Jump to content
Due to a large amount of spamers, accounts will now have to be approved by the Admins so please be patient. ×
  • 0
IGNORED

PC RGB to component converter & TV hack


Zebidee

Question

Hello Aussies, I'd like to explain how I've successfully hacked an interface board to convert PC RGB(S) output into component colour space for direct input to a TV that does not directly support either RGB or component inputs.

 

The PC is a very basic Windows XP64 machine (HP/Compaq core2duo E7500 3.0Ghz 2GB RAM) with an old X1050 64mb video card that outputs VGA + svideo. It is running CRT_emudriver and outputting +'ve composite sync.

 

The TV is a brand new Distar 21" CTV-1121 TV, built on a very common cheap generic Chinese chassis (bought in Thailand). "From the factory", it only has composite inputs. It has just one chip on the PCB, a jungle "monster" IC with OSD functions combined (8895CSN67GG9). This chip does not provide for RGB inputs at any stage. There is no RGB anywhere on this TV until you get out of the jungle and on your way to the neckboard. There were no component inputs on this TV either. However, there are IC inputs labelled "Y, Cb, Cr" at pins 19,20,21. This is where I hacked the component signal in.

 

Distar box.JPGDistar back.JPG

 

Now, I know that YCbCr is a digital standard, not analogue (YPbPr), but everything pointed to this TV being able to take analogue YPbPr (component) inputs, including unpopulated PCB traces and a blanked out spot on the TV case for the plug mounts to go. So I crossed my fingers and forged ahead with the assumption that whomever labelled the IC knew little about the difference between the two standards.

 

I couldn't find any datasheets that exactly matched my TV's jungle IC, but I found a few hits for a very similar one, 8893CRCNG7D65, which is also made by Toshiba and identical for my intents and purposes. Somehow I found an English language pinout:

 

http://www.datasheetq.com/datasheet-download/871940/1/Toshiba/8895CSNG7DN5

 

On the TV, I traced the unused jungle's Y/Cb/Cr inputs back to where they were terminated with 0.1uF capacitors and 100R resistors to ground. This is where component inputs would have been fed into the jungle, if they existed. Instead of soldering directly to the jungle chip, I figured it'd be much better hacking my inputs in here.

 

I carefully lifted the legs of the caps from the PCB (ie out of circuit), and added my component input wires there (ie nothing between my wires and the jungle "monster"'s component inputs).

 

[ATTACH=CONFIG]138354[/ATTACH]IMG_8615.JPG

 

I then connected a RGB2component circuit that I've been playing with, between the PC's VGA RGBS out and the TV's new component inputs (see below).

 

I've put up some screenshots below. If sometimes they seem a little dark, it is because I was focused on trying to capture "pixelation" or sharpness (in any case, I can still adjust the brightness/contrast, just like in a normal TV), which means my ancient iPhone 4S camera automatically used a small aperture (ie darker) to minimise "blooming" from bright areas.

 

You'll see that the image is a currently a little too wide - I will want to make some TV adjustments and tweak my video modes in CRT_emudriver.

 

IMG_8574.JPGIMG_8585.jpg

IMG_8595.jpgIMG_8605.jpg

 

 

Next post: Zebidee's RGB2component hack

Edited by Zebidee
Link to comment
Share on other sites

  • Answers 79
  • Created
  • Last Reply

Top Posters For This Question

Top Posters For This Question

Posted Images

Recommended Posts

  • 0

RGB to Component mod

 

To start with, my RGB source is a Win XP64 core2duo E7500 3.0GHz 2GB with a Radeon X1050 64mb PCI graphics card (VGA + Svideo out), running CRT_emudriver for 15KHz + video modes. It is setup to output composite sync (positive polarity) on VGA for all video modes.

 

So, this is the RGB2component circuit that I've been playing with. It sits between the PC's VGA RGBS out and the TV's new component inputs. The circuit was originally based on Ace's 5.0 circuit from the Sega16 forums, but also I've been looking at many other modders works, circuits, reading manuals/datasheets and revising basic electronics theory/practice. Apologies for the pencil-drawn schematic, I know nothing of how to use circuit drawing software like Eagle etc., another learning curve.

 

RGB2YUV_Zeb_v1-0.jpg

 

Features:

 

1. 75R termination resistors and 1uF caps on the PC RGB signal inputs (pins 13,14,15) to BA7230LS

2. 75R termination resistors and 220R electrolytic caps on component inputs to TV (Y, B-Y, R-Y outputs, pins 16,17,18 from BA7230LS)

3. Pedestal clamps (1uF caps & 1kR resistors) on unused inputs at pins 3,4,5 and 1uF clamp on 20 (video in), as specified in datasheet.

4. Decoupling capacitors (mixed ceramic/electrolytic 1uF/10uF) on +5v inputs (collector) to C945 transistors.

5. Decoupling capacitors (47nf and 10uF) on Vcc input (pin 24) to BA7230LS, as per datasheet

6. 1kR resistor and 220uF cap on PC CSYNC input, which goes directly into Y (pin 16) as it exits BA7230LS and enters the base of the C945 transistor for Y.

 

Discussion

 

One of the difficulties with converting RGB to component is that each signal influences the other. When component is decoded back into RGB in your TV, it relies on algebra equations where each value relies on the others. This means that even when you get the mix right, small levels of interference have an exaggerated affect.

 

I came to the conclusion that the modders who have trod this path before me seem to have missed a few important things. But that is always the nature of tech progress! I wouldn't be here without those modders. So I began with no assumptions beyond what I got from the datasheets/specs and used that as my base. I'm very happy with the results.

 

My circuit makes use of the BA7230LS IC which is designed to take RGBS inputs and roll them into a composite video signal. At an intermediate step the IC converts the RGB colourspace (in the "matrix") and spits out Y, B-Y and R-Y (without any sync). Normally the BA7230LS expects you to do some signal filtering and feed the signals back into itself so it can be further dumbed down to composite video. My circuit intercepts (rescues!) the BA7230LS outputs and uses them as the basis for component signals by combining sync with Y.

 

I didn't bother with a LM1881 sync combiner/cleaner or anything similar, I assumed that my composite signal (via CRT_emudriver, thanks Calamity!) from the video card was good.

 

I made a point of making sure that video signals had 75R termination, both into the BA7230LS and the TV.

 

I applied datasheet recommendations including pedestal clamps on the unused BA7230LS inputs and decoupling capacitors on Vcc input.

 

I noticed in many modder circuits floating around that some of the electrolytic caps were incorrectly placed. So I made sure that electrolytic caps correctly aligned with respect to ground, voltage source and signal direction.

 

I added decoupling capacitors to the C945 +5v inputs (at collector) as well to minimise signal interference and improve consistency. The extra decoupling caps also served to help mitigate and isolate ground interference.

 

Finally I was able to eliminate most remaining interference by keeping RGB inputs close to PC video ground, and linking ground from there directly to the TV (ground near video signal inputs).

 

I haven't gone beyond breadboarding this circuit, so I'm hoping to see further improvements as I cut off extra component legs, minimise wiring and solder things together for good connections. I'd better stop here and post. Constructive comments/criticisms welcome!

 

Here is how it looks on the breadboard. The thing on top of all the wires is the BA7230LS chip! The RGBS comes in at the right, then goes off to BA7230LS. The single white wire heading off to the middle is the CSYNC, going off to mix with Y. The white wires going from the BA7230LS a are all unused inputs, pedestal clamped. Y, B-Y, R-Y signals (coloured RGB) come out of BA7230LS, go left into that cluster of capacitors & resistors (the C945 transistors). The component signals come out from there and head off towards the TV (to the left). At the left-top you can see the +5v coming in (and ground going back). +5v is sourced via molex connector from PC.

 

Like most breadboards, the very top line is +5v, the next line down is ground. The rest is for components.

 

Going to update this pic in a minute, made a couple of small improvements.

 

RGB2YUVbreadboardv1-0.jpg

Edited by Zebidee
  • Like 2
Link to comment
Share on other sites

  • 0

Extra on TV Setup:

 

You need to use remote control to enter service mode and enable "YUV" there. You may also need to enable YUV voltage to AV from service mode.

 

[will add "enter service mode" instructions later]

 

I made further image improvements by entering normal "menu" and reducing sharpness there, as all it really does is introduce artifacts and make the image fuzzy. Default is 50 out of max 100. I reduced it to 0, but then convergence errors became obvious! So I scaled it back to 20 and it looked great. After this i needed to reduce the brightness slightly, but otherwise just left everything at default.

Edited by Zebidee
Link to comment
Share on other sites

  • 0

I expect RGB signals would be too high. You would need to bring them down to 0.7vp-p before they input to the BA7230LS. That might be as simple as a voltage divider with some adjustment pots.

 

Be a good idea to add a LM1881 or similar sync cleaning and converting circuit so that it could be used with different JAMMA boards as they come with different types of sync.

Link to comment
Share on other sites

  • 0

For JAMMA boards, I'd suggest a 1k pot in line with each RGB signal. With the 75 ohm termination resistor that makes a voltage divider, can bring a 5v signal down to 0.3v.

 

Either find the right level and then replace it with a normal resistor, or leave the pot there so that you can adjust it for different boards.

Link to comment
Share on other sites

  • 0
I use a THS7374 as a buffer and to ensure correct impedance, and place the arcade board under the correct load. That's why I use pots wired as a voltage divider as the drop needs to happen on the inputs, the 75R termination doesn't occur until the output stage.
Link to comment
Share on other sites

  • 0

Cheers for posting your rgb to component schematic, its one of the most thorough versions I could find using one of those BA chips.

 

I noticed at step 6:

 

6. 1kR resistor and 220uF cap on PC CSYNC input, which goes directly into Y (pin 16) as it exits BA7230LS and enters the base of the C945 transistor

 

Other versions of this schematic (aces etc) use a 680k. I plan to follow yours but keen to understand why the difference?

 

Also to clarify did you use 1uf or 0.1uf as part of your coupling caps?

 

 

Link to comment
Share on other sites

  • 0

Ta for that!

 

Ive found a few similar pcbs and designs, and it looks like the 680 vs 1k difference seems to be related to where the designer decided to join the c-sync up to the Y signal (ie 680 if connects to Y before it goes into the BA chip or 1k if it joins Y after exiting) ..but i could be wrong.

 

Looks like 1uf seems to be consistent cap value across designs ive seen too.

 

Zebidee's adjustments and design looks to use a few more caps but achieve a much better job of filtering the 5v in and using coupling caps to filter ground noise on the unused BA pins, (so less chance of interference).

 

I plan to try and make a scart version of Zebidee's so will post results (as soon as parts come in and i finish it).

 

Dek

Link to comment
Share on other sites

  • 0

I dont think it could be powered vga, but scart maybe (depending on the source ie retro console).

 

One of the tests will be to see how many amps the circuit draws (ie if its under or around 150mA). I bought a similar transcoder device that is powered via usb. If you want to see the other pcb designs just google the BA chip number and words rgb transcoder.

 

Just waiting on more parts to arrive (seems to be taking ages).

Link to comment
Share on other sites

  • 0

"I noticed in many modder circuits floating around that some of the electrolytic caps were incorrectly placed. So I made sure that electrolytic caps correctly aligned with respect to ground, voltage source and signal direction"

 

Im not sure the diagram posted above is entirely correct either:

- with respect to c-sync coming in, the positive end of the electrolytic cap should be connected to the csync in (c10). The negative end should rejoin the ba chip output.

- same with c17,18,19 the positive should be connected to the transistor output and the negative side going out to the tv.

 

 

Link to comment
Share on other sites

  • 0

Ok, finally got around to building one of these and compared to a transcoder i bought from ebay that uses the same chip.

 

My thoughts after building similar:

- the output is also very bright (i suspect this is due to over voltage somewhere.

- the colors apprear to have some over saturation.

- adding variable resistors on the output line for (Y) i could tune it down in brightness but the contrast appeared to change and became washed out. For the other lines, (R-y) and (B-y) playing around with a variable resistor on each line just washed out the colors.

- i added in lm1881 sync stripper which outputs 5v TTL csync signal. The tv didnt like thr 1k resistor on csync for me (flickered at the top of screen ) so used a 470ohm resistor in its place and picture was very stable after that.

- i used caps similarly to filter the 5 +ve but i only did this in one place then fed the power to both ba chip and lm1881- it seemed to work ok and no

interference.

- I suspect an extra resistor is need to reduce the amplification of the signal at each transistor but need to test this further.

 

More soon

 

 

 

 

 

 

 

 

  • Like 1
Link to comment
Share on other sites

  • 0

Further tests, iv done away with c945 transistors and while able to get a nice sharp picture using 500ohm trimpots /variable resistors in their place as a voltage dividers (with one leg to input, one to output,and one to ground) + a 75ohm resistor in line which then connects to the 220uf cap - im still not happy with it !! as the whites have a grey/blue tint when compared to composite.

 

Here is a sample with a snes input

 

Note: on the using the ace based version with c945 transistors, when adding extra resistors to adjust for colors it also seems to effect color timing (ie reds fade out slower during black screens)

 

 

Ive seen other examples of this circuit use a ths7374 amp/filter circuit instead of c945 transistors so will experiment with using those when they arrive in the post, then test with a range of rgb inputs.

10038eb7cc43d16830bc3c5cbf1050c3.thumb.jpg.822fdf9bca3ecb76472f36bd04961a83.jpg

  • Like 1
Link to comment
Share on other sites

  • 0

After experiementing with ths7374 amp, I was able to get a really good clear picture with crisp colors appearing as they should.

 

This.was based on a combination of the schematic above (which does a good job of properly connecting the ba chip and grounding the unused pins, filtering the 5v input), also aces v5 (which helped start the whole thing) and a modified ace schematic (which ive drawn all over ;) )

 

The ths7374 amp is VERY small and needs a breakout pcb (sop to dip). There are alternatives to this (google retro rgb ths7314 or ths7416 which are larger and easier to work with) - they all do the same job. In my opinion dont waste time with the transistor amps as they alter the color matrix slightly.

 

This schematic includes a sync seperator (Lm1881). For arcade boards you dont need it - as you already have a standalone sync. This can just be directly connected to Y at the same point sync is added to Y - which is just after the variable resistor in the schematic (you will still.need to add a resistor for sync, the value will depending if you are working with a 3-5v sync - Lm1881 uses TTL level for sync, which is why ive used 470R before connecting to Y as per retrorgb guides).

 

This schematic also expects 0.7v (p-p ) for input into the ba chip which may differ from arcade rgb signals. Again you may need to change the resistor values to bring it down. Ie the 75ohms on rgb input is to bring 1.4 p-p to 0.7p-p... this works well for vga/scart based rgb signals.

 

I havent defined the resistor values on the output as it may need to be tweaked per system or arcade board and to control for brightness (ie so it looks just right with the correct color mix) - 500R variable resistors (ie only use 1 input and 1 output and leave the remaining connector disconnected) seemed to do the trick for me.

 

This works great for a snes(pal) rgb signal and minor adjustments can be made using the trimpots while viewing to get the perfect picture on a component input tv.

 

Once you get the right color levels, measure the trimpots/variable resistor with a multimeter and then replace trimpots with set value resistors.

2d9921c8cc0cb0ad873c280ae330f92a.thumb.jpg.f4179b729e976e327510fdc3a4014992.jpg

1efb2defe3ef6d15e9ff60ee4ca6087b.thumb.jpg.fb19fa39550fbf7423cdef847c2c30ee.jpg

d763088298ee8c99ab16ba875a3911d7.thumb.jpg.71a5fb8b2277999de9f032b1a78cea9b.jpg

Edited by DekDek
More pics
  • Like 2
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • No registered users viewing this page.

×
×
  • Create New...