Post Go back to editing

Severe Harmonic distortion on Mixer (HMC951B) I,Q. Waveform gen. using ADF4159

Hello all,

We have implemented a FMCW radar, using ADF4159 and HMC951B and other ADI parts.

In our software setting, we tried both continue sawtooth and continue triangle ramping, frequency from 5.5 GHz to 7.5 GHz and time per ramp is 10 ms or 25 ms, respectively. 

First, in order to simplify the test case, we used cable to create time delay from Tx to Rx(to simulate a reflection situation ), ramp mode is "continue sawtooth" and time per ramp is 25 ms. we have found harmonic in full spectrum:


Then We opened Tx and Rx (without transmit and receive) and measured IQ from mixer. A periodic harmonic was found which the period is same as time per ramp. Any idea what causes the harmonic 

Time per ramp : 10 ms

Time per ramps : 25 ms

We also used spectrogram to measure Tx signal,  there are some interference appeared instantaneously at different frequency when ramp frequency from 7.5 GHz back to 5.5 GHz.

  • Inquiring radar expertise. Please allow some time to answer.

  • The following is example taken with our 24GHz FMCW radar chipset board EV-RADAR-MMIC2 containing the ADF5901 (2 channel Tx) , ADF5904 (4 channel Rx) and ADF4159 (Ramping PLL).

    Setup is below

    The ADF4159 is configured to generate 1ms 200MHz Triangular ramp 

    The delay cable has an effective length of 5.7m

    Based on the the beat frequency equation

    The beat frequency should be ~7.6kHz

    The resultant base band spectrum on Spectrum Analyser is below

    This is similar to what you see, spectrum with harmonics at the ramp rate (triangular ramp period 2ms, => 500Hz, 1kHz etc. tones)

    Taking the same signal and inputting it into oscilloscope the result is below

    Top plot is the time domain base band signal (yellow) and the ramp complete signal from the ADF4159 (blue) 

    The bottom plot is using MATH function to do FFT on the base band signal, the FFT is gated to be just done on a single ramp segment (the grey area on the time domain plot). The result is a single tone at 7.6kHz as calculated above for the setup.

    So for correct FMCW baseband analysis the FFT must be done on each ramp segment individually which is what the  oscilloscope is doing, the Spectrum Analyser is capturing the entire waveform and the dominant signal is the ramp transitions at the ramp rate.