Why SSTV pictures come out slanted

The cause, in one paragraph

SSTV has no clock of its own. The transmitting station paints one line at a fixed rate, and the receiving station has to draw lines at exactly the same rate. Both derive that rate from a crystal oscillator on a sound card, and no two crystals agree perfectly.

Suppose your card claims 48,000 samples per second and actually delivers 47,998 — an error of about 42 parts per million, which is entirely normal for cheap hardware. Every line you draw is very slightly too long. After 256 lines the accumulated error has shifted the bottom of the picture sideways relative to the top, and the image leans.

The longer the mode, the worse it looks. Robot 36 gives the error 36 seconds to accumulate. Scottie DX gives it 269 seconds, and the same clock error produces a slant seven times larger.

Which end is at fault

This is the useful diagnostic, and it takes one afternoon of listening.

Slant is not tearing

These get conflated constantly and they have nothing to do with each other.

• Slant is smooth. The picture leans like italic text, every line displaced a fraction more than the one above. Cause: clock rate. Fixable by calibration.
• Tearing is abrupt. Blocks of lines jump sideways, then recover. Cause: the decoder lost the sync pulse in noise or interference. No calibration helps; you need a better signal.
• Rolling — the whole picture wrapping vertically — means the decoder started mid-transmission. Harmless, and it resolves on the next picture.

Correcting it

Modern decoders track the sync pulse at the start of every line, measure how far it has drifted from where it was predicted, and stretch or squeeze the line rate to match. That happens continuously through the transmission, so the picture comes out square without anyone touching a calibration control.

Open Signal does this automatically, which is why slant is something you read about more than you see. Where it still shows up is the far end: a station whose transmit clock is off sends a picture that is genuinely the wrong shape, and no receiver can invent the missing information.

Why auto-slant is not the whole answer

Automatic correction fixes what you see. It cannot fix what everyone else sees, because your transmit clock is still wrong — and no amount of cleverness at your end reaches their screen. This is why experienced operators turn auto-slant off and calibrate properly: calibration fixes the cause, and it fixes transmit.

Calibrating against a time standard

The reference method uses a station whose carrier frequency is known to be exact — WWV and WWVH transmit on 2.5, 5, 10, 15 and 20 MHz.

1. Tune 4.995 MHz USB (or 9.995 MHz). You are deliberately 5 kHz low, so WWV’s carrier lands as a 5000 Hz audio tone.
2. Measure what that tone actually reads. If your sound card is fast or slow, it will not be 5000 Hz.
3. The error is your sample-rate error, directly: 5001 Hz instead of 5000 Hz is 200 parts per million.
4. Apply that as the receive sample-rate correction, then set transmit to match.

From the UK, WWV on 5 MHz is best at night and 15 MHz in daylight, and reception is variable — it is a transatlantic path. Any carrier you know to be accurate works the same way; the 5 kHz offset is just a convenient way to land it in the audio passband.

Older software with no calibration control takes the other route: receive a known-good picture, measure how many pixels the bottom is displaced relative to the top, convert that to parts per million, and apply it as a sample-rate offset.

Before you calibrate anything

• Rule out audio dropouts. A busy laptop that stalls the audio thread produces damage that looks vaguely like slant but is not consistent.
• Rule out levels. An overdriven signal degrades in ways that mask the sync pulses — see audio levels and ALC.
• Check it is not just one station before adjusting anything at your end. Calibrating your receiver to match somebody else’s broken transmitter breaks you for everyone else.