Abstract
A new two-probe waveguide slide screw load-pull tuner of which the probes share the same waveguide section; they are inserted diametrically at fixed depth into slots facing each other on opposite broad walls of the waveguide. The tuner does not have cumbersome adjustable vertical axes controlling the penetration of the probes and its low profile is optimized for on-wafer operations. The carriages holding the probes are moved along the waveguide using electric stepper motors or linear actuators. The calibration uses mutual de-embedding of initialized probes and is 20 or more times faster than the full probe permutations calibration.
Claims
1. A calibration method for a waveguide slide screw tuner having a test port and an idle port, and two tuning probes, a first tuning probe and a second tuning probe, inserted diametrically and laterally offset to each other into slots on opposite broad walls of the waveguide and movable at constant penetration along the waveguide, comprising the following steps: a) connect the tuner to a VNA pre-calibrated at a frequency F; b) move the tuning probes to initial positions, the first tuning probe to X1.0 and the second tuning probe to X2.0; c) measure s-parameters of the tuner and save in an init matrix [S0]; d) in a first move-measurement loop d1) move the first tuning probe from X1=X1.0 to at least X1=X1.0+/2 in N steps of X=/(2N), d2) measure s-parameters and save in a file S1 including Sij(X1,X2.0); e) return the first tuning probe to position X1=X1.0; f) in a second move-measurement loop f1) move the second tuning probe from X2=X2.0 to at least X2=X2.0+/2 in N steps of X=/(2N), f2) measure s-parameters and save in a file S2 including Sij(X1.0,X2); g) in a third move-measurement loop move the first tuning probe from X1=X1.0 to at least X1=X1.0+2 in steps of X and, for each movement of the first tuning probe, move the second moving probe from X2=X1M*X to X2=X1+M*X in steps of X, with M=0, 1, 2, 3 . . . and measure s-parameters and save in files SK including Sij(X1,X2), for 3K3+2M; h) in a numerical cascading loop for all permutations {X1,X2} if the first tuning probe is closer to the test port (X1X2) de-embed s-parameters of the second tuning probe: cascade s-parameters [S1]*[S0].sup.1*[S2] and save in file C1; if the second tuning probe is closer to the test port (X1>X2) de-embed the first tuning probe: cascade s-parameters [S2]*[S0].sup.1*[S1] and save in file C2; i) concatenate s-parameters of files C1 with C2, replace s-parameters with data from files SK and save s-parameters Sij(X1,X2) in a tuner calibration file C(F).
2. The calibration method as in claim 1 wherein the slots of the waveguide slide screw tuner to which the calibration method applies are at least one half of a wavelength (/2) long at a minimum operation frequency (Fmin).
3. The calibration method as in claim 1 wherein the tuning probes of the waveguide slide screw tuner to which the calibration method applies are controlled by remotely controlled stepper motors and appropriate gear.
4. The calibration method as in claim 1 wherein the tuning probes of the waveguide slide screw tuner to which the calibration method applies are metallic or at least partly metallized rods.
5. The calibration method as in claim 1 wherein the tuning probes of the waveguide slide screw tuner to which the calibration method applies are metallic or at least partly metallized slabs.
6. The calibration method as in claim 1 wherein the tuning probes of the waveguide slide screw tuner to which the calibration method applies are inserted contactless into the slots.
7. The calibration method as in claim 1 wherein the initialization coordinates of the tuning probes of the waveguide slide screw tuner to which the calibration method applies are X1.0=X2.0=0.
8. An impedance synthesis (tuning) method for the two-probe waveguide slide screw tuner calibrated using the calibration method as in claim 1 comprising the following steps: a) enter a frequency F; b) enter a target reflection factor Gamma; c) load s-parameters Sij(X1,X2) from calibration file C(F) in memory; d) in a search loop for all permutations {X1,X2} d1) compare S11(X1,X2) with Gamma; d2) if |S11(X1,X2)Gamma| is minimum, save X1.T=X1, X2.T=X2; e) move the first tuning probe to X1.T and the second tuning probe to X2.T.
Description
BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS
(1) The invention and its mode of operation will be better understood from the following detailed description when read with the appended drawings, in which:
(2) FIG. 1 depicts prior art: a typical generic automated load pull test system.
(3) FIGS. 2A through 2B depict prior art: a single probe waveguide impedance tuner;
(4) FIG. 2A depicts a front view of the entire tuner; FIG. 2B depicts a cross section of the tuning probe (typically a conductive rod) entering the waveguide slot.
(5) FIG. 3 depicts partly prior art: a Smith chart and two possible trajectories of impedance synthesis (tuning) to reach a target impedance starting from the origin of 501. Reaching TARGET-1 uses the prior art single-probe technique with horizontal and vertical control; reaching TARGET-2 uses the new two-probe method with horizontal-only and without vertical control.
(6) FIG. 4 depicts a cross-section through a waveguide with two conductive tuning rods (probes) entering from opposite sides into slightly offset slots.
(7) FIG. 5 depicts a front view of the waveguide tuner with two crossing over tuning probes.
(8) FIG. 6 depicts the impedance synthesis on the Smith chart using the two-probe impedance tuner with fixed probe insertion depth.
(9) FIGS. 7A through 7C depict possible relative positions of the two tuning probes: FIG. 7A depicts the probes far from each-other; FIG. 7B depicts the probes close to each-other entering mutual electromagnetic fields; FIG. 7C depicts the two probes partially or fully overlapping.
(10) FIG. 8 depicts the tuner calibration flow-chart.
(11) FIG. 9 depicts the impedance synthesis flow-chart.
(12) FIG. 10 depicts a TARGET reflection factor inside a calibration pixel.
DETAILED DESCRIPTION OF THE INVENTION
(13) This invention discloses a high frequency (RF, microwave, millimeter wave), computer-controlled impedance tuner, suitable for load pull measurements. The tuner (FIG. 5) uses a low loss waveguide transmission airline 57, which includes two broad top walls, two narrow sidewalls and two slots 46, 46A as shown in FIGS. 4 and 5, cut into the broad walls, one on the top and one on the bottom. The slots run parallel to the waveguide longitudinal axis and are positioned opposite to each other and slightly offset from the symmetry center line of the waveguide. The offset eccentricity is selected to allow the two tuning probes (typically metallic or metallized rods 42, 42A) to cross over (pass next to each other) without touching. This structure is chosen for economy of space because it uses slots of a total length 56 of one half of a wavelength plus the thickness of one tuning probe. An alternative configuration, where the tuning rods would not cross over, would, in principle, also work, but the slot (and the tuner) would have to be twice as long. The horizontal control of the carriages 53 and 53A can be made using stepper motors 54 and ACME screws 51 but can also be accomplished using linear electric actuators (see ref. 6), instead of motors 54 and ACME screws. The actuators have a motorized body, and their rotor axis is the extended horizontal ACME screw 51.
(14) FIG. 5 shows a front view of the waveguide tuner created by the cross section 41, 44 of FIG. 4: the slots 46, 46A can be seen as well as their effective length 56; the mobile carriages 53 and 53A are controlled by the ACME screws 51 and 51A along the waveguide 57, via the stepper motors 54, 54A; the coordinate of carriage 53 is X1 and of carriage 53A is X2 relative to a common arbitrary zero position that can be the test port 50. The distance of the first tuning probe 55 from the test port 50 is X1 and the electric distance between the two tuning probes is X=|X1X2|; the lower slot 46A is hidden in FIG. 5 as it is evident from the cross section 44 in FIG. 4.
(15) The total reflection factor 105 (FIG. 6) is the sum of all internal reflections in the waveguide and a function of the positions X1 and X2 of the two probes: S11(X1,X2) and is the vector-sum of the reflection factors 103 and 104 of both probes S11(X1) and S11(X2): S11(X1,X2)=S11(X1)+S11(X2), all referenced at the test port 50. The tuning probe 55 closest to the test port generates the primary reflection 100 i.e., S11(X1), 103. Since the reflection of the first probe 55 is selected not to be maximum (i.e. >0.9), but is, instead, selected to have medium value |S11(X1)|0.5-0.7, there is a signal portion traversing probe 55 towards the secondary probe 55A; this signal portion is then reflected back towards the primary probe 55; again, a portion of this reflected back signal traverses the primary probe towards the test port and adds to the total reflection. The signal reflected at the secondary probe 55A is also reflected back at the primary probe 55, and so on . . . creating the phenomenon of a multiple reflection. In mechanical terms this appears like a turbulence. This back and forth bouncing of signal is well described using signal flow graphs (see ref. 5). Instead, since the tuner is terminated with a matched load any escaping signal is not reflected back and therefore ignored.
(16) FIG. 6 depicts schematically the overall reflection factor synthesis mechanism: each probe creates at its own reference plane reflections .sub.1 or .sub.2, which are concentric reflection factor circles represented at test port reference plane as trace 100 on the Smith chart (FIG. 6). The total reflection factor trace 101 is created by a planetary epicycloid superposition of the two reflection factor vectors around the center 102 (one circle rotates around a point on the periphery of the other). When the probes cross-over the circles swap. The total reflection factor vector 105 is created by the vector sum of vector 103 (from the first probe) and the vector 104 (from the second vector) as described above.
(17) To be used in impedance synthesis the tuner must be calibrated and the calibration data must be used to generate the proper probe positions to reach a TARGET impedance. To create a reasonable accuracy each probe of the tuner must be characterized at least at 100 positions between X=0 and X=XMAX=/2; in terms of a typical circular trajectory 100 (FIG. 6) this corresponds to steps of 3.6. At 50 GHz using a waveguide class WR19 with internal dimensions of the cavity of a=4.8 mm, b=2.4 mm (FIG. 4) this corresponds to %=8 mm and mechanical steps of 0.04 mm. To calibrate all 100100=10,000 tuner states (calibration points) one would need at least 10,000 seconds or 2.8 hours, at 1 second per point. The new method disclosed here allows reducing this time to approximately 8.4 minutes.
(18) The calibration procedure executes as follows (FIG. 8): The tuner is connected to a pre-calibrated vector network analyzer (VNA); then the two probes are placed to initial positions, for instance probe 1 at X1=X1.0=0 and probe 2 at X2=X2.0=XMAX=/2; then scattering (s-) parameters are measured and saved in a init matrix [S0]; then, in a first measurement loop probe 1 is moved in a number N (typically 100) of steps X from X1=0 to X1=XMAX and s-parameters are measured and saved in a matrix [S1]=[S(X1,X2.0)] containing 4 sets of two-port s-parameters. In a next measurement loop probe 1 is returned to its initial position X1.0 and probe 2 is moved from X2=XMAX to X2=0 and s-parameters are measured and saved in a matrix [S2]=[S(X1.0,X2)]. The total data are saved in data files S1 and S2 comprising N sets of s-parameters each. In real tuning, though, there will be situations where the two probes will be near each-other or even overlap (FIG. 7B-7C). If we define near as the positioning of a probe at least one step left and one step right of the other probe, then we have a total of at least three positions where one probe affects the other, so a total of at least 3N positions; a third measurement loop is then, in the simplest scenario, as follows: Probe 1 moves from X1=0 to X1=XMAX in N steps of length X=/(2N); then probe 2 moves to three positions, one ahead, one opposite and one after probe 1: X2=X1X, X2=X1 and X2=X1+X, and each time we measure s-parameters and save in matrices [S3]=[S(X1,X1X)], [S4]=[S(X1,X1)] and [S5]=[S(X1,X1+X)]; the totality of these measurements are saved in data files S3 to S5 etc., comprising N sets of s-parameters each. This makes for a total of 2N+3N=5N measurements, instead of N.sup.2 measurements, a time reduction of N.sup.2/(5N)=N/5; in the case of N=100 the new calibration method is 20 times faster. In the case of N=200 and two adjacent calibration points, allowing to increase the calibration density and cross-over accuracy, the time improvement ratio becomes N.sup.2/(7N)29.
(19) In a next step the hitherto measured calibration data of the individual probes are used to generate the tuner calibration file as follows: for all X1, X2 permutations, symbolized as {X1,X2}, except for the probe-interacting or -overlapping cases, where X1XX2X1+X, cascade the matrices [S1], [S2] and [S0] as follows: if X1X2, cascade [S1]*[S0].sup.1*[S2] and save in a first portion of the tuner calibration file C1; if X1>X2, cascade [S2]*[S0].sup.1[S1] and save in a second portion of the tuner calibration file C2. In a final step, insert the s-parameter data matrices [S3], [S4] and [S5] into, and replace, the concatenated s-parameter data of C1 and C2 to create the overall calibration file C. Or, the s-parameter data in file C include sets of tuner s-parameters between the test port and the idle port for all N.sup.2 permutations of X1 and X2 positions every X=/(2N) steps including: (N.sup.23N) cascaded and de-embedded s-parameters of the individual probes and 3N directly measured s-parameter sets of probes overlapping or in proximity. In the case of 5 or more directly measured overlapping and adjacent points this number becomes N.sup.25N, N.sup.27N etc. . . . .
(20) Tuning is the procedure of synthesizing a TARGET reflection factor using the calibrated tuner, i.e., determining the best probe positions X1, X2 that create a desired reflection factor 110 (see tuning algorithm in FIG. 9 and interpolation details in FIG. 10). Tuning accuracy is the vector difference 112 between the user-defined target reflection factor 110 and the closest synthesized calibration point 111. Assuming the tuner is calibrated at 100100=10,000 settings (X1, X2): in a first rough approximation, on the Smith chart with a radius of 1 having a surface of , the size of each pixel is /10.sup.4 and, in a worst case scenario, as shown in FIG. 6, item 106, if a target is in the center of a contour defined by the surrounding pixel (four calibrated points), the possible error, i.e., the difference of the closest calibrated point from the target is 20 log(sqrt(/2)*10.sup.2))38 dB; in this case tuning is a simple search through the data points vector S11(X1,X2) of the calibration file C using an error function EF=|TARGET-S11| to minimize. This search could use optimized searching Real(S11) and Imag(S11) separately and should not last more than a few milliseconds. If the accuracy of 38 dB seems too coarse, then the number of calibrated points could be increased to 150 or 200 per axis with, still reasonable, calibration time and reach accuracies of 20 log(sqrt(/2)/3*10.sup.2))41 dB or 20 log(sqrt(/(2)/4*10.sup.2))44 dB etc. Even if every doubling of the number of horizontal steps doubles the, already inherently low, calibration time, it increases the accuracy by at least 6 dB.
(21) If the number N of calibration steps per tuning probe increases, in order to enhance the tuning accuracy and, depending on the form of the tuning probes (rods, slivers or else) it is possible that the interaction between probes spreads to more than plus and minus one step beyond overlapping. This is not probable, because the electric field is concentrated at the closest point 70 between the capacitive tuning probe and the opposite inner ground wall, but, depending on the number N and the step size X=/(2N) it cannot be excluded. In this case the number of calibration points before and after the probe overlapping should be increased to 2, or 3. In this case the calibration time increases as well, but the calibration procedure remains the same, as does the tuning algorithm.
(22) Obvious alternative embodiments of fixed penetration tuning probes, diametrically inserted into and sharing the same slabline of waveguide slide screw impedance tuners and associated calibration and tuning methods shall not impede on the core idea of the present invention.
