The circuit below shows an active low pass filter that is universally applicable. Only 4 equal (!) resistors needs to be changed to change the filter corner frequency fc. It uses only standard parts, is suitable for a large supply voltage range and needs only a modest 2mA of supply current.
R2, R3 and R4 set the input bias voltage at half the supply voltage. R5-R8 and C3-C6 are the low pass filter network. C3, C4, C5 are bootstrapped to provide a sharper roll-off. Q1 and Q2 are a series-feedback pair and function as a ‘super’ emitter-follower. A standard emitter follower has a gain of ≈ 0.90 … 0.95x. This ‘super’ emitter follower has a gain of ≈ 0.99x. Just like an OpAmp with feedback, in this series feedback amplifier, the feedback increases the input impedance, lowers the output impedance, stabilizes gain and lowers distortion. Not bad for a single €0.10 transistor that is added!
An OpAmp has a relatively good rejection of noise on the power supply rails (Power Supply Rejection Ratio). This feedback pair is not so good at that. For noise coming from the power rails. Q2 can be thought of as being a grounded base amplifier. To remedy this, the power rail and the bias network are decoupled. The voltage divider R2 and R3 are relatively high impedance and are decoupled with the large capacitor C2, The time-constant of R2//R3 * C2 is 5 sec, which might be a bit too long but it provides good attenuation of noise from the power rails. The incoming power is decoupled with R12 and C8. The input voltage can be anything between 8V … 24V. The power consumption is circa 2mA at 9V, rising to ≈5mA @ 24V. (Be aware of the working voltage of capacitors C2 and C7). The low-end corner frequency is set by C1 and R4 and is circa 35 Hz.
The formula to change the cut-off frequency fc is :
And here are the measurement results! As you can see the lower fc is indeed around 35Hz. The passband gain is (very close) to 0 dB. The upper -3 dB point is around 3500 Hz. It might seem the filter slope is gradual but don’t be fooled! The vertical axis is displaying over 100 dB….
And the obligatory PCB photo :
Do NOT use this filter if:
– you want the steepest roll-off. Use a Chebyshev instead or use a LC elliptical filter (see 40m DC receiver (2) for an example
– you want a high output drive. The quiescent current through Q2 is only 2 mA
very interesting Q1/Q2 combination
Hi Jan,
Q1 and Q2 form a compound emittor follower / unity gain amplifier. It offers a better performance than a simple emittor follower with Q1 alone. Of course, it could be replaced with a low-noise operational amplifier.
Kind regards,
Cor / pa3cor