In Mixers

Fairly often we receive questions (generally from grad students) about how to use the mixer as a phase detector. The basic idea is that the output of a mixer will inherit the phase from both the LO and the RF (or IF) signals. That is to say that the output signal will inherit the phase difference between the LO and input. This is why the quad hybrid on an image reject or single sideband mixer can be put on either the LO or RF port (a topic for another post). In a downconversion where the LO and RF signals are the same the output will be at DC, where the ‘phase’ translates into an amplitude. When the two signals are in phase with each other the voltage will be at a positive maximum, when they are 180° out of phase the voltage will be at a negative maximum, and when they are either 90° or 270° phased to each other the output will be purely imaginary, and the voltage will be (ideally) zero.

The wrench in the works is that the mixer is not perfect, and imperfections in the amplitude and phase balance of the baluns or mismatch in the diodes will lead to an inherent DC offset voltage even when the signals are totally in phase. This is referred to as the ‘DC offset voltage’ of the mixer. It is typically measured by terminating the RF port with a matched load and measuring the DC voltage at the IF port with a standard power LO input.

Typically it is recommended that the measurement be performed with a low pass filter at the IF port to eliminate the harmonics. The problem with this is that the mixer is designed for operation in a 50Ω environment, and a reflective low pass filter will present a very non-50Ω termination on the RF port, deteriorating the isolation that is provided by the mixer. Therefore Marki Microwave recommends that the DC offset measurement be performed with a good 50Ω termination at high frequencies. This can be provided with either an absorptive filter (such as the wavefade filters) or with a diplexer or bias tee, with the high frequency port terminated in a 50Ω load. We recommend that all phase measurements be performed in this manner, with the mixer matched at all ports.

As an example, below is the measured DC offset for our new ML1-0113L mixer using either a low pass filter or a terminated bias tee.

DC Offset with different terminations

As you can see the 50 ohm termination makes a significant difference in measured voltage. As a point of reference, and to demonstrate how awesome the new microlithic mixers are, below are comparisons of the new mixers with other high performance mixers in the Marki line. ML1-0113L vs M1-0212L DC Offset ML1-0220L vs M4-0220L DC Offset

 The superior balance of the microlithic ML1-0113 and ML1-0220 make them ideal  choices for phase detection applications. Other phase applications coming soon…

Recommended Posts
Showing 8 comments
  • Sabina
    Reply

    Hey guys, thanks for the helpful post. Wow… those microlithic mixers show tremendous improvement in DC offset, I can’t believe it. Do you see that kind of improvement in other areas? like spurs and isolations?

    • Doug
      Reply

      Sabina – yes the same balance that improves the DC offsets also improves the isolations and MOST spurs. The even by even, odd by even, and even by odd spurs are all canceled better in these mixers. The odd by odd products are never canceled, since that would eliminate the 1×1!

      • Mike
        Reply

        I have a 30-31 GHz mixer phase detector application. You mention the ML1-0220 is good. How good is the DC offset for the ML1-0832SM mixer as a phase detector when tested the same way?

        • Doug Jorgesen
          Reply

          Weirdly not as good as I expected. I tested a variety of mixers in this band, and the ML1-0832 was the highest DC offset with around 10 mV on some mixers and some ports. The legacy mixers I tested were in the 4-5 mV range, and the other Microlithics a little below that. The best one I tested was the ML1-0732, which had a DC offset of under 1 mV into the RF and under 3 mV into the LO with a 0 dBm input and the L diode. I’m not sure what role the surface mount transition plays in this, though, so if that is something that you need then it might be worth going with it anyways.

  • Sabina Belov
    Reply

    Wow, Doug! Thank you for the reply… very helpful. Cheers!

  • Steve Baker
    Reply

    I’m curious if for my application it makes any sense to use a mixer as a phase detector. I’m trying to detect motion on order +/- 0.1 mm in a duration of 25 ms with a 24 GHz RADAR with range approximately 1-m. Others have detected this using a VNA w/o issue, looking at phase difference between Tx and Rx signals.

    I’m concerned with these very small Doppler shifts that we may have substantial errors due to the DC offset, even with the
    Would a solution using a mixer-based phase detector have sufficient SNR and resolution. Is it reasonable to calibrate and correct for the DC offset? Would you try instead an I-Q solution for detecting phase? (If so, what Marki product would you use?)

    Thanks!

    • Doug Jorgesen
      Reply

      Great question Steve. The wavelength in free space of 24 GHz is 12.5 mm, so 0.1 mm would be about 3 degrees, or 0.025 radians. Sin of 3 degrees is about 0.05, and if you have two inputs of about 0 dBm you could get an output of at least 100 mV (I would expect), so you would be looking for a signal of +/- 5 mV. This sounds pretty detectable to me, depending on the stability of your experimental setup. If you look at the plots, this is below the ML1-0220/ML1-0113 phase offsets. So I would start with something like the ML1-0536LS.

      Now since your signal is periodic, you might be able to filter out the DC offset. You would need a DC block that goes down to 40 Hz, which means a blocking cap of about 80 uF.

      Good luck.

pingbacks / trackbacks

Leave a Comment

Start typing and press Enter to search