FAQ: How can I get the rms detector to accurately detect the power during a pulse modulated signal input?

Document created by CarlosC Employee on Nov 2, 2010Last modified by AndyR on Jan 31, 2012
Version 2Show Document
  • View in full screen mode


How can I get the rms detector to accurately detect the power during a pulse modulated signal input?




The "mean" portion of the rms detector is determined by the filter capacitor (CFLT).  A large capacitor gives the highest level of averaging and provides the most accurate measurement.  However, this comes at the cost of slower response times.  If the capacitor value is reduced, then a faster response time is possible, but the detector may be more susceptible to complex waveform variations.  It is a trade off.

The pulse response time of the ADL5501 with no external filter capacitors (CFLT or COUT) is about 6 us.  This is the fastest that the device will respond.  If the pulsed signal is faster than this, then the output voltage will not have time to respond before the signal is gone. If on the other hand, the pulsed signal is present for longer (say 20 us pulse width or longer), then the ADL5501 can readily keep up (see figure 30).  More capacitance (for more averaging) can be added, but the pulse response will also increase (see figure 32).  30.png32Untitled.png


If fast speed is a necessity and changing modulation is not of concern, then I would suggest looking at the log amps as they are much faster devices (e.g. AD8318, AD8319, or AD8317).

Note that this method does NOT measure the average power of both the ON and OFF states together.  If that is the desired measurement, then the averaging time constant must be increased.  Any external filtering (CFLT) added to a detector works in parallel with the following on-chip filter networks. Even though these devices have a two pole filter, it can be roughly expressed as the single dominant pole. The corresponding corner frequencies are good for estimating the detector response to low frequency content in the modulation. The settling time response may not be accurately estimated from these single pole approximations.