I'am using an AD9835 to generate a sine wave of 875kHz. The Clock of the DDS is 50MHz. This signal is amplified and travels through a sensor and is rectified and the resulting signal is filtered with a lowpass filter -3dB@10Hz, DC is subtracted and the result is gained 100x.
This results in a DC signal with a unwanted noise signal of 3.14Hz Vtt= 160mV. Tuning the DDS 875.001kHz and the noise signal is gone and system noise is 20mVtt. I guess the noise signal is caused by intermodulation.
Is there a way to predict which frequencies have large noise? I noticed all integer numbers of the 50MHz clock have a large noise amplitude.