**Q.**

I have been trying to null the LO Leakage of an IQ Modulator by adjusting the dc bias levels on the I and Q inputs. As I adjust these levels, I see no change in the LO Leakage. What am I doing wrong?

-----------------------------------------------------------------------------------------------------------------------

**A.**

You are getting the dc bias levels and dc offset levels mixed up. In almost all IQ modulators, the I and Q signals are differential and require a particular bias level, usually 500 mV. Conveniently, ADI DACs provide differential I and Q outputs that have a default bias level which is also equal to 500 mV. However, if you simultaneously adjust the bias level on the I and Ibar inputs, this will have absolutely no effect on the LO Leakage. In order to adjust the LO Leakage, you need to apply a *differential *offset voltage to the I and Q inputs. So in effect the I and Ibar inputs now have different bias levels (typical required offset adjustment voltages are in the low mV range)

Let's consider and example. Lets assume that you have a differential peak-to-peak voltage swing on the I and Q inputs of 500 mVpp. This means that the individual I and Ibar inputs each have a peak-to-peak swing of 250 mV. If the signals are biased to 500 mV, then the I and Ibar inputs will each be swinging between 375 mV to 625 mV with respect to ground.

To null the LO Leakage, you need to add a differential dc offset voltage between I and Ibar (and Q and Qbar). So, for example, you if you add a +2 mV offset voltage, the bias level of I will change from 500 mV to 501 mV and the bias level of Ibar will reduce from 500 mV to 499. Simultaneously adjusting the bias level of I and Ibar by 2 mV will have no effect.

Generally these differential offsets are applied in the digital domain or in the DAC. Some DACs such as AD9779, AD9122 and AD9788 have dedicated offset adjustment registers which can be used to null LO Leakage.