This is one of those posts I was sure I had written before, I have done this modification several times, and each time I have thought there was no point in taking lots of photos as I had already done that before.
Well this time I am going to take the photos and write it up.
The VIC20 is great, one of my favourite machines, but picture quality can be bit patchy.
One solution to this is an S-Video mod.
The VIC chip outputs
The VIC chip generates two separate video signals, chrominance and luminance. These are combined almost immediately on the VIC20 board to give a composite video signal which is sent to the AV port.
This is the only output available on the VIC20.
These traces are from the colour bars program in the Penultimate +3 cartridge.
That looks quite messy, but it contains a lot of information.
I have split that up, the top trace is the luminance signal. This looks like monochrome video. It includes the sync and the video signal in terms of brightness, but has no colour information.
The bottom trace is the chrominance signal, and is just colour information.
The middle trace is a combination of the two, composite video. (using the math function in the scope to add channels 1 and 2, the first time I have found a use for that)
The colour information is superimposed on top of the video information. This reduces the definition of the video signal, the clean edges in the luminance signal are obscured by the colour information (which I don't fully understand, but it it based on variations of a 4.43MHz clock, different colours have different waveforms)
The loss of detail is particularly noticeable when you zoom into pixels of individual characters, this is the top line of the Penultimate menu. Remember the middle trace is what you get, the other two are "look what you could have won".
On the VIC20, the pixel clock is same 4.43MHz as the colour information, so that is why if you have lots of alternating single pixels next to each other it could be confused for colour information. That is what causes the colour flashes sometimes on tight black and white text, and sometimes moire patterns.
(of course you can use this to your advantage, which is how Apple II colour works)
That got worse on a 40 column system like the Commodore 64 where the dot clock was double that at around 8MHz. They had to make a new font, where all the single pixel vertical lines from the PET / VIC20 font were thickened up to 2 pixels wide, which is why it doesn't look as nice.
S-Video is essentially the same as the VIC chip generates, separate chroma and luma, so it is not that difficult to modify a VIC20 to output S-Video.
But, is it worth it?
Well, in many cases, no.
IF you have a GOOD S-Video monitor or capture device or HDMI converter, then yes, but the S-Video inputs on some monitors are not particularly good and are probably just merged to composite video internally. The same with low end video capture and HMDI converters etc.
(update, keep reading to see the revised conclusions)
If you don't have a good way to display S-Video, then you are better off making sure the pots on the VIC20 are adjusted correctly and sticking with composite video.
With a little tweaking, you can go from an awful looking signal like this
or this
To something pretty decent.
It is still composite video, you effectively get the colour spray painted over a monochrome image, so you get fuzzy lines around the characters.
It can look quite bad, particularly when you have bright colours on black.
VIC20 S-Video Conversion
I have been asked to do an S-Video conversion on one of TFW8b's test boards, so let's go for it.
This is one of the later cost reduced VIC20s, so much so they even got rid of the tin around the VIC chip and associated circuitry. (there was a cartridge shield, but TFW8b uses this machine for testing cartridges, so that has been removed)
One option for an S-Video monitor is the Commodore 1901. I have to say I was a little disappointed with this monitor when I used it with the Commodore 128 80 column mode, mainly because the tube seems to be the 0.43mm dot pitch version (which is about the same as a potato).
However, with a correctly adjusted composite video signal, the result is pretty good, particularly on something like the VIC20 (which is a fairly lower resolution anyway).
One thing you do see with composite is jaggedly vertical lines, particularly on places like the border transitions.
Hmm, what's that smell?
Ah, it appears I didn't get around to replacing the RIFA capacitor in this one (something else I thought I had done).
Ooooh nasty.
Time for a modern X2 capacitor. (actually, I just noticed the date code is 2017, but that is modern to me, it was only a couple of years ago, right?)
That's better.
Now, where was I?
Ah yes, I was about to unplug the 1901 anyway, to show you the back.
The 1901 has 3 RCA phono sockets scattered randomly around the back panel.
The red one on the top left is chroma, the colour information.
The white one below that is audio.
The yellow on the right is luma, the monochrome brightness and sync (ignore the fact it has chroma written to it's side).
There are also spaces for various other options not fitted, including a large DIN socket(?) and what looks like a SCART socket (although the pinout doesn't look right)
The 1084S and other Commodore monitors also have the same three red/white/yellow inputs as S-Video could be had from later Commodore 64s, with an appropriate cable.
The Future Was 8 bit supply VIC20 / Commodore 64 video cables with three phono jacks on the end.
This is not the cable you need for S-Video, as these days when you see red/white/yellow jacks it is almost always video and stereo audio, which is how this cable is wired.
These are wired with yellow as video and red and white both connected to audio, however, those are ideal to adapt for S-Video for a Commodore monitor.
The pinout of the 5 pin DIN on the VIC20 is as follows.
This is from the VIC20 manual supplied with all revisions, however it was only correct for the original Japanese VIC-1001. That had a sort of 6V feed (although it was actually just 9V through a dropper resistor).
All VIC20s from the very first US versions of the VIC-1001 onwards had that changed to 5V.
The video circuitry changed with every revision, but by the time of the VIC20-CR, the video outputs were not high and low, they were just connected together.
The obvious solution here is to split up pins 4 and 5. Leave pin 4 as it is, composite video, but disconnect the chrominance signal from it to leave the luminance signal, and feed the chrominance to pin 5.
The relevant section of the schematic for an NTSC VIC20-CR is as follows.
On the NTSC version of the VIC20-CR, capacitor C13 connects the chrominance output (VIC chip pin 2) to the luminance output (VIC chip pin 3), before it is buffered by transistor Q2 and feed to the AV jack (pins 4 and 5).
The PAL version uses the same PCB, but they fitted C13 differently.
Rather than a ceramic disc, it is now an axial ceramic. The right hand pin still connects via FB7 to the VIC chip, but the left hand end now connects to the other side of variable resistor R10.
That means the chrominance is always at a fixed level, rather than being attenuated along with the luminance, as it is on NTSC VICs.
Either way, we don't want the signals mixed, so C13 has to go.
Whether you start with PAL or NTSC, you get the same result, no C13.
Congratulations, you now have a monochrome VIC20.
It is a cleaner picture though.
Gone are the jagged lines down the side (but also gone is the colour)
If you aren't bothered about colour and just want a clearer picture, congratulations, you're finished, you can take the rest of the day off.
Let's get the Colour Back
The next step is a difficult one, look away now if you don't want to see it, but you have to cut the track between pins 4 and 5 of the AV socket.
Next we need to get the chrominance signal to the now spare pin 5 of the AV socket.
For that, you need a 100nF capacitor and a 75Ω resistor.
Those can be fitted on the back of the board, between pin 5 of the AV jack and the closest end of C13, which is connected to FB7.
That is just a test fit to check the lengths, I covered it with heatshrink before soldering.
The cable needs to be modified to split the red and white wires, leaving white on pin 3, and moving red to pin 5. (Yes, the pin numbering is confusing, because they kept pins 1, 2 and 3 the same as the 3 pin DIN).
A bit of heatshrink also helps there.
I find it useful to label up things like this when they have been changed, in case you wonder why the video is in monochrome next time you use a standard video cable.
Or you use the modified cable on a standard VIC20 and wonder why you are getting noise on the audio.
Time to Test it
Looks good to me.
Nice and clear on the single colour characters.
(to save you scrolling back up, this is what the composite video version looked like)
The ready screen is looking good.
And notice the clean line down the side.
Mike Singleton's Snake Pit is a nice colourful game to test with.
That's looking nice and clear.
Without smearing around the sides.
(again a quick reminder of the "before" shot)
The VIC 20 version of Tut-Tut by Dave Stephenson and me is also a good one to test.
All very clear.
Yes, I'll take that.
Some Other Devices
In the past, I have had mixed results, SVideo mods can look very good with the right monitor, but I have seen a lot of rubbish implementations in the past on monitors I have tested. The same signal can look great on one monitor, and awful on another.
I thought I should back that up with some examples. My usual test monitor does not support SVideo, so I had a look around and borrowed an LCD TV/Monitor with an SVideo input on the side from another room.
Oh, that's actually quite good.
I tried some of the other tests.
Yes, OK, that's pretty good.
The colour is a bit washed out in some places, but it's certainly a lot clearer.
OK, so I will have to adjust the advice a bit, both the monitors I have tried, CRT and LCD seem to have pretty good results.
What about USB capture?
USB capture devices are usually awful with SVideo, right?
Well, it took a little effort to assemble the appropriate adapters.
But that again surprised me, pretty decent results from the cheap STK1160 "EasyCap" clone.
No complaints.
Nice clear, colourful pictures.
Conclusion
To conclude, yes, you can S-Video mod a VIC20, it is a fairly simple (and reversible) process, but does involve cutting one track (which could easily be repaired if you need the same signal on two pins).
Will it make your picture better? maybe. It depends on what S-Video device you want to plug it into, but all the ones I have tried with this mod on this VIC20 have preformed pretty well.
To conclude the conclusion, two questions:
Q: Is properly adjusted S-Video on a good S-Video monitor better than badly adjusted composite video?
A: Yes.
Q: Is S-Video on a bad S-Video monitor or capture device or HDMI converter better than properly adjusted composite video?
A: probably not.
Final Questions
Q: Do I still love the VIC20?
A: Of course.
Q: Does my workshop still smell of RIFA capacitor smoke?
A: Yes.
Adverts
Whilst testing this, I was using the VIC20 Penultimate +3 DCR cartridge, available from TFW8b.com
Note if you don't need the built-in SD2IEC drive, simply disconnect the cable, I didn't refit it for any of the testing as everything I needed was in the PU+3 menu already.
The VIC20 composite video / audio cable I modified was also from TFW8b.com
There is no keyboard connected, to make it easier to make the mods, so I was using one of my Joystick Emulator boards to navigate the menu on the Penultimate +3, these are available from my Tindie store:
My Tindie store contains all sort of kits, test gear and upgrades for the ZX80, ZX81, Jupiter ACE, and Commodore PET.
Patreon
You can support me via Patreon, and get access to advance previews of blog posts. You also get progress updates on new projects and other behind the scenes updates, as well as access to my Patreon only Discord server for even more regular updates.
