Differential matching of AD5356 dual mixer on CN-0140

Hi all,

I am designing a dual IF receiver and studying the Circuit From The Lab 0140 which is indeed very interesting for me. Maybe someone could explain how the two matching networks between Mixer/SAW filter/VGA have been calculated as I cannot get such values with the Analog Devices RF impedance matching calculator nor with mine.

The network between Mixer/SAW filter is a high-pass (right?) but according to my calculations those values are beyond the freq of interest (which is 153.6MHz).

Could someone shed some light on that?



  • 0
    •  Analog Employees 
    on Sep 9, 2011 1:01 AM

    Let me speak to each matching circuit (or filter) separately to avoid confusion.

    The circuit between the ADL5356 (mixer) and the AD8376 (VGA) aims to provide impedance match from the mixer to the SAW filter to the VGA.  The ADL5336 has an output impedance of 200 Ohms and needs to be matched to the SAW filter that has a natural matched impedance of 100 Ohms (the impedance is actually complex, but for the sake of simplicity, a simple real value of 100 Ohms is provided) .  This is done by transforming down from 200 Ohms to 100 Ohms through a series C and shunt L configuration (18 pF and 72 nH, respectively).  The output of the SAW looks to step up from 100 Ohms to 150 Ohms through a similar (yet reversed) shunt L and series C configuration (58 nH and 18 pF, respectively).   The suppliers of the SAW filter generally give recommended matching networks that can be tuned to account for parasitics.

    The circuit between the AD8376 (VGA) and the AD9258 (ADC) looks to provide 1) the optimal 150 Ohm load impedance for the VGA and 2) anti-aliasing filtering (AAF) before the ADC input.  Let me break up the purpose of each stage for clarity.  The 309 Ohm at the VGA output and the two 165 Ohm resistors at the ADC input combine to provide a 150 Ohm load for the VGA.  The shunt 1 uH inductors provide dc biasing for the VGA output stage and the 470 pF capacitors provide the necessary ac-coupling (and neither significantly affect the match or the filtering).  A fourth order low pass filter - the two stages of series Ls and shunt Cs (330 nH, 3.3 pF, 330 nH, and 20 pF) - is used to start to shape the AAF.  The band pass shape of the AAF is determined by the tank circuit - the shunt 56 nH and the last C stage of the low pass filter, i.e. the 20 pF shunt capacitor.  The process to design an AAF like this one is described in detail in the application note below:

    AN-1098: Methodology for Narrow-Band Interface Design Between High Performance Differential Driver Amplifiers and ADCs


    The sort of matching done between the mixer, saw filter, and VGA and the VGA’s AAF design are easier when aided by a software package that models and accounts for parasitics.  A simple Smith chart program or filter design tool may be helpful for a first pass, but fine tuning is most easily done in Agilent’s ADS or something similar to minimize empirical adjustment cycle time.

  • Dear CarlosC,

    Thanks a lot for the guide on the circuit, it is very detailed and I appreciate your effort. Naturally, I read a similar description on the Circuit Note itself.

    If I understand correctly, the RF impedance matching calculator offered by Analog Devices can only offer values to be considered as guidelines... nevertheless I could not manage to have something similar to the CN. Would you be so kind to post a screenshot of how you would use (and the result it would give) the tool for the maching between Mixer and SAW, and between SAW and VGA? At least I would be able to verify if I used it correctly.

    Is then ADS the only way or could you recommend a cheaper option?  I use RFSIM99 and calculate the matching using the single ended assistant and considering 1/2 on parallel branches of the network but could not get values similar to the ones on the CN.

    Thanks a lot anyway,


  • 0
    •  Analog Employees 
    on Sep 10, 2011 1:47 AM

    Hi Michele,

    The RF impedance calculator is not really applicable in this case.  The SAW filter requires a broadband match and the online tool tends to be too narrowband.  To further complicate the matter, the input match affects the output match and vice versa (making it rather cumbersome).  Nonetheless, I put in the s-parameters of the filter's center frequency just see what it would give.  While it gave a similar configuration to what we used in the Circuit from the Lab, the values were quite off from what was ultimately needed.

    So what is it that is needed?  A match and, more importantly, the correct passband shape.

    The starting point of the circuit design can be any of the following: the simple online matching tool, optimization tools within ADS or the like, or recommendations directly from the SAW filter manufacturer.  Regardless of the starting point, you will likely need to tune to your specific needs (for flat response in the passband).  Finding the right simulation tool is key to speeding up the inevitably empirical nature of the process.

    I am most familiar with ADS and that is why I suggested it, though it can be more than you need (or want).  I had never heard of RFSim99. I downloaded a version this afternoon and it works pretty well with some limitations which you may be able to work around, e.g. only S2P files can be loaded.  S4P files do not seem to be supported on the version that I had, but I may be wrong - it was my first time using it.  I ended up converting the S4P file that I had for the SAW filter to an S2P file.  I loaded the new S2P file in RFSim99 and below are the results of both.  They are virtually the same. You can get to the end point through both avenues.

    I hope this helps.

    Best regards,

    - Carlos

    (note: I attached both the S4P and S2P if you would like to repeat this on your own workstation).


    Schematic in ADS

    Schematic in RFSim99



    Raw SAW filter s-parameters in ADS

    Raw SAW filter s-parameters in RFSim99

    Matched SAW filter in ADS


    Matched SAW filter in RFSim99