New Koryuu test firmware 1.1-test0

[megari]

After a long wait, for which I do apologize, a test version of a new firmware version for the Koryuu has been released.

Please see the release notes and download the firmware images here: Koryuu test firmware 1.1-test0

[newsupermarkio]

Hi Megari,

I’m using the Koryuu with a Capcom CPS changer S-video output with Street Fighter 2 Turbo PCB.

The connection is:
Capcom CPS changer S-video -> Koryuu -> Component Video -> RetroTink Comp2RGB -> PVM

With the stock Koryuu firmware, I encountered a strange issue where after 5 minutes of operation, the video starts to flicker and eventually it loses all color.

A direct S-video connection into my PVM display is ok. The S-video input to a RetroTink 2x Pro to an HDMI display is ok too.
I have separately tested the RetroTink Comp2RGB -> PVM connection with a SNES -> HD retrovision cable -> RetroTink Comp2RGB -> PVM, and verified that works fine. Thus, I believe there may be an issue with how the Koryuu is handling the S-video output of the CPS Changer.

I suspect there may be some issue with the field reversal and updated the Koryuu firmware to version 1.1 to see if the issue goes away. The updated firmware did seem to help the issue, but after several minutes of operation, I could see the video start to flicker again under certain scenarios/screens. It seems that the issue gets cleared out between screen transitions, but certain video input patterns seem to trigger the flickering/color issue. The start up screen/attract mode with vertical scrolling seems to be problematic, as is Chun Li’s stage for some reason. I think the field reversal solution helps, but it seems there remains some residual issue that is still not fully resolved. Do you have any theories why this might be happening?

A side note on the firmware update procedure, as it took me a while to get this working with proper driver and programmer software:
1. I used the Olimex AVR-ISP-MK2 (https://www.olimex.com/Products/AVR/Programmers/AVR-ISP-MK2/open-source-hardware) with AVR-ICSP 10-to-6 cable (https://www.olimex.com/Products/AVR/Programmers/AVR-ICSP/)
2. I have a Mac with Windows 7 via Parallels
3. I had to manually install drivers as recommended by Olimex (https://www.olimex.com/Products/AVR/Programmers/AVR-ISP-MK2/resources/AVR-ISP-MK2.pdf, https://www.olimex.com/Products/AVR/Programmers/AVR-ISP-MK2/resources/DRIVER-MK2- AS-6-7-W10.zip)
4. I was not able to get Khazama AVR programmer (http://khazama.com/project/programmer/) to work. Instead I was able to use Atmel Studio 7 (https://www.microchip.com/mplab/avr-support/atmel-studio-7) to flash the Koryuu.

[megari]

Hi newsupermarkio,

I’m using the Koryuu with a Capcom CPS changer S-video output with Street Fighter 2 Turbo PCB.

The connection is:
Capcom CPS changer S-video -> Koryuu -> Component Video -> RetroTink Comp2RGB -> PVM

With the stock Koryuu firmware, I encountered a strange issue where after 5 minutes of operation, the video starts to flicker and eventually it loses all color.

Hmm, that’s odd. Losing color usually happens because lock on the chroma subcarrier is lost and/or the chroma signal becomes weak enough that the ADC chip’s “color kill” threshold is hit. This can be due to a variety of reasons.

Given that there is initially color in the output, but the colors start to flicker and the picture eventually falls back to grayscale, there may be some sort of a process that gradually degrades the chroma signal and/or subcarrier after lock on the sync signals has initially been established.

(CPS Changer apparently uses the CXA1645 video chip for its composite and S-Video outputs. This is the same chip as in some Sega Genesis/Mega Drive models, which are known to sometimes be problematic with the Koryuu, although S-Video modded units tend to have good S-Video output, in my experience. I wonder if there is a connection here…)

A direct S-video connection into my PVM display is ok. The S-video input to a RetroTink 2x Pro to an HDMI display is ok too.
I have separately tested the RetroTink Comp2RGB -> PVM connection with a SNES -> HD retrovision cable -> RetroTink Comp2RGB -> PVM, and verified that works fine. Thus, I believe there may be an issue with how the Koryuu is handling the S-video output of the CPS Changer.

Interesting. I wonder if having multiple converters/transcoders chained might contribute to the problem.

If I understand correctly, your PVM has S-Video and RGB inputs in some form or another. As the Comp2RGB seems to have a SCART output connector, there is probably an adapter to the BNC RGsB inputs on the PVM, correct?

I have an idea for exploring this a bit further. The Koryuu is technically capable of outputting an RGsB or RsGsBs signal in addition to YPbPr, so it should be possible to simplify the video chain a bit by connecting the Koryuu directly to the PVM (modulo RCA->BNC adapters, probably). Would you be willing to try a version of the current in-development firmware with RGsB output instead of YPbPr?

I would like to make sure I have the correct understanding of the issue. If you do not mind, could you confirm or correct my understanding, please?

With the current public test firmware (1.1-test0), is the process leading to the issue roughly as follows?

1. Initial sync is established: the picture is stable and has color.
2. Roughly 5 minutes passes, possibly with some game activity or attract mode on-screen.
3. The colors and/or picture start to flicker or glitch. (How? Wrong colors, image flickering between grayscale and color, image blanking out or its geometry distorting?)
4. Color is lost. The picture shown on-screen is now in grayscale. (Apart from missing color, is the picture otherwise stable?)

Another thought that came to my mind is whether it might be possible to try the composite output of your CPS changer with the Koryuu to see if it has a similar problem as with S-Video.

I suspect there may be some issue with the field reversal and updated the Koryuu firmware to version 1.1 to see if the issue goes away. The updated firmware did seem to help the issue, but after several minutes of operation, I could see the video start to flicker again under certain scenarios/screens. It seems that the issue gets cleared out between screen transitions, but certain video input patterns seem to trigger the flickering/color issue. The start up screen/attract mode with vertical scrolling seems to be problematic, as is Chun Li’s stage for some reason. I think the field reversal solution helps, but it seems there remains some residual issue that is still not fully resolved. Do you have any theories why this might be happening?

At present, I do not have any particularly strong hypothesis about the cause of the issue you are experiencing. It could be a variety of things. The gradual nature of the degradation of color may be a sign of an ongoing physical phenomenon, such as temperature changes in some component(s) affecting the chroma subcarrier somehow, but this requires more investigation to say anything for certain.

The field reversal issue is related to interlaced NTSC modes, which should cause visible, jarring artifacts in the picture (as if every set of two consecutive lines were swapped). Not sure if that can have anything to do with color issues.

Could you perhaps record a video of the issue? Just a smartphone-quality video should be plenty to get an initial understanding of the issue.

A side note on the firmware update procedure, as it took me a while to get this working with proper driver and programmer software:
–snip–

Yes, the Windows instructions in the junkerhq wiki are not very good at the moment when using the Olimex programmer (Khazama seems to be hardwired for a particular, different type of programmer). We are hoping to update the instructions soon. I have created a raw, non-narrated screencast of a relatively straight-forward process to update the Koryuu firmware in Windows using modern, free tools, but it needs to be turned into a proper tutorial.

Cheers,
megari

[newsupermarkio]

Hi megari,

Thanks for your reply.

The PVM I’m using is the Sony 2530 (ref: https://docs.sony.com/release/PVM2030.pdf, https://cdn.retrorgb.com/assets/monitormanual/SonyPVM_2530_Service_Manual.pdf). This actually has direct S-video and direct composite inputs. It also has a DB-25 pin RGB input. My intention with the Koryuu was to get composite or S-video sources into a component switch which then feeds into the RetroTink Comp2RGB (https://www.dropbox.com/s/rnoxcaxaqljq018/Transcoder%20Manual.pdf?dl=0) and then into the PVM via the SCART to DB-25 connection. That being said, the RGsB or RsGsBs options wouldn’t work in my case.

I just re-tested direct composite input to the PVM, and that is running fine with no issue on color.

I did try running the composite input through the Koryuu, and for some reason that output was pure gray from the start. I also had an issue with the running the composite signal through a RetroTink 2x Pro to HDMI display — it was also gray. It puzzles me why the direct composite connection to the PVM is ok with color, but running through Koryuu (or even RetroTink 2x Pro) is interpreting it as grayscale.

Regarding S-video, the direct S-video connection to the PVM is ok. The S-video connection through RetroTink 2x Pro to HDMI display is ok as well.
When running S-video though Koryuu, I can get color, but it has issues. With the original stock firmware, the issue showed up quickly, and eventually most of the video signal turns gray. Strangely enough, with the updated test firmware (1.1-test0), the onset of color loss in S-video is delayed, and the problem isn’t as bad as with the original stock firmware — that’s why I suspected something changed with the interlace handling modification in the blanking interval.

I also had looked up the video chip in the CPS Changer too (CXA1645, ref: https://console5.com/techwiki/images/2/2f/CXA1645P-M.pdf). From my observations/tests, the direct video output of the CXA1645 via composite or S-video to the PVM is ok (no flicker, color loss). Maybe there is something about how the Koryuu is handling the signal output.

When the issue occurs, the picture is stable. There is some strange color flickering and or complete color loss to grayscale.
I’ll see if I can get some video clips of the issues seen later this weekend.

Thanks and Best Regards,
newsupermarkio

[megari]

Hi supermarkio,

Sorry for taking a few days to respond.

I have gone through my old data on CXA1645 and other chips used on the Sega Genesis Model 2 (as it is the closest match to this case), and based on my old oscilloscope captures, it seems that the color burst amplitude tends to sometimes be low on these units, especially so for CVBS output (sometimes barely 0.1 Vpp). S-Video output is usually fine. RGB output is basically always fine. The low amplitude for CVBS output is puzzling, as the specifications for the CXA1645, for instance, state that during normal operation, the output color burst amplitude is always in the range of 0.2 – 0.3 Vpp, 0.25 Vpp being typical.

I will probably have to obtain a few samples of representative hardware using this type of chip to properly investigate this. I’m hoping to be able to find someone relatively local who could lend me a Sega Genesis 2. A Capcom CPS changer would probably be even rarer in Europe.

That said, we can try and troubleshoot this a bit more. I have an experimental firmware image I would like you to test, if you do not mind. Please download it here. There are a few variants in the ZIP file, but unless you are willing to use the on-board USART serial port for logging debug output from the Koryuu, you’ll probably want to use the default or default_ntp variant.

The experimental firmware has one major change: the OPTION button no longer toggles smoothing, rather than cycles through different settings for the color kill threshold. On the first press, the Koryuu uses the most lenient threshold, where it only reverts back to grayscale if the color signal is very, very weak. Subsequent presses increase the threshold step by step, requiring ever higher color signal strength for color to be displayed. After the highest threshold setting, the threshold cycles back to the lowest setting.

The goal is to try finding out 1) Whether the issue happens even with the lowest threshold (first press of the OPTION button after power cycling the Koryuu), 2) If it does not, how many presses after the first one are required for the issue to occur.

Would you be willing to spare some of your time to try this experiment while I try to figure out the next step?

Thank you,
megari

[newsupermarkio]

Hi megari,

Thanks for the reply!

Here is a video I took of the color issue I saw with the original stock firmware with the following connection:
CPS Changer -> S-video -> Koryuu -> Component -> RetroTink Comp2RGB -> SCART/DB-25 -> PVM 2530

https://goo.gl/photos/SR9uWteAQLa3pGEY6

Everything runs fine for a while, then the colors start to flicker and eventually it goes to gray. Perhaps this gives you an idea of what may be going on. As noted above, with the firmware 1.1-test0, the problem is not as bad as with the stock firmware for some reason — I don’t have a video of that yet. With a direct S-video connection to the PVM 2530, this issue is not seen.

Thanks for preparing a test firmware. I’ll take a look and provide some feedback later this week. This problem is puzzling me and it very may well be an issue with how the output of the CXA1645 is being handled.

Thanks and Best Regards,
newsupermarkio

[leewrigley]

@megari myself and another user on this forum have been trying to use the koryuu with svideo from pal n64s (using modded cables which fix the pal n64s video problems) we have however been experiencing very high levels of interfernce (checkerbaord like) much more than would be expected of well shielded svideo cables. after much testing, user @noctua on this forum discovered that the comb filter from the retrotink mini solved this issue using giving him the cleanest signal from an unmodified pal n64.

do you think that this kind of filter could potentially be added by firmware or do you think there would be any other solution for this problem that could be implemented in some way?

thanks

[megari]

@leewrigley That’s weird. An S-Video source requiring a comb filter sounds really odd, as the role of a comb filter is to reduce artifacts (such as dot crawl) when a composite signal is separated into luma and chroma components. One might think of this as a process that is fed a composite signal and outputs an S-Video signal (Y and C separated). A proper S-Video signal simply should not require combing, at least in theory.

That said, it is of course possible that the Koryuu has an issue with the S-Video output of modded PAL N64s. The ADC in the Koryuu does support a variety of shaping and combing filter settings, and it should be possible to make at least some of them to not only apply for CVBS signals. If you’re willing to experiment a bit, I can provide you a version of the firmware with tweaks to the filtering of S-Video.

If it is not too much trouble, could you take a short(ish) video clip of the interference you are seeing, please? A smartphone video is fine, as long as it is in focus. Seeing the video might make it clearer to me what sort of interference you are experiencing.

By the way, some N64 games (such as some from the 007 series) apply their own filtering to simulate film grain, which I have sometimes seen to be mistaken for noise (UltraHDMI captures showed conclusively that the “noise” was there in the original digital pixel data). Could you state which game(s) you are experiencing the bad image quality with, please?

[newsupermarkio]

@megari I tried the experimental firmware with the different color kill thresholds using both composite and S-video inputs.
Even with the most lowest threshold setting, I was still getting a grayscale signal output from the composite input.
Strangely with this experimental color kill threshold firmware, the S-video input also shows grayscale output much faster than stock or 1.1 firmware.

[megari]

@newsupermarkio I see. That’s a bit odd. In addition to making the color kill threshold adjustable, the firmware was also made to make the internal “in lock” status dependent on having not only horizontal lock, but also subcarrier lock established. I wonder if that caused the quicker fallback to grayscale.

Could you try this firmware image, which has the “in lock” status only dependent on hlock, please? Much appreciated.

Cheers,
megari

[newsupermarkio]

Hi @megari, I tried the linked Koryuu firmware image out with the “in lock” status tied to hlock, but it exhibits a new issue where the video signal is lost completely in some instances.

I uploaded two new videos at: https://goo.gl/photos/SR9uWteAQLa3pGEY6

In the newly uploaded videos, I compared the CPS Changer S-video output through the Koryuu and the RetroTink-2x Pro over several minutes of operation.
The RetroTink-2x Pro seems to handle the S-video output better than the Koryuu. There is no loss of color or video signal with the RetroTink-2x Pro.

From here [https://videogameperfection.com/forums/topic/composite-s-video-solution/page/2/#post-20336], it seems the RetroTink 2x series might be using the ADV7280, similar to the Koryuu? There must be some setting issue difference which is leading to the Koryuu not being able to handle the CPS changer S-video output compared to the RetroTink 2x Pro.

Thanks and Best Regards,
newsupermarkio

[Author]

Posts