Help with Composite out

A place to discuss general QL issues.
User avatar
dilwyn
Mr QL
Posts: 2753
Joined: Wed Dec 01, 2010 10:39 pm

Re: Help with Composite out

Post by dilwyn »

Derek_Stewart wrote:Hi,

I hope you have better luck with the HDMI upscaler. I could not get an upscaler to display all the QL screen.

I gave up and got a refund. Now using a GBS8220 to a VGA screen, all perfect no missing columns.
I have had mixed results with the scart-hdmi upscaler, which I bought at the time of the original discussion on here.

At the time, I used it with a Technika TV (my kitchen TV at the time!). Using a simple QL 8-pin DIN to SCART lead, I got a good quality picture, but the usual loss of a few characters at the edges of the picture. With the upscaler, it gave a good full 512 pixel display and I had nothing but praise for the device.

More recently, I've tried it with other TVs and monitors which have HDMI inputs. Mixed results - some do as well as the Technika, others fail to "upscale" at all, or give a poor quality picture.

Fiddling with settings on the TV sometimes helps me, one of the TVs has a setting (can't remember what it's called) which prevents or allows upscaling to happen. Not sure how it works, maybe it prevents the TV and upscaler both trying and fighting to adjust the display.

I also remember the late Lee Privett (former Quanta magazine editor) suggesting the use of a device from JS Technologies which went between a SCART connection and a SVGA input on a monitor instead of HDMI. It's fairly expensive at nearly £60, but if you asked them nicely (i.e. mention it's to be used with QL on ordering) they had a version which was slightly attuned to allow full QL display. http://www.js-technology.com/store/cate ... category=5. Lee wrote an article on the subject in Quanta mag once, I'll see if I can find it and scan it for people to see,

Things have of course moved on since those early upscalers such as mine and there's more choice and prices have largely come down.

EDIT: Found the article (from Quanta mag June/July 2011, and it's by Dr. John Sim, not Lee, sorry):
SimpleVideoScalar.pdf
See also an article at http://www.dilwyn.me.uk/docs/hardware/ql2vga.pdf


User avatar
bwinkel67
QL Wafer Drive
Posts: 1187
Joined: Thu Oct 03, 2019 2:09 am

Re: Help with Composite out

Post by bwinkel67 »

Martin_Head wrote:
bwinkel67 wrote:Now that I"m running off of NTSC color composite, the picture is better than RF. At least some of the artifacts do go away (like most of the shimmering). It still has the shadow effect where, for colors, the image is echoed a millimeter or two to its right (and there is no adjustments on the monitor to fix that).
The way analogue PAL and NTSC colour televisions work, is a high resolution black and white picture, and low resolution colours on top of it. If you just look at the colour parts of the television picture. You would just see coloured blobs. The coloured blobs have to overlayed onto the black and white picture in the correct place. Otherwise you would see a left or right mismatch, like you describe. I was never very 'up' on the NTSC system. And it's been a very, very long time since I did any theory on the PAL system. But the PAL system (I think it was PAD-D in the UK) has 'delay' lines to re-time the video signals to ensure that the colour, and the monochrome signals overlap correctly.

So this effect may be down to the cheap monitor. It could also involve this issue about the monitor thinking it's a PAL signal. Where it's adding, or not, the required delays to the signals.

For the resistors, take a look at http://www.dilwyn.me.uk/docs/hardware/scart.zip One of the examples uses 680 ohm resistors.
Thank you for that link. I've read before about how to reduce voltage so that's what he seems to be calculating. What I was curious about before I mod'ed my system is whether or not it was just a voltage issue on my color composite (pin 1), i.e. that it was at 5 volts. To cut down the voltage you need to resistance values who's ratio seems to cut down the voltage by that value...if I understand it correctly...maybe I'm confused. When I measure the voltage on the working color composite it shows 4 volts and varies very slightly. When I measure the voltage on the non-working color composite it shows 5 volts without varying. So I guess I'm wondering if I stuck a 100 ohm resistor on the line (with the assumption of the TV drawing 75 ohms giving me 3/4 ratio and thus 4 volts) would it give me a signal. Just hard to understand how composite works with regard to its voltage. Reading this makes it seem the voltage should be 1 volt, not 4 so again, not sure.

http://zone.ni.com/reference/en-XX/help ... nals_cvbs/


User avatar
tofro
Font of All Knowledge
Posts: 2685
Joined: Sun Feb 13, 2011 10:53 pm
Location: SW Germany

Re: Help with Composite out

Post by tofro »

What you are actually doing is creating a simple linear voltage divider with one resistor being the input impedance of your monitor (75 Ohms, as you say) and an added second resistor of your choice - I think you got that right.

Composite inputs are specified as 0.7 Volts (not 4!), so your added resistor needs to "eat" 4.3 from the QL's output of 5V, which results in a resistor ratio of roughly 6/1, --> roughly 460 Ohms.

If you use that, you will experience a relatively dark picture or no picture at all. Reason is, the above calculation is a blunt simplification, because it neglects that the QL's video output also has an output impedance which would have to be substracted fro the added resistor. We don't know that value, but can assume roughly 100 Ohms, resulting in a resistor value of 330-380 Ohms. The nearest values you can buy are 390 or 330 Ohms, so I'd try these two.

Tobias


ʎɐqǝ ɯoɹɟ ǝq oʇ ƃuᴉoƃ ʇou sᴉ pɹɐoqʎǝʞ ʇxǝu ʎɯ 'ɹɐǝp ɥO
Martin_Head
Aurora
Posts: 847
Joined: Tue Dec 17, 2013 1:17 pm

Re: Help with Composite out

Post by Martin_Head »

Unless there is something very different about the video out on the US version of the QL. Someone in the past has made a lot of modifications around the video socket. I suspect they cut the signal from the modulator to the composite video out (because they did not need it) and used the freed up pin as a 5 volt supply for an external unit plugged into the video connector. And those resistors made other modifications to the video signals (RGB, and sync) for the use of this external unit. Which I assume you don't have.

I would not recommend trying to bridge anything to that pin with the 5 volts on it until you have determined exactly what the modifications that have been made, do.

I meant to mention on the last post, that it looks like someone has drilled a hole through the printed circuit board, on the left hand resistor.

I had a look on the internet for a US QL user manual. To see what it said about the video signals, but I could not find one. So unless those modifications are factory fitted. I would be inclined to remove them all, and fix any cut tracks. To return it back to standard.


User avatar
bwinkel67
QL Wafer Drive
Posts: 1187
Joined: Thu Oct 03, 2019 2:09 am

Re: Help with Composite out

Post by bwinkel67 »

Martin_Head wrote:Unless there is something very different about the video out on the US version of the QL. Someone in the past has made a lot of modifications around the video socket. I suspect they cut the signal from the modulator to the composite video out (because they did not need it) and used the freed up pin as a 5 volt supply for an external unit plugged into the video connector. And those resistors made other modifications to the video signals (RGB, and sync) for the use of this external unit. Which I assume you don't have.

I would not recommend trying to bridge anything to that pin with the 5 volts on it until you have determined exactly what the modifications that have been made, do.

I meant to mention on the last post, that it looks like someone has drilled a hole through the printed circuit board, on the left hand resistor.

I had a look on the internet for a US QL user manual. To see what it said about the video signals, but I could not find one. So unless those modifications are factory fitted. I would be inclined to remove them all, and fix any cut tracks. To return it back to standard.
I bought this machine from Sharps in Mechanicsville, VA in 1989 from likely unused surplus of US machines after the end of Sinclair Research in 1986. It was definitely new.

The RF modulator works as that was all I ever used until I did the mod recently. I don't believe the resistors are a hack but a factory (or post factory for US) add-on. There are no traces cut. The hole in the circuit board does not look like someone drilled a hole. The motherboard says on it US 85 so it is specific to the states.

I have seen other machines (German) that have that blue wire connected to pin 1. I'm guessing that the resistors, since they have such small ohm ratings, are there to reduce interference to do with the RGB signal to satisfy the US FCC regulations.

My mod works and now pin 3 gives me both monochrome and color based on a toggle switch. The reason I'm still talking about this is because I'd like to know why there is a constant 5 volts on pin 1 and I thought perhaps it wasn't constant and just needed to be stepped down with the appropriate resistor. However, I tried putting a 390 ohm resistor on pin 3 and use that for composite and get no signal. When I measure output and compare pin 1 and pin 3, the first gives a 5 volt signal and stays steady and the latter gives a 4 volt signal and varies with screen changes. This is why I think pin 1 is just a 5 volt signal and gives me no picture (and pin 3 gives me both). Note that pin 3, regardless of monochrome or color, gives a 4 volt signal that varies. I realize the composite signal for the monitor needs to be around 1 volt but I measured it with no load and likely with resistance on the monitor end pin 3 will drop accordingly. I should actually test voltage on both pin 1 and pin 3 while it is connected on a running monitor to see what I get.

I should probably post on the general forum for any other US'ers to see what the back of their motherboard looks like. I appreciate your continued feedback on this. I keep learning more and more about analog video, which is kind of crazy across all the standards including composite, component, ega, cga, and vga. I have a decent picture now so it's not a need as much as it is interest in the topic.


Martin_Head
Aurora
Posts: 847
Joined: Tue Dec 17, 2013 1:17 pm

Re: Help with Composite out

Post by Martin_Head »

bwinkel67 wrote:I should probably post on the general forum for any other US'ers to see what the back of their motherboard looks like. I appreciate your continued feedback on this. I keep learning more and more about analog video, which is kind of crazy across all the standards including composite, component, ega, cga, and vga. I have a decent picture now so it's not a need as much as it is interest in the topic.
Probably a good idea. I know there some European people who collect various different kinds of QL's. Who probably have US style one's and know what signals should appear on the video connector.


User avatar
bwinkel67
QL Wafer Drive
Posts: 1187
Joined: Thu Oct 03, 2019 2:09 am

Re: Help with Composite out

Post by bwinkel67 »

Still playing around with video, though I've gotten used to using color-composite. Even with monitor (F1) mode in color-composite giving me a black&white image, the picture is usable. I think one difference between monochrome- and color-composite in monitor (F1) mode is that in color-composite, for red and blue, it just gives high contrast and a grid pattern making it look somewhat weird. But, once I run a game (Spook specifically), the picture is very watchable, almost normal, albeit no color. I should try Spook in F1 mode in both monochrome- and color-composite to get a better feel of the difference in the black&white image quality. As I said in the past, in F1 color-composite I do get an occasional monitor glitch (like a quick re-shift/align of picture) that shows brief flashes of color and a brief message saying PAL in upper corner.

However, the reason to post here is that I decided to compare the RF modulation signal to the color-composite again. I expected them to be similar but that wasn't quite true. They are almost identical in F2 mode with RF being slightly more shimmery (expected since it is an analog TV signal over channel 3) but still a good picture. Otherwise though, the two are identical in terms of image placement, etc...there is no difference in viewable area or screen size.

So I expected F1 to give me similar results, i.e. that color-composite and RF would give the same black&white high-contrast image. Certainly in initial appearance that seems to be the case. However, RF version gives different placement of the image on the screen as the left side shows the first character and the image is a little higher up. The RF image also does not show a PAL designation on top corner but it is definitely trying to show more than the TV can handle since it is more screen area and again, no color. Compare this to color-composite where the image area is bigger in F1.

So to test, when loading Spook in all 4 modes, things didn't quite align as expected:
  • (1) F2 color-composite: gives the expected smaller (TV) screen size and normal color with all characters visible
    • Spook gets bottom cut off so no menu options as pic cuts out after 8th top score
    (2) F2 RF modulation: gives exact same picture as above (F2 color composite) except a bit more shimmer
    • Spook does same thing, nothing after 8th top score
    (3) F1 color-composite: gives black&white picture (TV shows PAL in upper corner briefly) with high contrast
    • Spook shows the entire game board as the picture is squeezed a bit to appear fully (though in black & white)
    (4) F1 RF modulation: this does not match up with F1 color-composite which was surprising
    • Spook shows same picture as F2 but in black&white (i.e. after 8th top score we see nothing).
So that was a bit unexpected. I would have thought that (3) and (4) would give the same picture as I assumed that the RF signal was just taking the color-composite signal (from pin 9 of the MC1733) and sent it over the air on an analog NTSC channel via the RF modulator. Somehow it changes that signal causing it to display a smaller picture that is equivalent to the smaller TV mode of (1) and (2). So how does that do that? Is there another place where a color-composite signal is getting generated that you can access besides pin 9 of the MC1733?

I had been wondering if the mod I did had been at all necessary since the quality of F2 (TV mode) in either RF and color-composite was almost identical (the QL just has a really quality RF modulator), but now I know that F1 color-composite gives me the entire screen that F1 RF does not so that is a bonus. I'm still hoping that a better RGB-to-composite cable that is shielded will perhaps finally bring me color in F1 color-composite mode.


User avatar
bwinkel67
QL Wafer Drive
Posts: 1187
Joined: Thu Oct 03, 2019 2:09 am

Re: Help with Composite out

Post by bwinkel67 »

Test part 2. So looking at the monochrome-composite picture in detail (again using Spook, which is a pretty cool implementation of Pacman), it seems totally different. In places where the color-composite picture, in black&white, tries to display color (i.e. red or blue for instance) the area forms a grid pattern and looks to be moving slowly. Again, this is likely because the monitor is struggling displaying the color-composite image and doing its best. However, in monochrome-composite the colored areas are just solid gradations of grey. I don't believe they are simply trying to show color but instead truly create a monochrome black&white image. Everything is smooth, crystal clear, and crisp. Note that for me crisp means as clean as a modern LCD monitor on a laptop whereas sharp just means you see the changes in contrast too much when color areas transition and a clear grid pattern in in place of smooth color fill -- I know that may not be a standard definition but it's the best I can do.

I was hoping I could see some similarities in both and then for color-composite, perhaps add additional resistance values to soften the harsh sharpness of the color parts but I doubt that would work. The image size and placement for F1 monochrome-composite is identical to color-composite but that's where things end. Anyone know specifically how the composite signal is generated other than it comes on pin 3 of the RGB DIN socket? Does the MC1733 chip just process that picture differently and output it on a different pin?


User avatar
bwinkel67
QL Wafer Drive
Posts: 1187
Joined: Thu Oct 03, 2019 2:09 am

Re: Help with Composite out

Post by bwinkel67 »

F1 color-composite picture where monitor fails and shows black&white and claims PAL. You can see the grid pattern and those slowly are in motion:
color-comp-spook-f1.png
F1 monochrome-composite and the picture is smooth. The circular artifacts on the white part are from the iPhone camera and don't appear on the screen.
mono-comp-spook-f1.png
Also note that when you view image in full size that both pictures are about twice the size of what the monitor shows.

Interesting also is the edges of the picture where the monochrome-composite shows the block outline of the ghost clearly whereas in the color-composite some of that disappears. Looking specifically at the lower lip area really demonstrates this.

F2 color-composite picture added just so you can see how my monitor shows color when in TV mode. Note that the picture is a little wider since TV mode stretches the image but also removes the bottom portion (i.e. the stuff after the 8 high scores in Spook -- though here you are only seeing the ghost from top of picture).
color-comp-spook-f2.png


Martin_Head
Aurora
Posts: 847
Joined: Tue Dec 17, 2013 1:17 pm

Re: Help with Composite out

Post by Martin_Head »

In the first picture. Does the monitor/TV have a 'colour' control to turn the intensity of the colours up and down. If so does the patterning disappear or vary? This would indicate that the patterning is caused by the TV trying to 'decode' the colour signal, but having problems, and producing the patterns instead. Which would not be surprising if the TV is trying to decode a NTSC signal as a PAL signal.

If I have read what you say correctly, comparing RF and colour composite outputs in the F1 screen mode. The RF one switches the TV into PAL mode, where the colour composite one does not. If your TV, when it switches into PAL mode. Also switches into the UK mode of 625 lines (where the US mode is 525 lines). I would expect this to alter the size/position of the displayed picture. https://www.movavi.com/learning-portal/ ... etter.html this talks about the differences between the two systems.

When Sinclair designed the QL display circuitry, They bent the specifications of the TV signal a bit. Making the 'visible' part of the picture between the synchronising pulses, a bit wider than standard. That's why you can loose a character or two at the sides. Hence the F1/F2 Monitor and TV modes. Now I'm not too sure about NTSC, but in PAL there is a special signal between the end of the synchronising pulse and the start of the displayed picture called the 'burst'. It's used to synchronise an oscillator in the colour decoder. Which if not correctly synchronised will give you all sorts of colour problems (This is 4.4433MHz in PAL, 3.58MHz in NTSC, and I think there may be a 4.43MHz NTSC as well).
This may be where your monitor/TV is getting confused, If the burst signal and the start of the picture are a bit too close for the monitor to handle reliably, it may be misinterpreting the the signal, and choosing the wrong standard.

Here http://www.sinclairql.net/srv/QLSchematicIssue5.gif is a circuit of the QL, look in the top right hand corner for the video circuit. TR9 takes the RGB and sync signals, joins then together, and send them out of the mono composite pin of the video socket.
The RGB and the sync signal are also supplied to the MC1377, on pins 1,3,4,and 5. The colour composite signal comes out of pin9, and is fed to the modulator and pin 3 of the video socket (in a UK QL). I don't know about a US QL.
So the mono and colour composite signals are completely separate.


Post Reply