Amplifying DC Diode Detector Output

Description: The small DC voltages at the output of the diode detector are challenging to amplify accurately due to DC offsets introduced by any operational amplifier (op-amp) and low-frequency noise (1/f noise). A solution to these issues is to chop the DC signal using MOSFET switches. In the circuit described, a chopper integrated circuit (LTC1043) alternates between two square-law detectors: one produces a positive voltage while the other generates a negative voltage. The capacitor (Cosc) sets the chopping frequency to approximately 90 Hz, which should be kept relatively low, avoiding 60 Hz and 120 Hz. The op-amp following the diode chopper has a gain of 181, amplifying the 90 Hz square wave generated by the chopper. Any offset introduced by the op-amp is removed by a de-chopping switch at the output. An RC network (R3, C3) acts as a low-pass filter, establishing a bandwidth of about 10 Hz, allowing for a reasonably fast response while minimizing noise. A second op-amp functions as a conventional DC amplifier, increasing the voltage to drive an external meter. The gain can be adjusted (1 + R6/R7) to provide a suitable output range for the meter. The input offset voltage of this op-amp is amplified by its gain and appears at the output, which can be significant if the gain is high. The sensitivity limit of this circuit is primarily affected by the thermal noise of the 50K resistance of the rectifying germanium detector diode, which adds about 10 nV/√Hz of noise. Within the 10 Hz bandwidth, the thermal noise from the 50K detector resistance is approximately 0.1 µV. At -60 dBm input RF voltage, the detector diode generates about 0.1 µV DC voltage, making inputs lower than -60 dBm indistinguishable from noise. A narrower filter (<10 Hz) could improve sensitivity but would slow down output settling time. For accurate RF noise measurement, additional filtering is necessary. While this sensitivity may seem inadequate compared to modern radio receivers that can detect RF signals significantly smaller than one microvolt, it highlights the limitation of square-law detection. To detect signals smaller than -60 dBm, an in-line linear amplifier should be added before the square-law detector, although this may introduce limitations such as restricted bandwidth and poorly defined input resistance. The square-law detector discussed here features a wide RF bandwidth, low input standing wave ratio (SWR), and stable DC gain, with a -3 dB bandwidth exceeding 500 MHz. Some op-amps integrate the chopper internally, and while some external storage capacitors are needed, chopping is otherwise seamless for the user. The TLC2652 is an example of a chopper-stabilized op-amp with an input offset voltage of 1-3 microvolts, drifting only 0.03 µV/°C, though it exhibits higher noise levels, with input noise equivalent to one microvolt within the same 10 Hz bandwidth as the LTC1043. While simpler, it does not achieve the same ultimate sensitivity. Its single-ended input is suitable for use with a single square-law detector diode. Another approach to reduce noise and offset involves AM modulating the RF source, resulting in an AC signal output from the detector diode that matches the modulating frequency. A low-noise op-amp can amplify this small signal to drive a synchronous demodulator, similar to the previously described chopper switch. The averaged output from the chopper provides a DC voltage proportional to the RF input amplitude, effectively rejecting DC offsets from the device under test and the amplifier. If the RC product of the averaging filter at the output of the chopper is long, it can achieve a very narrow equivalent noise bandwidth.

The circuit described effectively addresses the challenges of amplifying small DC voltages generated by a diode detector. The use of a chopper integrated circuit (LTC1043) allows for the mitigation of DC offsets and low-frequency noise by alternating between two square-law detectors, thus improving the accuracy of the output signal. The design includes a low-pass RC filter to ensure that high-frequency noise is filtered out while maintaining a reasonable response time for the meter. The incorporation of a second op-amp for DC amplification further enhances the output voltage, making it suitable for external measurement devices.

The sensitivity of the circuit is critically determined by the thermal noise associated with the detector diode and the op-amp, which can limit the ability to detect weaker RF signals. To achieve better sensitivity for lower RF inputs, the addition of a linear amplifier before the square-law detector is recommended, despite its potential drawbacks. The overall design showcases a balance between sensitivity, bandwidth, and noise performance, making it a robust solution for applications requiring precise measurement of small DC voltages derived from RF signals.The very small DC voltages at the output of the diode detector are difficult to amplify accurately. DC offsets introduced by any op-amp are the main culprit. Low frequency noise (1/f noise) is another. Chopping the DC with MOSfet switches is an excellent way to get around these problems. In the circuit below, a chopper integrated circuit ( LTC1043 ) switches between two square-law detectors: one detector gives a positive-going voltage and the other gives a negative voltage. Cosc sets the (square wave) chopping frequency to about 90 Hz - you want to keep this chopping frequency fairly low, but well away from 60Hz and 120Hz.

The op-amp that follows the diode chopper has a gain of 181, and amplifies the 90 Hz. square wave that the chopper provides. Any offset that the op-amp adds is eliminated by the de-chopping switch at its output. The RC network (R3, C3) at the output is a low-pass filter, and sets the bandwidth at about 10Hz. This allows reasonably fast meter response, while eliminating noise. A second op-amp acts as a conventional DC amplifier, and boosts voltage to drive an external meter. Gain can be adjusted (1+ R6/R7) to provide a reasonable output range for your meter. This op-amp`s input offset voltage is amplified by the op-amp`s gain, and appears at the output. If gain is high, this can be a significant offset from true zero. What`s the sensitivity limit of this circuit The primary source of noise is the warm 50K resistance of the rectifying germanium detector diode. The op-amp shown adds about 10nV/rt(Hz) as well. Within the 10 Hz. bandwidth, thermal noise of the 50K detector resistance is about 0. 1uV. At -60 dBm input RF voltage, the detector diode generates about 0. 1uV DC voltage, so inputs less than -60dBm are lost in noise. A narrower filter (<10Hz) would help, but would require patience while the output settles to a new reading. When measuring RF noise, more filtering is required to determine its amplitude with any accuracy. You might consider this sensitivity to be terribly poor. A modern radio receiver can detect RF signals much smaller than one microvolt across its 50 ohm input resistance, perhaps in the -160dBm ballpark.

This points out the major limitation of square-law detection: if you want a more sensitive detector, amplify first. If you need to detect signals smaller than -60dBm, your only choice is to add an in-line linear amplifier ahead of the square-law detector.

The disadvantage of doing this is that the amplifier may have limited bandwidth, poorly defined input resistance, and its gain may be ill defined. The square-law detector described here has very wide RF bandwidth, low input SWR, and all gain is well defined, stable DC gain.

The -3dB bandwidth point is greater than 500MHz. Some op-amps incorporate the chopper inside. Apart from the external storage capacitors, chopping is transparent to the user. The TLC2652 is a chopper-stabilized op-amp that sports an input offset voltage of 1-3 microvolt. This offset drifts by only 0. 03uV/degree(C). However, it is a bit noisy. In the same bandwidth (10 Hz) as the LTC1043 chopper above, input noise equivalent is one microvolt. While much simpler, ultimate sensitivity is not as good. Its single-ended input is appropriate for use with a single square-law detector diode. Another noise (and offset) reduction technique involves AM modulating the RF source. The output of the detector diode is in this case an AC signal whose frequency is the same as the modulating signal. A low-noise op-amp can amplify this small signal to drive a synchronous demodulator (very similar to the chopper switch described above).

The averaged output from the chopper contains a DC voltage proportional to the amplitude of the RF input. DC offsets of the device-under-test, and of the amplifier are rejected. If the RC product of the averaging filter at the output of the chopper is long, very narrow equivalent noise bandwidt