Coleco AV mod weird colors.

[phoenixdownita]

I performed the AV mod to my Coleco (the OpAmp based mod) and although it works the colors are somewhat weird.

 

The output from the RF modulator is actually very good and no color mess can be seen from that end, but from the AV mod it looks like the colors are in need of a tint adjust (a la 7800 if you know what I mean) and moreover such adjustment is different on different part of the screen and sometimes it depends on what's on screen.

 

I am using the composite output on an LCD (I also tried a plasma same effect).

I used Antarctic Adventure as a reference.

 

In specific with Antarctic Adventure:

some colors are weird, like seals are brownish instead of orange, the flags on the trail are cyan rather than green, the Australia flag at the end of the fist level is reddish rather than blue.

 

Even more at the beginning of each stage the game shows the map that then fades away from the inside out (it blackens out really). Well in that screen the top and bottom borders of the map are brownish while left and right borders more like goldish color, as the map blackens the left and right borders get darker until they match the top and bottom brownish color.

 

Usually the tint skew is more obvious on the bottom part of the screen rather than the top.

 

None of this effects is visible when using RF out, but obviously it is more disturbed and blurry so I'd prefer the AV route.

 

Given the mod is a simple OpAmp on the composite out of the modulator itself what can be causing what I see?

 

I attach the schematic that I used.

 

Any help appreciated.

Edited March 29, 2015 by phoenixdownita

[eslapion]

The schematic shown here has serious problems. The most obvious is the 47uF capacitor.

 

If you look at dedicated video amplifiers circuits available on Digikey, the THS73XX family datasheets all recommend a linking (DC isolation) capacitor of 330uF or 470uF on the output side.

 

This makes perfect sense as the impedance load of 75Ohms combined with this capacitor acts as a high pass filter with a cutoff frequency of 45Hz. TV have a refresh rate of 30Hz so the value of the capacitor must be at least doubled.

 

The other nonsense is the 10kOhm resistor on input. The input impedance of most op-amps on positive input is already near infinite so adding 10kohms is ridiculous.

 

Finally, the LM318 is a totally antiquated video op-amp. Today, you pay less for better op-amps.

[ChildOfCv]

Well, TVs draw a full frame 30 times per second, but they have a 60Hz sync due to interlacing.  But since the signal components are at much higher frequencies, I'd consider that to be irrelevant anyway.  That is, the sync signals determine frame rate and they're in the microsecond ranges.

 

2 hours ago, eslapion said:

The other nonsense is the 10kOhm resistor on input. The input impedance of most op-amps on positive input is already near infinite so adding 10kohms is ridiculous.

There must be something to it though, because the datasheet shows doing just that in almost all example circuits.  http://www.ti.com/lit/gpn/lm318-n

Perhaps it sacrificed common mode rejection for speed, and the 10K resistor does make a difference.  However, also of note is that the datasheet uses 5pF for slew rate compensation, not 27.

 

Well anyway, a more modern op amp would probably allow you to wire up a voltage follower with a simpler circuit.

 

That said, the circuit is posted as someone's AV mod, so apparently somebody got it to work correctly.

 

[+-^CrossBow^-]

This same basic scheme is still used today and I think the color shifting is specific to the console as it seems to vary from console to console. I've only installed the mod a few times myself in CVs and didn't notice anything with the colors in the past. But now that I'm actually working on a CV as a we speak, I'm going to fire up Antartic Adventure to check for this myself. I'm using the kits from Retrofixes.com only issues with them and others I've used is the jailrbars that tend to show up on the background colors mostly. But that varies from game to game and is more pronounced from my Atarimax SD than it is from actual carts.

 

[R.Cade]

Sounds like the weird things that can happen from overdriving the signal and/or a serious impedance mismatch.

 

I did a simple transistor mod to my Coleco, but I remember that the top of the screen was a little brighter than the bottom until I adjusted the voltage down.

 

 

 

Edited October 8, 2019 by R.Cade

[ChildOfCv]

20 hours ago, eslapion said:

The other nonsense is the 10kOhm resistor on input. The input impedance of most op-amps on positive input is already near infinite so adding 10kohms is ridiculous.

I think I found the reason for this:

Quote

Note 3: The inputs are shunted with back-to-back diodes for overvoltage protection. Therefore, excessive current will flow if a differential input voltage in excess of
1V is applied between the inputs unless some limiting resistance is used

If you never expect to leave the feedback range of the op-amp, then chances are the resistor won't be needed.  But they added it for the general case.

[eslapion]

On 10/8/2019 at 1:17 PM, ChildOfCv said:

I think I found the reason for this:

If you never expect to leave the feedback range of the op-amp, then chances are the resistor won't be needed.  But they added it for the general case.

The question remains: Why is it people use this seriously antiquated solution as a video amplifier for the Colecovision ?

I am presently working on RF modulator replacement for the VIC-20 and C64 (VAMP-20 and VAMP-64) and I will likely use the THS7314 and/or THS7374 which cost very little and do an amazing job.

 

A collaborator had a programmer develop a 'test' cartridge for the CV and it allows the display of test colors so I will probably investigate the video chip TMS99x8 Y R-Y B-Y output signals to create a converter. Converting to what exactly is to be determined. Of course it will be nothing like the F18 which generates VGA signal.

Edited October 16, 2019 by eslapion
typos

[ChildOfCv]

4 hours ago, eslapion said:

The question remains: Why is it people use this seriously antiquated solution as a video amplifier for the Colecovision ?

From what I saw of the original solution, the creator was a tinkerer and not an engineer, and just knew op-amps and followed the cookbook.

4 hours ago, eslapion said:

A collaborator had a programmer develop a 'test' cartridge for the CV and it allows the display of test colors so I will probably investigate the video chip TMS99x8 Y R-Y B-Y output signals to create a converter. Converting to what exactly is to be determined.

A better solution is always welcome

 

The biggest gotcha with the signals is that the B-Y pin sends out a negative pulse during the color burst period, and many TVs (and presumably YCbCr input processors) take that same time to synchronize on the blue level.  So for component output, at least, you have to suppress that or you get some ultra-blue images.  I've seen a Hackaday project that uses a sample and hold circuit to do that.

[eslapion]

15 hours ago, ChildOfCv said:

The biggest gotcha with the signals is that the B-Y pin sends out a negative pulse during the color burst period, and many TVs (and presumably YCbCr input processors) take that same time to synchronize on the blue level.  So for component output, at least, you have to suppress that or you get some ultra-blue images.  I've seen a Hackaday project that uses a sample and hold circuit to do that.

One of the various solutions I have considered is the use of TI's LMH1251 converter IC. During my research, I have found a document from TI which tells how to convert the output of a TMS99x8 to RGB but the conversion equations seemed almost identical to those used in any modern display equipment having component inputs. It also includes a DC level resetter which uses the old 4066 logic IC probably to address what you mention here.