1
\$\begingroup\$

In High school and University courses, I have only seen active filters made around OpAmps. I'd like to know something about active filters built with BJT amplifiers.

Example of BJT filter compared to OpAmp version

Figure: Example of BJT filter compared to OpAmp version

I can find lots of information in books I have about OpAmp filters, but nothing about BJT filters. I searched a lot also on the Internet, but I found very little information.

My question is how to design filters with BJTs and what are the advantages/disadvantages versus OpAmp circuits (costs, performances, noise...)?

Moreover, can you please suggest me a good book or some lectures to better understand these kind of filters?

\$\endgroup\$
  • 2
    \$\begingroup\$ Unless you are doing custom RF filters >> 1MHz why would you want a less accurate filter with low gain and thermal drift? Quad Op Amps are only 0.25$ in single supply with <50mV near Rail2Rail Low input offset voltage: ±1 mV • Rail-to-rail output • Unity-gain bandwidth: 1 MHz • Low broadband noise: 30 nV/√Hz • Low input bias current: 10 pA • Low quiescent current: 70 µA/Ch • Unity-gain stable • Internal RFI and EMI filter \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Jan 29 at 23:09
  • 2
    \$\begingroup\$ Alessio, the only place where I've seen some discussion of using BJTs in place of opamps is in Don Lancaster's "Active Filter Cookbook." It's an old book and some terms used there will need to be "interpreted" into more modern contexts (things have changed over the years.) But it covers these. Opamps today are so much more common, cheap, and don't have some of the liabilities of early opamps. So it's almost always better to use an opamp today. However, a BJT is dirt cheap, has only three pins instead of at least 5, and can be used effectively. Read "old stuff" if you are serious about this. \$\endgroup\$ – jonk Jan 29 at 23:14
  • \$\begingroup\$ Alessio, I should correct myself -- a little. Sallen & Key in their TR-50 paper from 1954 (developed from an MIT contract from the US Air Force) discusses circuits using vacuum tubes. To a lesser degree, so does their 1955 paper published in IRE Transactions -- Circuit Theory. Their use of cathode followers is very close to what you want, as well. \$\endgroup\$ – jonk Jan 30 at 0:38
1
\$\begingroup\$

a bipolar running at 10mA will have Rout (Zout!) of 2.6 ohms; and you can run this circuit on 1v or 1.5 volts, unlike an opamp. Thus the bipolar STOPBAND will be far superior to the opamp.

Note for each of your schematics, the C2 becomes a shunting path for high frequencies, and the only filtering remains the C1 (thus one-pole passive becomes the behavior).

The 0.6 volt offset can be reduced, with a NPN/PNP Darlington.

The inherent noise of a bipolar is set by the rbb', which is the non-useful resistance of the bulk silicon material between base-contact node and base-collector region; you should easily have this less than 1,000 ohms, or 4 nanoVolt/rtHz noise density. In 1MHz bandwidth, you'll have 4 microVolts RMS (scaled up by pi/2 or so, for rolloff) of total integrated random (and white) noise.

Thus with 1 volt PP input, and 4 microVolts RMS noise, you can expect about 100dB SNR.

Do be aware of Finite Power Supply Rejection, likely set by the Early Voltage.

| improve this answer | |
\$\endgroup\$
0
\$\begingroup\$

The easiest approach is to take filters designed for op-amps configured at unity gain, (as in your second schematic), and apply them directly to BJTs in emitter follower configuration (as in your first schematic).

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ ...but you should know that the exactness of the desired filter response will be not as good as the opamp based realization - due to a larger output resistance and a smaller input resistance (if compared with an opamp). Howevr, a simulation program can be used to correct the filter response... \$\endgroup\$ – LvW Jan 30 at 16:06
0
\$\begingroup\$

In order to answer this question, we first have to get an overview of the different principles for realizing active filters:

A) Filter stages

1) with finite and fixed gain

2) with (nearly) infinite amplification

3) with impedance converters (GIC)

B) Active component simulation of passive filter structures (GIC technology, integrator stages).

If one knows the properties of the bipolar transistor (BJT), it becomes immediately clear that actually only the realization A1 is promising:

Sallen-Key structures with a finite and fixed gain value, whereby there are circuits for both positive and negative gain values.

Remark: In 1955 R.P.Sallen and E.L.Key published a fundamental paper in IRE-CT2 in which bipolar transistors were used for realizing active filter topologies.

| improve this answer | |
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.