Embedded Files
Created: Dec. 2023
Introduction
Introduction
Detectors used in power measurement and radio astronomy require temperature invariant output. However, the basic diode detector suffers from a temperature drift that is especially severe at low input power (fig. 1). To quantify the severity of the drift: At -20 dBm input power, the output voltage changes ∆Vo ≈ 10mV over the temperature range -25C ~ +75C (fig. 1).
We will limit our discussion to large-signal detectors which are intended for input power higher than -20 dBm, because the small signal variant has an opposite thermal characteristic.
Fig. 1: At -20 dBm input power, the output voltage changes ∆Vo ≈ 10mV over the temperature range -25C ~ +75C
There are dozens of solutions to the temperature drift problem, and here're two simple ones.
Solution 1
Solution 1
Use a shunt diode as a voltage divider, aka. Ericsson-Waugh detector (fig. 2). At -20 dBm input power, the variation in output voltage is beneficially reduced to ∆Vo ≈ 1mV over the temperature range -25C ~ +85C Disadvantage: this detector requires a very high resistance load >= 10 MΩ, e.g. digital multimeter.
Link: AN1328, https://docplayer.net/30617697-Application-note-1328.html
Fig. 2: Adding a shunt diode can beneficially reduce the drift to ~1 mV over the temperature range -25C ~ +85C for a -20 dBm input signal
Solution 2
Solution 2
Bias the diode at 500 uA (fig. 3). This bias current is chosen as a compromise between temperature drift and sensitivity. For a 3V supply, the resistances RL are ~5kΩ. At -20 dBm input power, the output voltage changes ∆Vo ≈ 1mV over the temperature range -25C ~ +75C
Disadvantage: The bias creates a quiescent voltage which requires removal by using a combination of op-amp & reference diode.
Link: "Large signal detectors for cellular handsets and base stations" (PDF embedded below)
Fig. 3: Another way to reduce drift is to bias the diode at 500uA. ∆Vo ≈ 1mV over the temperature range -25C ~ +75C for a -20 dBm input signal
d08_large sig det--temp, diff error.pdfSolution 3
Solution 3
Add an NTC thermistor as part of the output load; i.e. 10 kΩ thermistor // 470 kΩ load. This is arguably the simplest temperature compensation scheme but a variation of 3mV ~ 7mV still exists at 0 dBm over the temperature range -30C ~ +70C. So, the drift should presumably be even worse at -20 dBm. Maybe the temperature compensation is marginal because the values of the thermistor and fixed load are not optimum.
Link: Le Polozec, X. & Muffat-Joly, S., "A simple Device for Decreasing Thermal Dependency of a Diode Detector", May 2015, https://www.researchgate.net/publication/277017879_A_simple_Device_for_Decreasing_Thermal_Dependency_of_a_Diode_Detector
Fig. 4: Alternatively, connecting a thermistor in parallel with the load resistor can also minimize drift
Page updated
Google Sites
Report abuse