RGB/YPbPr Intellivision Thoughts

[the_crayon_king]

Just so everyone knows I am moderately confident I could retime sync to fit into whatever narrow tolerance the RT5X is wanting. Provided I can get my hands on one. 

 

[Icelvlan]

I wonder if anyone can send one over for testing. I think this is the last thing. Don’t HDRV cables use the same type of sync as well?

Edited June 15, 2021 by Icelvlan

[the_crayon_king]

16 hours ago, Icelvlan said:

I wonder if anyone can send one over for testing. I think this is the last thing. Don’t HDRV cables use the same type of sync as well?

The HDR is using sync on composite video not composite sync. I know that terminology is confusing. So far things I need to possess to be truly finished with this project: 

PAL Intellivision (this will need slightly different sync/timing correction compared to NTSC). 

HDR (I would want to try the RGB and composite as it's video/sync source and see if it works). I am still working on boards for composite.  

RT5X

A LTO would be nice to have but I have gotten this far without one.

 

I would rather not borrow things (because I could very well break them) but I could buy them or do trades or something. I really want a PAL motherboard/console since that should be easy to get implemented into the current code. 

[the_crayon_king]

Presenting Chili Fries V1:

 

I realize now this looks a lot like the board I posted the other day; I assure you there were dozens of minute changes. 

 

The board costs me 2.5 times more than the V5. This is mostly because of the two op amps.

This doesn't include the cost of misc components like the encoder or the RCAs, TRRS jack, S-Video din, etc. 

That doesn't mean this would be 2.5 times more expensive overall.

I could have saved a good deal of money by not including YPbPr but since this is what I usually play games over this was not something I wanted to do.

Nor did I want to have a addon to re-add YPbPr. The price of having YPbPr always active was a little over 1$ per board so I think it is better this way. 

 

PAL actually means PAL region this time. Set the two PAL jumpers for PAL or the one NTSC jumper for NTSC. 

(PAL still needs some coding; I can't do anything until I get a PAL console). I will probably make two different firmwares one for PAL and one for NTSC. That way EXT can be used for something more interesting than just setting PAL/NTSC. 

 

ALTCOL would now be the jumper for the alternate pallet.

 

I am still on the fence on whether I should make this or not. I was just wanting to show this is more or less what the board would look like. 

Chili Fries is 1.56x3.24 inches vs Baked Potato being 1.03x2.04 inches. Keep in mind this board should still be technically inferior to the V5 board but may be useful for people wanting to use the HDR, Composite, or S-Video. Testing will happen 2 weeks from now (if I do order). 

 

Let me know if you have any questions.

Edited June 16, 2021 by the_crayon_king

[+-^CrossBow^-]

10 hours ago, the_crayon_king said:

Presenting Chili Fries V1:

 

I realize now this looks a lot like the board I posted the other day; I assure you there were dozens of minute changes. 

 

The board costs me 2.5 times more than the V5. This is mostly because of the two op amps.

This doesn't include the cost of misc components like the encoder or the RCAs, TRRS jack, S-Video din, etc. 

That doesn't mean this would be 2.5 times more expensive overall.

I could have saved a good deal of money by not including YPbPr but since this is what I usually play games over this was not something I wanted to do.

Nor did I want to have a addon to re-add YPbPr. The price of having YPbPr always active was a little over 1$ per board so I think it is better this way. 

 

PAL actually means PAL region this time. Set the two PAL jumpers for PAL or the one NTSC jumper for NTSC. 

(PAL still needs some coding; I can't do anything until I get a PAL console). I will probably make two different firmwares one for PAL and one for NTSC. That way EXT can be used for something more interesting than just setting PAL/NTSC. 

 

ALTCOL would now be the jumper for the alternate pallet.

 

I am still on the fence on whether I should make this or not. I was just wanting to show this is more or less what the board would look like. 

Chili Fries is 1.56x3.24 inches vs Baked Potato being 1.03x2.04 inches. Keep in mind this board should still be technically inferior to the V5 board but may be useful for people wanting to use the HDR, Composite, or S-Video. Testing will happen 2 weeks from now (if I do order). 

 

Let me know if you have any questions.

Well the ability to have s-video alone and I'm sure a much better and most consistent composite output is excellent to hear! The new board size is about the same size and I think still smaller than Yannaros's current design. But still both are small enough to install just about anywhere. The caps on your new board are the only concern I have, as I wonder how much taller this makes the board?

 

Would it be possible to use ceramic or tantalum in place of the electrolytic in this case?

 

Another option to keep the pricing low if you were to offer these... Give options? As an example I don't really need switches or AV jacks since I have plenty of those on hand with other kits already or purchased on hand anyway. Basically a populated board only option for example?

 

[the_crayon_king]

6 hours ago, -^CrossBow^- said:

Well the ability to have s-video alone and I'm sure a much better and most consistent composite output is excellent to hear! The new board size is about the same size and I think still smaller than Yannaros's current design. But still both are small enough to install just about anywhere. The caps on your new board are the only concern I have, as I wonder how much taller this makes the board?

 

Would it be possible to use ceramic or tantalum in place of the electrolytic in this case?

 

Another option to keep the pricing low if you were to offer these... Give options? As an example I don't really need switches or AV jacks since I have plenty of those on hand with other kits already or purchased on hand anyway. Basically a populated board only option for example?

 

Those caps make the overall board thickness about 1/3rd an inch. I will change the caps (to tant) just so I don't have to worry about height. 

This board will have to have some sort of options. I would like to source a multiout or something and just put everything on it.

 

The price of these boards really mostly reflects the time it takes me to put them together plus cost of components. Baked Potato would take much more time than Chili Fries but Chili Fries will require many additional components. That's a fun sentence lol. Anyway, IDK how the price for that board will work out.

 

The cheapest I could make a RGB kit would be a DAC ladder into an op amp (likely the THS7374). That is if I can get the fabs to populate the THS.

Something like that I would sell for 25$. Less for bulk, less for no components. 

I have about a dozen projects I have been sitting on waiting for JLPCB to get into gear and now I am scrambling to revalidate stuff from years ago. 

I am researching how accurate I can make the DAC which would apply to the NTSC encoder thing and this.

I am having trouble finding 0.1% resistors that this fab will populate. So that is what I am working on right now.

 

Just an update but the second batch of V5 and that little NTSC expansion board for it should be here later today. 

Edited June 16, 2021 by the_crayon_king

[the_crayon_king]

So, I goofed up on the expansion board since I forgot to add pads for the inputs. < This would be the board that was meant to add onto V5 and add Y/C and CV. 

I will probably use this to test and make a full board off of the results. It will still be open source so there will be no need to order from me. 

This was the general idea: https://i.imgur.com/lRtewtR.jpg

But on all the inputs there should be pads. Also I had to add a little addon resistor thing to make up to 375 ohms for 3 pulldowns (I don't have a 375/374 ohm resistor on hand). These pulldowns basically set the P-P of the video.

 

This is what S-Video (via V5 expansion) looks like:

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

https://i.imgur.com/tbeOsm7.jpg
https://i.imgur.com/b7NLyzj.jpg

https://i.imgur.com/b7NLyzj.jpg

 

To me it looks almost as good as YPbPr but not quite. Still worlds apart from RF, and trumps any composite mod out there for this system. 

 

This will be the bar that I measure "Chili Fries" against.

I will be changing Chili Fries to be CPLD > R2R DAC > THS7374 > RGB out split (one to TV one the NSTC ENCODER)> NTSC ENCODER.

I think that could be manageable for most people and I will try to keep the price the same.

If Chili Fries has a visibly worse video I will not produce it and instead continue on the expansion board path. 

Using this board will free up about 50% of the code space and technically has more bits 8:8:8 vs 7:7:7; The limiting factor will be power accuracy (so far 1%) and DAC ladder resistor accuracy (also currently 1%).

I would like to get the ladder down to 0.1% or 0.25% but there will be some hurdles since I can only work with what that fab provides. 

 

[the_crayon_king]

So I have been working on three different boards in tandem.

An expansion board for the V5 board (since I goofed up the first one). < does the same as Chili Fries [when used with the V5] (should be better than Chili Fries). 

Chili Fries which is done more or less. < uses a dead part but can put out RGB, YPbPr, S-Video, Composite. No LPF from the THS (no room). 

The another cheaper option which has yet to be named. < R2R DAC into THS7374 (RGB only but expandable for everything else).

I have found 0.1% resistors btw (which was the hold up before). 

 

That is the cheaper board. I should be able to sell it for 25-30$ I am waiting to get the parts confirmation from the PCB manufacture (they said wait a week and ask again).

It puts out two different SRGB signals so one can be used for the encoder and the other can be used for normal RGB output.

This board will sort of be more future proof since the end encoder for the other video formats can be easily changed down the road.

 

I have two tentative fixes for the 5X. In lieu of owning one I will just have to implement these fixes and get confirmation from someone else.

I could exchange a board with this fix to someone that has one of the current 5Xs. 

[Icelvlan]

Cool hopefully someone can test 5X

[the_crayon_king]

16 minutes ago, Icelvlan said:

Cool hopefully someone can test 5X

I still haven't gotten around to making the sync changes. I wanted to do two and set the other one as a back up (on the EXT pin). So if my corrected sync still doesn't work that the other one will work.

The reason I wouldn't use the other one as default is because it is a very broken looking sync signal (although it seems to work [it is what Yannick's board is doing]). I will do this before the week is up. 

That broken sync (which would be activated by jumping EXT should also work for PAL. I call it broken because there is like 1300us where sync is completely flat where that gap would normally be 190us. So you can imagine why I wouldn't want to use it unless I have to. 

 

 

Uh well in other news V5 is still the final board. Those other versions are now abandoned since they cost more and are less accurate than what I am already doing. 

So that means if you want/need S-Video or Composite you will have to use an addon board. 

I went over the last design and added a THS7374 (so as to not degrade RGB once split). Then I added the BH7236AF (for composite and s-video) then finally a way to convert the RGB output from the BH7236AF into PB and PR. So that if you absolutely wanted to you could put out everything all at the same time. 

 

YPbPr being converted in analog (instead of digital) means it would be less accurate.

If you wanted Composite or S-Video then you can no longer use the SOG or the YPbPr jumpers on the V5 mainboard (video won't work if you try). 

 

Confusingly now there are jumpers for PAL and NTSC that need to be set on this expansion board. So "PAL" on the expansion board refers to the PAL region and "PAL" on the V5 (mainboard) refers to color pallet. 

I will change this on the V5 board once I deplete my current stock. 

 

Link to the board below. BOM is attached. 

https://oshpark.com/shared_projects/eGXT27kM

BOM_CPLD EXPANSION - BH7236AF_2021-06-22.csv

[the_crayon_king]

I changed the sync to a simpler looking one (removed the inverted pulses inside of Vsync).

I also took the sync logic of : Sync = 1XXX1 NOR 11XXX and made that trigger on bridging the EXT pin. This is what Yannick's board does [which works with the 5X?].

 

Anyway, it is my hope that the default (retimed) mode works as-is on the 5X.

 

EXT not set (default sync) [191.1us gap for vsync]:  

 

EXT set (1XXX1 NOR 11XXX) [1274us gap]:

 

So boards ordered from now on will be setup like this.

The 1XXX1/11XXX mode should also work on PAL consoles. I think it should be more obvious why I didn't want to use that as the default.

 

Anyway, the point being is one of these two modes SHOULD work on the 5X going forward.

I don't have one to confirm but if someone wanted to order one to try that would be sweet. If it doesn't work I would refund and you would be out like 2$ in shipping. 

[+-^CrossBow^-]

There is one visual anomaly that remains on @YannAros current boards that I hard modified to work. That is that the very upper portion of the image is skewed a bit towards the right. It is like the vsync isn't quite hitting where it should be and it is visible because of this. However, it is only doing this in what would have been the overscan area of the image anyway. So it doesn't actually affect the game play field at all. Many games use a black background or black border around the game field and this visual anomaly isn't shown in those instances. 

 

It will appear on all of the LTO menu screens and title screen but again doesn't cause any video drop outs or anything like that in the LTO. It is also visible on most of the games that use a title screen that use the Mattel style screen before beginning the game. But once in the game, the anomaly either isn't present due to background and border being black, or it is present but not causing any visual issues within the game play field.

 

The videos I've posted in the past in this thread I think show the effect I'm talking about. But here are some screencaps from my YT vids that demo the boards:

 

Here is the LTO title through Crayon's v5 board:

 

And here is the same title screen through Yannick's current version of the board where I've made the hardware modifications to the board to get it stable on OSSC:

 

But, this effect isn't present when using the cheap SCART to HDMI converters and through the OSSC it is a bit less pronounced but still present though darker and blends in more on the screen.

 

My guess is that the anomaly on the top on Yannick's board could be due to the longer delay Crayon mentioned in his last post?

 

Both shots were captured with the boards connected directly to my Tink2x SCART using a set of Insurrection Industries Genesis 2 RGB SCART cables. OSSC with Yannick's board has the same effect only you don't see the white/grey colors along the top nearly as bright and so it just looks cut off along the top instead.

 

And yes the brightness is that different between the two boards in this setup. Yannick's also has 75Ω resistors on the outputs for the R,G,B but for whatever reason, the image isn't darkened as much with those cables as it does through Crayon's. I also did modify my cables temporarily to remove the resistors on those lines inside the SCART housing and the brightness did come up more in line with what you see on Yannick's. Yannick's board also has vias behind and in front of those resistors so you found that it was too dark for whatever reason, you could put jumper wires in those vias to bypass his output resistors but I've not found that needed in my installs yet and in fact when I did this originally, the image was a bit too bright and washed out as his older revision board produced.

 

 

 

[the_crayon_king]

@-^CrossBow^- On Yannick's fix, in addition to Vsync lasting longer than it should it is also happening 192.1us sooner than it should be. Glitches at the top/bottom of the screen would be related to Vsync.

I have been looking over the logic I am not sure how feasible it would be to make on a non-programable fix. If someone could think of an IC that changes the width of a signal then that may work?

 

Refer to https://tinyurl.com/ye9ktjhg

When using a cable that doesn't have internal 75 ohm resistors it should hit that full scale and the colors should more or less match the pallet that I programed in there.

 

The reason his looks brighter without removing those resistors could be because it is putting out over 1.4v p-p? That is the most logical thing I can think of. Still the video looks fine colorwise so I may be wrong about that. There isn't any good way to test that without using an oscilloscope or simulating the circuit.

[+-^CrossBow^-]

9 hours ago, the_crayon_king said:

The reason his looks brighter without removing those resistors could be because it is putting out over 1.4v p-p? That is the most logical thing I can think of. Still the video looks fine colorwise so I may be wrong about that. There isn't any good way to test that without using an oscilloscope or simulating the circuit.

Well there is a small fix done on the colors. His board due to the way it is designed, allows for resistors to be changed to basically tweak each of the main colors within the color bars on the diagnostic. In my testing and observation, his colors appeared to be very close to emulation minus the light green on the right hand side of the bar being a touch too bright. So the screen shots of his board I've shared and videos all show the colors with a resistor changed out to bring that green shade down a touch. As you have seen with his board, it is largely all done through passives and an amp to bring out the RGB signals. So his board is nearly 3x the size of your v5 but still able to be fit easily in both model and model 2 systems.

 

I'm still waiting on my client to send me their Intellivision that is wanting the works so they will be getting one of your v5 boards in it as they want both RGB and YUV outputs installed. On theirs I will be removing the RF modulator to place the RGB output there and redo the channel select switch on the bottom to flip between YUV and RGB outputs. As they are sending me a model 1, it will have full RCAs along the back for the YUV output. 

 

I say all of that because my plan is to video the process of that install to not only show how to install your V5 board, but also show both YUV and RGB etc. into a model 1 unit. I will also be putting one of Yannick's (I've a small batch of his boards on hand), that I will also video the process of installing into an SVA spare that I recently got up and running again. From there I hope to get some good quality pics taken and see if I can draw up a PDF for the installation of each. They are very similar in the way they install in regards to wiring so should make things easier for any that want to follow suit.

 

[the_crayon_king]

@-^CrossBow^- I don't remember if I mentioned it but I was making a non-programable version at one point (still might). The resistor summing part is where I got overwhelmed; that and I refuse to use more than a few diodes. 

 

I am just doing thought experiments over Yannick's board in order to distract myself. 

 

I don't know how TV's and scalers really react to p-p in excess of 700-714mv. Maybe it auto adjusts?

 

Here is a simulation of Yannick's circuit: https://tinyurl.com/yexo5hr5

Given the regulator VCC could be 4.8 - 5.2V that would be 96% accurate +/- 30mv of error (on max p-p) VS a more accurate reg 4.95 - 5.05 99% accurate +/- 10mv error (max p-p). So that is one area that could be changed.

The darker the color (higher the resistance) the less the % of error.

I don't know the specs of the diode he is using so that could really effect the output sim. 

I know it is a Schottky so it shouldn't be dropping the voltage by much. 

 

If you take a decimal value of say 255 (really it's 256) and you know the p-p of video (700mv)

then: 700/256=2.734375; That is the ratio I use when doing resistor summing anyway.

 

With a potentiometer, multimeter, and an oscilloscope you would just need to manually tune each color to match the decimal values

Lets take IMBEZERK's grey: RGB = 162, 172, 162; Voltage would be : RGB = 442, 470, 442 (mv)

So you would put up a grey screen (or any screen with grey really) and manually tune until reaching the p-p that matches. (then that would be the needed resistor values for grey). 

These numbers would be before amplification or after amplification when terminated.

 

You could also manually tune by eye but you need to know the resistor value to make 700mv/714 first. Then add a pot in series since you know every other value will be that resistor value or higher. 

 

He may have already done that and obviously the output video looks fine regardless. Like I said I am just trying to distract myself.

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

On a PDF I have added step by step instructions on the sale page. Feel free to use that text in any way you see fit.

I trust you to make better videos/instructions than I possibly can but let me know if you need anything to assist in that. 

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

 

I added all the needed components text to the expansion board (the open source version):

https://oshpark.com/shared_projects/vOnMDVR7

 

I did a rough outline around sections 3 and 1. Section 2 is the one left alone (in the middle). Refer to earlier in this thread or the text on the board page for information relevant to these sections.

All the little capacitors are 0402 size. All the resistors besides the two 10k resistors are 0603. The two 10k resistors are 0402. 

The 220uf capacitors are: "CASE-B_L3.5-W2.8-H1.9" C1411 might fit as well or just using electrolytic. Just make sure it's 220uf of higher and rated 4V or higher. Also mind the polarity. 

Edited June 25, 2021 by the_crayon_king

[+-^CrossBow^-]

Yeah, his board is heavily populated with a lot of SMD resistors and diodes scattered across. The hardware fix uses two 4148s that I've soldered into place where needed along with a resistor. The other resistor that I'd been changing out is a standard 1/4w resistor that I have to fold the leads inward to solder it down into place of the SMD that gets removed to adjust for the green slightly. 

[the_crayon_king]

6 hours ago, -^CrossBow^- said:

Yeah, his board is heavily populated with a lot of SMD resistors and diodes scattered across. The hardware fix uses two 4148s that I've soldered into place where needed along with a resistor. The other resistor that I'd been changing out is a standard 1/4w resistor that I have to fold the leads inward to solder it down into place of the SMD that gets removed to adjust for the green slightly. 

This should show off some of the expected values when using IMBEZERK's pallet. 

 

In this file if you change the decimal values for RGB then you will get everything else automatically. (so any pallet can be used here). 

The basic logic should work on any situation where you have the decimal RGB values but want the other values for RGB or YPbPr. 

I'm just glad I will no longer have to work this out long hand.

 

You guys can PM if you have a need for this file.

 

If that is the pallet you are going for you would have to tweak the resistor values on the earlier posted file to get an idea of the resistors needed. 

I don't know how accurate the sim is vs the actual PCB.

The real way would be to use an oscilloscope multimeter and potentiometer.

 

I will probably keep this up until I get a formula for the Rvalues in there.

[the_crayon_king]

So, the excel format was limiting.

Instead, I have spent the better part of the day making a program that writes the code I need for me.

If you have the MV of a signal, decimal values, or hexcodes; this program will display that color, convert to everything else, and also create the code (and auto copy it to the clipboard). 

 

All I would need to do is enter a single color from a pallet and this thing will literally write the code for me.

Also no, I won't share this. The good news is now I can program pallets in a few mins instead of an hour. 

[towmater]

Successfully installed a Crayon King RGB board in this Intellivision. Before I start tearing the 2600 module apart, I just wanted to find out if it is known to not be working with RGB? It could just be corrosion.

[the_crayon_king]

1 hour ago, towmater said:

Successfully installed a Crayon King RGB board in this Intellivision. Before I start tearing the 2600 module apart, I just wanted to find out if it is known to not be working with RGB? It could just be corrosion.

 

The system changer would not work with this. It should still be working through RF. 

 

If you absolutely wanted to have this working over RGB you would have to do a RGB mod on the 2600 module then maybe sum the two rgb videos together?

That is the best I can figure anyway. I don't know if Tim Worthington's Atari RGB mod works with clones (like the system changer). 

 

[+-^CrossBow^-]

4 hours ago, towmater said:

Successfully installed a Crayon King RGB board in this Intellivision. Before I start tearing the 2600 module apart, I just wanted to find out if it is known to not be working with RGB? It could just be corrosion.

As @the_crayon_king mentioned the system changer will only work through the RF output or a composite output if you have one installed. Similar to the Colecovision expansion module and the VCS adapter for the 5200, the actual onboard graphics processor IC on the main consoles are not used and so the RGB mods will not work with the add on 2600 clone expansions. 

 

 

[towmater]

I threw a couple of RCA jacks onto the system changer.

[Icelvlan]

Hello,

 

How are the other boards coming along?

 

[the_crayon_king]

14 hours ago, Icelvlan said:

Hello,

 

How are the other boards coming along?

 

I finished work on all the derivatives of the V5 board a while ago. Although I did finish the designs they ended up being impractical or otherwise undesirable for me to make.

 

So instead of overcomplicating the V5 design (in order to support Composite or S-Video) I just made an addon board that would put out everything else: https://oshpark.com/shared_projects/vOnMDVR7

Not only will that expansion board work on the V5 board, it should work on most consoles where you have RGB and a clock and want Composite, S-Video, or YPbPr. 

You guys can order those without involving me. If there is a point that might need tweaked it would be the 30 ohm resistors. There might be a better resistor divider ratio. You are just trying to make 1.4V p-p into 1V p-p. 

 

If there was enough interest I could order a batch but it seems like it is one of those things that not a lot of people would need/want. 

[Icelvlan]

Cool, has anyone tested with 5X yet?