is there a simple solution for an Cartesian Feed-Back Loop?
Or is there a easy (or better) way to use the envelope Feedback, for example with ADL5511?
The task is linearisation for TETRA-Systems.
Yes there is a simple solution. Actually simplicity is in the eye of the beholder, so let's just say there is an all-analog solution and it uses the AD8302, which is a gain and phase detector. Still, it is easier than digital predistortion.
First, start with the premise that a linearized PA will have a constant gain and a constant phase shift -- you can think of the phase shift as the delay through the PA rather than a phase shift caused by a low pass filter.
The easiest way to get constant gain is to wrap an AGC loop around the PA. There is usually a variable-gain amplifier (VGA) such as the ADI AD5330 in front of the PA already to set the output power, so you include the VGA in the gain-control loop and you measure the overall gain from the input of the VGA to the output of the PA. This is the overall gain you want to hold constant. In practice, you'll set the VGA's gain to achieve some specified output power level and then sum in the voltage from the gain-control loop for linearization.
To keep the phase constant, you would add a voltage-controlled phase shifter to the input of the VGA and control it with a phased-based AGC loop.
Figure 5 in this article shows this approach:
Because a phase-shifter may be difficult to implement, you could use just an AD8302-based gain-control loop by itself for some improvement.
Note that a typical transmit AGC loop control the output power, NOT the gain of the PA. The AD8302 meaures both the gain and phase shift in the PA and the circuits shown in Figure 5 in the above reference control the gain and phase in the PA circuit, not the PA's output power.
One other design note: You set the time constants for the control loops based on the modulation envelope, not the carrier frequency. For example, if you were linearizing an AM transmitter (double sideband with carrier) and the lowest modulation frequency was 100 MHz, you'd want a timeconstant of 10 Hz or less in the feedbacl loop to avoid "pumping" the carrier.
Read the AD8302 data sheet carefully: The AD8302's internal filters are set up for RF, not a modulation envelope, so you may be adding external filter capacitors.
thanks for your comments
I discussed the topics with the designer - in the next design loop, your proposal will be considered.
Retrieving data ...