Connect with us

Crystal frequency for monochrome video signal?

Discussion in 'Electronic Equipment' started by DaveC, Jan 30, 2013.

Scroll to continue with content
  1. DaveC

    DaveC Guest

    80's vintage German printing equipment (offset press industry) uses a video
    plug-in card (made by the manufacturer of this equipment) to generate
    parameter display for the operator. The display is a standard baseband video
    tube monitor. (It is possible, being German and sold in the USA market, that
    the video may be NTSC or PAL.)

    There is no video signal on the BNC output connector.

    This is used equipment being resurrected, so operational history is unknown.

    There is a place on the video card labeled "Q2" that is the right shape &
    size for a crystal can. The pads look like it was ripped off the board: a
    short lead soldered in one pad; a hole in the other pad where a lead was
    soldered (poorly, apparently!). (Rough handling is a distinct possibility:
    the client is a used-equipment dealer and the fork lift is their main

    The board is populated with 80's technology, mainly 74LS' :: the crystal pads
    connect to an 'LS04 inverter/driver and then to an 'LS96 parallel-to-serial
    converter. The 'LS96 spec sheet says that it can be driver up to 25 MHz.

    The board uses a 8275 CRT controller, and in the datasheet it says: "CCLK is
    a multiple of the dot clock and an input to the 8275."

    Maybe these clues will tell someone what frequency this crystal needs to

    What frequency crystal should I be looking for?

  2. tm

    tm Guest

    Can you feed in a test signal from a signal generator and see what you get
    on the display? A line in NTSC is about 64 us. If you have 80 characters x 7
    dots, that's 560 dots per line or about 0.114 us per dot. That gives about a
    9 MHz clock frequency. Maybe you can find a good signal generator and start
    out in that range. At least it would give you a clue as to what the video
    format should be.
  3. Guest

    Followups set to .

    Suggestion: Identify the "output" of the LS04, remove the LS04, hook up
    a function generator to the "output" trace, and start turning the knob.
    At some point, something resembling video should start coming out of the
    output. Continue turning the knob until the period on the video is
    correct. At a guess, the answer is probably somewhere between 1 and
    20 MHz.

    Another way to do it: Grab the first random crystal between 1 and 20 MHz
    you can find and solder it in. Look at the video output with a scope.
    You will probably see something resembling either PAL or NTSC video;
    compare the period of what you see to the standard, and change frequency

    An analog color TV will have a 3.579545 MHz crystal in it. An old PC
    motherboard (286 and below) will probably have a couple of crystals on
    it; one is often 14.31818 MHz. lists some
    common PAL and NTSC crystal frequencies.
    The "AC Characteristics" section gives the minimum CCLK period as
    480 ns, which is 2.08 MHz. That doesn't mean you need a 2 MHz crystal
    max; CCLK is one-eighth of the dot clock, so the crystal would be 16 Mhz
    max. From the block diagrams on the data sheet, if the parallel bus on
    the LS96 you found is hooked up to a couple of ROMs, then the clock
    input to the LS96 is probably the dot clock.

    Matt Roberds
  4. Tilmann Reh

    Tilmann Reh Guest

    If it's monochrome, we don't need to talk about NTSC or PAL and their
    particular color carrier frequencies...
    In Europe, especially Germany, horizontal frequency was 15.625 kHz and
    vertical frequency 50 Hz in those days.

    As others have already suggested, supply a reasonable clock to it,
    measure the sync outputs and then change the frequency accordingly.
    Chances are good that it's a standard and even frequency, like (for
    example) 16 MHz.

  5. Martin Brown

    Martin Brown Guest

    I'd try 13.5MHz first but anything in that ballpark and output the video
    to a multisync monitor and you should get some sort of picture.

    Old monitors don't like being driven too slowly for long periods.
  6. Tauno Voipio

    Tauno Voipio Guest

    Not quite.

    In PAL, the subcarrier is 4.433618 MHz and the line rate is 15625 Hz,
    the ratio is 283.75512, not an exact multiple.

    When I studied the thing in the 60's, the explanation was to find
    a frequency at as non-integer rate as possible, to get rid of
    Moire effects.
  7. Tauno Voipio

    Tauno Voipio Guest

    There is another consideration:

    The scan system is resonated on the third harmonic to the line rate
    to create the S-correction for the scan, slower on the edges and
    faster at the middle. This is to compensate for the varying distance
    between the electron gun and the screen. This means to keep the
    line rate within a few percent of the nominal. The method was
    populas with the monochrome tubes, but colour things often use
    more sophistacated methods (parabolic correction, etc).
  8. Guest

    You mean fractional multiplier ?

    The B&W contains spectral peaks at multiplies of line and field rate.
    For (stationary) images, there is no energy between the spectral

    The whole idea of both NTSC and PAL (but not SECAM) is to code the
    chrominance signal into these "empty" spaces and thus the subcarrier
    must be at some submultiple of the line rate. In PAL there is an
    additional 25 Hz frequency shift, thus the same phase relationship
    occurs every 4th field.

    While this spectrum interleaving works pretty well for stationary
    images, any movement will spread the spectral lines and luminance and
    chrominance can no longer be perfectly separated, causing cross
    chrominance and cross luminance problems. For this reason ties with
    small details should not be used in TV studios if it is expected that
    the signal could be transported through NTSC/PAL, since a tie with
    only small B/W stripes would cause a quite colourful result :).
  9. In PAL & NTSC the colour carrier was a multiple of the line rate.
    This is not a correct explanation -- and I'm certain your numbers are wrong
    (you've rounded off the line rate).

    The subcarrier HAS to be a multiple of the line rate -- specifically, an odd
    multiple of half the line rate -- or the sidebands of the color signal will
    not properly interleave with the sidebands of the luminance signal.

    By the way, moire is not capitalized. It is not a person's name.
  10. The B&W contains spectral peaks at multiplies of line and field rate.
    This is not correct, unless every line is like every other line. The normal
    variation in vertical details causes the peaks to "smear" somewhat.
  11. Rich.

    Rich. Guest

  12. tuinkabouter

    tuinkabouter Guest

    As far as I know it is
    283.75 fH + 1/2 fV = 283.75 fH + 25 Hz = 4433618.75 Hz

    I don't recall why. But Google has the answer:
    To make the dot pattern that results from the colour subcarrier almost
    invisible, we need to satisfy this equation:

    4*fc - 2*fr fl = line frequency (15625 Hz)
    fl = ----------- fc = colour subcarrier frequency
    n fr = frame rate (50 Hz)

    This will ensure that dark and light dots cancel each other as much as
    possible between alternating lines and between alternating frames.

    The number n must be odd, and high enough to get a high enough colour
    frequency. It was chosen to be 1135.

    We get 4*fc - 2*fr = 1135 * 15625
    fc = 1135 * 15625/4 + 25 = 4433593.75 + 25 = 4433618.75 Hz

    Note: when colour television was first on the air, we did not have the
    25 Hz offset yet. And although the dot pattern should have cancelled over
    the screen, it was visible and you could tell, from watching your old
    black and white screen, that a colour transmission was on.
    After adding in the 25 Hz "time compensating" offset, this was no more.

    The "integration time" for the screen is 4 frames, or 80 milliseconds. If
    you could photograph the screen with that as the exposure time, the dot
    pattern would be absolutely invisible.
    If you took a picture with an exposure time of 2 frames (40 ms), or in
    other words exactly one complete screen, you could see the residual dot
  13. Fred Abse

    Fred Abse Guest

    That's yet another version of the widely-held misapprehension about
    horizontal output harmonic tuning.

    The *leakage inductance* of the flyback transformer is resonated at either
    the third (monochrome), or fifth (color) harmonic of the "flyback
    frequency", which is the reciprocal of twice the flyback time, somewhere
    around 3, or 5 times 50kHz for NTSC/CCIR 525/625 line TV, assuming 10
    microsecond flyback.) This has the effect of flattening the peaks of the
    (half-sine) flyback pulses, and can be seen as either one or two small
    dips in peak flyback voltage. In early tube designs, this was done with a
    small winding underneath the HV winding, which was resonated with a
    capacitor. In later designs with diode-split windings, it was done by
    carefully controlling interwinding capacitance.

    S-correction is a separate issue, achievable with a suitable capacitor in
    series with the actual scanning current.
  14. Guest

    I have quite often used the following example what the B&W signal
    looks like:

    There are quite often repeating hills every 15625 Hz with a tree
    standing at every 25 Hz starting from the top of the hill

    With severe wind (image movement) the tree branches will be mixed
    with each other, making it impossible to separate luminance and
    chrominance properly.
  15. wrote in message
    True, but you're missing the point of what I said. "Movement" is sufficient,
    but not necessary. Changes in vertical detail produce same effect. That is,
    no ordinary object is the same from line to line.
  16. halong

    halong Guest

    Do you have pictures ?
  17. Mark Zenier

    Mark Zenier Guest

    The Intel MDS blue box system used it, if someone wants a design

    Mark Zenier
    Googleproofaddress(account:mzenier provider:eskimo domain:com)
  18. Mark Zenier

    Mark Zenier Guest

    Yes. (Overdesigned slug). There were enough of them around that somebody
    probably has scanned the schematic and posted it on on of the document

    Yea, there are reasons that nobody used the 8275 much. I lucky to
    find a 1984 databook with its datasheet on the top of my pile of boxes
    of databooks.

    Yuck. It was one of many Intel peripheral chips from the late '70s
    that took the wrong road. It is a programmable video counter chain
    (like the 6845, the one used on the IBM video cards), but instead of
    outputting all the address bits to feed an external memory, it had two
    internal 80 character display buffers that were loaded by program or DMA.
    (So the driver had to babysit it, making sure it got a new line of text
    every milisecond or so). In operation, it only output the data byte, the
    row counter, and some attribute bits to feed the character font ROM.
    (Using it for graphics, at one display line per row of data would
    either have saturated the micro's databus, or isn't even be possible).
    It only works at the level of characters, the video serializer and
    dot-per-character counters were external TTL. It only ran at 2 or 3 MHz.

    Mark Zenier
    Googleproofaddress(account:mzenier provider:eskimo domain:com)
  19. Guest

Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day