OSSC v1.xx series beta firmware

[uchristo]

An update:
My native Amiga PAL HiRes wirhth ypbpr doesn‘t toggle anymore but syncs perfectly without delay. Unfortunately my 720p screen (Picasso96 with minimigRTG won‘t sync. I‘ll have to do some further testing….

[marqs]

Did that screen work with the v0.XX series FW? If so, please take a picture from the infoscreen (press INFO on remote) of the new FW if you have another display that works with it (perhaps the info should be dumped into debug console output too).

[uchristo]

Yes, this mode worked before on fw0.90 as 764-p

Details:
45,99kHz 60.20Hz
Mode Preset: 720p_60
IMode (FPGA): 764-p 60.23Hz
Ccnt / frame: 448210
Firmware: 0.90a

This is my Config

Picasso 96 (Amiga Settings)
Resolution 1280×720
Colour depth 256
Pixel Clock 75,62
No Interlace
No Doublescan
horiz vert
Framesize 1570 800
Bordersize 0 0
Position 0 0
SyncSize 64 4
Sync Pol off off
Frequency 48KHz 60Hz

working OSSC Mode on 0.90-audio
Video in Proc
Video LPF Auto
Sampling Opt
480p in sampler VESA 640×480 60Hz
Allow TVP HPLL2x On
Allow upsample 2x Off
Advanced Timing (720p)
H.Samplerate 1650
H.Sample frac 0
Output Opt H.synclen 59
H.backporch 238
H.Active 1280
V.Synclen 5
V.Backporch 39
V.Active 720
Sampling Phase 180

Sync Opt
Analog Sync LPF 10MHz
Analog STC LPF 4.8MHz
Analog sync Vth 56mV

Forum Software will not allow me to submit the debug output because it estimates spam, “contains contacts”

• This reply was modified 1 year, 8 months ago by uchristo.

[marqs]

@uchristo: found one issue which is a likely cause for your problem. It’s now fixed so please redownload the .jic and try again.

[uchristo]

installed the new build (debug version).
nios2-terminal gave me this:
Sync LPF value set to 0x0
PCM1862 found
RC data read (54 bytes)
Profile 0 data read (1916 bytes)
SetInputMode(00,00)
Input mode is RGB24
Output mode is RGB24
SetupAFE()
Reg[61] = 00
Reg[62] = 18
Reg[63] = 10
Reg[64] = 0C
Reg[65] = F0
Reg[66] = 00
Reg[67] = 3F
HDMITX_SetAVIInfoFrame: VIC=2, ColorMode=0, Aspect-Ratio=4:3, ITU709=No, ITC=No, pixelrep=0
SetAVIInfo():
CTS: 27000
### DIY VIDEO DIGITIZER / SCANCONVERTER INIT OK ###

RCODE: 0x02e0, 1
### SWITCH MODE TO AV1_YPbPr ###
THS source set to 1
SOG detected

totlines: 1381 (cur) / 0 (prev), pcnt_frame: 1048575 (cur) / 0 (prev)
Sync up

Lines: 1381 p
Clocks per line: 759
Hswidth: 127 Vswidth: 0 Macrovision: 0
Error: no suitable mode found, defaulting to 240p

OSSC won’t change inputs after this anymore nor reflects the incoming ir-code with green led.

Non-Debug build constantly toggles between 764-p and 764-i
23.01. kHz/ 46kHz / 92kHz and 60.23Hz/120.47Hz
Non-Debug-Build will allow further input after initial selection of source.

• This reply was modified 1 year, 8 months ago by uchristo.

[BuckoA51]

@uchristo if you need me to whitelist you to be able to post anything let me know.

I was able to spend some time with the firmware today and test problematic games and systems and here’s what I found.

1) PAL Atari Jaguar – This seems to be fixed now, the console syncs up immediately no matter if in 50 or 60hz mode. Finally I can play Kauzumi Ninja in 50hz like God intended 🙂

2) Sega Megadrive – Bugs Bunny in Double Trouble still is unstable (https://videogameperfection.com/forums/topic/the-curious-case-of-bugs-bunny-in-double-trouble-genesis/)

3) Sticking with the Genesis, there seems to be regression regards interlace content, see the image below that I captured from Sonic 2’s 2 player mode:-

4) Atari 2600 – Seemed to behave exactly as before, certain games (such as Entombed, as reported here – https://videogameperfection.com/forums/topic/ossc-and-atari-2600/) still randomly lose sync.

5) Super Nintendo (dejittered 1 chip system) also played just fine, same as before.

• This reply was modified 1 year, 8 months ago by BuckoA51.

[marqs]

@uchristo: initial debug info (line count etc.) after selecting a source may not be valid, but should normally settle into fixed numbers after couple rounds. Based on non-debug build behaviour the sync frontend seems to detect irregular field changes, and as a result the detected mode does not stabilize. Without seeing sync waveform it’s hard to say why this happens (trilevel sync, equalization pulses, etc?). Perhaps you can try reducing ‘Vsync threshold’ from sync menu.

It’s worth noting that sync unstability (previously indicated by red LED) is now encoded into green LED so it can hide remote control activity indicator. Also, if the connection to JTAG UART is lost for whatever reason, the debug build freezes.

@BuckoA51: I believe Bugs Bunny and certain Atari 2600 games just break sync regularity and consequently line-locking, thus no way around it using OSSC HW. I’ll try to reproduce your issue with interlaced content.

[uchristo]

I’ve got an old ( Scope) with 60MHz….. Has been some time since I last used it…. Should I trigger on Y from YPbPr since this includes the Sync signals or shall I trigger already on my FPGA Board? It’s a MiST based on CycloneIII

Alternatively, I could probably send you my second MiST ?

• This reply was modified 1 year, 8 months ago by uchristo.

[marqs]

The Y channel at the end of cable is probably easiest as I don’t see native component output on the schematic linked. Alternatively I could share sof/stp files which would allow using built-in logic analyzer of Quartus if you are interested trying it out.

Before going there, try still the latest image and also reducing ‘Vsync threshold’ from sync menu. This new image has the deinterlace issue fixed as well.

[uchristo]

Both modes are working now!
Thanks a lot!

[BuckoA51]

Do we download the new image from the same link on the first page?

[marqs]

@uchristo: with just by updating to latest or did you need to additionally adjust ‘Vsync threshold’?

@BuckoA51: I keep updating the same file until any clearly identified issues are resolved. Then it’s time for v1.00 release with the usual SD image.

[BuckoA51]

Thanks Marqs,

Interlace on the MD still seems to be having issues, though it’s better than it was. Here’s a picture I took, you can notice the corruption on the little Sonic and Tails portrait boxes.

I took another screen shot and you can see that row of black pixels around Tails portrait has shifted. I’m not sure if that gives you any clues. A correct image there should be a border of black pixels on the top and bottoms of the portraits.

Update – Sometimes if you continue to watch the attract mode, it will reproduce the interlace levels perfectly. Probably about 1 time in 4 at a rough guess.

Update 2 – I’m seeing the same behaviour on the older OSSC firmware, so I think we can chalk this down to capture card quirk/bad scaling and say this is fixed, it’s working as well as it does on the release firmware and looks fine on a monitor.

• This reply was modified 1 year, 8 months ago by BuckoA51.
• This reply was modified 1 year, 8 months ago by BuckoA51.
• This reply was modified 1 year, 8 months ago by BuckoA51.

[BuckoA51]

Played all afternoon on PAL C64 via Koryuu to test PAL 288p, no issues detected whatsoever.

[uchristo]

Hi Marqs,
initial message was more detailed, but forum software asked me whether I really want to submit my message…

I did not need to adjust VSync Threshold. From time to time, the 288p (313-p) screen will lose sync for less than a second. But it’s not really reproduceable.

[Author]

Posts