Following envelopes with transistors

[jatalahd]

Since my auto-wah design (?topic=106988) was not a very successful one, I decided to convert it to a regular envelope controlled filter. Because I am a transistor fanatic, I wanted to make the envelope follower part without op-amps. I kind of succeeded, so just thought to share it here. The current version of the schematic is here (drawn up to the stage from where the control signal would be passed to the filter):


The envelope signal is constructed using CB, RBR2 and the B-E diode of Q1 and it wiggles at the base of Q1. This combination acts as a clamp circuit which produces a "one-sided signal" as shown here (red trace is the input signal, green trace is the "one-sided envelope"):


BJT Q1 is deliberately biased into saturation, but the gain should still be related to the value of RC, 47k in this case. So, the signal at the collector of Q1 is the amplified version of the varying envelope signal at the base of Q1. Please note, that the gain is not linear, so the waveform at the collector is only a representation of the envelope instead of an exact copy of it. The sensitivity control is not intended to be controlled by the gain of Q1, but instead JFET J1 at the input is a source follower, where the signal is attenuated to the appropriate level using RB1.

When Q2 is switched on by the varying envelope signal, it charges the capacitor CE, which holds the steady-state envelope with small ripple. Maximum voltage is determined by the operating voltage V1. The charging time is controlled by REL and the discharge time by RE. BJT Q3 is a buffer, which halves the steady-state envelope representation to max V1/2. BJT Q4 is used for creating an inverted envelope image. All the steady-state envelope signals are visualized here:


The inverted and non-inverted envelope representations can be, for example, used for driving JFETs having a suitable pinch-off voltage. If the mid-rail voltage (generated by the op-amp) is set at 4.5 V, then one should use a JFET with pinch-off voltage close to 4.5 V. Here I am using BF245B with Vp = -3.6 V so the mid-rail voltage should be lowered below 4 V. The idea is to use JFETs as voltage-controlled current sources to drive the LEDs of any resistive optocoupler. The filter connected to this circuit is intended to have its ground use the mid-rail voltage. That is why the input voltage has its ground at the mid-rail as well.

I created a video to demonstrate the envelope waveforms with the help of an oscilloscope: http://youtu.be/zuM-hs-3VCI

As a final note, of course it is more easier and more robust to design an envelope follower using op-amps. This just demonstrates one alternative method and different ways to use transistors in circuits.

[duck_arse]

v. nice, jatal, (parts of) might be what I've been looking for.