Click here for our original post on Schiffman phase shifters
Cluck here if you like chickens
Click here to learn about a new 180-degree phase shifter based on Schiffman's idea
Bernard Schiffman is in our Microwave Hall of Fame!
New for March 2025. Bernard Schiffman is known for his April 1958 IRE paper, "A New Class of Broad-Band Microwave 90-Degree Phase Shifters." During that month the top two songs on the radio were "Tequila" by The Champs and "Sweet Little Sixteen" by Chuck Berry. Tequila was penned by Mexican American Danny Flores, Chuck Berry was a brown-eyed handsome man, and Schiffman was Jewish. What would life have been like in 1958 if Rosevelt had appeased Hitler, and Germany had won WWII? Science would have been set back 50 years, and we'd all be listening to "oldies" music that involved yodeling...
Schiffman's phase shifter uses shorted coupled lines, which have a nonlinear phase characteristic that can be exploited to provide a nearly flat 90 degree phase shift when compared to a single transmission line. The phase shift has a finite bandwidth, but here is something you should know: the amplitude response of the network (return loss, insertion loss) have no such limitation as it is fully matched to Z0 at all frequencies. The necessary condition for this is that the characteristic impedance of the shorted couplers, SQRT(Z0o*Z0e), must equal Z0. That is certainly not a hardship.
In his paper Schiffman reviewed six variations of his phase shifter which he calls Type A through Type F. Here we will only review his four best works. Interestingly, in the final paragraph of Schiffman's paper he gives credit to fellow Hall-of-Famer Seymour Cohn for the "basic idea".
Some of Schiffman's networks are physically long and may have appreciable loss. It is doubtful that a Schiffman phase shifter would ever be used in a MMIC or RFIC design because chip area is so precious. Also, any design that relies on obtaining an exact coupling factor should be examined for the effects of manufacturing tolerances on performance before you fabricate any hardware...
Type A network
The image below is what most people consider the consummate Schiffman phase shifter, which means they stopped reading his six-page article on page two. Schiffman configured a normal transmission line of 270 degrees (length "3l" in his paper) with a shorted, coupled line of 90 degrees (length "l" in his paper) to get 90-degree relative phase over a modest bandwidth.
For some reason, Schiffman insisted on configuring his phase shifters as three port networks. That's fine, so long as you don't care that the common port is 25 ohms (Z0/2).
Schiffman used the ratio ρ ( Greek letter rho, equal to Z0e/Z0o) to describe the coupled line. Most of us are used to thinking about coupled lines in terms of coupling factor. Below, ρ and coupling factor are compared in a plot. When ρ is 1, there is zero coupling. When ρ is 3, coupling is 3 dB (half power, weird how that worked out...) High values of ρ may be hard to achieve depending on the media you are using. Most Lange couplers are over-coupled (more than 3 dB), with ρ maybe 4 to 5, corresponding to coupling of 2.2 to 1.8 dB.
Here is our schematic of Schiffman's Type A circuit. We separated the legs so that it is a four-port network. One way to use this to create a switchable phase shifter is to configure the two arms between a pair of SPDT switches. We used ρ=2.28 (coupling -4.1 dB), as suggested by Schiffman, but you can tweak that parameter to your heart's desire. 2.28 is a value that will be easy to attain, even in microstrip.
The type A response is shown below. Bandwidth is quite modest, at 57%, but for use in a narrow-band phased array that's plenty. We judged the bandwidth by adding markers where phase error is +/- five degrees, but phase error is in the eye of the beholder.
Type B network
Following his description of the Type A design, Schiffman proposed an "error correcting network" which we cribbed from his article, below. It also has non-linear properties but does not provide 90 degree phase shift by itself. A new variable is introduced: "m". It almost goes without saying that "l" refers to 90 degrees of length. In any case, the fifty ohm line needs to be twice as long as the shorted coupled line (2ml as opposed to ml). As he notes in the figure, there are two ways to connect the network to the Type A design, to realize improved bandwidth.
Below is Schiffman's Type B circuit, with the error correcting network added and the values of ρ optimized. The value of "m" is three. You will notice that both legs have significant lengths of 50 ohm line, 270 degrees on the left and 540 degrees on the right. This is an intermediate result... remove the 270 degree on the left and shorten the line on the right by 270 degrees and you will get the same phase result n a much smaller footprint. Here, ρ values are 1.18 and 3.
Check it out, Type B has improved bandwidth compared to Type A:
Type C network
The Type C network is derived from Type A, with the error correcting network connected to the opposite legs compared to Type B, and m=2 instead of 3. The ρ values are re-optimized to 2.35 and 5.83 in our example. ρ of 5.83 represents a 1.5 dB coupler.
Here is the Type C response. We are now up to 135% bandwidth, more than double that of Type A. You could certainly use this in a 6-18 GHz "wideband" application.
Type F network
Notice we did not review the Type D and Type E networks. Schiffman himself said they were not all that useful. Maybe he only put them in the paper to increase the page count? On the opposite end of the spectrum, imagine if you had access to Beethoven's trash can? He was such a perfectionist he probably threw away some amazing works.
The Type F network is not a simple modification from Type A. Schiffman had to roll a second spliff to come up with this one. Just kidding! Here, we have a 450-degree line in one leg and two 90 degree coupled sections cascaded in the other leg. ρ values are 1.6 and 6.2; 6.2 is a pretty sporty -1.4 dB coupler.
The Type F network has bandwidth that falls between Type B and Type C.
We created a 180-degree phase shifter that is based on Schiffman's technique. Read about it here.
Reference
B. M. Schiffman, "A New Class of Broad-Band Microwave 90-Degree Phase Shifters," in IRE Transactions on Microwave Theory and Techniques, vol. 6, no. 2, pp. 232-237, April 1958.