0
\$\begingroup\$

I am not experienced with log amps, and am trying to learn about them.

Every log amp article has the base of the logging transistor grounded.

schematic

simulate this circuit – Schematic created using CircuitLab

Why not have the transistor as set up as common collector or common emitter?

schematic

simulate this circuit

or

schematic

simulate this circuit

\$\endgroup\$
3
\$\begingroup\$

The common collector approach has the issue that the base current flows into the input of the opamp through the transistor and causes an error that depends upon the hfe of the transistor. Also as shown it will not function - you need to put the collector to a negative voltage.

The common emitter design can be used as in the AD8304 (Datasheet).

The reference voltage to the positive input of the opamp cannot be at zero volts though. In the AD8304 it is raised to about 0.5V to allow operation of the logging transistor. If the input current is from a photodetector this is usually not an issue.

As George points out in his answer a diode connected transistor can also be used fairly successfully. They work much better than ordinary diodes and avoid the excessive gain within the feedback loop of the normal common base approach.

This large gain in the loop creates difficulties in maintaining stability that in turn causes the settling time to be extremely slow at low light levels (I had one design that would work down to -90dBm but take more than a minute to settle).

\$\endgroup\$
  • \$\begingroup\$ Didn't see the AD8304, and thanks for the clear explanation about the issue with CC. Would it be too much of a generalization to say that in general emitter currents are less dependable than collector currents because they depend on Vbe and hfe and collector currents only depend on Vbe? \$\endgroup\$ – DavidG25 Oct 18 '17 at 20:14
  • \$\begingroup\$ @DavidG25 - that sounds pretty much correct. \$\endgroup\$ – Kevin White Oct 18 '17 at 22:20
2
\$\begingroup\$

I've never made a log amp. But it works on the ideal diode equation. I thought I should add the ideal diode equation, from wiki

I_c ~ e(V_d/V_T)-1,

where V_T is the thermal voltage* kT/e (e here is the charge of the electron)

If you look at the circuit you'll notice that the collector is a virtual ground, so the grounded base looks almost like a diode connected transistor (b-c shorted.) Now I have played with diode connected transistors and they are very close to ideal over many orders of magnitude in current. Much more so than a normal diode. So my answer (which is part guess) is that it makes for a more ideal diode.

*if you're making a log amp you should be thinking hard about the temperature of the diode/transistor in the FB loop.

\$\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.