Tweet
Loudthud and Trobbins have it correct.
Ordinary silicon rectifier junctions like the 1N40047 and others not specified as specially fast recovery have the issue. Being majority and minority charge carrier junctions (for which, check on some semiconductor physics texts), the active part of the semiconductor material is full of charge carriers during conduction. When the voltage reverses and sets up the conditions for the junction to block current, it can't quit conducting until the charge carriers are swept out of the active region or neutralized by recombination. The time this takes is the the major component of the specified reverse recovery time.
What is quirky is that the shutoff is not a smooth shutoff, but a nearly instantaneous stop when the charge carriers get swept out or neutralizes. It's this super-abrupt stop that shocks the inductance and capacitance of the wires and other parts connected to the diode junction into ringing as RF sources. You get this little spike of oscillation. It can be conducted or even radiated as a radio signal and picked up by the audio circuits.
The pulses of RF happen at 2x the line frequency, or 8.6mS for 60Hz, 10mS for 50Hz. If this gets into the audio as pulses, they are amplified and put out as a pulse train at the same rate. The human ear is very good at picking out the repitition rate of pulses as a frequency on its own. The ear still hears the buzz that is where the majority of the energy is, but still perceives the pulse repitition rate as a hum at 120 (or 100) Hz.
Capacitors do tune the ringing frequency of the parasitic inductances and capacitances, primarily running the ring frequency down so low that it doesn't radiate well, or as noted, tuning the ring to a frequency that is not well picked up by the rest of the circuitry. However, it is possible to put a resistor/capacitor snubber in that will not only run the frequency down lower but also eat the energy in the ringing, effectively spoiling the Q of the wires so that much less energy gets transmitted out. The only down side to snubbers as opposed to lowering frequency with just a cap is that the snubber style needs tuned to the actual wiring.
Fast recovery, ultra-fast recovery, and FAST!! recovery diodes work by having such a short recovery time that there is much less energy in the pulse that excites the ring. I read a whole paper on this topic somewhere. I'll try to find it.
Ordinary silicon rectifier junctions like the 1N40047 and others not specified as specially fast recovery have the issue. Being majority and minority charge carrier junctions (for which, check on some semiconductor physics texts), the active part of the semiconductor material is full of charge carriers during conduction. When the voltage reverses and sets up the conditions for the junction to block current, it can't quit conducting until the charge carriers are swept out of the active region or neutralized by recombination. The time this takes is the the major component of the specified reverse recovery time.
What is quirky is that the shutoff is not a smooth shutoff, but a nearly instantaneous stop when the charge carriers get swept out or neutralizes. It's this super-abrupt stop that shocks the inductance and capacitance of the wires and other parts connected to the diode junction into ringing as RF sources. You get this little spike of oscillation. It can be conducted or even radiated as a radio signal and picked up by the audio circuits.
The pulses of RF happen at 2x the line frequency, or 8.6mS for 60Hz, 10mS for 50Hz. If this gets into the audio as pulses, they are amplified and put out as a pulse train at the same rate. The human ear is very good at picking out the repitition rate of pulses as a frequency on its own. The ear still hears the buzz that is where the majority of the energy is, but still perceives the pulse repitition rate as a hum at 120 (or 100) Hz.
Capacitors do tune the ringing frequency of the parasitic inductances and capacitances, primarily running the ring frequency down so low that it doesn't radiate well, or as noted, tuning the ring to a frequency that is not well picked up by the rest of the circuitry. However, it is possible to put a resistor/capacitor snubber in that will not only run the frequency down lower but also eat the energy in the ringing, effectively spoiling the Q of the wires so that much less energy gets transmitted out. The only down side to snubbers as opposed to lowering frequency with just a cap is that the snubber style needs tuned to the actual wiring.
Fast recovery, ultra-fast recovery, and FAST!! recovery diodes work by having such a short recovery time that there is much less energy in the pulse that excites the ring. I read a whole paper on this topic somewhere. I'll try to find it.
) In which case it seems to me that it is likely switching noise. Though I have no prior experience in the matter. I would try the faster switching diode as trem suggested. I don't know the formula, but different value snubbers might be worth experimenting with. But see the next paragraph...
Comment