+ Post New Thread
Results 1 to 10 of 10

14th March 2012, 22:48 #1
 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 AdvanceWhen 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 ?

14th March 2012, 23:36 #2
 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 allpass 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.

Advertisement

14th March 2012, 23:38 #3
 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
GoldsmithWhen 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 ?

Advertisement

14th March 2012, 23:52 #4
 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
KeithI started life with nothing and I've still got most of it left. (Seasick Steve)
1 members found this post helpful.

Advertisement

15th March 2012, 00:03 #5
 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
1 members found this post helpful.

15th March 2012, 06:11 #6
 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.

Advertisement

15th March 2012, 14:23 #7
 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 audioband (300 Hz  3 kHz) phase shifting, I obtained the original paper describing its operation [Saraga, W., "The Design of WideBand 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) < 1e9) fprintf('C%d = %6.2f pF\n', i, C(i)*1e+12); elseif (C(i) < 1e6) fprintf('C%d = %6.2f nF\n', i, C(i)*1e+9); elseif (C(i) < 1e3) 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 imagerejecting 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 FR4). 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 teflonsubstrate PCB blank and a scalpel, cutting around the outline of a printed template!
Good luck :)
2 members found this post helpful.

27th November 2013, 18:58 #8
 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

27th November 2013, 19:18 #9
 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 100Hz1.1MHz 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.

27th November 2013, 19:27 #10
 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.
+ Post New Thread
Please login