• New signal generator

    Pavel • 10/14/2019 at 06:12 • 3 comments

    A couple of months ago I've dug this TV up again and played with it, using the generator from previous log, trying to measure signals inside the TV to understand what is going on. Unfortunately at one time I shorted something and fried some chips in the generator - those became very hot; afterwards the signal generator was no more outputting any periodic signal, just constant high or low on different leads.

    Thus the need for a new generator arose. This time I made it more  adjustable -- one can set start of horizontal blanking interval, timing of vertical and horizontal sync pulses, and  horizontal line time with switch banks.

    On the picture above, top row is made of only 74hc163 4bit synchronous binary counters (grey box), and mentioned switch banks are red at bottom.

    In between there is logic needed to facilitate adjustment (XOR and NAND gates -- two 74hc86 and one 74hc30).

    Functional blocks are inside rectangular shapes, ellipses indicate output/test leads.

    Functional parts by colour:

    In red box is the power supply -- 7905 IC with capacitors.

    In grey box is the counter which provides all timings by dividing the clock signal. Actually, it is in two parts - one counts to 1024 clock cycles for line timing, these are the 3 rightmost ICs hidden under foil; to the left is line counter which provides timings for frame. It counts up to 256, so there are 256 lines per frame total. 216 of them are used for image (intended to be only 200, but I've got my calculations slightly off and soldered some wire in the wrong place).

    In the light-green box on the bottom-left, near switches, is the clock circuit, uses one of the two quartz crystals (16MHz and 20MHz) - this drives the counters. Main clock is 16MHz, as this gives base line time of 64 microseconds, or frequency of 15625 Hz, which is perfect fit for the TV. Due to having less lines than standard PAL signal (256 vs 312), this gives me  61 frames per second. The tv seems to be fine with it. The 20MHz is experimental, it boosts line frequency by 25%,and it seems like it is on the edge of what tv can handle -- the picture is fine at first, but after a minute or two starts to distort. 

    From right to left:

    In rightmost light-grey box is the sync signal timing adjustment, one can use it to adjust when this signal is after start of the line. It has one output lead to check the signal is right shape on oscilloscope.

    Next to it, in green box is adjustment of start of line blanking interval. 

    In the blue box are two 74hc00 ICs. On one of them two gates wired as SR latch, which generates blanking interval from Start of Blanking and Start of Line signals. Checked on lead in light-blue circle. Other gates are used to generate crude test signal from this blanking and carry out from the rightmost 74hc163 counter - checked on lead in blue circle. Yet other gates on these ICs are forming complementary signals for line sync circuitry - they come off to a red/black pair of wires.

    Yellow shape contains circuitry needed for Vertical sync adjustment.

    Light-blue box has another SR latch, forming Vertical blanking interval. Other gates are wired to combine Vertical, Horizontal blanking, and two other inputs, which are picked off by two yellow wires. The wires have pins which can be inserted into headers next to counters. This way a grid of rectangular shapes can be created on the screen. Sizes of the shapes are 2^n dots x 2^m lines. The white and black circles are complimentary outputs, one gives white rectangles on black background, while another is black rectangles on white background.

    The transistor-resistor pairs are essentially RTL invertors, used here to interface between 5V level of this board with ~12V voltage used by TV's circuitry.

    The wire from clock circuit to counters has wire with complementary signal wound around it, and another wire which is connected to ground, to make some shielding from EM interference. The dot counters are also covered with Aluminium foil (sandwiched between...

    Read more »

  • Signals for TV. Sync generator.

    Pavel • 07/10/2018 at 13:25 • 1 comment

    Here is my description of my further tinkering with TV.

    So, i bought this little TV almost a year ago, dove into it  for a while, threw up some "unneeded" parts (tuner, most notably) and added  two wires to it, so I could feed composite signal into it. Fed some crude generated signal, saw white and black stripes and then put this project on hold.  And now my interest in it reemerged, so I dug this TV up and restarted activity on it.

    This time I decided to give a go for working with logic chips, as mere diode-transistor logic would be too slow for generating complex signals, so there are some fine patterns could be seen on the screen. Also, this way, the board is much more compact, has much less interconnect complexity, hence less chance of making a mistake. And works faster, yeah.

    I conceived counter-based circuit for outputting blanking and sync signals for crt tv. The signals are loosely based on PAL standard, but not exactly. It supposed to be non-interlaced 50 Hz 312 line composite monochrome video. For this project 16 MHz chrystal and bunch of 74HCxx ICs are used, 10 in total.

    Here is the schematic:

    (This image is exported from Digital logic simulator made by H. Neeman https://github.com/hneemann/Digital (Logisim clone))

    Base clock signal is 16 MHz, for each line there is 1024 clock cycles; for each frame there is 312 lines. Frame and line blanking intervals and sync pulses are generated by combinational logic from appropriately divided clock pulse. 

    Outputs L0 through L8 comprise dot counter for a line, so 512 dots (1024 clock cycles, 64 us) per line can be addressed. Among them, first 384 dots (768 clock cycles, 48 us)  are used for actual image, and other 128 dots are used up for blanking interval and sync pulse (4 us blank only, next 4 us blank and sync, and last 8 us again blank only).

    Outputs F0 through F8 are a line counter for a frame, counter resets at count 312, so there is 312 lines per frame, giving us 312l*64us=19,968 ms per frame, or ~50.08 frames per second (acceptable for TV). Among these 312 lines, first 288 are for image, and other 24 are for frame blank and sync (4 - blank, 4 - blank+sync, 16 - blank).

    Overall frame generated looks like this:

    The frame is scanned left to right, top to bottom. White color reprsents the area of image, grey is blank signal, black is blank+sync.

    Here is the actual board (quite messy):

    The chips are directly soldered to the board, all have capacitors between Vcc and GND (for lower row chips these caps are underneath the board). While making the board I learned a lesson about why direct solering of chips is bad practice -- I fried one of my 74hc163 synchronous counters (fourth in the top row), and had to desolder it and use ripple counter which I had instead. Test signal is also somewhat differ from schematics -- I used TC output (carry) from the first chip, and ANDed it with the third bit of line counter.

    Here is oscilloscope view of the signal generated (already hooket up to TV):

    Each horisontal division is 10 microseconds, each vertical division is 1 Volt, ground is the first solid line under downward-going sync signal. The signal levels are quite not standard, but my TV can cope with it (slight tuning of brightness and contrast is all it needs for the picture to be what it supposed to be).

    And, for the result is actual picture of the screen of TV fed with this signal:

    The picture looks almost like I thought it should look, overall. But there are some deviations: the horisontal resolution varies across the screen -- the distance between bars should be the same, but in actuality, in some places they are closer together, and in other the bars are further apart. It seems there are some problems with horisontal scan circuitry, it gives sawtooth signal with some interference. Also, not apparent on still photo, the picture as a whole wiggles slightly, changes size. The lines themselves are not stable and slightly tremble up and...

    Read more »

  • Feeding signals to tv

    Pavel • 09/02/2017 at 14:03 • 0 comments

    I have a small board based on 555 timer ic which can generate square wave signal with frequencies from ~30Hz up to ~64kHz. Here is the picture of this board hooked up to tv and oscilloscope (generating its maximum frequency), tv is shut off:

    As one can see, the signal is almost clear square wave. When tv is turned on, some noise from pre-amplifier is getting mixed in, and main signal gets distorted. Also the pattern is visible on tv screen: