+ Post New Thread
Results 1 to 10 of 10
  1. #1
    Advanced Member level 5
    Points: 19,467, Level: 33
    Achievements:
    Created Blog entry Created Album pictures 7 years registered
    goldsmith's Avatar
    Join Date
    Dec 2010
    Location
    Tehran - IRAN
    Posts
    3,984
    Helped
    743 / 743
    Points
    19,467
    Level
    33
    Blog Entries
    8

    90 degree phase shift for a wide range of frequency ( exactly , 90 degree !)

    Dear All
    Hi
    Thank you for your time .
    I know how can i create 90 degree phase shift for an specific frequency , but how to create it for a wide range of frequency ?
    Thanks in Advance
    When you sensed that i can help you , feel free to ask me .
    ************************
    Experience is a good professor of mine !
    ************************
    The only thing that i couldn't understand yet , is why humans should fight together ? why not helping each other to have a better world ?

  2. #2
    Super Moderator
    Points: 260,253, Level: 100
    Awards:
    1st Helpful Member

    Join Date
    Jan 2008
    Location
    Bochum, Germany
    Posts
    45,450
    Helped
    13829 / 13829
    Points
    260,253
    Level
    100

    Re: 90 degree phase shift for a wide range of frequency ( exactly , 90 degree !)

    In the analog domain, a differential phase shift between two all-pass chains is the only available means. It has been discussed previously at Edaboard, the latest in this recent thread. Please refer to the first link in post #9.
    https://www.edaboard.com/thread241020.html#post1032377

    In digital signal processing, a Hilbert transformation can be approximated in a restricted frequency range by a FIR filter. As all FIR filters, it involves an additional signal delay.

    Referring to the question title, neither the analog nor the digital wide band phase shift are exact by nature. They are approximations with a ripple depending on the implementation effort.


    1 members found this post helpful.

    •   AltAdvertisement

        
       

  3. #3
    Advanced Member level 5
    Points: 19,467, Level: 33
    Achievements:
    Created Blog entry Created Album pictures 7 years registered
    goldsmith's Avatar
    Join Date
    Dec 2010
    Location
    Tehran - IRAN
    Posts
    3,984
    Helped
    743 / 743
    Points
    19,467
    Level
    33
    Blog Entries
    8

    Re: 90 degree phase shift for a wide range of frequency ( exactly , 90 degree !)

    Dear FvM
    Hi
    Thank you for your reply .
    I'll read that thread completely , and tell the result .
    Appreciate
    Goldsmith
    When you sensed that i can help you , feel free to ask me .
    ************************
    Experience is a good professor of mine !
    ************************
    The only thing that i couldn't understand yet , is why humans should fight together ? why not helping each other to have a better world ?



    •   AltAdvertisement

        
       

  4. #4
    Super Moderator
    Points: 50,045, Level: 54
    Achievements:
    7 years registered
    keith1200rs's Avatar
    Join Date
    Oct 2009
    Location
    Yorkshire, UK
    Posts
    10,877
    Helped
    2075 / 2075
    Points
    50,045
    Level
    54

    Re: 90 degree phase shift for a wide range of frequency ( exactly , 90 degree !)

    It depends on the frequency but for RF I think one method uses transformers http://www.seboldt.net/k0jd/phase_notes.html

    Keith
    I started life with nothing and I've still got most of it left. (Seasick Steve)


    1 members found this post helpful.

    •   AltAdvertisement

        
       

  5. #5
    Super Moderator
    Points: 260,253, Level: 100
    Awards:
    1st Helpful Member

    Join Date
    Jan 2008
    Location
    Bochum, Germany
    Posts
    45,450
    Helped
    13829 / 13829
    Points
    260,253
    Level
    100

    Re: 90 degree phase shift for a wide range of frequency ( exactly , 90 degree !)

    It depends on the frequency but for RF I think one method uses transformers
    Besides different circuit technologies available for audio or RF frequencies, it mainly depends on the meaning of "wide range". The principle problem of generating an "exact" wide range phase shift is basically the same. There are additional means provided by distributed circuits, but they are also approximations of an ideal 90° shift in some regard.


    1 members found this post helpful.

  6. #6
    Super Moderator
    Points: 52,942, Level: 56

    Join Date
    Apr 2011
    Location
    Minneapolis, Minnesota, USA
    Posts
    12,949
    Helped
    2580 / 2580
    Points
    52,942
    Level
    56

    Re: 90 degree phase shift for a wide range of frequency ( exactly , 90 degree !)

    The amount of phase change is directly related to the time constant. (R*C, or L/R).

    As long as there is any amount of series resistance, the amount of phase change will never reach 90 degrees, in either a capacitor or coil.

    Current lead is built into the way a capacitor behaves when exposed to a sine wave. At 3/4 cycle the charged capacitor starts discharging into the falling sine wave.

    By the same token, current lag is built into the nature of an inductor exposed to a sine wave.

    With any given capacitor or inductor, you can increase the phase change by lowering the series resistance.

    If you could achieve zero resistance, then phase change would be minus 90 degrees for every capacitor, and plus 90 degrees for every inductor, at every frequency.

    Of course this is not possible with real components. A real coil has some ohmic resistance and some capacitance.

    A real capacitor has some ohmic resistance and some inductance. In general, capacitors yield more 'purity' of response.


    1 members found this post helpful.

    •   AltAdvertisement

        
       

  7. #7
    Full Member level 6
    Points: 4,149, Level: 15
    Achievements:
    7 years registered
    thylacine1975's Avatar
    Join Date
    Dec 2010
    Location
    Canberra, Australia
    Posts
    337
    Helped
    227 / 227
    Points
    4,149
    Level
    15

    Re: 90 degree phase shift for a wide range of frequency ( exactly , 90 degree !)

    Heya Goldsmith (et al)!

    I 've needed broadband 90 degree phases shifts on a number of occasions, and I've had a *great deal* of success with a configuration I first dicovered in the ARRL Handbook, and described in detail (in the "The Phasing Method of Image Rejection" section) of http://michaelgellis.tripod.com/mixerscom.html.
    [I've reproduced it here for convenience:]


    Although it's often discussed in the context of audio-band (300 Hz - 3 kHz) phase shifting, I obtained the original paper describing its operation [Saraga, W., "The Design of Wide-Band Phase Splitting Neworks," Proc IRE, Vol 38, p 754 (1950)] and coded the equations into the following MATLAB script:

    Code C - [expand]
    1
    2
    3
    4
    5
    6
    7
    8
    9
    10
    11
    12
    13
    14
    15
    16
    17
    18
    19
    20
    21
    22
    23
    24
    25
    26
    27
    28
    29
    
    n      = 6;           % Number of RC stages within network
    fLower = 1000000;     % (Hz) - Lower network design frequency
    fUpper = 10000000;    % (Hz) - Upper network design frequency
    R      = 220;         % (Ohms) - Value of resistance used throughout the network
     
    m = 1 - (fLower/fUpper)^2;
    K = ellipke(m);
     
    for i = 1:n
        [sn, cn, dn] = ellipj(((2*i - 1)*K)/(2*n), m);
        RC(i) = dn/(2*pi*fLower);
    end
     
    C = RC/R;
    fprintf('For a %d Hz - %d Hz %d stage RC network (R = %d), Cn is as follows:\n\n', ...
    fLower, fUpper, n, R); 
    for i = 1:n
        if (C(i) < 1e-9)
            fprintf('C%d = %6.2f pF\n', i, C(i)*1e+12);
        elseif  (C(i) < 1e-6)
            fprintf('C%d = %6.2f nF\n', i, C(i)*1e+9);
        elseif  (C(i) < 1e-3)
            fprintf('C%d = %6.2f uF\n', i, C(i)*1e+6);
        elseif  (C(i) < 1)
            fprintf('C%d = %6.2f mF\n', i, C(i)*1e+3);
        else           
            fprintf('C%d is physically unrealisable!\n', i);    
        end
    end

    As the script suggests - I've successfully built a 1 MHz - 10 MHz network using this approach! I used this as a part of an image-rejecting HF mixer design and I recall obtaining +/- 2 degree phase accuracy over the design bandwidth. This figure can apparently be improved with more stages, but I haven't personally tried this since the level of accuracy obtained was sufficient for my application.

    I built a second version operating at a few 100 kHz a while later, with similar success.

    So that's the LF stuff - for the microwave end of the spectrum, you cant go past a stripline "quadrature hybrid coupler" (descibed here: http://www.microwaves101.com/encyclo...idcouplers.cfm) for simplicity and performance. I've found that EM simulation and optimisation (with ADS, CST etc) is essential for best performance, as is constructing them on a predictable substrate (e.g. Rogers instead of FR-4). I've made 2.4 GHz devices with 1 GHz bandwidths this way with little difficulty, and even had modest success with nothing more than a teflon-substrate PCB blank and a scalpel, cutting around the outline of a printed template!

    Good luck :)


    2 members found this post helpful.

  8. #8
    Newbie level 2
    Points: 16, Level: 1

    Join Date
    Nov 2013
    Posts
    2
    Helped
    0 / 0
    Points
    16
    Level
    1

    Re: 90 degree phase shift for a wide range of frequency ( exactly , 90 degree !)

    Dear thylacine1975,
    Thanks a ton. This is exactly what we were looking for. Also, as an aside, we have now ordered a copy of the ARRL Handbook.

    As another aside, this is probably the most helpful forum post I've ever come across. Who includes MATLAB code in a comment?! You do.

    Thanks again,
    Brandon



  9. #9
    Advanced Member level 5
    Points: 20,688, Level: 34
    Achievements:
    7 years registered

    Join Date
    Jan 2011
    Posts
    3,383
    Helped
    1209 / 1209
    Points
    20,688
    Level
    34

    Re: 90 degree phase shift for a wide range of frequency ( exactly , 90 degree !)

    It occurred to me a while ago that you could make a good low frequency wideband phase shifter by upconverting the signal to RF, using a relatively narrowband phase shifter, then downconverting back again. So a 1MHz bandwidth at 100MHz (easy to do) would become a 100Hz-1.1MHz bandwidth (crazy). Has anyone tried this? It would be pretty complicated and would require full image rejection, but in principle I can't see why it wouldn't work.



  10. #10
    Newbie level 2
    Points: 16, Level: 1

    Join Date
    Nov 2013
    Posts
    2
    Helped
    0 / 0
    Points
    16
    Level
    1

    Re: 90 degree phase shift for a wide range of frequency ( exactly , 90 degree !)

    I've seen/heard of similar things being done. For example, this technique was used enable the use of COTS variable phase shifters in bands where COTS phase shifters don't exist. I haven't heard about it for lower frequency, like you're proposing, but I haven't really worked below ~200MHz before now.



--[[ ]]--