ColecoVision Component Video Mod

[ikrananka]

Over the past few years I have aspired to obtain a ColecoVision component video mod that works well on modern HD LCD TVs.  Firstly I had hoped to buy one from Viking Video Games (doubledown) but unfortunately he has stopped supplying standard video modded CV's to order years ago.  I then thought I'd try the 8bitdomain mod which, along with composite and s-video mods, includes a component video mod.  I installed this mod around a year ago but unfortunately it didn't work well with my LCD TVs - posted my experience with it here:

http://www.atariage.com/forums/index.php?showtopic=124242

Discussed this with 8bitdomain who said that, in the near future, he would be releasing a new mod specifically designed for HD TVs - a year later and this has not materialised.  So, I decided that the only way this is going to happen for me is to try and do it myself - problem is I have very little knowledge of electronic circuits - but I am handy with a soldering iron and a multimeter and reckon this should be a fun challenge

So over the past few months I have been doing as much research on the internet as I can and have found some very useful information.  There are lot's of details describing how to do mods for composite and s-video but little tangible information about a component video mod.  There are a few threads that discuss, or touch on, this subject but none of the ones I have found run to completion with any component video mod schematics and/or screen shots, e.g.:

http://nfggames.com/forum2/index.php?topic=2925

http://www.atariage.com/forums/index.php?showtopic=105191&st=50

From what I have read there does appear to be a need to amplify the Y, R-Y and B-Y signals from the TMS9928 VDP and there are typically two ways of doing this, one using transistors and the other with op-amps.  Now the op-amp circults are much more complex than the transistor based ones and so I reckon I'll start with the transistors and move to op-amps later if necessary.  The other thing I have concluded is that I really need to be able to adjust the strength of each signal (Y, R-Y & B-Y) to get the picture right and so I need to include pots on each line.

Based on the above I have knocked up a simple schematic (attached).  Probably going to head out and buy the components next weekend and knock up a proto board.  In the meantime any constructive comments woud be most welcome.  Things I am questioning are:

• Do I need 75 ohm resistors on all three output lines or just on Y?
  
• Do I need capacitors on the B-Y and R-Y lines to remove the DC component of the signal?
  
• Are the pots in the right place, i.e. adjusting the voltage to the transistors, or should they be on the signal lines?
  

Many thanks for ANY help.

[viletim]

An emitter follower video amp (your circuit) has no voltage gain. The 75 ohm resistor on the output forms a divider with the one inside the TV and you end up with a gain of 0.5. Not much good if you don't have at least 1.4Vpp of signal to start with.

Go with the op-amp circuit, much easier to understand and adjust. Use a non-inverting configuration with a gain of two. The colecovision's has +/- 5v rails to use.

Instead of a pot, just try different fixed resistor values...experement with 75-150 ohms. Go for a fast amp....the video will exceed 5Mhz.

[ikrananka]

Thanks for the great reply - much appreciated.

An emitter follower video amp (your circuit) has no voltage gain. The 75 ohm resistor on the output forms a divider with the one inside the TV and you end up with a gain of 0.5. Not much good if you don't have at least 1.4Vpp of signal to start with.

To be honest I took the basis for my circuit from your post here:

http://nfggames.com/forum2/index.php?topic=2925.msg19897#msg19897

It is also similar to the circuit used in 8bitdomain's component video mod except that in his mod the pulldown is after the transistor and the R-Y and B-Y line have capacitors in them instead of resistors.  There's also an inductor after the 75ohm resistor in the Y line.

I'm assuming that the Y, B-Y and R-Y voltages are lower than you previously thought - is that correct?  How does one interpret the p-p voltages for each line from the TMS9928 datasheet?

Go with the op-amp circuit, much easier to understand and adjust. Use a non-inverting configuration with a gain of two. The colecovision's has +/- 5v rails to use.

Instead of a pot, just try different fixed resistor values...experement with 75-150 ohms. Go for a fast amp....the video will exceed 5Mhz.

OK,  I'll move over to an op-amp based circuit using the non-inverting gain of two as you suggest.  Would I not be best to use pots on the output lines so that I can adjust them live and see the results immediately on screen?  I'm also curious about if it would be better to put the pots on the +5V supply line instead - much in the same way as the Bally Midway Vidiot board does.

So, I'll search around for a suitable op-amp.  The a1k0n ColcoVision composite video mod uses an LM318, while the Vidiot board uses LM359's.  My preference would be to keep the circuit compact and so would lke to use a single chip containing three op-amps.  I've found a few but am having difficulty interpreting their specifications to determine if they're suitable.  If I compile a short list could you help point me to which ones should work well?

Once I've plumped on an op-amp I'll draw up a new circuit and order the items.

Thanks for your help.

[viletim]

To be honest I took the basis for my circuit from your post here:

http://nfggames.com/forum2/index.php?topic=2925.msg19897#msg19897

It is also similar to the circuit used in 8bitdomain's component video mod except that in his mod the pulldown is after the transistor and the R-Y and B-Y line have capacitors in them instead of resistors.  There's also an inductor after the 75ohm resistor in the Y line.

I don't reccomend amplifiers like this anymore... Unless you know exactly what you are doing the result will relatively poor. Also take into account the components already inside the colecovision. There will undoubtably be some pull down reisistors on the video lines already.

I'm assuming that the Y, B-Y and R-Y voltages are lower than you previously thought - is that correct?  How does one interpret the p-p voltages for each line from the TMS9928 datasheet?

The highest voltage will be a white, the lowest voltage at black (disregard the sync on Y for the moment). The difference between them is the peak to peak voltage of the waveform. The black voltage will be the DC offset. The tolerances seem to be pretty loose - look at the difference between min and max values.

OK,  I'll move over to an op-amp based circuit using the non-inverting gain of two as you suggest.  Would I not be best to use pots on the output lines so that I can adjust them live and see the results immediately on screen? 

Then replace the 4.7k feedback resistor with a 5k pot. Leave Rs at 75 ohms. The op-amp gain is adjustable between 1 and 2. Use small trimpots to reduce noise.

[/quote]
I'm also curious about if it would be better to put the pots on the +5V supply line instead - much in the same way as the Bally Midway Vidiot board does.

It doesn't work like this.

So, I'll search around for a suitable op-amp.  The a1k0n ColcoVision composite video mod uses an LM318, while the Vidiot board uses LM359's. 

Both old parts from the '80s. You can do much better...

My preference would be to keep the circuit compact and so would lke to use a single chip containing three op-amps.  I've found a few but am having difficulty interpreting their specifications to determine if they're suitable.  If I compile a short list could you help point me to which ones should work well?

Something like the LM6172 or MAX4451 (both come two amps per package) would be ideal.

[ikrananka]

Thanks for all of the extremely useful feedback - I'll be drawing up a new schematic in a day or two and would appreciate any comments on it before I order the parts.  Reckon I'll go for the LM6172 as it is available in DIP config making it easier to solder up on my proto board.  I also have a few questions about the LM6171/6172 below.

To be honest I took the basis for my circuit from your post here:

http://nfggames.com/forum2/index.php?topic=2925.msg19897#msg19897

It is also similar to the circuit used in 8bitdomain's component video mod except that in his mod the pulldown is after the transistor and the R-Y and B-Y line have capacitors in them instead of resistors.  There's also an inductor after the 75ohm resistor in the Y line.

I don't reccomend amplifiers like this anymore... Unless you know exactly what you are doing the result will relatively poor. Also take into account the components already inside the colecovision. There will undoubtably be some pull down reisistors on the video lines already.

It does have pulldown resistors but they are further down the trace lines and inaccessible under a large RF daughterboard.  Interestingly the ColecoVision treats each one differently:

Y => 43uH inductor => 390ohm pulldown
B-Y => 390ohm pulldown
R-Y => 1000ohm pulldown

The difference may have something to do with conditioning for the downstream conversion to composite video.

I reckon I'll have to pick up the signals directly from the TMS9928 outputs and add my own pulldowns to each line - in fact the TMS9928 calls for 470ohm pulldowns on each Y, B-Y and R-Y output -  so I'll start there.

My preference would be to keep the circuit compact and so would lke to use a single chip containing three op-amps.  I've found a few but am having difficulty interpreting their specifications to determine if they're suitable.  If I compile a short list could you help point me to which ones should work well?

Something like the LM6172 or MAX4451 (both come two amps per package) would be ideal.

As I will have three signals to amplify should I use two dual amps (2x LM6172) or one single amp (LM6171) in parallel with a dual amp (LM6172)?  Bit concerned that the LM6171's slew and output currents are different to the LM6172 - can't tell if this would make any visible difference.  Also, I can't see anything on the LM6172 datasheet that indicates if it is a problem if only one of the internal amps is used (i.e. if I were to try and use two dual amps).

[viletim]

Thanks for all of the extremely useful feedback - I'll be drawing up a new schematic in a day or two and would appreciate any comments on it before I order the parts.  Reckon I'll go for the LM6172 as it is available in DIP config making it easier to solder up on my proto board.  I also have a few questions about the LM6171/6172 below.

To be honest I took the basis for my circuit from your post here:

http://nfggames.com/forum2/index.php?topic=2925.msg19897#msg19897

It is also similar to the circuit used in 8bitdomain's component video mod except that in his mod the pulldown is after the transistor and the R-Y and B-Y line have capacitors in them instead of resistors.  There's also an inductor after the 75ohm resistor in the Y line.

I don't reccomend amplifiers like this anymore... Unless you know exactly what you are doing the result will relatively poor. Also take into account the components already inside the colecovision. There will undoubtably be some pull down reisistors on the video lines already.

It does have pulldown resistors but they are further down the trace lines and inaccessible under a large RF daughterboard.  Interestingly the ColecoVision treats each one differently:

Y => 43uH inductor => 390ohm pulldown
B-Y => 390ohm pulldown
R-Y => 1000ohm pulldown

The difference may have something to do with conditioning for the downstream conversion to composite video.

I reckon I'll have to pick up the signals directly from the TMS9928 outputs and add my own pulldowns to each line - in fact the TMS9928 calls for 470ohm pulldowns on each Y, B-Y and R-Y output -  so I'll start there.

Just remember to disconnect the pins from the rest of the circuitry.

My preference would be to keep the circuit compact and so would lke to use a single chip containing three op-amps.  I've found a few but am having difficulty interpreting their specifications to determine if they're suitable.  If I compile a short list could you help point me to which ones should work well?

Something like the LM6172 or MAX4451 (both come two amps per package) would be ideal.

As I will have three signals to amplify should I use two dual amps (2x LM6172) or one single amp (LM6171) in parallel with a dual amp (LM6172)?  Bit concerned that the LM6171's slew and output currents are different to the LM6172 - can't tell if this would make any visible difference.  Also, I can't see anything on the LM6172 datasheet that indicates if it is a problem if only one of the internal amps is used (i.e. if I were to try and use two dual amps).

I wouldn't worry about it. Either use the LM6171 or two LM6172 with the unused amplifier configured as a voltage follower (-ve and output connected together). This way there is only a gain of one and it can't oscillate.

[ikrananka]

After a few months of research and planning I finally landed on a schematic design that I felt comfortable with (attached). This schematic is based heavily on the schematic from the TMS9928 "Interface to Color Monitors" application report (attached) plus viletim's help in this thread. Reading the text of the report it seemed to make a lot of sense to use the Sync Separation and DC Restoration parts while simply running each restored signal line (Y, R-Y and B-Y) to a dedicated op-amp should provide well conditioned and amplified component video signals.

So I spent a lot of time identifying what I thought were suitable parts and ordered them. To simplify the circuit I replaced the TI schematic's sync separator circut with a single EL1881 sync separator chip. Following viletim's advice I used more modern op-amps (LM6172) compared to the older ones (LM318) in the TI report. I also noticed what I believe are some typos in the TI report's schematic around the CD4066B chip's pin numbering and adjusted as follows:

Before => After
2 => 3
3 => 2
5 => 13
13 => 5

I also noticed that the TI schematic didn't show any power supply to the CD4066B chip. For my first pass I decided to include power and so added +5V to pin 14 and routed pin 7 to ground.

I also included trimmer pots on the gain feedback lines to each op-amp to enable me to tweak each signal.

So I built the circuit and hooked it up to a 26" LCD TV. Much to my disappointment the picture was jerking badly from left to right with the jerking being much worse at the top of the screen and getting better towards the bottom. In addition, the colours were wrong with black being a very strong blue colour and the other colours just generally being wrong (except white and blue seem okay). See attached pictures. Adjusting the trimmer pots brings about little discernable difference. I wondered if my adding power to CD4066B was contributing to the problem, however disconnecting the power and ground from pins 14 and 7 made little difference to the image.

Have I overcomplicated this? Is the TI schematic overkill? Have I interpreted it incorrectly? Should I just be looking at running each raw TMS9928 output straignt through an op-amp without additional conditioning? Unfortunately I am not an EE and do not have an oscilloscope to analyse the signals - not that I know how to use an oscilloscope or interpret its signals anyway!!!

I'm willing to try anything but really need some expert/experienced advice.

Much appreciate any help anyone can give.

[ikrananka]

TI report attached:

[viletim]

ikrananka,

To convert the component video to RGB or encode it to baseband video it needs to be DC restored. No need to go to the trouble is this case, all you need to do is make the signal characteristics suitable to dive the TV's input.

So remove all trace of that clamp circuit, it's very broken anyway.

Instead, place a resistor of about 10k between the +ve input of each op-amp and ground. That should do the trick for biasing. Sorry,  I inadventantly left that out of the circuit I posted previously.

[ikrananka]

ikrananka,

To convert the component video to RGB or encode it to baseband video it needs to be DC restored. No need to go to the trouble is this case, all you need to do is make the signal characteristics suitable to dive the TV's input.

So remove all trace of that clamp circuit, it's very broken anyway.

Instead, place a resistor of about 10k between the +ve input of each op-amp and ground. That should do the trick for biasing. Sorry,  I inadventantly left that out of the circuit I posted previously.

villetim,

Firstly many thanks for your help - much appreciated.

Secondly, could you have a quick scan over the attached schematic and let me know if I have interpreted your suggestions correctly.

Should I try 10k pots instead of fixed resistor values so that I can adjust the bias?

Many thanks

[papa_november]

Has any new progress been made on this front?