Details for the next generation of Flickerfixers from Individual Computers

[keropi]

Originally Posted by Schoenfeld

[...]
Example:
You're choosing 49.09MHz in the config tool. On a PAL machine, this will produce an exact pixel clock of 48.868MHz, and on an NTSC machine, it will produce 49.318MHz pixel clock.[...]

thanks for the explanations
correct me if I am wrong but does this mean that the values on the config tool are not in reality exact since they get calculated differently depending on pal or ntsc clocks?

also I had an idea about the H/V offsets... would it be possible when you test a screenmode to have the arrow keys work as a centering means and automatically pass the values to H/V offset fields when you press apply? It would make finding those values way more easy IMHO

[Schoenfeld]

Originally Posted by keropi

thanks for the explanations
correct me if I am wrong but does this mean that the values on the config tool are not in reality exact since they get calculated differently depending on pal or ntsc clocks?

Yes, that's correct. I do have an idea how to make an automatic detection of the actually-installed crystal, but that requires a realtime-clock, and not everyone has that installed. So for now, it's an average value, slightly incorrect for both PAL and NTSC. However, since it's PLL circuits on the monitor side as well, it's well within tolerance.

Originally Posted by keropi

also I had an idea about the H/V offsets... would it be possible when you test a screenmode to have the arrow keys work as a centering means and automatically pass the values to H/V offset fields when you press apply? It would make finding those values way more easy IMHO

Changes are not instant - we cannot write "directly into the FPGA", but only into the flashrom. Then we need to trigger an FPGA-reconfig, which in turn will re-read the config from flash. In any case, you'll loose sync for a fraction of a second.

Jens

[Retrofan]

Originally Posted by Schoenfeld

SuperPlus: This mode was released after the Indivision AGA MK2 was released. You can't expect me to support this and maybe more modes for all times in the future. I might do it, but at this point, we have "known bugs" to fix first. I'm happy to collect information, but I won't open a new construction site when the old one is not yet finished.
Jens

It was working before the config here, don't know what should I change to have it again.

I've got a problem. I managed to get the Indie working with the screenmodes that I regularly use (standard NTSC & NTSC Interlace modes ONLY) with the initial release of the config tool. I downloaded the updated tool the other day on this site, followed the installation instructions and the new cores that came with the package were flashed. Now, I get no output at all from the Indie. I have tried several different monitors; a VGA TFT and a DVI TFT. The VGA TFT was previously working with the Indie. I have a 1084 on the 'video out' and can see that the NTSC modes are being output from the Amiga; ie the test patterns from the new Test button are displayed on the 1084 but not on the Indie. I've tried almost all of the supplied modes starting with VGA 640x480 which was working before. What can I do?

Note: The Indivision logo is displayed on bootup and new cores are flashed when I attempt try out different modes so there is still something going on, just no output.

thanks,

Brian

[Schoenfeld]

This might be a problem caused by both versions of the core(s) installed in the DEVS:Indivision directory. The tool *should* choose the latest cores by their version numbers (which are encoded in the filenames), but there may be a bug.

Please clean the DEVS:Indivision directory and make sure that you replace the config files with the ones that I have provided in the new (rather large..) archive. Then start the configtool and choose "reflash all" from the drop-down menu.

Jens

[NovaCoder]

Originally Posted by Schoenfeld

Novacoder: If a screen looks "totally offset", then the screenmode recognition is obviously failing. For a proper bug report, please configure the OSD of the flickerfixer to "always on" and enable the OSD debug feature. This will display an additional line with information of the Amiga-screenmode such as number of lines, number of pixels and the H/V frequencies of the Amiga mode. These numbers are used for screenmode recognition. Compare them to the standard WB settings. As I wrote yesterday, there's still this mystery of Euro72 having two different screen sizes on two computers, which is not solved yet. If you have a similar effect even on one computer but with different pieces of software, we might get more hints on why this is happening. Note that the screenmode recognition is happening inside of the FPGA, and that's re-configurable. A new screenmode recognition entity can be put in a core update.

Jens

Managed to get rid of the corrupt graphics by choosing a 640x480 mode, the colors are still not 100% but much better than they were. Not sure what is wrong with the colors, they are very close (about 95% correct) but you can see they are not quite right.

Also created a new 1280x720 mode for HD720, that looks pretty nice.


This is as good as I can get it looking with PAL at the moment -> [ Show youtube player ]


Yep we need SuperPlus sorted please Jens, it used to work really well with the original core.

Thanks

[gibs]

How to go back to the out of box settings please ?

[pintcat]

@gibs: All your settings are saved in the prefs files. Try to replace them in ENV: and ENVARC: with the unchanged ones from the archive. After restarting the config tool you should see the original settings.

[Yarc]

Oh dear! After the update my Mk2 does not work as well as it did. This is a little worrying. With the version prior to V10 I had it setup great and could switch graphics modes on my A1200 and the Mk2 would adjust accordingly. Now, whenever I change graphics mode all I get is a black screen and have to cycle the power. I have tried going back a version, but the Mk2 just wont work the same as it did. To be sure, I deleted all the installed files and did a fresh install, and each time selected to reflash all cores. Any ideas?

[Yarc]

Originally Posted by Yarc

Oh dear! After the update my Mk2 does not work as well as it did. This is a little worrying. With the version prior to V10 I had it setup great and could switch graphics modes on my A1200 and the Mk2 would adjust accordingly. Now, whenever I change graphics mode all I get is a black screen and have to cycle the power. I have tried going back a version, but the Mk2 just wont work the same as it did. To be sure, I deleted all the installed files and did a fresh install, and each time selected to reflash all cores. Any ideas?

UPDATE:

Well, I've managed to get back to a working system but only after installing the very first core (AGA_MK2_78MHz.rbf) and then re-installing the first beta release (folder AGAMK2_cfg_install) prior to the current V10 version. Not sure what is going on here, but my monitor does not like the latest software/cores. It looks like the new version is putting something into the cores that does not work for me, and whatever it is cannot be removed by the software unless I go back to the originally supplied core and flash that. This then presumably removes the problem.

What is strange is that installing the version prior to the current V10 and reflashing all cores does not properly revert back!? This is how it seems to me as a user.

I'm not sure why the new cores don't work for me, but at least I now have a working system albeit minus the recent bug fixes and test button.

Yarc.

[Yarc]

Originally Posted by Yarc

UPDATE:

Well, I've managed to get back to a working system but only after installing the very first core (AGA_MK2_78MHz.rbf) and then re-installing the first beta release (folder AGAMK2_cfg_install) prior to the current V10 version. Not sure what is going on here, but my monitor does not like the latest software/cores. It looks like the new version is putting something into the cores that does not work for me, and whatever it is cannot be removed by the software unless I go back to the originally supplied core and flash that. This then presumably removes the problem.

What is strange is that installing the version prior to the current V10 and reflashing all cores does not properly revert back!? This is how it seems to me as a user.

I'm not sure why the new cores don't work for me, but at least I now have a working system albeit minus the recent bug fixes and test button.

Yarc.

Damn! Still not right. It seemed to be working as it did, but a few minutes later wont switch graphics modes. Maybe the Mk2 has a problem?

[NMI]

Yeah, I got exactly the same issue with the new cores.

Only PAL-hires (non-interlaced) works OK, everything else makes the Indy hang and don't output anything, the only way to recover is to power-cycle.
Also, pressing "apply" in the config-tool makes the Indy hang, forcing a new power-cycle.

When I install the latest update and do "update all" it updates cores 0-5 and 7 (not core 6?) then the Indy stops sending output to my monitor (tested multiple times with 3 different monitors).

I'm currently using the latest config-tool with the older beta cores, and it appears to work OK.

[Yarc]

Originally Posted by NMI

Yeah, I got exactly the same issue with the new cores.

Only PAL-hires (non-interlaced) works OK, everything else makes the Indy hang and don't output anything, the only way to recover is to power-cycle.
Also, pressing "apply" in the config-tool makes the Indy hang, forcing a new power-cycle.

When I install the latest update and do "update all" it updates cores 0-5 and 7 (not core 6?) then the Indy stops sending output to my monitor (tested multiple times with 3 different monitors).

I'm currently using the latest config-tool with the older beta cores, and it appears to work OK.

Good call! This seems to work, what a relief. As you say, there must be an issue with the latest cores. Now the test button works as it should.

[johnim]

hi guys in getting the same on mine. works great with v24 cores and config v10

[Schoenfeld]

Those of you who have trouble with the V2.5 cores, please report if you have a PAL or an NTSC crystal in your host machine. PAL machines use 28,37516MHz and NTSC machines use 28,65656MHz.

The difference is that on some FPGAs, the PAL crystal generates a too-low VCO frequency for the internal PLL. It's off by less than 1%, but it might cause trouble. Only two output pixel clocks are "on the edge": 59,8621MHz and 149,6552MHz.

If you suspect that this is the cause for your machine, then delete these two files:

coreMK2C25_B21.rbf (the 149,6552 MHz core)
coreMK2C25_E21.rbf (the 59,8621 MHz core)

In any case, please make sure that the V2.4 cores are removed from the directory, and that you don't mix config files from the old tool with the new release.

Jens

[Yarc]

Originally Posted by Schoenfeld

Those of you who have trouble with the V2.5 cores, please report if you have a PAL or an NTSC crystal in your host machine. PAL machines use 28,37516MHz and NTSC machines use 28,65656MHz.

The difference is that on some FPGAs, the PAL crystal generates a too-low VCO frequency for the internal PLL. It's off by less than 1%, but it might cause trouble. Only two output pixel clocks are "on the edge": 59,8621MHz and 149,6552MHz.

If you suspect that this is the cause for your machine, then delete these two files:

coreMK2C25_B21.rbf (the 149,6552 MHz core)
coreMK2C25_E21.rbf (the 59,8621 MHz core)

In any case, please make sure that the V2.4 cores are removed from the directory, and that you don't mix config files from the old tool with the new release.

Jens

Thanks Jens for getting back. I could not find a coreMK2V25_B21.rbf file but deleting the _E21.rbf file seems to have done the trick.

[Schoenfeld]

Originally Posted by Yarc

Thanks Jens for getting back. I could not find a coreMK2V25_B21.rbf file but deleting the _E21.rbf file seems to have done the trick.

Great - thanks for the feedback. And sorry about the B21 core - that's not in the release package, because it's above 135MHz. Anything above 135MHz is totally experimental and therefore not released. With these high pixelclocks missing, any attempt using 1600x1200 will result in very low H/V frequencies, so that mode should be considered experimental as well.

Jens

[lurch]

Latest config tool puts my screen slightly off to the right, just slightly when in workbench.

The first version of the config tool worked 100%, so will try going back to that.

Don't get the corrupt graphics as with the original core at 1024x768 @ 60 if I change this to 75 graphic corruption problems.

800x600 works great for games/whdload, although I can't get any resolution lower than that with the two newer cores but could with the original core. With the latest ones I get a black screen, and the monitor reports out of range.

My machine is a PAL machine.

[johnim]

hi i removed the _E21.rbf file reflashed with v25 cores

but workbench loads fully then the screen goes off

i have a pal system with pal hires laced screen on a dell 1280x1024 monitor with blizzard 1230mkiv with cwb3.9

[NMI]

Hi, I removed the old beta cores and config files, then installed the new tool and cores, removing coreMK2C25_E21.rbf.

Unfortunately it gave no improvement at all. "Reflash all" still results in a black screen after core 7 has been flashed, as does pressing "apply" or trying to change the resolution from PAL-hires (non laced).

Using a PAL A1200 with a ACA1231, WB3.1.