CPS2 digital AV interface

[sosafire]

Hello everyone, are there any existing instructions for installing the Taitoc Labs flexi, I can't seem to find any!

[marqs]

Hi everyone. I was thinking of buying the "CPS2 Digital AV Interface DIY Kit" board for a Neo Geo MVS MV1C and I would like to know if it is giving good results or if there may be problems, if it is necessary to install some firmware or if it is just a matter of soldering cables.

I suppose it depends where you buy it from. If you refer to this, it comes with firmware flashed to match the system selected on the order form. Only known potential issue is 59.19Hz refresh rate which some displays don't support properly or at all. That can be resolved with DFO mod, though.

Hello everyone, are there any existing instructions for installing the Taitoc Labs flexi, I can't seem to find any!

Not sure, I'm not even able to find the flex cable from their product page. I'd assume installation is pretty self-explanatory as it is supposed to just replace a bunch of wires which are documented on github page of the project.

[Emerl]

I have a MV1C with the CPS2 digi AV, and am running into an issue with the direct mode on the RT4k. I get a black screen and no sound, but it detects the resolution (320x224). 480p and up are fine (with the exception of 1920x1440p which has some weird waviness to it). Is this a known issue, or is there something I'm missing? DVI works with 240p, but then I don't get sound.

[marqs]

I have a MV1C with the CPS2 digi AV, and am running into an issue with the direct mode on the RT4k. I get a black screen and no sound, but it detects the resolution (320x224). 480p and up are fine (with the exception of 1920x1440p which has some weird waviness to it). Is this a known issue, or is there something I'm missing? DVI works with 240p, but then I don't get sound.

The only difference in DVI mode is metadata (and lack of audio), so most likely the issue is on how RT4k handles the signal. If I had to guess, it does decimation via the HDMI RX chip which results to too low pixel clock frequency for it to handle.

[Callisto]

I tried to see if this also works for the NeoGeo CD (CD-T01, NEO-CDM3-2). I could find all the hookup points but did't get any picture. I checked and I saw no activity on shadow and dark. Shadow traces back to the NEO-MGA-T. I assume with no activity on VS(shadow) I won't get any picture then. Might there be a workaround or will the CD-T01 not be compatible because of it?

[marqs]

I tried to see if this also works for the NeoGeo CD (CD-T01, NEO-CDM3-2). I could find all the hookup points but did't get any picture. I checked and I saw no activity on shadow and dark. Shadow traces back to the NEO-MGA-T. I assume with no activity on VS(shadow) I won't get any picture then. Might there be a workaround or will the CD-T01 not be compatible because of it?

You should get some picture even without dark bit. At least it should sync and display OSD if the video/sync characteristics match MVS/AES.

[Callisto]

That was my assumption, but the shadow signal is also just low all the time and this is connected to one of the sync inputs of zee CPS2dig. I used the input to the LS05 for it, but could also probe it again at the R150 resistors. Might be a bridge somewhere. Will simply lift the pin on the NEO-MGA-T and probe that one to isolate the issue. I really hope I just got a bridge somewhere which I didn't see.

https://freeimage.host/i/F3x2Vwu

[marqs]

VS pad is used for vertical sync only with cps2 firmware - on Neogeo fw csync is expected on HS pad. Where have you tapped csync from and is it 3.3/5V level?

[Callisto]

VS pad is used for vertical sync only with cps2 firmware - on Neogeo fw csync is expected on HS pad. Where have you tapped csync from and is it 3.3/5V level?

Pin 10 on the CXA1645P. Its 5V. I guess I should try again and use a coax cable this time.
..
Nothing. The RT4K detects some 294 MHz Pixel clock but that's it.

[marqs]

VS pad is used for vertical sync only with cps2 firmware - on Neogeo fw csync is expected on HS pad. Where have you tapped csync from and is it 3.3/5V level?

Pin 10 on the CXA1645P. Its 5V. I guess I should try again and use a coax cable this time.
..
Nothing. The RT4K detects some 294 MHz Pixel clock but that's it.

That would be around double the expected pixel clock. Maybe you could check if you've tapped 12MHz & 6MHz clocks correctly.

[Callisto]

They are 6MHz (C2) and 12MHz (C1). But they are sawtooth shaped (C1 and C2 with some 300 - 500mV offset, peak at 4.2V for C1 and 4.8V for C2). They are all generated by the NEO-GRC but have a resistor inline (B7, R150) for C2 before it connects to the LS273 clk pin, and B29 (R150) for C1 before it connects to the NEO-SFT. If I probe before the resistor (or what is B, ferrite bead?) I get a much cleaner rectangular shape pulse (-0.8 to 5.2V). I will rewire it and test again today or during the next days.

..
Old (after the resistors):


New (before the resistors, labels are wrong):

No luck again.

[marqs]

It seems NEO-CDM3-2 uses NEO-GRC chip like MV1A which requires additional settings for OSSC due to different sync. Perhaps that is also the reason why it doesn't currently work with cps2_digiav since the firmware has been verified only with MV1C (NEO-GRZ chip).

[Callisto]

It seems NEO-CDM3-2 uses NEO-GRC chip like MV1A which requires additional settings for OSSC due to different sync. Perhaps that is also the reason why it doesn't currently work with cps2_digiav since the firmware has been verified only with MV1C (NEO-GRZ chip).

Yes, then that's issue here. Do you plan to eventually add support for this?

[marqs]

Yes, then that's issue here. Do you plan to eventually add support for this?

Sure, just need to get such unit into my hands to see what is the issue and then fix it.

[marqs]

@Callisto I found some old pictures of NEO-GRC sync timings (see below) and adjusted the sync alignment logic so that it should be less reliant on exact timing of equalization pulses. A test firmware can be downloaded here. Saving of settings in that may not work at there's also a larger rework in progress.

[marqs]

Firmware 0.94 have been released for CPS2 and Neo-Geo. The main visible change is addition of HDR and VRR flag options. The firmware will be later ported to other supported platforms (and NGCD eventually).

[Callisto]

I installed the cps2digital in an AES system and haven’t ordered a new one yet to test it in the NG CD. Will do it probably soon. But I have another question for in the meantime. How does the cps2digital handle this issue: https://github.com/fliperama86/neopico-hd/issues/2 I didn’t really notice “stuttering” until now. Just that the sound sometimes seems to clip or distort (MV1C) which just might be normal.

[marqs]

I installed the cps2digital in an AES system and haven’t ordered a new one yet to test it in the NG CD. Will do it probably soon. But I have another question for in the meantime. How does the cps2digital handle this issue: https://github.com/fliperama86/neopico-hd/issues/2 I didn’t really notice “stuttering” until now. Just that the sound sometimes seems to clip or distort (MV1C) which just might be normal.

As written in README, the output is framelocked so no stuttering, tearing or variable latency. Obviously you need a monitor which properly supports the refresh rate.

[marqs]

Installation instructions have been now added for NEO-CDM3-1 / NEO-CDM3-2. The process should be similar for other CD revisions.

[Konsolkongen]

Do the CPS2 Digital AV output specific HDMI SPD infoframe? So that scalers can identify and set presets accordingly?

[marqs]

Do the CPS2 Digital AV output specific HDMI SPD infoframe? So that scalers can identify and set presets accordingly?

The SPD Infoframe is already used for Freesync. I don't know if adding some bytes or using reserved ones would break compatibility, but I'm afraid it most likely would.

[Konsolkongen]

Thanks for letting me know

It locks to the input refresh rate of the signal, right? So in Neo Geo’s case that’s something like 59.18Hz.
That could probably be used to auto identify the system.

[marqs]

Thanks for letting me know

It locks to the input refresh rate of the signal, right? So in Neo Geo’s case that’s something like 59.18Hz.
That could probably be used to auto identify the system.

Yes, it only supports framelock operation. In case on Neo Geo, refresh can be 59.19Hz (MVS and early AES), 59.60Hz on (most AES and CD) or 59.94Hz (DFO) which could complicate such detection.

[Konsolkongen]

Thanks for letting me know

It locks to the input refresh rate of the signal, right? So in Neo Geo’s case that’s something like 59.18Hz.
That could probably be used to auto identify the system.

Yes, it only supports framelock operation. In case on Neo Geo, refresh can be 59.19Hz (MVS and early AES), 59.60Hz on (most AES and CD) or 59.94Hz (DFO) which could complicate such detection.

I got a DFO in my MV1C for 59.18 and 59.94Hz. Hopefully that will still work. If not I have the old 24MHz crystal still.

Do you think it is possible to extend the wires a bit so can place the CPS2 digital to the right of the MV1C, rather than on top?

Mine is installed in an AES case so I’m not sure there is enough room above the motherboard. Still has the RF shields.

[marqs]

I got a DFO in my MV1C for 59.18 and 59.94Hz. Hopefully that will still work. If not I have the old 24MHz crystal still.

The board itself should be fully compatible as long as you select matching option (for audio sync) from the menu.

Do you think it is possible to extend the wires a bit so can place the CPS2 digital to the right of the MV1C, rather than on top?

Mine is installed in an AES case so I’m not sure there is enough room above the motherboard. Still has the RF shields.

If you shield the clock wires properly, then it most likely works reliably on a bit longer distance.

[Konsolkongen]

I got a DFO in my MV1C for 59.18 and 59.94Hz. Hopefully that will still work. If not I have the old 24MHz crystal still.

The board itself should be fully compatible as long as you select matching option (for audio sync) from the menu.

Do you think it is possible to extend the wires a bit so can place the CPS2 digital to the right of the MV1C, rather than on top?

Mine is installed in an AES case so I’m not sure there is enough room above the motherboard. Still has the RF shields.

If you shield the clock wires properly, then it most likely works reliably on a bit longer distance.

Thank you!