|
Post by mrbombermillzy on Apr 2, 2016 15:00:10 GMT
Register 6 notes:
My monitor can cope with 37 visible vertical chars before sync is lost.
Register 7 notes:
A sync position of 38 is adequate for my monitor to display 37 characters vertically (296 pixels). The vsync pulse being much smaller than the h. one (only 4 scanlines).
I still need to work on some adjustments so that I can possibly add 1 or maybe 2 extra chars to the v. Displayed char count.
|
|
|
Post by mrbombermillzy on Apr 2, 2016 15:11:02 GMT
General notes/conclusions:
On old crt equipment at least, there is a constant battle of compromise between monitor stability and char corruption.
The way I see it, you can either buy a modern tft/flatscreen that has either a digital rgb input (or perhaps invest in a converter for a rgba screen), or you can replace your VDC vram chips for ones that require less refresh intervals before losing/partially losing their contents. The latter solution is probably a bit much.
I will conduct further testing when I have time, but that is something I really dont have atm, so I am writing this to hopfully help others to test the boundaries of their own equipment.
It would be great to hear about the results others have had using these values that I have provided as a starting point for further enhancing the C128 VDC resolution.
Perhaps if you could let me know in this thread what monitor you have and the vdc version and revision (if you know it) and a description of how well your setup coped with the chosen resolution.
With this information, we can begin to formulate a best safe average resolution for use in games, etc.
Give it a go, but be careful!
Phew! Thats it from me for a while now. :-)
|
|
|
Post by hydrophilic on Apr 4, 2016 8:08:32 GMT
Thank you, mrbombermillzy for all your detailed notes! You have so many notes, I will *have* to print them out and compare them side-by-side with my testing! Yes, I will watch out for smoke coming from my monitor I'm mainly interesting in finding out max resolution as a function of char-width. Reports by you and Tokra indicate maximum width of screen using standard 8x8 cells (or 8x2 cells)... but the fact that VDC goes "crazy" (corrupt cells) when there is only little more than 80 (standard) cells seems to me there is some internal (and UNDOCUMENTED) limit on "number_of_cells" (#cells). I hope my testing will isolate this hidden/magic number. But the VDC is really a mysterious piece of silicon, so I don't have hopes... Even if I don't find anything new, thanks for the info, and I'm sure it will be valuable to the *next guy* who finally does figure out what the Hell is happening inside the VDC...
|
|
|
Post by mirkosoft on Apr 24, 2016 0:54:28 GMT
I tried to use 1024x296(i) with CGA2VGA converter, image looks wide, I attached screenshot, but really it looks unusable or it has to be so? Miro 17
|
|
|
Post by mrbombermillzy on Apr 24, 2016 16:05:58 GMT
That looks much more corrupted than what I get! Have you set the vram refresh down as low as it will go? Also what VDC version/revision do you have?
|
|
|
Post by mirkosoft on Apr 24, 2016 17:35:10 GMT
Hi!
I used DCR - so rev2 or 8568. Also results were different if was SCPU running at 20MHz. I was following instructions in this thread.
Miro
|
|
|
Post by mrbombermillzy on Apr 25, 2016 17:49:18 GMT
Try incrementing the vram refresh rate (register 36) in steps of +1 (from 0) until your monitor starts to get shaky. Mine can only get to about 5 before it starts having trouble. You should begin to see the corruption getting less problematic as you raise the number of refreshes per scanline value. However, as the characters get less corrupt your monitor will have more trouble stabilizing as you are now adding extra refresh time to the horizontal frame timings. This is the balance you have to find on YOUR monitor. Let me know what results you get.
|
|
|
Post by hydrophilic on Apr 26, 2016 5:54:40 GMT
Great photo (screen shot) mirkosoft! I finally got some time-off from work in the coming days (so I plan to test VDC limits and work on SD2IEC firmware). Thanks for the info, mrbombermillzy, when ever I try "extreme" modes, I normally set register 36 to zero... it seems like I need to test with less "extreme" values (like the default value!).
|
|
|
Post by mrbombermillzy on Apr 26, 2016 18:12:34 GMT
I suggest incrementing from 0 upwards as just leaving the value at the default may not always get your monitor to sync. The character corruption does get eliminated so dont worry about that not clearing up...I would rather you get corrupt characters to start with rather than send your monitor out of sync..much safer! :-)
|
|
|
Post by mirkosoft on Apr 26, 2016 18:37:34 GMT
I mean in my case is problem that it is not displayed on real Commodore monitor (I own 1084) 'cause I don't want risk damage. Before mode switch I turn off 1084 and output is only CGA2VGA converter which can display up to 1600 wide modes and connection to PEXHDCAP card allows me to create screenshots or record video - like you could see in comparison of SCPU128 (SCPU part is recorded on RGBI2SV - needs not extra large resolution) vs Mega 65 (M65 part is downscaled from 1920×1200 to 1440×900 and scaled to 1280×800). So, here's Q if it is not problem of adapter...
Miro
|
|