NeoVGA: Lagless, pixel-perfect line doubler for Neo-Geo MVS and AES

[Lemony Vengeance]

Looking good, Mike

 

[mikejmoffitt]

I got MV-1C going strong. Using shielded wire for the clock signal (coaxial snipped from an old Wi-Fi antenna) I was able to get it completely stable. For the other wires, length is not so critical as they are all 6MHz max. I fixed one more bug (SHAD and DAK were mixed up, oops) and I have it automatically detecting if it's on an MV-1C or not so I don't have to have settings to correct for horizontal timings.

My new PCB is much MUCH better on interference, it's pretty much all gone. Here are some pretty shots of RBFFS on the MV-1C. Real Bout Special has been a great testing game because #1 it uses the full 320x224 (most games black out the side columns) and #2 it has a few backdrop patterns that in the past have induced clock stability issues. When I see this game working well, it's a great sign! KoF '99 is a great test for the SHAD and DAK channels as well.

Once I get the parts for it, I'm going to populate PCB #2 and send it off to somebody for a test install.

 

[Lemony Vengeance]

You had mentioned previously that you would be making these compatible with CPS2.. is that still the plan?

 

[mikejmoffitt]

Yeah, that's what I'm going to tackle next. The CPS2 was having the exact same problems as the MV-1C regarding stability, so I think by keeping the clock really short and using the coaxial cable I can get it on there too. Almost all of the actual development work is already complete for the CPS2/1 though

Here's it running during a brief bout of stability (with some horizontal shift not yet dealt with):

 

[Xian Xi]

Would making the trace for the clock thicker work better?

 

[Lemony Vengeance]

awesome! Any thoughts about CPS1 or 3 implementations? I'd love to consolize the entire CPSX collection. (getting greedy here)

 

[mikejmoffitt]

Would making the trace for the clock thicker work better?

It can improve it a little, but it doesn't help the actual wire being run to the board. Keeping the wire short and/or using coaxial cable is the real solution.

I think CPS1 will "just work" as the pixel format is the same. I'll dig up one of my broken CPS boards at home and see if I can fix it up for testing. I've never touched a CPS3 before so I can't say, but I know CPS3 has much higher color depth and I have no idea if the pixel bus is on-board.r

 

[GTRetro87]

I got MV-1C going strong. Using shielded wire for the clock signal (coaxial snipped from an old Wi-Fi antenna) I was able to get it completely stable. For the other wires, length is not so critical as they are all 6MHz max. I fixed one more bug (SHAD and DAK were mixed up, oops) and I have it automatically detecting if it's on an MV-1C or not so I don't have to have settings to correct for horizontal timings.

My new PCB is much MUCH better on interference, it's pretty much all gone. Here are some pretty shots of RBFFS on the MV-1C. Real Bout Special has been a great testing game because #1 it uses the full 320x224 (most games black out the side columns) and #2 it has a few backdrop patterns that in the past have induced clock stability issues. When I see this game working well, it's a great sign! KoF '99 is a great test for the SHAD and DAK channels as well.

Once I get the parts for it, I'm going to populate PCB #2 and send it off to somebody for a test install.

Looking good for the MV-1C!!

 

[distropia]

Mike, your PM inbox is full

 

[GadgetUK]

Awesome work! The output looks fantastic!

 

[mikejmoffitt]

I just cleared my inbox. Give it another go!

I need to fix one bug in the VHDL regarding 240p output ( it's a silly mistake ) but the hardware should be more or less ready to go. I'm going to talk to a few places that do fabrication and get a quote for what it'll cost me per board. I'd rather not keep building all of these by hand!

 

[mikejmoffitt]

Well, I fixed up the remaining 240p output. Looks great on both RGB and YPbPr, tested on this little 8" PVM at home.

Now all four video modes are doing great and I can start looking at manufacture.

 

[pixeljunkie]

So badass!

 

[vinnieboy1]

Great, can't wait

 

[fenikso]

When you implement CPS2, will it require some sort of software update?

 

[mikejmoffitt]

The CPS2 support is a different bistream, so that'd be a matter of having different "firmware" flashed. It's something you'd choose when you buy it, but it isn't hard to update yourself if you have the $10 USB blaster clone easily found on eBay.

 

[mikejmoffitt]

Well, this little cheapo $20 VGA-to-HDMI converter does a fantastic job with the NeoVGA output on my family's LG TV:

However, the same LG TV rejects the 480p component. I'll have to do a little more testing with that... again, not hardware related, so it's not a big roadblock.

 

[McColbo]

Looks great, I'm in the stages of building a consolized MV-1c and this would be awesome!

 

[Niko]

Wow this looks great man! Good Job!

 

[GTRetro87]

Well, this little cheapo $20 VGA-to-HDMI converter does a fantastic job with the NeoVGA output on my family's LG TV:

However, the same LG TV rejects the 480p component. I'll have to do a little more testing with that... again, not hardware related, so it's not a big roadblock.

Ok, I'm ready to order. Let us know when we can cause my "Atari 5200" CMVS is dying to play on HDMI!

 

[bwi]

I'm so all over this when you start shipping

Nice work Mike

 

[mikejmoffitt]

Alright, I've hit a snag... two, actually. On BOTH HDTV's in my family's house, the Component video is acting funny. It is as if green and blue are swapped around! I've wired it correctly (it wouldn't sync if Y wasn't on the green cable, of course!) and can "correct it" by swapping blue and green in my code, but on my little PVM the original setup I have is perfect. It's a mystery to me.

EDIT: I'm a moron - it was much simpler. Nothing is wrong with the component video. I just confused the five green bits with the five blue bits when I wired the board up. D'oh!

The other problem is my auto-detection for MV-1C systems is broken on my MV-1FZ. Looks like it's back to the drawing board on that. For the first release I might just have a little jumper pad, similar to the NESRGB configuration pads. I'm anxious to get something released already.

 

[Xian Xi]

Alright, I've hit a snag... two, actually. On BOTH HDTV's in my family's house, the Component video is acting funny. It is as if green and blue are swapped around! I've wired it correctly (it wouldn't sync if Y wasn't on the green cable, of course!) and can "correct it" by swapping blue and green in my code, but on my little PVM the original setup I have is perfect. It's a mystery to me.

I'm a moron - it was much simpler. Nothing is wrong with the component video. I just confused the five green bits with the five blue bits when I wired the board up. D'oh!

The other problem is my auto-detection for MV-1C systems is broken on my MV-1FZ. Looks like it's back to the drawing board on that. For the first release I might just have a little jumper pad, similar to the NESRGB configuration pads. I'm anxious to get something released already.

Let me know when you want me to test a proto on either MV1C or AES. As for the color swap, I once had that problem with dirty sync, added an LM1881 and fixed it right up.

 

[buaku]

I'm anxious to get something released already.

I'm anxious to throw money at you!

 

[mikejmoffitt]

Let me know when you want me to test a proto on either MV1C or AES. As for the color swap, I once had that problem with dirty sync, added an LM1881 and fixed it right up.

Naw, it was something much goofier. I had just wired the five green bits to the blue input, and vice versa, on my PCB. Just a gaffe from working too quickly. I'm pulling sync digitally, direct from where it's generated on the graphics chipset, so fortunately there's no way for it to get dirtied from a lack of buffering or the resistor divider.

I'll send you a PM shortly.