MOD GUIDE - Universal RGB-to-YPbPr/Component Circuit & Mod (8/24/2014)

[Helder]

Nice board! I guess I must have missed that 2k resistor on the schematic so I will give that a shot and see if changing the 15ohm to something higher might help. The Blue isn't the issue it's the green instead of black and I post a screen shot if you guys want to see what the issue is.

[thesteve]

screenies help

[NightWolve]

My test calls for more resistance on green so still wanna see what happens if you try that idea.

1) Red = 500/.299  = 1672 Ohms
2) Green = 500/.587 = 852 Ohms
3) Blue = 500/.114 = 4386 Ohms

Just pointing out on blue that there was a mistake on the schematic and you wound up building a board with it. If my test/idea is no good, you wanna default back to Steve's generic values and just up resistance on green in your case.

[thesteve]

important point to make
the resistor values will have a much higher impact at high light (near white)
also the pots are for green balancing
the red pot sets the cross point between green and red
the blue pot sets the cross point between green and blue

[Helder]

Sorry I haven't posted any results just been a busy weekend, but tomorrow is a free day to get cracking on it. Also I heard that there are ways to make a similar circuit with an op-amp and some resistors, ever look into this steve?

[Helder]

My test calls for more resistance on green so still wanna see what happens if you try that idea.

1) Red = 500/.299  = 1672 Ohms
2) Green = 500/.587 = 852 Ohms
3) Blue = 500/.114 = 4386 Ohms

Just pointing out on blue that there was a mistake on the schematic and you wound up building a board with it. If my test/idea is no good, you wanna default back to Steve's generic values and just up resistance on green in your case.

Just to let you know I tried to values with some pots and it was sooo off on every color. I also tried tweaking each pot to see if maybe it would improve but I could never get the colors to even remotely look like they did with the normal resistors in place.

I'm going to try steve's method but I added a pot in the Green input to see if that will help and also increased the 15ohm to 33ohm, I'll post results soon.

[NightWolve]

Ah, thanks for trying!

Alright, just to review here, I trust that the two 10K pots for the Red-Y/Blue-Y lines were set right in the middle to 5k, 5k, etc. when you started it off ? You do have to adjust those pots to achieve the proper tinting especially with a dramatic change in Luma/Y as was done here, and they're not supposed to be exactly the same actually. Strength levels are supposed to be .71 for Red-Y and .57 Blue-Y according to the spec. So, output resistance of Blue-Y would be higher in a normal design, it should be weaker than Red-Y, but in Steve's case with the 10K pot adjustments and no output resistors, you'd fiddle around with just those pots.

Another test would've been to have had just the Luma/Y output from your PCB/board connected to the TV's Composite jack to see if you're getting a good black'n'white picture - that would be the best test for a good black'n'white picture. Should be no excess green, or blue, or red, etc. Then after connecting Luma/Y normally to the YPbPr/Component jacks with Red-Y and Blue-Y also, you'd have to adjust those 10K pots for tinting.

But yeah, I assume in your test you tried plenty of adjusting of those pots and didn't just leave 'em alone, right ? I don't see you mentioning any adjusting of those, just your custom ones on the RGB input lines.

Well, anyhow, that formula was based on that company's op-amp and a sweet spot 324 Ohm feedback resistor, so no expectations it'd apply to the generic transistor with a 500 Ohm feedback resistor. There still should be a way to compute exact resistance values for our Luma/Y, but this might not be it is all.

Also I heard that there are ways to make a similar circuit with an op-amp and some resistors, ever look into this steve ?

Yeah, that's what this is that I posted earlier!



Here's where that math info I wanted to test with came from (http://cds.linear.com/docs/en/datasheet/6559f.pdf - good research material from this PDF!):



The company's full page on their circuit titled, "High Speed RGB to YPBPR Converter":

linear.com/solutions/1312

They're trying to sell you on their chips, LT6559 and LT1395, both op-amps, and that "high speed" RGB-to-YPbPr circuit is an example on how to use them.

(Single 400MHz Current Feedback Amplifier)
(Triple Video Amplifier)


Self Reminder: I have recently created PCE/TG-16 video testing software that I asked Bonknuts/Tom to produce for me both for flashcarts and as a CD-R for CD drives. All the software does is when it's booted is make an all white screen, all pixels max white, then every time you press button I, it goes to all Red, then all Green, and all Blue, etc. This will allow you to view a steady RGB signal when using an oscilloscope on the HuC6260 video chip on the RGB pins and hopefully help tweak circuits like this one, etc. I will update the thread's OP with info and download links later when I get a chance.

[Helder]

I didn't mess with those pots at first but the difference wasn't major so I didn't bother so much. So that op-amp circuit can be used with any op-amp chip? I have quite a few laying around might give that a go as well. I have some free time today so I will see what my last changes give as results.

[NightWolve]

The exact resistor choices are expertly tweaked based on their chips, so there would be changes with other chips. I dunno what luck you would have. Also, as mentioned earlier, they left H/V Sync up to you. Something like Steve's Sync amp would work I guess, and then you'd connect it to Luma/Y after the 75 Ohm resistor on the final path to the TV.

Oh right, they also assume the RGB signals coming in are 75 Ohm ready, like say from a SCART output source, meaning, 1Vp-p (I guess ?), that is, properly amplified, so if not, you'd have to remove those 3 pull-down resistors at the start, the 80, 86 and 76 ones. Dunno how well this would wind up amplifying weak signals even after that. If you did a basic RGB amp first and THEN fed the lines into this design, then no problem.

Man, can't wait to get going to try this out for myself! My new scope is ready, willing and able, just need to make the time for this project finally. Might just see about using my TurboExpress since my Duo is out of commission waiting for the recap job. Would've been nice to have had a working TG-16 to mess with as well. I won one via NEC's Turbo Titans contest, but it broke on me... The games would load, but the colors were all wildly inverted/crazy, etc. and I guess I just threw it away after I upgraded to a Turbo Duo for the $99 deal back in the 90's.

[Helder]

After tweaking things to where it looks good and getting rid of that Green bleeding the downside is that the Green is lacking or super weak in alot of places. When everything is as it should be with the correct color balance the Green bleeds into the black, below is a picture showing a great example:



Up close (notice the green in the eyes and hand outline):


what do you guys think might need to be done or altered in the circuit to fix it ? I tried messing with the 15ohm resistor and all that did was mess with the sync so that's not anything that could fix it. There must be something in the mix that is causing the bleeding but everything else is great.

I tried to find those amps you linked above and they aren't that easy to find unless you get them in China so I looked for something similar and found that the TSH344 and TSH343 used in conjunction could output Component.

Datasheets:

st.com/../technical/document/datasheet/CD00079538.pdf
st.com/../technical/document/datasheet/CD00079537.pdf

[thesteve]

what input caps are you using?
i was using 100UF as my scope liked the signal better

[Helder]

I'm using the same 100uf caps as you and the schematic, could changing the cap on the green line affect the bleeding? maybe go to 220uf?

[thesteve]

the bleed is caused by luma sag
either you have a cap reverse biased or need bigger caps
note the inline caps in your scart cable (if used ) will cut the input cap values near in half

[Helder]

I'll try larger caps and I'm using some RCA sockets so that shouldn't be the issue with the Scart. I'll post results soon on different cap sizes.

[Helder]

Well I tried a 470uf and it didn't change anything but one thing I also noticed before and again when I was testing with Sonic 2 is that in certain parts of the level where there seems to be more Blue the green actually changes hue a bit then back to normal once I leave that area. Any other suggestions to stop the bleeding?

[thesteve]

a few sugestions
try it with just luma hooked and see if its still bleeding
try it with 1 input color unhooked and see if it stops bleeding
that should determine where your having an issue

[Helder]

I will try that but before that I changed some of the other resistors to pots on the green line and have had some good results but still bleeds. The only way it doesn't bleed is if I over saturate either Red or Blue but mainly Red. I'm thinking that Red might need some resistor tweaking, I'll post results later today.

Edit: Tried the Luma by itself and I can see a heavy outline on the lettering which is the bleeding green when I connect the other colors one by one. So looks like something on the Green/Luma is the culprit.

Edit 2: It turns out it was the 1k R1 on the Red input that was causing the bleeding, I put a pot on there and tweaked it till the bleeding disappeared and adjusted the other pots to get the correct colors. Now that I know the values I'm wondering how I can remove the 10k pots and put specific numbered resistors in their place and how to rewire them since the pots fed 2 location at a time. I will post the values I used once I documented them all so everyone can use them for the Genesis.

Edit 3: To get no bleeding simply change the Red Input Resistor to 2.4k and the Green Input Resistor to 1.5k. I also have the 100 ohm pot at 45ohms which looks great. One last question so I can refine a new board with these values, can 22uf ceramic caps be used in place of 100uf and 470uf? or do these have to be electrolytic caps or even tantalums?

[thesteve]

if using the sync stripper a 1K resistor from sync input pin to ground is a drastic improvement
dont know how it would effect it running direct chip sync from a PCE

[Helder]

I've never had issues with sync unless this somehow adds to the bleeding on the Red channel?

[thesteve]

It does not, however if red is bleeding, a resistor from the input red to ground will help

[Xenogears]

Well I also have a special bleeding problem. It´s white that goes out to the right. I did an RGB-Mod, using the THS 7314 Amp with 3,6M Ohm on +5Volt, 82nF CerCap to the Inputs and 330uF ElCap + 75 Ohm to the outputs of the IC.



To Sync I use C-Sync in line with a 220uF ElCap. On my 2nd Duo, this mod worked fine, I swapped the Mod and it also worked fine. So my conclusion - it´s from the Duo / 6260...

Any Idea where to look for ?

[thesteve]

are you saying your running as an RGB mod, or running an RGB amp into the component board?
it looks like you may just not be loading 1 (or more) of your signals causing an echo

[Xenogears]

Hello The Steve. Its like this:



I go from 6260 R / G / B to THS-7314 Chip. I use the circuit enclosed. It works in serveral Engines (I´m from Germay and RGB-Scart21 is common here). For some Reason this PC-Engine DUO has her own mind... could shadowing come from the IC-Circuit ?

P.S.: your Component Mod is great ! Especially used on LED-Flatscreens it looks superb ! Would it be okay to translate and post it into the german Forum Circuit-Board ? (www.circuit-board.de)

[thesteve]

yes, but i suspect thats not your issue
does the composite output have issues as well?

[Xenogears]

...yes. Let me check...

Just resoldered the Composite Video Connection - no bleeding. But still Bleeding if I use C-Video as Sync. I also tried to let the Kynar Wires go different paths - no change.... Also desoldered R, G, B and resoldered it, looked hard for solder blobs or bridges...nothing. 

The problem is, my Oszilloscope is 600 km far away - out of reach... :-(

[thesteve]

check your grounds and cables first
try a 1K resistor from the RGB inputs to ground

[Xenogears]

To be certain - 1K before THS to GND or 1K after THS (where the Signal goes to TV) ?

[thesteve]

should be before

[Xenogears]

Soldered in 1K to GND for each Color - no change, still "white bleeding" in RGB-Mode. If I connect to TV by Composite, the bleeding is gone. I never had this Issue before...

[thesteve]

and this amp/cable has been tested on another system?

[Xenogears]

For Certain. I swapped the Amp PCB with a working one into a 2nd PC-Engine Duo. The Test-System worked perfekt with the "bleeding Amp", the bleeding System was still blurring white with the "non bleeding Amp". I used the same RGB-Cable and also made a 2nd one with R/G/B to 75Ohm on GND. No reaction. What I might have found out - the Issue comes from the green line. If I swap colors the issue is still visible on the screen but not in white (...and not that strong). If I disconnect the green signal, it looks fine to me (but might depend on the missing color, dimming the brightness)

[thesteve]

the odd thing is your having an issue with white while using RGB
that means your getting the effect on all colors together

[crans]

I use this for my RGB only need's.

Some things changed from original is use of 5.1M resistors for each color input to ground.
Using 0.1uf on all inputs and one bridged power and ground rail's. For the output i use 75Ohm and 220uf.

Not sure what it will help just my use of this amp

[Xenogears]

I will try it out this week. But it must be something from the Engine, as thr 3.6 MOhm Amp works perfekt in other units. @thesteve: yes, this is really strange - bleeding / shadowing with white I had never before...

[thesteve]

its not bleed, its actually echo (bleed streaks decreasing from left to right, echo has more of a image copy shifted right overlayed on the pic)
is it possible you getting some composite crosstalk?

[crans]

Just checking in noticed I put 1uf not .1uf as I should have.

[Xenogears]

Hello guys, back from sickbay this Weekend I will again encounter the "Shift-screen-Battle". Thesteve - your last Idea fits perfect to the issue. I used the Testsuit 240p and found out, that the Image is in deed double layered, shifted to right and not only bleeding. Update will follow...

UPDATE - PROBLEM SOLVED but still weired -

It was missing Ground ! Inside the 8-Pin-Din-Input, there is a small Ground-Connection getting into contact with the Shield-round Metal Part of the DIN-Plug. If the Ground-Connection is missing, you get the Problem. Weired: even if you connect GND and Shild inside the cable, the issue will still come up. You have to install the Ground Connection!!!

[Keith Courage]

I am trying to install one of these in a TG16 docking station for the first time. I've installed this board in many DUOs and TG16 systems without any issues.

Anyways, I am having trouble finding a good ground. I've even tapped the ground from the expansion port and it doesn't help. The problem I am having is the picture will not stay steady no matter how I adjust the brightness gain pot. I had the same issue in a DUO in the past before I found a good ground.

Anyone have any suggestions?

[Keith Courage]

Found the issue. Thought I'd post in case anyone else ever has this issue. Not sure how or why but the ground from the expansion port on the TG was not getting the same ground provided from the Hu6260 chip. I simply tapped a wire from the chip to the expansion ground and problem solved. First time I've encountered that.

[poponon]

This is amazing and I can't wait to get building one. Question though - would it be possible to make this into a standalone universal rgb to yuv transcoder if you source the 5v input from an AC adapter ?

[NightWolve]

Sure, it is pretty universal. Here's essentially a fan-made turbo booster using this circuit:

https://www.pcengine-fx.com/forums/index.php?topic=14876.msg392378#msg392378

It started with TG-16, but he added the SNES Multi-AV cable for units that had RGB available all along.

[poponon]

that's awesome and more along the lines of what I'm thinking

In particular I'm thinking if I could hook up the circuit to a female SCART input so I could use it for multiple consoles and maybe a 15khz mamebox in the future. SCART in particular so I could still use the consoles with my upscaler when on flatscreens. I was hoping I could build something sort of like the csy2100 or the tc1600 transcoders. Just thought it'd be a fun project especially considering the tc1600 isn't being produced anymore (the developer passed away). And the screens people are posting look pretty mint

[thesteve]

thats exactly how i had it set up when i was building this circuit
was swapping between PCE and GENI scart cables

[poponon]

did you attach both SCART pins 8 and 16 to the circuit 5v input ? or just one

sorry if the question is really basic, I'm basing it off these:

[thesteve]

Just 8

[poponon]

awesome, just ordered all the remaining parts. I'll post some pics of the end product once completed

[lastcallhall]

So, I have a question... I bought a few of these chips from Turbokon, and I set to installing 2 of them into a Genesis and a Master System. On both systems, with two different chips, my colors are coming out very dark. The whites look like a medium grey and the remaining colors are dark, as well. I tried tweaking the luma signal but that only served to make the colors look washed out, and not brighter. So, I removed the 1k resistors that are connected to the RGB inputs and replaced them with 1.5k resistors. no change. I'm tapping into a CXA1145 encoder on both systems. I haven't tried it with my SNES yet, but I wanted to get this resolved before I move on.

Here's a question: For sync, I'm tapping into the sync pin and not the composite pin. Would that make a difference ? I'm using the older encoders for this, so I thought without the LM1881 I could use the sync signal without issue (I'm using sync out, not sync in). I'd really appreciate any help on this. Thanks!

Edit, I have also tried this circuit on a Genesis:



This circuit works fine, and on the same television, the colors look correct. I could build this circuit, but since I have the chips from TK, I want to use those instead. Thanks.

[thesteve]

what version of the board do you have?
the original board works well with 1K input resistors and no caps
the sync must be C-Sync when it gets to the circuit (LM1881 or any other TTL sync source is fine
the new board needs the 75ohm resistors installed on the board and 75ohm resistors with caps between the CXA and the inputs (as found in the GENI SCART cable)

[lastcallhall]

Hey Thesteve, thanks for writing me back.

To answer your questions:

It is the old version of the board. I bought about 15 of them a few weeks back.

I have placed 1K and 1.5K resistors on the board, both to the inputs themselves, as well as to the input solder points of the resistors (which had been suggested by Turbokon). The video is still dark either way.

I've tapped C-sync from the AV out as well as directly from pin 11 of the CXA1145 - there is no change in quality.

I'm in the process of building the circuit I already posted to show off the difference.

Thanks again!

[thesteve]

well never had a dark pic issue.....that i can think of