search

marqs85 / ossc

Open Source Scan Converter
http://junkerhq.net/xrgb/index.php/OSSC
GNU General Public License v3.0
469 stars 63 forks source link

Loss of synchronization on firmware versions 1.7-1.9 #111

Open LobAnovich opened 4 months ago

LobAnovich commented 4 months ago

Since I updated the firmware of my ossc (China 1.6 upgraded) not sequentially, I can’t know exactly which version of the firmware the problems with loss of synchronization started with. It worked without problems with PS1 on older firmware versions (up to 0.9), after upgrading and installing firmware 1.7, problems began with loss of synchronization (RGBsync on composite,RGBcomosite sync, RGBsync on luma) and on version 1.9 they became so frequent that it is simply not possible to use the PS1 . I tried different video cables and different versions of the console. All the consoles I tried were region J. Problems in any modes (game menu, BIOS, games (J, U)).

marqs85 commented 4 months ago

Does the issue occur with other consoles than PS1? Can you also post which NTSC-J models you have tried? I have not heard similar reports of PS1 issues with 1.09 firmware, but presumably most are using PAL/NTSC-U consoles. There might be also an issue with the HW upgrade mod or the specific board. You can always downgrade back to 0.9 if that works for you.

LobAnovich commented 3 months ago

At the moment, I have tested all types of video signals with all possible types of synchronization, different connecting AV and HDMI cables and power supplies, only on one PS 1 (pu18,ntsc j), with which problems arise. I haven’t changed the settings of ossc itself since installing firmware v.1.x. Multiplication does not affect the problem. I can check the same thing on other copies of PS1, as well as Saturn, Dreamcast (All copies of ntsc j are in stock), but for this I will need a little more time. Are there any recommendations to check any specific settings in ossc?

marqs85 commented 3 months ago

No specific recommendations, just use default settings. Was the 'upgrade' mod verified with any source to ensure the problem is not in the upgrade?

LobAnovich commented 3 months ago

I have drawn up a check plan, but for now I am far from the place where the device is located. In a week, I think I will be able to carry out checks according to this plan. As for the initial malfunction or malfunction during modification, I highly doubt it. But further tests will show the truth

LobAnovich commented 1 month ago

Finally, I was able to test it on three versions of PS1 (5500,7000,7500) from the Japanese region, both in stock condition and with PS1PSU. On two of the three, all capacitors have been replaced. All three have the same problem with losing sync. Three different video cables. Two different power supplies for ossc. Reset ossc to factory settings. The problem does not go away. I haven’t tried just rolling back the firmware and reflashing 1.9

UPD. Downgrading from 1.09 to 1.08 did not solve the problem. Reverting from 1.08 back to 1.07 completely eliminated the loss of synchronization. Some changes in version 1.08 that were not changed in 1.09 led to loss of synchronization (luma, composite, csync) with RGB

marqs85 commented 1 month ago

First of all, let's make sure my understanding on results using the firmware versions is correct:

I can only think of 2 causes:

LobAnovich commented 1 month ago

First of all, let's make sure my understanding on results using the firmware versions is correct:

  • v1.07-1.09 (after performing the HW mod): issues with 3 different PSX revisions. Saturn, Dreamcast (VGA?) mentioned but not explicitly stated they have the same issue.
  • v0.90 (before performing the HW mod): PSX works OK.

I can only think of 2 causes:

  • same sync processing issue that has been reported with MSX on the latest fw. According to the report only 1.09 is affected, though. This is expected to be resolved eventually once I receive HW that reproduces the issue.
  • the China variant is somehow different and incompatible with the HW. Without further details on it, the only feasible fix is to revert to 0.90, possibly also removing the wire added by the HW mod if it's part of the problem.

I updated the previous post, where I added information that the problem completely disappeared with a return to version 1.07. But stayed on version 1.08

If necessary to solve the problem, then I will be able to check Saturn model 1 (va0 and va0.5) and model 2 (I don’t remember the board revision) in the near future. I can try to check the Dreamcast, but it connects to the TV via vga2hdmi and I don’t remember if I have an rgb video cable. There is definitely no VGA cable for dreamcast

LobAnovich commented 1 month ago

The version from Ali has a simplified power supply circuit and I can, alternatively, assume that the problem may be related to interference in the power circuits. Although I tested it with several different power supplies, there wasn’t really a very high-quality one among them. So, were any changes made to firmware 1.08-1.09 that, hypothetically, could depend on the quality of the power supply?

marqs85 commented 1 month ago

No sync related changes were introduced in 1.08, but FPGA fitter seed was changed to meet timing. In the unlikely case that it had some effect on certain boards, I created a test version of the latest dev firmware using the seed from 1.07. It doesn't produce identical logic due to other changes, though, but may be still worth a try.

LobAnovich commented 1 month ago

I installed 1.10-t2 today and it got even worse. First the synchronization was lost, and then the image disappeared altogether. I rolled back to 1.07, but the synchronization there is also unstable. I tried several different power supplies and rgb cables, the problem remained. Upon closer inspection, I found out that the image was outside the permissible range in several multiplication modes, including 1x. Haven't returned 1.10-t2 yet to check for changes in work from changing settings

marqs85 commented 1 month ago

Regarding permissible range, do you mean some of the sampling settings do not make sense? 1.10 should ignore saved presets from previous firmwares and use defaults, but if for some reason that does not work you can try explicitly resetting the settings.

LobAnovich commented 4 weeks ago

All tests were done on stock settings. I reset the settings after installing each new firmware. The range was exceeded when the multiplier was switched. It’s strange that even in 1x mode the range was exceeded.

But, as I wrote earlier, I need to thoroughly check all the settings, including on the TV, in order to understand the pattern

marqs85 commented 4 weeks ago

Perhaps I misunderstood, could you elaborate what you mean by "outside the permissible range"? Something reported by the display?

The inconsistency in the results indicate the mod wire is perhaps not properly soldered down, or picking noise somewhere from the PCB.

tecteun commented 3 weeks ago

I have almost the same issue, when using anything newer than 1.8, I get frequent out of sync issues, using in this case a PS2 in PAL format. I've also tried the dev builds that were floating around. The display shows the frequency hopping between 49.9 and 50.0. It happens random, and for about a second, then syncs back to the good frequency (49.9 apparently). I also have seen that it was stable for a longer period. I'm back to 1.8 and that fixes the issue.

So maybe this is related? Or it could be a separate issue.

Anyway, let me know if there is something I can test.

marqs85 commented 2 weeks ago

This should be now fixed in c91e1ad00c3c12b81dc3ddd234bec3d7fc3be748 which in included in just released v1.10 firmware.

LobAnovich commented 1 week ago

Thank you for your work and help ! I don't plan to use lumacod, but exporting profiles is important to me. Is it possible to release a modification of version 1.10 without lumacode, but with working export? At any time, if I need lumacod, I can flash ossc with standard firmware