Manfred's FT736 impulse noise problem

Brief description of the problem: The receiver noise, instead of being pure white noise, has random pulses on it with much higher amplitude than average. At zero signal it's not really audible. In SSB mode, weak signals are heavily interfered by this impulse noise. Signals above roughly S7 become clean, thanks to AGC action reducing all noise, including these impulses. Reducing the RF gain control in SSB on weak signals also has the effect of making the noise pulses less troublesome, as it avoids saturating the AF LPF module on the tall peaks, but the pulses themselves remain present.

In FM mode the impulse noise is particularly easy to hear and see, when a weak carrier is being received. Stronger signals eliminate the noise impulses through limiter action.

The noise happens on all four bands in the same way. It's definitely generated inside the radio, as it also happens when using two different signal generators, and when using the radio in the middle of the desert, a long distance from any human settlements or power wires. TX is fine.

It cannot be pickup of interference from some internal source, such as the digital circuitry, because that would produce constant level pulses. But my pulses appear in a strength proportional to the receiver's white noise. It seems rather that the IF gain pulses up and down randomly!

The problem started roughly in 2003, after using the radio for over 10 years 24/7 in my fully automatic digital sat station, without any such noise. Since it first appeared, it seems to be constant and stable. It does not react to vibration nor temperature changes.

The effect is most troublesome when receiving digital signals, because the the data is corrupted by the pulses.Packet radio operation is impossible, unless the signals are very strong. At least 10dB higher signal strength is required for decoding, than before the problem started.

This is an oscilogram  showing about two and a half seconds of the impulse noise, riding over white noise, taken directly at the discriminator, while applying a weak carrier that produces incomplete quieting:




This is the detailed oscillogram of a single impulse, taken at the same place, surrounded by normal white noise. The pulse width and shape seems to be given by the discriminator's and IF bandwidth characteristics:


And this is how it sounds, also directly at the discriminator:    Poroteo.mp3

Any help is welcome!

Manfred