Attenuation is acieved by subtracting the output (current) of the LM13700 from the input signal. C1 provides DC blocking and should be large enough to allow effective gain reduction down to the lowest frequency of interest. R1 sets the input current to the LM13700 to be no larger than the half the diode bias curent (1mA). R2 (2 of) set the feedback current around the second op-amp such that the LM13700 can provide complete cancelation of the input signal with aprox. 1mA amplifier bias current. RV1 is trimmed for mimum feed through of the control voltage at the output. V1 is the negative-going retified audio signal. V2 is the threshold level, adjusted by RV4. R5 sets the maximum threshold level. RV2 adjustes the ratio, with R3 setting the maximum ratio (highest end-to-end side chain gain). The combination R3 and RV2 set a current I1 proportional to V2-V1 which is driven into the current mirror. I1 is mirrored through R5, generating a proportional (negative) voltage V3 which is buffered by an op-amp. The value of R5 also affects the side chain gain. D1 limits the complience of the current mirror to avoid saturation of the op-amp. C2 holds a voltage representing the current attenuation, 0v representing unity and going increasingly negative for greater attenuation. R6 and RV5 limit its discharge rate (the attack time), with R6 setting the minimum attack time and limiting the current into the op-amp to a safe level. C2 is charged (the release phase) by a current I2 supplied from a constant-current source. Note that this means the attack is logarithmic whilst release is linear. During the attack phase, ie when C2 is being discharged, the release current I2 is cut off by pulling the base of the current-source transistor to +12v. This cut-off state is initiated whenever V3 falls aproximately 250mV below V4. Within the subcircuit supplying release current I2; The combination of RV3 and R4 sets the (ultimate) release rate, C4 sets the hold-off time which must elapse between the end of the attack phase (ie a discharge of C2) and the start of the release phase (the application of the release current I2). During this time the charge on C2, and so the attenuation level, remains constant. C3 sets the rate at which the base of I2's source transistor is allowed to return to +6v, ie, the rate at which the release current is ramped up after the hold-off period elapses. The voltage on C2 is further R-C filtered and buffered by an op-amp follower. This voltage, V4, is converted to a current I3 by a transistor and R7, and driven into the LM13700 as amplifier bias. R7 also affects the side-chain gain, and is set to give a maximum current of aproximately 1mA. Note that there are two diode drops to be overcome between V3 dropping below 0v and current flowing at I3 (one at C2's discharge diode and one at the transistor sourceing I3). This implies an aditional increase in the threshold signal voltage, which will depend upon the ratio setting (RV5). V-out1 is negative-going rectified output for gate or VU applications. V-out2 is negative-going attenuation signal, dropping to about -9v for maximum attenuation, for gain-meter applications. I-in1 is a current input to the LM13700 for noise-gate and setup applications. For setup, apply an AC signal between 0 and -12v via at 10k resistor and trim RV1 for minimum breakthrough.