Post Go back to editing

Phase detection using passive or active mixer

Dear community,

I wish to make a phase detection of two sinus signals. I am a beginner in the field and seeking for some advices.

My setup would be the following:

- DDS (i think of using the AD5932) to deliver my reference signal. I want to have a frequency range variable but the limit of the AD8343 is already enough for my application,

- This signal goes thought my DUT. Since this one is tied to the ground, I am measuring the current flowing thought a reference and transposing it into voltage using the AD8220 instrumentation amplifier

I want to make a phase detection on the signal delivered by the DDS and the one measured at my DUT. I thought of using a mixer for it (the AD8343 or the LT5526) 

I thought of feeding my reference signal to the RF input and the received signal from the DUT to the LO input. But i read that i should be careful to the reflexion of the LO signal on the RF and the output. So this would be my first question which filtering should i use to avoid this effect?

the output signal will finally be feed to an ADC and processed with a Micro controller to extract the phase and amplitude change. 

My second question more theoretical probably, but what is the difference between a Passive mixer and an active one? Does it make a difference for sinusoidal signals? 

I thank you in advance for your help and your advices :)

Best regards,

Parents
  • Hello there,

    For relative amplitude and phase measurement application, you might take a look at the AD8302. This device has the mixer implemented as "an exclusive-OR style digital phase detector" (ref. datasheet page 15, near the bottom of the page).

    Either the Double-balanced Mixer or the EX-OR will function in similar way in this type application, and the phase measurement range is also similar, 0 to + or - 180 degrees. You may find the AD8302 more favorable compared to the DBM approach, because the phase response is quite linear over a broad phase measurement range. The linearity in phase response is because of the AD8302 internal limiters preceding the EX-OR. The same triangle shape response can be obtained with DBM but it means external limiters are necessary, whereas AD8302 already has such circuitry inside, as part of the log detectors used for relative magnitude measurement. 

    Notice the phase measurement range of DBM or EX-OR style of phase detectors has a phase measurement ambiguity issue, whenever this simple type of mixer or EX-OR arrangement is used. The ambiguity is seen on AD8302 datasheet, Figures TPC 25 through 29, in the triangle shape of the phase response. For best phase measurement accuracy, it's best to avoid operating at the top or bottom of the triangle shape response. This might mean adding a simple delay or phase lead or lag arrangement to one input, to shift the phase operating point.  If the whole 360 degree measurement range is needed, then other circuits must be looked at, or a workaround for the simple DBM / EX-OR circuit must be devised. A nearby MCU can be a big part of the workaround scheme if necessary. Circuits which give 360-degree measurement range are sequential logic phase detectors, and IQ demodulators, the latter of which usually require more extensive calibration for this purpose. There's probably other circuits also, maybe other readers will chime in on the subject. 

    Hope this helps.   -Bruce H. 

  • Hello Bruce,

    Thanks a lot for you answer! It is really appreciated :) One question though: i thought EX-OR method was more accurate for square signals and DBM for sinus signals? or i misunderstood and mixing it up with XOR detector?

    Also when you write:

    it's best to avoid operating at the top or bottom of the triangle shape response

    you mean for the future use of the signal?

    My idea was to extract the phase and magnetitude of the resulted signal and feeds it to my uP to make my measurement. For now i just want to extract this information and do not wish to use the signal for further implementation(for now at least), so maybe this is not so muchofan issue for now? or also by the reading of the VMAG and VPHS i should take care of this?

    Again thanks a lot for your answer and for your help Slight smile

Reply
  • Hello Bruce,

    Thanks a lot for you answer! It is really appreciated :) One question though: i thought EX-OR method was more accurate for square signals and DBM for sinus signals? or i misunderstood and mixing it up with XOR detector?

    Also when you write:

    it's best to avoid operating at the top or bottom of the triangle shape response

    you mean for the future use of the signal?

    My idea was to extract the phase and magnetitude of the resulted signal and feeds it to my uP to make my measurement. For now i just want to extract this information and do not wish to use the signal for further implementation(for now at least), so maybe this is not so muchofan issue for now? or also by the reading of the VMAG and VPHS i should take care of this?

    Again thanks a lot for your answer and for your help Slight smile

Children
  • Hi there,

    Perhaps I could answer your questions here:

    i thought EX-OR method was more accurate for square signals and DBM for sinus signals?

    True, EX-OR is a logic operation. But inside the AD8302, both RF inputs get limited, i.e. the RF inputs get amplified and sliced, provided the input signal is relatively strong enough to overcome the input noise. Limiting process is part of the cascaded log amp architecture. Some of the Analog Devices log detectors will buffer and output the limited signal (see for example AD641, AD8306, AD8309, and HMC813), but most other Analog Devices RF log detectors do not output the limited signal. 

    it's best to avoid operating at the top or bottom of the triangle shape response

    you mean for the future use of the signal?

    You notice in the datasheet, the phase response curves are relatively linear in terms of VPHS vs phase difference, UNTIL the phase is near an extreme, such as 0 degrees, or +/- 180 degrees. These are extremes to avoid, because internally the EX-OR output pulses become very narrow, and rise/fall times on the EX-OR internal to the device start to contribute increasingly to the total phase error. This is why datasheet graph TPC25 shows more error for higher input frequencies: 

    Hope this answers your questions.    -Bruce H.