New 3DO RGB Mod Possibility.

[the_crayon_king]

Would it be possible to integrate a 68 position PLCC socket into the design?

Like this one:https://www.mouser.com/ProductDetail/3M-Electronic-Solutions-Division/8468-21B1-RK-TR/?qs=sGAEpiMZZMs6akeJfFiy2rc%2ftN2YUFFe

This would make the mod infinitely easier, just snap on and solder wires to the output connector. The NeoBIOSMasta (https://www.neogeofanclub.com/projects/neobiosmasta-og) uses this technique.

 

Edit: Though this may not be possible due to components too close to the IC.

 

Anything is possible., I think the board as is would be the next step since it would still be technically compatible with the other 3DO revisions.

 

I have been too sick to update but I see now there is the same issue with sync across all these 3DO RGB kits.

That sync issue being that the image isn't centered on screen.

 

It can be adjusted at the TV end but that is not a great solution.

Basically H/V sync is slightly out of time with RGB (I think).

 

What all the boards need is a monostable multivibrator.

I think the idea is to delay H/V sync (which is already delayed) until it rolls over to being correct again. (really talking out my ass here so I dunno if that is even close to correct)

 

The otaku board had a solution for this and the specs are out there (in German). Whatever the solution involving the monostable multivibrator is, it should be the same for all these 3DO rgb kits, only slightly different if using the lm1881 (but I may drop that out of my design).

 

Ill check my notes on this and update when I have more info.

 

Old : https://i.imgur.com/LxVviOP.jpg

New: https://i.imgur.com/UsXEL6a.jpg

 

I would normally source this type of stuff but I have no clue where I got it from.

Ill scrub the Otaku board I have here to see what we can learn from it.

Edited November 5, 2017 by the_crayon_king

[the_crayon_king]

Some notes., VSYNC should get XORED (or gate) together with HSYNC to make C-Sync.

Just need to run HSYNC through a mono-stable multi-vibrator.

 

It should be noted that the old schematic shown on retro-rgb does not correlate to the Otaku boards sync whatsoever.

 

I scrubbed the board to see what they did. Looks like they were adjusting the voltage to a single pin by somewhere between 22k-50k

VCC into 25k pot into 22k resistor. I am assuming they had a reason to not just use a 50k pot.

I cannot correlate the chip they used to any normal mono-stable vibrator and the chiseled the dang info off of there.

 

The German version posted as "new" just needs some slight adjustments to be usable.

Dump all the inverters, add an XOR gate.

May need an inverter on OE depending on some stuff. OE would be blanking.

Ill need to read up on this type of chip to see where a potentiometer could be placed to adjust timing of Hsync.

[citrus3000psi]

Glad you are making progress with this :-) Yeah I'm not sure about Blank either. Like you said, the timing could be off by using the Back Porch signal. I've noticed that the BT9101 and BT9103 both use Blank signal. I will try and trace my BT903 Blank pin to the CLIO chip. Hopefully the same pin on VA536 motherboards will provide an active Blank signal.

 

Do you have a picture of the RGB mod chip you found? It sounds like they might have used a BU3616K chip which is about the same size as the ADV1725 but simpler in design and application.

 

 

Do your HSYNC and VSYNC trace back to the CLIO as well? I'm curious to know if the CLIO sends out a 240p signal and these encoders are interlacing them to 480i. I'd like to maybe remove the VP536A entirely....

 

This might also solve the sync issues that @the_crayon_king is having.

Edited November 9, 2017 by citrus3000psi

[Taijigamer]

It looks like the VP536 and BT9103 just get CLK and 24bit RGB from the mobo/ CLIO, the rest of the timing is managed by the encoder. Both encoders receive 2 clks and create either 240p or 480i (288p or 576i for PAL) depending on the progressive pin state. The only way to get RGB on these consoles is to insert a new RGB capable encoder and feed the necessary signals to it.

 

The problem with horizontal shift is the Hsync is being delayed relative to the processed RGB. the pipeline delay within the encoders is minimal but in the Sync stripper or XOR chip it is more so the 2 parts of the signal are out of time. Maybe just using Luma as sync would eliminate the delay but it is undesirable.

 

As we can't speed up the Hsync, we need to delay it by a complete cycle (64us). As the horizontal shift is about 8us delayed, we need to delay by about a further 56us.

 

With the BT encoders, there are 2 further possible solutions. The BT9101 outputs CSYNC natively so this could be used as is with a 480 Ohm resistor to attenuate the TTL signal.

 

The BT9103 has the same pinout as the BT9107 which is RGB capable so these could be swapped out on the mobo giving native RGB. I still need to test this one.

[the_crayon_king]

It looks like the VP536 and BT9103 just get CLK and 24bit RGB from the mobo/ CLIO, the rest of the timing is managed by the encoder. Both encoders receive 2 clks and create either 240p or 480i (288p or 576i for PAL) depending on the progressive pin state. The only way to get RGB on these consoles is to insert a new RGB capable encoder and feed the necessary signals to it.

 

The problem with horizontal shift is the Hsync is being delayed relative to the processed RGB. the pipeline delay within the encoders is minimal but in the Sync stripper or XOR chip it is more so the 2 parts of the signal are out of time. Maybe just using Luma as sync would eliminate the delay but it is undesirable.

 

As we can't speed up the Hsync, we need to delay it by a complete cycle (64us). As the horizontal shift is about 8us delayed, we need to delay by about a further 56us.

 

With the BT encoders, there are 2 further possible solutions. The BT9101 outputs CSYNC natively so this could be used as is with a 480 Ohm resistor to attenuate the TTL signal.

 

The BT9103 has the same pinout as the BT9107 which is RGB capable so these could be swapped out on the mobo giving native RGB. I still need to test this one.

I think it is the pipeline delay of the VPA. Output sync is in line with the video out but RGB is a video input.

doujindance has a similar kit putting out H/V sync those being put direct out to the TV and instead are adjusted on the TV end (do not want).

 

I think the simplest solution is to delay H sync and then mix (logic gate) V sync for TTL C sync. Then resistor/jumper to toggle between.

there are tons of older solutions but im trying to wrap my head around how to do it with just one or two ICs and a potentiometer.

 

I think the true solution here is something called a programmable delay block for Hsync then mix with V sync to make corrected C sync. Leaving direct outputs for corrected H/V sync as well if possible. This would do away with the LM1881 and will take up about the same space. That means the blanking this is out the door but Im not really sure it will be relevant for most setups.

 

Check the datasheet but I think this is right ?

 

Edited November 10, 2017 by the_crayon_king

[citrus3000psi]

I designed a board over the weekend.

 

 

 

 

I looked at how the otakus board was handling the horz shift. Its funny that they remove the chip markings .

Anyways, I did a little digging, the otakus is based originally on Tim's design:

 

 

That mystery IC is a ne558d quad timer. Now one thing that has me a little confused is that that the otakus board runs on 3.3V. Why they are running it on 3.3, IDK. But the NE558D isn't spec'd to run off 3.3V, but I guess it does in a real life scenario.

I wonder if their XOR gate was only rated up to 3.3V...

Edited November 13, 2017 by citrus3000psi

[the_crayon_king]

I designed a board over the weekend.

 

 

 

 

I looked at how the otakus board was handling the horz shift. Its funny that they remove the chip markings .

Anyways, I did a little digging, the otakus is based originally on Tim's design:

 

 

That mystery IC is a ne558d quad timer. Now one thing that has me a little confused is that that the otakus board runs on 3.3V. Why they are running it on 3.3, IDK. But the NE558D isn't spec'd to run off 3.3V, but I guess it does in a real life scenario.

I wonder if their XOR gate was only rated up to 3.3V...

 

 

This is what I scrubbed from the otaku board:

https://i.imgur.com/Yo7Obu8.jpg

 

That may solve some mysteries I didn't learn much from it.

 

You board looks great and like a viable solution (the LTC6994 is not). If that is using the ADV7125 I believe you can switch blue and green like I did that way you can have it at a 45 degree angle. Looks like you fit everything on there regardless so that may not be worth the effort. C1, C3, C4 are all too close for my comfort. Kudos if you can solder that close and not snap a bridge. I always seem to...

 

So on Tim's design I see R7 and R5 in parallel why not just use one resistor there ?

Edited November 13, 2017 by the_crayon_king

[citrus3000psi]

Few things to add, that is not a transistor not a transistor, but a voltage regulator. The small IC is an XOR gate.

 

Compare your scrub to Tim's schematic

[the_crayon_king]

Few things to add, that is not a transistor not a transistor, but a voltage regulator. The small IC is an XOR gate.

 

Compare your scrub to Tim's schematic

 

Lol Im doing that now. I do see it. It looks very much the same. I look this over further in the next few hours.

From what I can tell it is the same design with a few modifications here and there. That 3.3v thing could be relevant to everyone's issues with these otaku boards no ?

[citrus3000psi]

 

Lol Im doing that now. I do see it. It looks very much the same. I look this over further in the next few hours.

From what I can tell it is the same design with a few modifications here and there. That 3.3v thing could be relevant to everyone's issues with these otaku boards no ?

 

Very well could be.

 

 

FYI, Tim's diagram mislabels the Pin1 on the LM1881. As that should be CSYNC, not HSYNC.

[the_crayon_king]

 

Very well could be.

 

 

FYI, Tim's diagram mislabels the Pin1 on the LM1881. As that should be CSYNC, not HSYNC.

 

It looks like he is using the back porch to derive Hsync instead of the Hsync pin ?

 

I can see from the datasheet that this circuit is in the "Long-time delay" configuration.

 

Is there a substitute for the NE558/NTE926 ? I don't see a reliable source for those chips.

 

Maybe 3x LM555 ?

Edited November 14, 2017 by the_crayon_king

[citrus3000psi]

 

It looks like he is using the back porch to derive Hsync instead of the Hsync pin ?

 

I can see from the datasheet that this circuit is in the "Long-time delay" configuration.

 

Is there a substitute for the NE558/NTE926 ? I don't see a reliable source for those chips.

 

Maybe 3x LM555 ?

 

There is no "hysnc" pin on the lm1881. But the backporch has this info, so that's why Tim is using it.

 

The Ne558 is now an obsolete part, I just sourced some from ebay. Yeah triple 555 would work fine.

[the_crayon_king]

Sorry for the delay I am very busy with non console stuff. BOO!

Lets actually try the LTC6994CS6-2 before we count it out...

 

344k is base

 

55us = 344-344=0k
55.5us= 348-344=4k
56us = 352-344=8k
57us = 357-344=13k
57.5us= 361-344=17k

 

That should be enough deviation one side or another and will remove the need for a potentiometer.

Assuming the IC will work at all here is the schematic will design the board over the holidays:

Go here to check math: http://www.linear.com/tools/Products/LTC6994/LTC6994.php

Edited December 24, 2017 by the_crayon_king

[Taijigamer]

That's cool that u are making progress on this. I may have been off with my original statement of 56us. I was looking at this page http://www.sega-16.com/forum/showthread.php?25988-RGB-LEFT-RIGHT-sift-issues-theory-and-fixes

And mistook 8pixel for 8us. So the shift needs to be more like 62us.

 

My 3DO is out of action at the moment as I tried to swap the encoder for a BT9107 but it didn't work :-(

 

I think if I revisit my RGB mod design, I would make it similar to yours with horizontal correction and TTL selection.

[the_crayon_king]

That's cool that u are making progress on this. I may have been off with my original statement of 56us. I was looking at this page http://www.sega-16.com/forum/showthread.php?25988-RGB-LEFT-RIGHT-sift-issues-theory-and-fixes

And mistook 8pixel for 8us. So the shift needs to be more like 62us.

 

My 3DO is out of action at the moment as I tried to swap the encoder for a BT9107 but it didn't work :-(

 

I think if I revisit my RGB mod design, I would make it similar to yours with horizontal correction and TTL selection.

 

Well it looks like the LTC6994-2 can't be used here refer to:

 

"The LTC6994-2 will delay both the rising and falling edge of the input signal. The output only responds to input changes that persist longer than the delay period. The LTC6994-2 is ideal for switch debouncing, delay matching and guaranteeing that a signal meets minimum pulse width requirements."

 

After looking at it till my head hurts the LTC6994-1 doesn't change timing exactly.

 

Rising-Edge Delayed:

plainly when HSYNC happens is the same how long the pulse last only get wider.

 

 

Falling-Edge Delayed:

Delays Hsync but at the cost of making it more narrow.

Or to say if Time-delay is longer than the width of HSYNC then it won't make a transition at all.

-------------------------------------------------------------------------------------------------------------------------

I think I understood that correctly but maybe somehow that could work but idk doesn't sound very promising.

I'ma have to draw a picture.

 

The top two should delay sync by the time between two sync pulses minus the length of a sync pulse.

I'd be real curios to see what that would do in a circuit... am I missing anything ?

 

Edited December 26, 2017 by the_crayon_king

[the_crayon_king]

So potential solution .We are delaying the transition to 5v making HS (0v) wider; so if we use that positive edge as a trigger then we can

use a LTC6993-1 or a Negative output pulse, Positive edge triggered, non retriggerable one shot to do this:

This would require the LTC6994-1 and the LTC6993-1.

Schematic for reference: link

 

Edited December 26, 2017 by the_crayon_king

[citrus3000psi]

So potential solution .We are delaying the transition to 5v making HS (0v) wider; so if we use that positive edge as a trigger then we can

use a LTC6993-1 or a Negative output pulse, Positive edge triggered, non retriggerable one shot to do this:

This would require the LTC6994-1 and the LTC6993-1.

Schematic for reference: link

 

 

 

 

Interesting idea. I started looking at line delays the past few days and I was eyeing the LTC6994-2. I'm not able to get anywhere with the shift fix using the quad timer. It works, but I get a vertical scroll issue that I can't seem to resolve. This is the same issue the the original otaku RGB mod has.

 

https://youtu.be/lkf5Z1w8Be8

[the_crayon_king]

 

 

Interesting idea. I started looking at line delays the past few days and I was eyeing the LTC6994-2. I'm not able to get anywhere with the shift fix using the quad timer. It works, but I get a vertical scroll issue that I can't seem to resolve. This is the same issue the the original otaku RGB mod has.

 

https://youtu.be/lkf5Z1w8Be8

 

Well at least its centered lol. Seeing as the otaku board uses the same timer id expect the same result

 

Yea no the LTC6994-2 wont work unless I am misreading the datasheet:

"Also unlike the LTC6994-1, the output pulse width (of the LTC6994-2) can never be less than tDELAY ."

That means its positive or negative pulses can't be shorter than the delay on the line. (or well that's how I interpreted it)

The pulse width of sync is very short and we need a long delay.

 

Did you ever test the dimensions of your board ? I wanted some input on if I got the size right but I suppose it would be faster to just print them and adjust.

 

Realistically I need to take the 220uf capacitors off of my design since the cables I want to interface with should already have capacitors in the lines.

The oldest version of a 3DO rgb mod used a monostable it's the one shown on retrorgb.

 

The schematic here shows the design I want to try but I doubt Ill have enough time

R3 and R4 are the only two that need changed. R3 should be a 500k pot.

R4 just needs to be set against the width of hsync I don't know the specific length but you could use another 500k pot and set from 0.1us all the way to 10us width.

[citrus3000psi]

 

Well at least its centered lol. Seeing as the otaku board uses the same timer id expect the same result

 

Yea no the LTC6994-2 wont work unless I am misreading the datasheet:

"Also unlike the LTC6994-1, the output pulse width (of the LTC6994-2) can never be less than tDELAY ."

That means its positive or negative pulses can't be shorter than the delay on the line. (or well that's how I interpreted it)

The pulse width of sync is very short and we need a long delay.

 

Did you ever test the dimensions of your board ? I wanted some input on if I got the size right but I suppose it would be faster to just print them and adjust.

 

Realistically I need to take the 220uf capacitors off of my design since the cables I want to interface with should already have capacitors in the lines.

The oldest version of a 3DO rgb mod used a monostable it's the one shown on retrorgb.

 

The schematic here shows the design I want to try but I doubt Ill have enough time

R3 and R4 are the only two that need changed. R3 should be a 500k pot.

R4 just needs to be set against the width of hsync I don't know the specific length but you could use another 500k pot and set from 0.1us all the way to 10us width.

 

Sounds like you are right about that LTC6994-2.

 

I will mess around with the timers and see what I can come up with. Your design looks promising.

 

I have soldered my design in and the fitment is fine:

.6mm holes with 1.27mm spacing

Hole spacing across the chip is 24mm.

 

[the_crayon_king]

 

Sounds like you are right about that LTC6994-2.

 

I will mess around with the timers and see what I can come up with. Your design looks promising.

 

I have soldered my design in and the fitment is fine:

.6mm holes with 1.27mm spacing

Hole spacing across the chip is 24mm.

 

 

So your design IS using the ADV7125. I had thought it might be that other IC. I like how its dwarfed by that quad timer.

 

I picked out the parts needed in here: http://www.digikey.com/short/q8cj24

 

There are alot of extra stuff in that cart but still faster than looking for them manually, This is if you are using 0603 sized stuff.

I am taking all the 0402 stuff out of my designs if possible I can't handle the eye strain.

 

Also thanks, I think I got the spacing now. I was going off of the package size for the PLCC68 and just building around it.

 

Two question's:

 

What kind of clearance do you have under that circuit board ?

Do you have any scope images of the combined H and V sync ?

[citrus3000psi]

 

So your design IS using the ADV7125. I had thought it might be that other IC. I like how its dwarfed by that quad timer.

 

I picked out the parts needed in here: http://www.digikey.com/short/q8cj24

 

There are alot of extra stuff in that cart but still faster than looking for them manually, This is if you are using 0603 sized stuff.

I am taking all the 0402 stuff out of my designs if possible I can't handle the eye strain.

 

Also thanks, I think I got the spacing now. I was going off of the package size for the PLCC68 and just building around it.

 

Two question's:

 

What kind of clearance do you have under that circuit board ?

Do you have any scope images of the combined H and V sync ?

 

I do have a design for other chip, but space started being an issue (and that chip is obsolete) so I switched to the ADV.

 

Thanks, I will check out the parts. I'm using these pin headers https://www.ebay.com/itm/11mm/171044943199 which offer quite a bit of clearance. The board could have components on both sides without issue using these legs.

 

Here is an image of sync after going through the quad timer.

 

 

And here is a video of me playing with the 6994-2. (No quad timer, just straight H/V) The results surprised me as the screen shifts left? Heres a video of what I'm talking about:

 

[the_crayon_king]

 

I do have a design for other chip, but space started being an issue (and that chip is obsolete) so I switched to the ADV.

 

Thanks, I will check out the parts. I'm using these pin headers https://www.ebay.com/itm/11mm/171044943199 which offer quite a bit of clearance. The board could have components on both sides without issue using these legs.

 

Here is an image of sync after going through the quad timer.

 

 

And here is a video of me playing with the 6994-2. (No quad timer, just straight H/V) The results surprised me as the screen shifts left? Heres a video of what I'm talking about:

 

 

Im going of memory here but delay on Hsync shifts the image to the left.

 

This may explain it better than I could:

https://www.extron.com/company/article.aspx?id=adsp

 

The signal of the quad timer is noisy could explain the video issues ?

 

So as your adjusting to the left (delaying h sync) the video should wrap back around from the right at some point.

I didn't think you could/can adjust past the width of Hsync with just an LTC6994-2 but maybe I am misunderstanding the datasheet.

 

What that page is saying would be like adjust the sync till you get a delay of 1 cycle

 

So say subtract like 50k from whatever the approximate resistor should be then have a 100k pot in series

that should let you adjust video from the right to left (i think)

it is possible that a specific resistor could be used and the pot be done away with entirely.

[citrus3000psi]

 

Im going of memory here but delay on Hsync shifts the image to the left.

 

This may explain it better than I could:

https://www.extron.com/company/article.aspx?id=adsp

 

The signal of the quad timer is noisy could explain the video issues ?

 

So as your adjusting to the left (delaying h sync) the video should wrap back around from the right at some point.

I didn't think you could/can adjust past the width of Hsync with just an LTC6994-2 but maybe I am misunderstanding the datasheet.

 

What that page is saying would be like adjust the sync till you get a delay of 1 cycle

 

So say subtract like 50k from whatever the approximate resistor should be then have a 100k pot in series

that should let you adjust video from the right to left (i think)

it is possible that a specific resistor could be used and the pot be done away with entirely.

 

 

That makes a lot of sense about 1 cycle delay.

 

Yes the Quad timer outputs a noisy line. My initial reason for trying the 6994-2 was to use it as a Noise Discriminator (found on page1 of the datasheet) to try and clean up the signal. But so far have been unsuccessful this way.

 

I'll keep messing with stuff and see what I can figure out.

[the_crayon_king]

 

 

That makes a lot of sense about 1 cycle delay.

 

Yes the Quad timer outputs a noisy line. My initial reason for trying the 6994-2 was to use it as a Noise Discriminator (found on page1 of the datasheet) to try and clean up the signal. But so far have been unsuccessful this way.

 

I'll keep messing with stuff and see what I can figure out.

 

I would think that the LTC6994-1 and LTC6993-1 might work out better but you should be able to use the quad timer in combo with the LTC6994-2 have you tried that ?

 

I would be curious of the 6994-1 and 6993-1 mostly for the space savings.

[citrus3000psi]

 

I would think that the LTC6994-1 and LTC6993-1 might work out better but you should be able to use the quad timer in combo with the LTC6994-2 have you tried that ?

 

I would be curious of the 6994-1 and 6993-1 mostly for the space savings.

 

Still can't get the quad and 6994-2 to accomplish anything useful. I just ordered the 6994-1 and the 6993-1 just now from digikey. I'll see if I'll have any luck with that. I'll keep you posted.

Edited January 17, 2018 by citrus3000psi