TECHNICAL FIELD

The present invention relates to apparatus for eliminat¬ ing noise and/or hum in a signal. The arrangement in¬ cludes an attenuation circuit connected between a signal input and a signal output and a detector and a non-linear filter circuit connected in series between the signal in¬ put and a control input on the attenuation circuit.

BACKGROUND PRIOR ART

Noise eliminating apparatus are previously known in music contexts. These apparatus, however, are normally rela¬ tively expensive to buy and do not fulfill the high re- quireinents placed on good musical standards, except in the case of the most expensive apparatus. This applies, for instance, in the case of tones or sounds which nor¬ mally shall be ^' toned down or moderated slowly and con¬ tinuously under one noise level, but which are attenuated abruptly and unexpectedly when a noise-gate turns off and another control (RELEASE) is on minimum, which results in a dramatic and abrupt termination of respective tones. Another example resides in tones which shall normally have a soft and gentle start, i.e. a soft rise to the in- tended amplitude but, which start abruptly, i.e. rise quickly to the intended amplitude, when the noise-gate turns on and another control (ATTACK) is on the minimum setting.

A further example resides in those situations where the noise-gate turns on and off several times in sequence when the level of the input signal is close to the gate threshold level. This may be becuase the signal strength varies as a result of interference between the different tones of a musical chord. This can sound extremely aw-

ful .

A rapid tone start can also be distorted to some extent when the gate is not able to respond until the tone start 5 time-lapse has passed the gate threshold value (i.e. the value at which the gate is triggered). This results in a click sound at the start of a tone.

SUMMARY OF THE INVENTION

10 In accordance with the invention, the apparatus for elim¬ inating noise or hum described in the introduction is characterized in that the non-linear filter circuit has a variable time constant. More specifically, the time con¬ stant has 1) a minimum value tl when the amplitude of the

15 input signal lies slightly above a given pre-determined level Al,' 2) has a minimum value t2 when the amplitude of the input signal lies between the threshold values Al and A2 (A2<A1 ) ; and 3) a value t3 which is lower than the maximum value t2 and higher than the minimum value tl

2.0 when the amplitude of the signal lies slightly beneath the said threshold value A2.

These and other characteristic features of the invention are- set forth, in the following claims. 25

BRIEF DESCRIPTION OF THE DRAWINGS

The invention will now be described in more detail with re erence to the accompanying schematic drawings:

3 ^" 0 Figure 1 illustrates a known apparatus which comprises an attenuation circuit, detector and non-linear filter cir¬ cuit;

Figure 2 illustrates the time constant of the filter cir- 35- cuit in diagram form;

Figure 3 illustrates one embodiment of the detector;

Figure 4 illustrates one embodiment of the attenuation circuit. .

Figure 5 illustrates one embodiment of the filter circuit with three branches connected in parallel;

Figure 6 illustrates a low pass filter which is included in a first branch ;

Figure 7 illustrates a detector incorporated in a second branch ;

Figure 8 illustrates a detector incorporated in a third branch ;

Figure 9 illustrates a second embodiment of the filter circuit which incorporates a series-connected low pass filter;

Figure 10 illustrates one embodiment of a non-linear ele¬ ment incorporated in a first low pass filter;

Figure 11 illustrates a characteristic for a second low pass filter;

Figure 12 illustrates a characteristic for a limiter;

Figure 13 illustrates a variant of the input section of the filter circuit according to Figure 9;

Figure 14 illustrates another variant of the input sec¬ tion of the filter section according to Figure 9; and

Figure 15 illustrates a modification of the uppermost branch of the circuit illustrated in Figure 5.

DESCRIPTION OF PREFERRED EMBODIMENTS The apparatus illustrated in Figure 1 includes an atten¬ uating circuit 13 which is connected between an input terminal 11 and an output terminal 12, and also a detector 14 and a non-linear filter circuit 15 which are mutually connected in series between the input terminal 11 and the control input 131 of the attenuating circuit 13. The filter circuit has a variable time constant t, the depen¬ dency of which on the amplitude A of the input signal supplied to the input terminal 11 of the apparatus will be apparent from Figure 2-

The time constant t has a minimum value tl when the ampli¬ tude A of the input signal lies slightly above a given pre ₇ determined level _s Al.

The time constant has a maximum value t2 when the ampli¬ tude of the input signal lies between the threshold values Al and A2 (A2<Al) .

The time constant has a value t3 which is lower than the maximum value t2 and higher than the minimum value tl when the amplitude A of the input signal lies slightly beneath the aforesaid threshold value A2.

The time constant t of the non-linear filter is thus very short (down to about 0.1 ms) when the input signal is very strong; the time constant t is progressively lower with decreasing signal strength and has its maximum value (>100 ms) somewhere between the threshold values Al and A2 of the signal. When the signal strength falls from the threshold value Al to the threshold value A2, the time

constant is held substantially constant. The time con¬ stant becomes progressively shorter with further decreases in signal strength (down to ca 10 ms) . The aforementioned click sound encountered at the start of a tone is hereby suppressed, as is also the aforesaid repeated turn on and turn off of the noise gate as a result of, e.g., interfer¬ ence between the different tones of a musical chord.

The detector 14 illustrated in Figure 1 and in slightly more detail in Figure 3 has connected in series in one branch of the circuit or network an amplifier 141, a delay circuit 142 having a delay of about 20 ms, and a.c./d.c. circuit. 143 and a diode 145. A branch connected in paral¬ lel with the units 142, 143, 145 incorporates an a.c./d.c. circuit 144 and a diode 146. ^" The output terminal 148 of the detector is earthed over a resistance 147. The diodes 145 and 146 are actually incorporated in the a.c./d.c. circuits themselves and have only been shown in the draw¬ ing in order to express clearly that the. voltage of the output signal is the greatest voltage of the two output volatges from the a.c./d.c. circuits 143 and 144.

This improved detector also senses signals of very low frequency (beneath ca 50 Hz) as signals of constant ampli- tude, despite the fact that the curve form of the signals may have a pulse train configuration. The curve form of the input signal is- smoothed more effectively in this man¬ ner, and the input signal will not affect the control sig¬ nal fed to the attenuating circuit 13, i.e. the occurrence of intermodulation distortion at low frequencies is avoided,

A detector of this kind can be enlargened with a plurali¬ ty of parallel-connected branches of the same kind as the uppermost branch in Figure 3. In this case, however, the delay circuit shall have mutually different delays

suitably distributed linearly between zero and maximum delay (in the illustrated example 20 ms).

Smoothing of the input signal envelope becomes more effective with an increased number of branches. Further¬ more, the level of the limit frequency of the downstream (non-linear) LP-filter can be increased with increasing numbers of branches, resulting in a more rapid gate re¬ sponse.

The attenuating circuit 13 incorporated in the apparatus il¬ lustrated in the apparatus illustrated in Figure 1 and shown in slightly more detail in Figure 4 comprises a series-coupling incorporating a delay circuit 132, having a delay time of from 0.1 ms to 10 ms, and an attenuating pad 133. Since the detector 14 is connected directly to the input of the apparatus illustrated in Figure 1, ^" the ^• apparatus incorporating the described attenuating pad 13 will have a negative reaction time (ignoring the fact that the signal is delayed). A negative reaction time means that the tone-start transient is not "clipped off" and that any tendency towards '^clicking-" is eliminated provided of course that the delay time is of sufficiently long duration,

The filter circuit 15 in the apparatus illustrated in Figure 1 and shown in more detail in Figure 5 comprises three parallel-connected branches of whichr

A first branch . includes a low pass filter 151 which has a linear characteristic in the case of an input signal of low frequency or amplitude and a high time-constant value, whereas in the case of an input signal of increasing fre¬ quency and/or amplitude said filter has a progressively decreasing time-constant value, and a resistor 152; this

means in practice that the output signal will essentially follow the curve form of the input signal, although small and simultaneous rapid variations in the input signal are filtered off while large and rapid variations, on the other hand, pass through the filter circuit; the resistor 152 (high value Rl) determines the inertia when the am¬ plitude of the input signal lies between the threshold value A2 and the level Al;

a second branch includes a detector 153 for detecting a signal having an amplitude which lies beneath the afore¬ said threshold value A2, and a resistor 154 (low value R2) which determines the shortest possible RELEASE-time; and

a third branch including a detector 155 for detecting an input signal having an amplitude which lies above the aforesaid determined level Al, and a resistor 156 (low value R3 ) which determines the shortest possible ATTACK- time. The following operative features are obtained with the aforedescribed filter circuit 15:

th first branch results in an output signal which is approximately proportional to the input signal in the case of slow time lapses;

the second branch causes the output signal to fall rap¬ idly to zero when the input signal is excessively low, and

the third branch causes the output signal to rise rapid¬ ly when the input signal is excessively high.

The low pass filter 151 incorporated in the filter cir- cuit illustrated in Figures 5 is shown in more detail in

Figure 6. The filter comprises a non-linear element 1511 in a series branch, and a capacitor 1512 in a parallel- branch. In turn, the non-linear element comprises a par¬ allel-coupling of a diode and a resistor and has a high impedance for signals of low amplitude and a low impedance for signals of high amplitude. The non-linear element 1511 may, alternatively, comprise a single zener diode instead of a parallel-connected diode and resistor.

The detector 153 incorporated in the filter circuit il¬ lustrated in Figure 5 is shown in more detail in Figure 7. The detector includes two series-connected compara¬ tors 1533 and 1535. The low input terminal (-) of the comparator 1533 is connected across a resistor 1531 to the input terminal of the filter circuit and across a re¬ sistor 1532 and a diode 1534 to the output of the compar¬ ator 1535, while the high input terminal (+) of the com¬ parator 1533 is connected to a point having the potential A2, i.e. the aforesaid threshold value, while the output of said comparator is connected to the low input terminal (-) of the comparator 1535. The output is also connected to the branch point between a resistor 1537 (R), con¬ nneecctteedd ttoo pplluuss ppootteeintial (V ), and a capacitor 1538 (C) connected to earth.

The relationship between the magnitude of the resistors 1531 and 1532 determines the hysteresis of the circuit. It is often desired to exclude the hysteresis, in which case the resistor 1531 is replaced with a short circuit, and the branch incorporating the resistor 1532 and the diode 1534 are removed. Hysteresis, however, reduces the risk of the noise gate repeatedly turning on and off. The risk of repeated activation and deactivation of the noise gate can also be reduced by ensuring that the volt-

age of the capacitor in Figure 5 (at the very bottom of the Figure to the right) falls to a sufficiently low value (when the detector illustrated in Figure 7 is acti¬ vated (so that a long period of time expires before it rises solely with the aid of the resistor Rl) . The input signal has often fallen to such a low level after this long period of time as to render the risk of repeated activation and deactivation minimal. This facility can be provided, for example, by connecting a very large re- sistance between the plus-voltage and the point at which the resistors Rl , R2, R3 and the capacitor in Figure 5 meet.

The high-input terminal of the comparator 1535 is con- nected to a point having the potential (l-l/e) ^* V , i.e. the potential to which the voltage of the capacitor 1538 rises after the time constant RC, while the output of the comparator is connected to the output of the detector.

The comparator 1533 compares the amplitude of an input signal with the threshold value A2. If the amplitude of the input signal lies beneath this value, the capacitor 1538 is charged via- the resistance 1537. After time RC the comparator 1535 will lower its output voltage. How- ever, should the amplitude of the input signal exceed the value A2 during the time period RC, the capacitor 1538 is rapidly discharged and the output of the detector retains the theorectically infinite impedance that it shall pos¬ sess when the amplitude of the input signal exceeds the threshold value A2. The time RC is conveniently made equal to the time period of the lowest expected frequency of an input signal which shall not result in alternating activation and deactivation of the circuit.

The detector 155 incorporated in the filter circuit il-

lustrated in the filter circuit illustrated in Figure 5 is shown in more detail in Figure 8. The detector in¬ cludes a comparator 1551, the low input terminal (-) of which is connected to a point having the potential Al, previously referred to as the "determined level", while the high input terminal (+) of the comparator 1551 is connected to the input 1550 of the apparatus, and the output of which comparator is connected across a resistor 1552 to the plus-potential and across a diode 1553 to the output 1554 of the detector. When the amplitude of the input signal exceeds Al, the output signal becomes high. In other respects the output impedance of the detector is theoretically infinite. The resistor 156 in Figure 5 is actually incorporated as the resistor 1552 in Figure 8.

Figure 9 illustrates a modification of the filter circuit 15 incorporated in the apparatus illustrated in Figure 1. This modified filter circuit -incorporates a plurality of series-connected non-linear low pass filters. A diode 90 and a capacitor 91 on the input side of the filter cir¬ cuit form a smoothing circuit which roughly smoothes the curve form of the input signal. This is followed by a buffer amplifier 93 (e.g-. an emitter follower). A down¬ stream non—linear low pass filter 94 has a series imped- ance Zl and a shunt capacitor C; the function of these components is to filter off rapid changes of low ampli¬ tude of the input signal; c.f. the low pass filter 151 in Figure 5. The filter is followed by a buffer amplifier 95. A down-stream non-linear filter 96 has a series im- ^* pedance Z2 and a shunt resistor R. The filter 96 is operative in compressing the working range of the appara¬ tus, i.e. the range within which the apparatus has neither maximum nor minimum attenuation, see Figure 11 which illustrates the output signal u_ as a function of the input signal i. This means that the signal from the

output of a down-stream low pass filter 97 can be changed very rapidly when the amplitude of the input signal to the filter 96 lies outside the working range of the fil¬ ter circuit 15. In other regards the low pass filter 97 assists in smoothing the remainder of the a.c. signal supplied to the whole circuit. A down-stream limiting circuit 98 allows only those voltages which lie within the working range of the apparatus to pass through, see Figure 12, which shows the input signal i as a function of the output signal u for the circuit 98, with maximum attenuation at 120 and minimum attenuation at 121.

The circuit 98 is not needed when the voltage-controlled attenuation circuit has the form of a, e.g., field effect transistor, since a field effect transistor is not dele- teriously effected when the voltage lies slightly outside the working range (point 120 to point 121). In actual fact the voltage in the case of a field effect transistor should lie slightly outside the working range at maximum attenuation and minimum attenuation, in order to achieve large maximum attenuation and distortion-free minimum attenuation.

The impedance Zl incorporated in the low pass filter 94 illustrated in Figure 9 is shown in detail in Figure 10. This impedance Zl includes in a series- branch two zener diodes connected in negative feedback 101, 102 and a re¬ sistor 103 which is connected in parallel with the diodes. When suitably dimensioned, the impedance Zl need comprise solely one zener diode, oriented as the zener diode 102 in Figure 10.

Figure 13 illustrates a conceivable variant of the input part 90-C shown in Figure 1. Extending from the input terminal 1300 is a series-branch which includes a diode

1301 and a zener diode 1303 and a resistor 1304 connected in parallel, and a second series-branch which includes a diode 1302, a buffer amplifier 1308 and a zener diode 1305. The branch node or point between the diode 1302 and the buffer amplifier 1308 is connected to earth, via a parallel coupling comprising a capacitor 1306 and a re¬ sistor 1307. The coupling junction between the zener di¬ odes 1303 and 1305 is connected to the output terminal 1301 of the input section and to earth via a parallel- coupling comprising a resistor 1309 and the capacitor C. The time constant of the parallel-coupling 1309-C may, for instance, be ten times longer than the time constant of the parallel-coupling 1307-1306. The diode 1303 cor¬ responds to the diode 101 in Figure 10, the diode 1305 corresponds to the diode 102 in Figure 10, and the resis¬ tor 1304 to the resistor 103 in Figure 10.

It can be said purely as a manner of illustration that "large" signals pass via the zener diode 1303 and rapidly charge the capacitor C; signals of average magnitude pass via the resistor 1304 and charge or discharge the capaci¬ tor after a long time constant,- small signals pass via the lower branch, and the zener diode 1305 discharges the capacitor C after a short time constant.

When comparing the operating methods of the apparatus il¬ lustrated in Figure 5 and Figure 13, it can be said by way of illustration that the uppermost branch in Figure 5 corresponds to the resistor branch 1304 in Figure 13; the intermediate branch in Figure 5 corresponds to the zener diode branch 1305 in Figure 13 and the lowermost branch in Figure 5 corresponds to the zener diode branch 1303 in Figure 13. The branch incorporating the zener diode 1305 in Figure 13 can be exchanged for the intermediate or centre branch in Figure 5 when an addi-

tionally short time constant t3 in Figure 2 is desired.

The apparatus illustrated in Figure 9 can be improved by _j connecting in parallel with the capacitor 91 a series- 5 branch which incorporates a zener diode 141 and a resis¬ tor 142, see Figure 14. The zener voltage of the zener diode shall lie slightly above the level Al of Figure 2. When the voltage of the capacitor 91 lies above the zener voltage, the time-constant dependency on the value of the 1.0 resistor 142 is shortened. This also shortens the so- called release-time without affecting the remaining time constants.

In the circuit shown in Figure 13 this series-branch can 15 be coupled in parallel with the capacitor 1306 in order to achieve the same result.

The branch incorporating the detector 155 in Figure 5 can be excluded when the uppermost branch in Figure' 5 is

20 arranged in accordance with the illustration of Figure 15. The zener diode 1570 fulfills the same function as the non-linear element 1511 in Figure 6, and the zener diode 157L fulfills roughly the same function as the comparator 1551 in Figure 8.» This circuit solution affords a

25 smoother transition between the time constants tl and t2 in Figure 2.