Improving Composite Output of 48K Spectrum

edited April 2013 in Hardware
I've been playing around with this Spectrum 48k I got inside a dK'tronics keyboard, because well I've already got two working other ones so why not...

...and I have become quite obsessed with improving the composite video out modification. Like, really obsessed. I've done a lot of googling and found some things that I don't get why they would work, and then tried them and found out they didn't seem to do anything.

The first thing I did was remove the contents of the UHF modulator box, and wire the composite signal wire to the centre pin. That was a good start, but I was sure it could be improved.

Then I saw this site talked about adding a 100uF capacitor in series with the signal out:
http://samsmods.blogspot.co.uk/2011/10/zx-spectrum-48k-composite-video-mod.html

Can anyone explain the reasoning behind this? It made no difference at all, from what I can tell. The idea of adding it is linked as coming from here:
http://user.tninet.se/~vjz762w/rebuild.html

From that latter site, I tried using an NPN transistor on the signal line which actually made a nice difference.

The other thing I did was add a simple low pass filter to the output of the amplifier, using a 270pf capacitor and 100 ohm resistor, that definitely removed some of the noise from the signal.

I got the value of the capacitor by using a calculator on a website that showed you the frequencies it would block, so I looked up the maximum frequency of a composite video video (which is 5.5Mhz I think?) and then the nearest capacitor value I could easily get hold of (as in, had to hand).

Calculator was here:
http://sim.okawa-denshi.jp/en/CRlowkeisan.htm

I suspect the reason I have a load of 270pf capacitors lying around is from when I built an Atari 2600 composite mod. A good sign I was on the right track I thought. :)

So now I feel I have something quite good, but there is still a fair amount of noise on the signal.

There is an interesting description here on how the video signal is generated:
http://www.1000bit.it/support/manuali/sinclair/zxspectrum/sm/section1.html

From looking at the output of the ULA luminance pin, the noise is there already. Also using FFT (on the composite out) there seems to be a prominent peek at about 13.3Mhz, which is close to the 13.8Mhz the main crystal is generating, but that could be a coincidence. Or maybe it is being introduced to to that crystal's proximity to the modulator box, I'm not sure.

Anyway, I need to do some more digging, but I'm wondering if the 4.4Mhz oscillator that drives the chroma carrier wave may be a cause of the noise? Looking at how dodgy the output of the 14Mhz oscillator is, could this be a factor in it?
Post edited by RobeeeJay on

Comments

  • edited March 2013
    This is very interesting RobeeeJay,

    A while back I was thinking of building a small board for the ULA to sit on - that would take the ULA output and encode it properly for use with SVideo - output.

    But then this mcleod_ideafix fella goes and creates the ULA+ that everyone has been talking about for ages and its clear that this is the way of the future - better graphics modes and RGB-Out - ..... that mcleod_ideafix is my hero!

    Although.... that SVideo idea would been a good one...... sigh..... damn better designs!
  • edited March 2013
    The capacitor should be there to prevent to much DC drain from the transistor (emittor follower) that handles the composite video. F.i. when the output is accidently shortcutted. Or connected to some 'life' wire. Don't try.
    Probably a capacitor is also found in the input stage of the monitor/tv for similar safety reasons.

    The synchronisation pulses in composite video are negative, so below '0V'. This 0V level is established artificial and somewhat arbitrairy 'behind' a capacitor, and is affected by the C and R values in the 'load'. A change in the zero level affects the amplitude of video and sync pulses.
    The best findings for one particular monitor may not be the best for an other!
    An added transistor stage, when well designed, may create a better defined zero-level. Plus a better matching to the 75 Ohm impedance that the video input of your monitor/tv expects. Over a 75 Ohm coax cable of course....

    Consider the 'video' part of the signal not as sinusoidal but as steep pulses. The needed bandwidth is not so much depending on the base-frequency of the pulses, but on the speed with which the changes occur. Filtering will also change the shape of the peaks (tops) of the signal. An R/C filter is not that steep, 5.5 MHz won't hurt too much, better kill glitches at the source.
    Some of the noise may reach the LM1889 over the +12V and +5V lines.

    It is a complicated matter, RobeeeJay. But since you're obessed anyway I thought I could mention (briefly!) a few aspects that might help you to interprete the effects you notice during your experiments.
    You can look at signals, so using a scope?

    Description for S-Video:
    http://www.sps-wagner.de/Sonstiges/zx.html
  • edited March 2013
    RobeeeJay wrote: »
    I tried using an NPN transistor on the signal line which actually made a nice difference.

    Just in case you didn?t know: there are specialized video-quality operational amplifiers ( like the LMH6723, among many others ), which do the job much better.
    'Interface 1bis' for the Sinclair ZX Spectrum
  • edited March 2013
    Dan Antohi wrote: »
    Just in case you didn?t know: there are specialized video-quality operational amplifiers ( like the LMH6723, among many others ), which do the job much better.

    That's handy to know, ta.

    What effect does the LMH6723 have on the video signal? Does it clean it up at all?
  • edited March 2013
    roko wrote: »
    The capacitor should be there to prevent to much DC drain from the transistor (emittor follower) that handles the composite video. F.i. when the output is accidently shortcutted. Or connected to some 'life' wire. Don't try.
    Probably a capacitor is also found in the input stage of the monitor/tv for similar safety reasons.

    That (sort of) makes sense. Wouldn't it block DC though? This is what I don't get about it really, I did a lot of fiddling and I think it blocked the signal completely. And when I look at it on the scope the composite is a DC signal.

    roko wrote: »
    The synchronisation pulses in composite video are negative, so below '0V'. This 0V level is established artificial and somewhat arbitrairy 'behind' a capacitor... An added transistor stage, when well designed, may create a better defined zero-level. Plus a better matching to the 75 Ohm impedance that the video input of your monitor/tv expects. Over a 75 Ohm coax cable of course....

    That makes sense, my monitor is quite forgiving (Commodore 1080S), but I think the main difference the transistor made was a more consistent black.

    roko wrote: »
    Consider the 'video' part of the signal not as sinusoidal but as steep pulses. The needed bandwidth is not so much depending on the base-frequency of the pulses, but on the speed with which the changes occur. Filtering will also change the shape of the peaks (tops) of the signal. An R/C filter is not that steep, 5.5 MHz won't hurt too much, better kill glitches at the source.
    Some of the noise may reach the LM1889 over the +12V and +5V lines.

    Thanks, I may try filtering at a lower frequency and see if that makes any difference. And I'll look at how much noise is coming from the +12V line, I think the +5V is quite clean, but I haven't checked the +12V.

    I guess Sinclair decided an unregulated PSU and building an invertor+step up thingy on the board was cheaper or more reliable than a custom PSU that delivered all the right lines.

    roko wrote: »
    It is a complicated matter, RobeeeJay. But since you're obessed anyway I thought I could mention (briefly!) a few aspects that might help you to interprete the effects you notice during your experiments.
    You can look at signals, so using a scope?

    Yes, I am, and it is, but it's interesting. And your explanations are very helpful thank you! I'll dig further this weekend and maybe try to map out where all the noise is coming from.

    (Didn't understand your link, I can't read German, but I understand some of the basics of the video signal).
  • edited March 2013
    I don't think I've ever seen a TV which doesn't have a DC blocking capacitor on the inputs.

    The first thing that happens to the video signal is that any DC component is removed by the blocking capacitor. Then the sync peaks are detected and used as the reference for the DC clamping circuit.
  • edited March 2013
    Indeed the video coming from Tr2 over R53 exists of fluctuating DC levels. A capacitor separates the AC part.
    Notes I once made say that at R53 the DC 'component' = 2V. I think I ment that sync pulses go down from that level, and video starts (going up) from that level. Or maybe I was just doing calculations for a next transistor stage, but that does not much affect the general picture.
    This DC level as such is not crucial, but it determines the energy in the sync pulses, from which the monitor determines the reference level (see Guesser).
    Which in turn (in general and a.o.) determines the brightness of the video. ("consistent black")
    The size of the series capacitor (22-100uF) and the load of the given monitor play a role in this balance.
    Keep us informed.
  • edited March 2013
    roko wrote: »
    Indeed the video coming from Tr2 over R53 exists of fluctuating DC levels. A capacitor separates the AC part...
    ...The size of the series capacitor (22-100uF) and the load of the given monitor play a role in this balance.

    This is the bit I'm struggling with, in this composite video mod here:
    http://samsmods.blogspot.co.uk/2011/10/zx-spectrum-48k-composite-video-mod.html

    He has a 100uf capacitor in series.

    Now I can understand that this would work if the signal was a mixture of AC and DC signals, but the voltage for the composite out signal never drops below 1.8v, so the current never flows in the other direction.

    So I can't see how this would do anything but block the signal.
  • edited March 2013
    RobeeeJay wrote: »
    So I can't see how this would do anything but block the signal.

    The signal is a mixture of AC and DC components. The blocking capacitor removes the DC offset so that the average DC voltage is zero. :smile:
  • edited March 2013
    guesser wrote: »
    The signal is a mixture of AC and DC components. The blocking capacitor removes the DC offset so that the average DC voltage is zero. :smile:

    I understand now, thanks. :) Will see it in action come the weekend.
  • edited March 2013
    OK. This 100uF capacitor can (must!) be seen as 'grounded' via the monitors input(-impedance). Now different DC voltages charge this capacitor to different levels. When the 'next' level is lower than the previous, some discharging HAS to take place. This results in an 'opposite' current in the wires toward the capacitor. Which current creates the 'negative' part of the AC that can be taken from the capacitor.
    Four sentences used sofar. Several WOSsers can explain better, I doubt if they can be more brief....:-)
    Edit: Ah, Guesser can. Now is'nt 'Guesser knows' a contradiction in terms? :-))
  • edited March 2013
    The first thing you should probably consider, if you haven't already, is reaplacing all the capacitors on the motherboard. One of the big ones at the bottom-middle of the board (C44) puts some particularly noticeable noise lines into the video signal.

    http://www.worldofspectrum.org/forums/showthread.php?t=30498
    Joefish
    - IONIAN-GAMES.com -
  • edited March 2013
    roko wrote: »
    When the 'next' level is lower than the previous, some discharging HAS to take place. This results in a 'opposite' current in the wires toward the capacitor. Which currents create the 'negative' part of the AC that can be taken from the capacitor.

    And now I understand better. :) TY!

    joefish wrote: »
    The first thing you should probably consider, if you haven't already, is reaplacing all the capacitors on the motherboard. One of the big ones at the bottom-middle of the board (C44) puts some particularly noticeable noise lines into the video signal.

    I'll take a look at that, ta. So far I haven't replaced anything because I really want to understand where it is coming from.
  • edited March 2013
    http://samsmods.blogspot.co.uk/2011/10/zx-spectrum-48k-composite-video-mod.html

    I didn't understand what is that flying cable over the motherboard.

    And, what about this? Is it worst to keep the rf modulator onboard?

    I did this mod and it works fine!
    videofixn.JPG
  • edited March 2013
    DaRkHoRaCe wrote: »
    http://samsmods.blogspot.co.uk/2011/10/zx-spectrum-48k-composite-video-mod.html

    I didn't understand what is that flying cable over the motherboard.
    That's the reset switch. If you have a rubber key machine this won't be present.
    DaRkHoRaCe wrote: »
    And, what about this? Is it worst to keep the rf modulator onboard?

    No, it makes no difference, just less tidy I suppose.
  • edited March 2013
    As soon as that resistor and 5v line are cut the modulator may as well not be there anyway, dont see the need to create extra work by removing all the innards
  • edited March 2013
    +12V Power Rail

    As the +12V supply comes from a switch mode power supply, a filter network is needed to remove the high frequency ripple. A filter is constructed using resistors or inductors and capacitors.

    Where electrolytic capacitors are used, these should be replaced with new capacitors, as the existing capacitors will have degraded over time. Sometimes you can improve the filter by fitting a larger value capacitor (and as modern electrolytic capacitors are smaller that should be easy, so for example fit a 22uF in place of a 10uF) and / or adding 100nF ceramic capacitors in parallel with the electrolytic capacitors (electrolytic capacitors are not good at dealing with high frequency signals).

    Composite Video Signal

    This is a very complex analogue signal that is actually made up of many different components.

    Most electronic circuits that process low power AC signals have a single power rail with a common "ground" (0V). Because of the single power rail, the circuits cannot process the negative parts of a AC signal directly. So instead, AC signals are combined with a DC level (in audio this is typically approx. half the supply voltage). Think of the AC signal as siting on top of a steady DC level.

    When a AC signal has to be fed from one stage to another, or from one piece of equipment to another item of equipment, as the different circuits may have different DC levels, they have to be isolated as far as the DC is concerned. Hence the "in-line" "coupling" capacitor.

    I hope this helps :D

    Mark
    Sinclair FAQ Wiki
    Repair Guides. Spanish Hardware site.
    WoS - can't download? Info here...
    former Meulie Spectrum Archive but no longer available :-(
    Spectranet: the TNFS directory thread

    ! Standby alert !
    “There are four lights!”
    Step up to red alert. Sir, are you absolutely sure? It does mean changing the bulb!
    Looking forward to summer in Somerset later in the year :)
  • edited April 2013
    I found this:
    http://piters.tripod.com/uvyrgb.htm

    May be you can use it, i dont think that it wil fit in 48k case maybe in a ZX Spectrum+ case.
  • edited April 2013
    Interesting, thanks for linking it. I'm not sure I want to go as far as adding RGB support, if I want that I can just plug in my toasttrack, +2 or +3 into my monitor. :)

    I did some more probing, and I think most of the noise is coming from the 5V rail into the ULA, but I'm not sure. I need to eliminate any other possible causes of that noise first.

    I need to try powering the ULA from a bench supply to see if that makes a difference, but I'm away from my Speccy atm.
Sign In or Register to comment.