to be more... unless the noise eliminates teh banding, I don't understand these result.

[Pasky]

Where do you suspect the interference is coming from? The composite video signal as sync? Or is it the green color line? Does disconnecting R, G, or B remove the banding? Or are you talking about banding in the composite video signal???

[CkRtech]

Long time since commenting on this one - I have a scope + one unmodded Turbo Duo + one unmodded TG16. I am not going to be around that stuff for another week, but I would probably do the following if I started scoping/comparing -

Using a multi-channel scope comparison of both systems simultaneously running the same software, I would shoot the breeze by looking for... -

1: Noisy power (at regulator and possibly voltage inputs of Hu chip)
2: Disconnecting a single color to see if banding disappears (as Pasky mentioned).
3: Comparing output of composite video as sync vs. composite sync as sync.
4: Comparing color output of each individual signal of both systems running Artemio's suite. (Run it on both consoles and scope Red from each console to the same scope and look for noise)
5: Changes in the signal after adding filter caps where mentioned elsewhere (I think maybe I linked to some places earlier in this thread?)
6: Changes after adding filter caps to all power inputs to the Hu chip.

Personally, I am curious as to what the output of the raw sync signal is without any sort of amping. I think most people just use composite video, but I think that the xrgb and take anything from 0.7v/1v up to 5V for the sync pulse.

My list above is just a bunch of guesswork and is not meant to demonstrate that I know what I am talking about.

[RGB32E]

Personally, I am curious as to what the output of the raw sync signal is without any sort of amping. I think most people just use composite video, but I think that the xrgb and take anything from 0.7v/1v up to 5V for the sync pulse.

Tim's AV-DRIVER has a sync buffer specifically tailored for the Hu6260 CSYNC output. I've observed zero banding when using the AV-DRIVER using PCE CSYNC. Pac-Land works really well for checking banding artifacts, as it has a lot of bright color fields along with black outlines.

[rtw]

As promised here are two videos from my setup

https://mega.co.nz/#!VUk1CZZK!IqqCseiNt ... YSEfG6-nT4
https://mega.co.nz/#!xclD2KZb!jLawSmmkF ... HPDdOBAGJc

Now tell me if I have banding or not...

edit: changed host

[Fudoh]

2nd video stops loading at about 23mb... but what I can see looks pretty good.

You got a sync dropout though in the Adventure Island video. There are certain games that will be unplayable on your setup. Do you have Ys 1/2 ? Try running the intro. Your setup should go berserk. To cure this you'll have to buffer the sync signal.

[RGB32E]

It would help if you disabled scanlines when checking for horizontal banding... I didn't notice any egregious banding in your scanlined video. What was the signal chain?

[rtw]

It would help if you disabled scanlines when checking for horizontal banding... I didn't notice any egregious banding in your scanlined video. What was the signal chain?

Oops, I can redo the video without scanlines...

[rtw]

Changed the hosting to: MEGA

@Fudoh
. At which point in the movie does it glitch ?
. Sorry no Ys 1/2 CDs around here, anything I can try in the HuCard department ?

Which Framemeister mode should I use when redoing the video: STANDARD ?

[Pasky]

Running it through a frame meister to test for artifacts is a bad test in my opinion. It can filter out some anomalies and noise since it's re-processing the signal. I had a genesis that would produce jailbars on my television and through my PEXHDcap but through the XRGB it looked fine. If you can, try to run it straight to the capture card.

[rtw]

Running it through a frame meister to test for artifacts is a bad test in my opinion. It can filter out some anomalies and noise since it's re-processing the signal. I had a genesis that would produce jailbars on my television and through my PEXHDcap but through the XRGB it looked fine. If you can, try to run it straight to the capture card.

Sorry all CRTs are gone, only one left is in my cab. This is my future, and it does not look too bad

The capture cart is connected to the HDMI output of the Framemeister, the card is an: XCAPTURE-MINI

Fudoh stated that he got artifacts on his Framemeister so that is what we are comparing

[Fudoh]

New Adventure Island is my primary test for sync dropouts on HuCards, while Ys in on CDs. The dropout can be seen at 0:11 right before you hit the snail under the cherry extra item. You can try to adjust the sync level on the Mini to get more (or fewer) dropouts.

The intro of Winds/Lords of Thunder can cause dropouts as well, but it's less messy than Ys1/2.

For 240p content like this set the Framemeister to PICTURE instead. Minimal filtering and processing.

[CkRtech]

Tim's AV-DRIVER has a sync buffer specifically tailored for the Hu6260 CSYNC output. I've observed zero banding when using the AV-DRIVER using PCE CSYNC. Pac-Land works really well for checking banding artifacts, as it has a lot of bright color fields along with black outlines.

Yeah! Still looking forward to this one.

[rtw]

New Adventure Island is my primary test for sync dropouts on HuCards, while Ys in on CDs. The dropout can be seen at 0:11 right before you hit the snail under the cherry extra item. You can try to adjust the sync level on the Mini to get more (or fewer) dropouts.

The intro of Winds/Lords of Thunder can cause dropouts as well, but it's less messy than Ys1/2.

For 240p content like this set the Framemeister to PICTURE instead. Minimal filtering and processing.

Here are the videos in PICTURE mode:

https://mega.co.nz/#!5ZUlkYaI!vA7s-j9md ... 2TMokAcH1s
https://mega.co.nz/#!lYkyXYIA!Yvxy66wXp ... WMEVO54LjQ

I still want to know if there is any banding

With banding out of the way I can start to worry about sync dropouts!

[Fudoh]

No visible banding for me