fooling with m2emu network

[dekkit]

Updated 14/5/26 - various edits to the post tl summarise experiences.

Firstly thank you sailorsat for the massive efforts to debug the emulator and try and troubleshoot the network component also to the community for keeping this thread active for one of the best arcade games ever made.

I recently setup another 2 player Daytona emulation rig (with newer hardware ) and was really surprised to see how differently (ie worse) the network performed on the newer rig when compared to my older pair of laptops.

Note: I use a cross-over cable between the two PCs (no hub/ router) to minimise lag.

Original PC Hardware

Old pair of laptops (both running at 640x480 @60hz resolution)
Laptop 1
 2 Cores, @2.26 GHz, Penryn (was master) win7 32-bit
Laptop 2
2 Cores, @1.66 GHz, Merom (was slave) winxp 32-bit

Both had 2GB ram
Using a cross over cable , 100Mbit network (so no jumbo frames)

Performance
Performance was great - with all the recommended settings (network sync =1 and did not need to use the m2 lagfix).  No freezes/pauses, very minimal purple car flickering (occasional after a few races and not enough to spoil the fun)

Newer PC Hardware
A pair of matching/identical HP Elite 8300  'All-in-one' pc's (running at 1920x1080 @60hz)
Were a different story initially but thankfully now are running better (it wasn't as good as the original setup for a while, but after lots of tinkering its finally there running widescreen and at a sharp resolution).

i5-3470. @3.6 GHz running on win7 64bit
Cross over cable, 1gb network (originally with jumbo frames =off)

Performance (initial)
The settings I carried over from the original setup resulted in the dreaded 1-2 pause people mention all over this thread but other than that the cars didn't flicker too much if at all - however the in game freezes eally killed the gameplay.  It seemed to happen randomly.  Sometimes not at all but then a few games in a row.

After turning off network sync(=0), and using the m2lagfix it was stable and removed the freezes but introduced a constant heavy flickering of the other players car (seen from p1 and from p2).  Still more playable but hardly fun.
 

Performance (tweak network adaptor + lagfix)
After reverting back to my original settings eg. network sync =1 the problem disappeared for a while until eventually the 2 second freezes returned.

I followed a gaming ' lag reducing' guides to turn off a bunch of network adaptor settings, turned on jumbo frames, changed the network cable ... amended the UDP cache size (increased windows buffer to 1mb or something)..and still the 2 second freeze problem persisted but now I do not get any purple car flickering at all.

Using the lagfix again but m2network with network sync on(=1), the freeze now seems to have been reduced to some very minor occasional glitches (like a very brief stutter).  Playing around the m2lagfix settings to see if there is anything can be done with the minor glitches and will update this post in case it helps anyone.


Performance  Update (with hacked lagfix with reduced wait times  on both machines)

Using the recompiled m2lag fix with stall detection reduced to a wait of 64ms has resolved all glitching issues now.
I installed the hacked lagfix on both PC's (with throttle=1). See my other post for link.

It seems to have given the emulator the packets it needs for smooth game play.  Very happy with the results now
Note - I'm running just 2 X players using a cross over network cable.

[buttersoft]

Sailorsat, sorry to be pain, but do you have outputs working ok with your 1.1c version?  I can see outputs in mamehooker but they never change state.

I was told, ages ago in another thread, that you needed the scripts from https://github.com/njz3/vJoyIOFeederWithFFB/tree/master/gameassets/nebula_m2_1.1 but that dir is 404 now.

EDIT: ok, new location for those scripts found (thanks @vincerplay) and outputs working again - https://github.com/njz3/BFF_gameassets/tree/dd185d82ea3829033a4d99a65767b3a4739778cc/image_njz3/Emulators/Model2%20v1.1a/scripts

I'd love to figure out how to get STCC leader lamps running next, but that looks like it's going to be much trickier.

EDIT2: @vincerplay modified a version of outputblaster to receive outputs from STCC and Sega Rally Champsionship - https://fastupload.io/1b9ad0d3638c2684
That file is all that is needed to get outputs from m2emu 1.1c. It works with the two games listed, does not work with Daytona anymore, and you would need to test other games.

[SailorSat]

I have outputs for all games in m2em woring, but I use my own tool to get them.

Basically the offset to read them is 'EmulatorEXE + 0x174CF0'

The only Exception is Daytona*, which uses 'M2EM_RAMBASE + 0x824' (and requires the lua patch)

Not very that helpfull I guess.

[buttersoft]

Not very that helpfull I guess.

I wouldn't say that, it's all interesting

[Monkee]

Just discovered this thread and it sounds great but sadly the download link seems to be down. 

[SailorSat]

try https://www.rev-girls.de/m2em/emulator_multicpu_11c.zip

[Monkee]

Thanks a lot SailorSat! I cannot wait to try all those amazing arcade racers in multiplayer just like in my childhood! 

[WolfRagnell]

Sailorsat, sorry to be pain, but do you have outputs working ok with your 1.1c version?  I can see outputs in mamehooker but they never change state.

I was told, ages ago in another thread, that you needed the scripts from https://github.com/njz3/vJoyIOFeederWithFFB/tree/master/gameassets/nebula_m2_1.1 but that dir is 404 now.

EDIT: ok, new location for those scripts found (thanks @vincerplay) and outputs working again - https://github.com/njz3/BFF_gameassets/tree/dd185d82ea3829033a4d99a65767b3a4739778cc/image_njz3/Emulators/Model2%20v1.1a/scripts

I'd love to figure out how to get STCC leader lamps running next, but that looks like it's going to be much trickier.

EDIT2: @vincerplay modified a version of outputblaster to receive outputs from STCC and Sega Rally Champsionship - https://fastupload.io/1b9ad0d3638c2684
That file is all that is needed to get outputs from m2emu 1.1c. It works with the two games listed, does not work with Daytona anymore, and you would need to test other games.

Yo currently adding race leader lights to my setup as I learn the ins and outs. the link for the output blaster sega rally & stcc file is down do you know where else I could find it, or if you could upload it for me? tried searching around to no luck,would love the file especially for sega rally cheers!

[buttersoft]

Hi @WolfRagnell, i've bene moving househad no time to check the forums. I think Vincerplay & MightyMike updated Boomslangz new OutputBlaster to let M2Emu push outputs to MameHooker. Try Boom's channel in the TeknoParrot Discord.

And my god that sounded like a lot of gibberish even to me. If it doesn't make sense or you still get stuck, msg me in a week or two!

[themotherbrain]

try https://www.rev-girls.de/m2em/emulator_multicpu_11c.zip

Hi! 

I've been setting up my own pair of Daytona racecabs and it has led me here to this thread. I have managed to get everything going just great and I am now just down to the flickering car issue. I have download the 11c multicpu exe and throttle linked in this thread but I do not understand what I am supposed to do with it.

If I am using IP 192.168.1.4 for player 1 and 192.168.1.5 for player 2 could you possibly show me how to add this into my files and what values to use please? I assume I keep these files in the model2link folder. Do I need to have these files running on both pcs? Do I need to download anything else to make this work?

If you wouldn't mind giving a little nudge across the finish line id appreciate it so much! Also I wanted to personally say thanks for working on this emulator!    Daytona is one of my absolute favorites and am looking forward to having a bunch more linked up soon! I can't even imagine the effort it takes to figure this stuff out so I want to say I truly appreciate the work you've done!

Many thanks in advance for any help that can be offered! 

[Endprodukt]

Hi @WolfRagnell, i've bene moving househad no time to check the forums. I think Vincerplay & MightyMike updated Boomslangz new OutputBlaster to let M2Emu push outputs to MameHooker. Try Boom's channel in the TeknoParrot Discord.

And my god that sounded like a lot of gibberish even to me. If it doesn't make sense or you still get stuck, msg me in a week or two!

Do you have further information if they made it work? I'm fighting with the same problem of no state change in m2emu.

[buttersoft]

Do you have further information if they made it work? I'm fighting with the same problem of no state change in m2emu.

No, i have not set up FFBBlaster yet! And TBH i'm not sure it's meant to cover M2emu, but maybe it is?

I do, however, have outputs working fine with FFBplugin and Outputblaster. All my files for those two elements, and my complete scripts directory from the m2emu folder, are attached. Using these with the 1.1c update by sailorsat gets me working outputs in mamehooker.

MY setup is mostly aimed at STCC as that game is my fav, so you might need to try the different DLL files in there and see. The one with ronames in the title will give individual rom ini's in mamehooker, not just the generic m2emu.ini or whatever it is that covers all roms.

[ddraigcymraeg]

Hi All,
The original link to M2lagfix download just hangs for me. Can anyone provide another link? Thanks.
Just catching up with all the work done here great job! Looking to get Sega Ski Super G to work for 2 players 4 tops, as well as the other games.

[buttersoft]

Is M2lagfix still needed with the 1.1c version of the emulator and framesync enabled? I'm not saying it's not, just asking...

[SailorSat]

don't think so.

[vertex]

Do you have further information if they made it work? I'm fighting with the same problem of no state change in m2emu.

No, i have not set up FFBBlaster yet! And TBH i'm not sure it's meant to cover M2emu, but maybe it is?

I do, however, have outputs working fine with FFBplugin and Outputblaster. All my files for those two elements, and my complete scripts directory from the m2emu folder, are attached. Using these with the 1.1c update by sailorsat gets me working outputs in mamehooker.

MY setup is mostly aimed at STCC as that game is my fav, so you might need to try the different DLL files in there and see. The one with ronames in the title will give individual rom ini's in mamehooker, not just the generic m2emu.ini or whatever it is that covers all roms.

Using your m2emu files and the Sailorsat version, Mamehooker works, but when I exit, sometimes the lights stay on. I have to close Mamehooker and restart it. This doesn't happen with Mame. Is there a specific reason for this? Or is there a way to fix it?
mamehooker keeps listening to m2emu outputs, but exiting it doesn't delete the information and sometimes the lights stay on.

[buttersoft]

when I exit, sometimes the lights stay on. I have to close Mamehooker and restart it. This doesn't happen with Mame. Is there a specific reason for this? Or is there a way to fix it?

Firstly, glad to hear you got it working

Secondly, sometimes the lights stay on. I imagine it's because when you exit, the game never sends a message to your arduino or whatever turn off the output. Happens to me with most games. MAME must have a close protocol that sends a 0 or something. And Mamehooker must do so on startup, or just re-initialise the arduino or something.

If you really needed to do something about it, you could write an autohotkey script that fires each time you close a problem game, and closes and re-opens mamehooker... That would be the simplest way i can think of to do it. Or maybe your frontend could do it for you, IDK.

[dekkit]

what my "m2lagfix"-tool does

• check status of header; the tool won't do anything until status-2 is reached.
• received frames get buffered, sent to the emulator and (if enabled) a copy is sent to the "control tower"
• if the emulator does NOT respond within 64ms, the buffered frame will be sent to the emulator again.
• double-frames sent by the emulator will be split to get rid of "lost" frames.
• if no frame gets received in 256ms, we have a network stall. (nothing will be done)

each event gets noted in a log file. no response in 64ms = LAG; double frames = DOUBLE; network stall = STALL

Hi Sailorsat - apologies for digging up an old post but I've been going thru the code of the m2lag fix and just want to check if the version on the GitHub is the latest version of your vb code? 

I couldn't see anything in the code the handles the double frames condition (or writes something  to the log when it sees a double). Also the value for the stall condition seems different to your post (128ms vs 256ms unless I read it wrong)?

Do you recommend using the Throttle=1 line in the ini if using lagfix on all cars/ emulator instances? 

 I'm thinking it might be inadvertently causing some glitching in my setup which has lagfix running on each car

Eg when throttle = 1
 car 1 - emulator doesnt send a packet in 128ms, so lagfix detects a throttle and sends a dummy packet for car 2.
Car 2 - meanwhile lagfix on car 2 also realises it didnt receive a packet in 64ms (the LAG condition), so lagfix also creates it's own dummy packet  and sends it to the car 2 emulator
Car 1 - meanwhile car 1 emulator now sends a new 'real' UDP packet which lagfix then immediately sends to car 2.

If I've read the code correctly, is it possible that the car 2 emulator could receive a burst of 3x UDP packets in quick succession (possibly with conflicting information) when throttle =1 and each car has its own instance of lagfix?

Source code for lagfix link: https://github.com/SailorSat/m2lagfix

Lagfix is a smart workaround, many thanks for sharing the source code

[dekkit]

Ok I think I've now resolved the issue for my setup and it has removed all intermittent glitches and lags (however small).   In short, the fix for me was to reduce the Stall detection wait time from 128ms down to 64ms in m2lagfix for my setup.

The reason i can think of why i'd need to do this is because is because i migrated my setup to PC with faster CPUs (but who really knows)

I found the original m2lagfix sourcecode within SailorSats respository https://github.com/SailorSat/m2lagfix and recompile it in a version of VB6 professional i found in archive.org.
In the m2lagfix source code - I just replaced line 224 "If GetTickCount - STALL_Tick >= 128 Then" with "If GetTickCount - STALL_Tick >= 64 Then"

I've put my code changes in a fork next the original m2lagfix (named _stall64ms) and dropped a compiled exe (m2lagfix_stall64ms.exe) here in my github:
https://github.com/dekkit/m2lagfix_stall64ms/

Please note on windows 10 triggers an alert in defender (i'm not sure if there is a way around it - but have shared the source code in case someone can compile or wants to port this to a newer language without this issue)

Sharing in case this helps anyone else experiencing or scratching their heads trying to resolve the random glitches in there setup.

[DJ_Izumi]

Sorry for the necro question, I came across this thread trying to resolve the 'cars flicking in Daytona with network play' thing and the conclusion seems to be 'there's no real fix' but have any of you tried it on a VRR/FreeSync capable display?

With FrameSkip and FrameSync to 1, I had flickering.  Vsync was off since I was using 180hz FreeSync displays, but I decided to turn it on and voila?  The framerate is a bit funny, it's 58fps, then jumps for a moment every few seconds, but nothing flickers and the audio doesn't seem to desync either.

[araujofilipe]

Sorry for the necro question, I came across this thread trying to resolve the 'cars flicking in Daytona with network play' thing and the conclusion seems to be 'there's no real fix' but have any of you tried it on a VRR/FreeSync capable display?

With FrameSkip and FrameSync to 1, I had flickering.  Vsync was off since I was using 180hz FreeSync displays, but I decided to turn it on and voila?  The framerate is a bit funny, it's 58fps, then jumps for a moment every few seconds, but nothing flickers and the audio doesn't seem to desync either.

Honestly, this topic will never be closed until the issue is resolved; it’s been nearly 13 years since the first post, and a lot has happened since then, MAME emulating Model 2 decently, and just a few days ago, teknoparrot also running Model 2 games great (I hope teknoparrot sorts this out once and for all). Still, it would be interesting if you posted a video of the result and a mini-tutorial here.

[Maillouski]

Sorry for the necro question, I came across this thread trying to resolve the 'cars flicking in Daytona with network play' thing and the conclusion seems to be 'there's no real fix' but have any of you tried it on a VRR/FreeSync capable display?

With FrameSkip and FrameSync to 1, I had flickering.  Vsync was off since I was using 180hz FreeSync displays, but I decided to turn it on and voila?  The framerate is a bit funny, it's 58fps, then jumps for a moment every few seconds, but nothing flickers and the audio doesn't seem to desync either.

Honestly, this topic will never be closed until the issue is resolved; it’s been nearly 13 years since the first post, and a lot has happened since then, MAME emulating Model 2 decently, and just a few days ago, teknoparrot also running Model 2 games great (I hope teknoparrot sorts this out once and for all). Still, it would be interesting if you posted a video of the result and a mini-tutorial here.

Agreed. I'm currently fooling around with this custom ARM HW that is able to run Daytona at full 60FPS without a sweat. It runs natively on it. Minimal linux Kernel + Daytona Emulation code with many fixes. It natively boots into Daytona. There's just the Kernel and the emulator. No Operating System. It acts just like a real arcade.

No network issue. No flickering/lagging/synchro issue.

It currently workj with USB steering and I'm going to develop a PCB to interface with the DriveBoard, IO Board PreAmp, and the most difficult, the original CRT (if you don't want to use LVDS or HDMI output). The soundboard will not be needed anymore.
Oh, native 16:9 also. Color is perfect, too.

I want to build this so I can still enjoy my real cabinet without the fear of getting a Soundboard or CPU stack to stop working.

[DJ_Izumi]

Honestly, this topic will never be closed until the issue is resolved; it’s been nearly 13 years since the first post, and a lot has happened since then, MAME emulating Model 2 decently, and just a few days ago, teknoparrot also running Model 2 games great (I hope teknoparrot sorts this out once and for all). Still, it would be interesting if you posted a video of the result and a mini-tutorial here.

Nah I spoke too soon.  I thought it was resolved but it's more like it made it 'less bad'.  With enough time it occurs and it seems nothing is different at all.  So no fix, please ignore my above post.

In the end the only suitable solution was to  use DaytonaThrottle between Machine 1-4.  There are very short audip drops every 15s or so, but it's tolerable.

[araujofilipe]

Honestly, this topic will never be closed until the issue is resolved; it’s been nearly 13 years since the first post, and a lot has happened since then, MAME emulating Model 2 decently, and just a few days ago, teknoparrot also running Model 2 games great (I hope teknoparrot sorts this out once and for all). Still, it would be interesting if you posted a video of the result and a mini-tutorial here.

Nah I spoke too soon.  I thought it was resolved but it's more like it made it 'less bad'.  With enough time it occurs and it seems nothing is different at all.  So no fix, please ignore my above post.

In the end the only suitable solution was to  use DaytonaThrottle between Machine 1-4.  There are very short audip drops every 15s or so, but it's tolerable.

Man, that's a shame, . Anyway, I hope SailorSat find a better solution than DaytonaThrottle. Until there try running the Link System through MAME; I think there's potential there.

[araujofilipe]

Sorry for the necro question, I came across this thread trying to resolve the 'cars flicking in Daytona with network play' thing and the conclusion seems to be 'there's no real fix' but have any of you tried it on a VRR/FreeSync capable display?

With FrameSkip and FrameSync to 1, I had flickering.  Vsync was off since I was using 180hz FreeSync displays, but I decided to turn it on and voila?  The framerate is a bit funny, it's 58fps, then jumps for a moment every few seconds, but nothing flickers and the audio doesn't seem to desync either.

Honestly, this topic will never be closed until the issue is resolved; it’s been nearly 13 years since the first post, and a lot has happened since then, MAME emulating Model 2 decently, and just a few days ago, teknoparrot also running Model 2 games great (I hope teknoparrot sorts this out once and for all). Still, it would be interesting if you posted a video of the result and a mini-tutorial here.

Agreed. I'm currently fooling around with this custom ARM HW that is able to run Daytona at full 60FPS without a sweat. It runs natively on it. Minimal linux Kernel + Daytona Emulation code with many fixes. It natively boots into Daytona. There's just the Kernel and the emulator. No Operating System. It acts just like a real arcade.

No network issue. No flickering/lagging/synchro issue.

It currently workj with USB steering and I'm going to develop a PCB to interface with the DriveBoard, IO Board PreAmp, and the most difficult, the original CRT (if you don't want to use LVDS or HDMI output). The soundboard will not be needed anymore.
Oh, native 16:9 also. Color is perfect, too.

I want to build this so I can still enjoy my real cabinet without the fear of getting a Soundboard or CPU stack to stop working.

wow wow, this has potential! Please document it on YouTube or GitHub; it could be the solution for many who are still struggling with the Link System, If you manage to enable native resolution for CRTs, FFB, and output lights, I’ll ditch MAME and the Model 2 emulator the very next day.

[araujofilipe]

A native Daytona USA recomp project was recently released that offers support for FFB, widescreen, and what interests us most, a link system support; I haven't tested it and don't know if it's stable, but I think it's promising, considering it's open source. I hope SailorSat can explore this project.

Link: https://github.com/alphanu1/daytona-arcade-recomp

[Maillouski]

Sorry for the necro question, I came across this thread trying to resolve the 'cars flicking in Daytona with network play' thing and the conclusion seems to be 'there's no real fix' but have any of you tried it on a VRR/FreeSync capable display?

With FrameSkip and FrameSync to 1, I had flickering.  Vsync was off since I was using 180hz FreeSync displays, but I decided to turn it on and voila?  The framerate is a bit funny, it's 58fps, then jumps for a moment every few seconds, but nothing flickers and the audio doesn't seem to desync either.

Honestly, this topic will never be closed until the issue is resolved; it’s been nearly 13 years since the first post, and a lot has happened since then, MAME emulating Model 2 decently, and just a few days ago, teknoparrot also running Model 2 games great (I hope teknoparrot sorts this out once and for all). Still, it would be interesting if you posted a video of the result and a mini-tutorial here.

Agreed. I'm currently fooling around with this custom ARM HW that is able to run Daytona at full 60FPS without a sweat. It runs natively on it. Minimal linux Kernel + Daytona Emulation code with many fixes. It natively boots into Daytona. There's just the Kernel and the emulator. No Operating System. It acts just like a real arcade.

No network issue. No flickering/lagging/synchro issue.

It currently workj with USB steering and I'm going to develop a PCB to interface with the DriveBoard, IO Board PreAmp, and the most difficult, the original CRT (if you don't want to use LVDS or HDMI output). The soundboard will not be needed anymore.
Oh, native 16:9 also. Color is perfect, too.

I want to build this so I can still enjoy my real cabinet without the fear of getting a Soundboard or CPU stack to stop working.

wow wow, this has potential! Please document it on YouTube or GitHub; it could be the solution for many who are still struggling with the Link System, If you manage to enable native resolution for CRTs, FFB, and output lights, I’ll ditch MAME and the Model 2 emulator the very next day.

Yes. I will release everything on Github once tested.

I’m about 90% done with the IO Board replacement PCB.

[DJ_Izumi]

This is what I was tinkering with the emulator for and this thread was helpful in getting it to a 'functional enough' state.  This is in my basement for now but it's a 4 Player LAN Party exhibition setup for Vintage Computer Fest Montreal where it'll be running typical LAN party games from 20 or more years ago.  But I thought it'd be fun to have Daytona USA ready to go as a bonus cause it'd be fun to see 4 players actually enjoying it's networked function.  There's player instruction cards for the gamepad mappings, all is just copied from the Xbox Live Arcade port layout.  And for fun, the RGB in the peripherals changes to match the player car when Daytona is running.

[Maillouski]

Hi ! Here's the first prototype of my IO board replacement for use with an emulator.

https://github.com/Maillouski/M2-IO-Emulator-Board

It even got a video amp if the input video signal is not strong enough to drive the CRT correctly.

[araujofilipe]

Hi ! Here's the first prototype of my IO board replacement for use with an emulator.

https://github.com/Maillouski/M2-IO-Emulator-Board

It even got a video amp if the input video signal is not strong enough to drive the CRT correctly.

 

[araujofilipe]

This is what I was tinkering with the emulator for and this thread was helpful in getting it to a 'functional enough' state.  This is in my basement for now but it's a 4 Player LAN Party exhibition setup for Vintage Computer Fest Montreal where it'll be running typical LAN party games from 20 or more years ago.  But I thought it'd be fun to have Daytona USA ready to go as a bonus cause it'd be fun to see 4 players actually enjoying it's networked function.  There's player instruction cards for the gamepad mappings, all is just copied from the Xbox Live Arcade port layout.  And for fun, the RGB in the peripherals changes to match the player car when Daytona is running.

what a great setup, did you use the DaytonaThrottle solution?

[Maillouski]

This is what I was tinkering with the emulator for and this thread was helpful in getting it to a 'functional enough' state.  This is in my basement for now but it's a 4 Player LAN Party exhibition setup for Vintage Computer Fest Montreal where it'll be running typical LAN party games from 20 or more years ago.  But I thought it'd be fun to have Daytona USA ready to go as a bonus cause it'd be fun to see 4 players actually enjoying it's networked function.  There's player instruction cards for the gamepad mappings, all is just copied from the Xbox Live Arcade port layout.  And for fun, the RGB in the peripherals changes to match the player car when Daytona is running.

what a great setup, did you use the DaytonaThrottle solution?

No. And it's not yet tested in the cabinet yet. Stay tuned. Firdst two prototype boards ordered.

[Zebidee]

Hi ! Here's the first prototype of my IO board replacement for use with an emulator.

Suggestion - Pop a schottky diode (for example, SS14) on the sync output (from U2, 74HCT86) after R47 (470R). Reverse biased, with cathode (stripe) on sync and anode on GND.

It'll "clamp" the "bottom" of your sync pulses to GND, makes the signal more consistent and detection easier for the monitor/display.

That was just something I saw looking at your photos, didn't go into schematics. I'm quite familiar with that 74HCT86 IC and video amps too, though I see you have njz3 to help already.

So I won't go on, but happy to help/look closer/advise where it might be useful. ATM I am busy busy with GreenAntz v2.4 development, and not otherwise involved in your m2emu project, but that is all this same video amp/sync stuff anyway so it won't cost me much to share tips/experience with you.

[Maillouski]

Hi ! Here's the first prototype of my IO board replacement for use with an emulator.

Suggestion - Pop a schottky diode (for example, SS14) on the sync output (from U2, 74HCT86) after R47 (470R). Reverse biased, with cathode (stripe) on sync and anode on GND.

It'll "clamp" the "bottom" of your sync pulses to GND, makes the signal more consistent and detection easier for the monitor/display.

That was just something I saw looking at your photos, didn't go into schematics. I'm quite familiar with that 74HCT86 IC and video amps too, though I see you have njz3 to help already.

So I won't go on, but happy to help/look closer/advise where it might be useful. ATM I am busy busy with GreenAntz v2.4 development, and not otherwise involved in your m2emu project, but that is all this same video amp/sync stuff anyway so it won't cost me much to share tips/experience with you.

Wow, cool! Thanks for this suggestion!

[Zebidee]

Well, Thanks!   

While I'm here, here is a bigger #tip then!

You are using literally only using half of the 74HCT86's potential. Only two of the four XOR gates are used. This will give you composite sync output but only for double positive (and double negative) sync, and not for csync input. If you are happy with this then that is great, though you could achieve the same outcome with less (two single XOR gates  (e.g. SN74LVC1G86, or the dual-gate option in one package SN74LVC2G86) or a single XOR with an inverter (e.g. SN74LVC1G04).

To access the 74HCT86's full potential, borrow Tomi Engdahl's composite sync circuit, which uses all four XOR gates to generate composite sync from whatever polarity combo you can to throw at it. Yes I know the schematic might look weird with four separated gates but that is actually what you have once you lump it together into a quad XOR package. The only extra parts you need are 2 resistors (2.2k) and two capacitors (22uF).

This will work regardless of the input sync polarity. It uses one full XOR input gate (2 inputs) for each H + V input. The resistor and cap on the second XOR input delays the pulse slightly, so that the XOR condition is satisfied and the gate registers regardless. The only real limitation is that it won't produce csync from Csync *input*, though you can rectify this (on a case-by-base basis) by blanking (pulling high) the Vsync input.

GreenAntz v2.3+ uses this basic sync circuit, but with proper AC coupling, buffering and filtering to tidy-up and clamp the pulses, and some Magic Sauce™ added so that it also works with csync input too, no manual blanking required.

Works with VGA TTL-level sync input (~3.5 to 5v). It works with Mister FPGA (~3.3v sync) too, though for GreenAntz I had to tweak the input impedance and AC coupling to sync reliably at the lower levels (have now revised the GreenAntz design to make this standard). You can push that input sensitivity even lower, if needed, by using a lower supply voltage to the 74HCT86. I expect this circuit would not work for HD modes with tri-level sync.

Peace 

Attached: Tomi Engdahl's unaltered original circuit. The 74HCT86 is a essentially more modern version of the 74LS86 specified.