Disclosure of Invention
Aiming at the defects of low frequency, additional noise influence caused by unbalanced quadrature coupler (QHC) output and the like in the prior art, the invention provides an optimization method of a millimeter wave low-noise amplifier and phase shifter combined system, aiming at the defects of an LNA (low noise amplifier), a PS (packet switched amplifier) combined system and a VMPS (virtual packet switched), combining the circuit structure of an LNA in RFFE (radio frequency field effect), realizing the broadband noise matching of the LNA by multiplexing the QHC, and simultaneously generating a quadrature signal for vector modulation to realize active phase shift so as to achieve the purpose of saving area.
The invention is realized by the following technical scheme:
the invention sets transformer as the main part of the orthogonal coupler in the combined system, calculates the required parallel capacitance and the initial value of output impedance of the orthogonal coupler according to the working frequency, optimizes the feedback part of the LNA transformer to make the square root n and the coupling coefficient k of the ratio of the equivalent inductance of the two coils of the LNA transformer1And maximizing, calculating the transconductance of the MOSFET under the condition of optimal noise, and adjusting the direct current bias point and the size of the MOSFET to enable the actual transconductance to reach the optimal value.
The combined system comprises a quadrature coupler, a low noise amplifier group and a programmable gain amplifier which are connected in sequence, wherein: the quadrature coupler generates I, Q two paths of signals through single-end input, amplifies the signals through a CG-level low-noise amplifier group, and synthesizes and outputs differential signals after being modulated by a programmable gain amplifier.
The two coils of the transformer are equivalent to an inductor L3Identical and minimized while ensuring maximum coupling coefficient.
Preferably, when the LNA and QHC are not amplitude matched, n is reduced and g is readjustedmAnd the process is repeated until the condition that the LNA is matched with the QHC amplitude is met.
Technical effects
The invention integrally solves the problems that the bandwidth of the existing VMPS is greatly influenced by QHC or other multiphase generating networks, the impedance matching network between the LNA and the antenna and the VMPS brings large insertion loss and area overhead, and the QHC output is unbalanced to bring extra noise influence.
Compared with the prior art, the invention inserts the LNA into the VMPS and optimizes the noise and amplitude matching of the LNA input end, so that the QHC simultaneously realizes broadband impedance transformation, noise matching and orthogonal signal generation, reduces the requirement of an additional LNA impedance matching network, saves the area and improves the bandwidth. The QHC used for vector synthesis does not require output precision calibration, subject to subsequent PGA control. The invention connects QHC and LNA, avoids noise caused by insertion loss of the impedance matching network before LNA, and makes specific analysis on noise matching of LNA.
Detailed Description
As shown in fig. 1, the present embodiment relates to a combined system, which includes a quadrature coupler QHC, a low noise amplifier LNA, and a programmable gain amplifier PGA connected in sequence, where: the quadrature coupler QHC generates I, Q two paths of signals through single-end input, amplifies the signals by a CG-level low noise amplifier LNA, and synthesizes and outputs differential signals after modulation by a programmable gain amplifier PGA.
And a two-turn or three-turn transformer is preferably adopted between the low-noise amplifier group LNA and the programmable gain amplifier PGA to realize impedance matching.
As shown in fig. 2 and 3, the quadrature coupler QHC includes: the main body part and input end IN, straight-through end THRU, isolation end ISO and coupling end CPL connected with the main body part respectively, the main body part includes: the QHC coils are respectively arranged between the input end IN and the straight-through end THRU, between the isolation end ISO and the coupling end CPL, have the same equivalent inductance value and are mutually coupled, the first capacitors are respectively arranged between the input end IN and the coupling end CPL and between the straight-through end THRU and the isolation end ISO, and the second capacitors are respectively arranged between the input end IN, the straight-through end THRU, the isolation end ISO and the coupling end CPL and the ground and have the same equivalent capacitance value.
The phase difference between the straight-through end THRU and the coupling end CPL is (90 +/-2) °, and the straight-through end THRU and the coupling end CPL are respectively connected with two same low noise amplifiers LNA.
The QHC coil, the first capacitor and the second capacitor meet the following conditions:
wherein: l is
3Is the equivalent inductance of two coils, k
2Is the coupling coefficient between two coils, R
sIs the characteristic impedance of QHC, C
1And C
2The capacitance values of the first and second capacitors are respectively provided, thereby ensuring the electricity under the conditions of odd mode and even modeThe magnetic wave propagation speed is the same.
As shown in fig. 1 and 2, the CG-stage LNA includes two identical LNA, and negative feedback (-a) from gate to source is implemented by using a transformer of the LNA, so as to improve effective transconductance (Gm-boosting), reduce noise figure, and reduce dc power consumption.
Noise figure of each low noise amplifier
Wherein: k is a radical of
1The coupling coefficient of the transformer used for the Gm-boosting LNA, n is two coils (L in FIG. 2)
1And L
2) Square root of ratio of sensitivity, g
mIs MOSFET transconductance, R
sIs the output impedance of QHC, γ is the noise parameter of the MOSFET, g
d0Is the drain-source conductance at a drain-source voltage of 0, delta is the gate noise figure of the MOSFET,
omega is angular frequency, C
gsIs the parasitic capacitance of MOSFET gate source.
The feedback coefficient A of the low noise amplifier is nk1。
When noise factor F
CGMinimum optimum quadrature coupler output impedance
Wherein: alpha is g
mAnd g
d0Ratio of the noise to the noise, the corresponding minimum noise figure
It can be seen that by increasing gmOr increase A (i.e., nk)1) The minimum noise can be reduced, the former can be realized by adjusting the direct current bias point and the size of the MOSFET, and the latter can be realized by enhancing the coupling of the transformer and improving the ratio of the equivalent inductances of the two coils. While taking alpha and g into accountmIn direct proportion, both methods will reduce Rs. At the same time, the effect of feedback on the input impedance is taken into accountIn response, the input impedance of the LNA may be approximately expressed as 1/gmAnd CgsParallel and divide by (1+ A). Thus, both methods reduce the output impedance of the LNA, which makes noise matching and amplitude matching non-contradictory.
In summary, the method for optimizing the QHC and LNA combined system according to this embodiment includes the following steps:
firstly, designing a transformer which can be realized by a process as a QHC main body part to ensure that two coils of the transformer have equivalent inductance L3Equal and as small as possible while ensuring a coupling coefficient greater than 0.7.
And secondly, calculating the parallel capacitance required by the orthogonal coupler according to the current working frequency.
And thirdly, calculating the output impedance of the orthogonal coupler as an initial value according to the current working frequency.
Fourthly, designing a transformer which can be realized by the process to be used as a feedback part of the LNA first-stage MOSFET so that the square root n of the ratio of the two coil equivalent inductances is as large as possible and the coupling coefficient k1Greater than 0.75.
Fifthly, calculating transconductance g of the MOSFET under the optimal noisem,optIncreasing the direct current bias voltage at the grid end of the MOSFET and increasing the width-to-length ratio to enable gmAnd gm,optAre equal.
And sixthly, when the input impedance of the LNA is different from the optimal QHC output impedance, returning to the fourth step, and decreasing n by taking 0.05 as a typical step, and repeating the steps until the condition that the input impedance of the LNA is the same as the optimal QHC output impedance is met.
As shown IN fig. 3, the basic structure of the quadrature coupler implemented according to the above method is that the coil disposed between the input terminal IN and the through terminal THRU is made of metal 1 and metal 3, and the coil disposed between the isolation terminal ISO and the coupling terminal CPL is made of metal 2 and metal 4. The two coils have the same equivalent inductance, and two layers of metal of the same coil are connected by a via hole 5.
The total thickness of the two coils needs to be as close as possible so that the equivalent inductance values of the two coils are the same. For increased coupling coefficient, the width of the coil was set to 1/5 of the transformer inner diameter.
Passing toolIn practical experiments, under the specific environment setting that the working frequency is 30GHz, R is useds=10Ω、k2=0.73、L3The above process was run at 58pH parameters and the results at the completion of the design are shown in figures 4 to 6.
The noise matching results are shown in fig. 4, where the LNA input impedance curve is tangent to a circle with a noise figure of 2.5 dB.
The output phase and insertion loss results of QHC are shown in fig. 5, and the operating frequency range of QHC is 10GHz to 40GHz within the phase error range of 1 °. The insertion loss of the two output ends is 3.5dB when the insertion loss is the same, and the additional 0.5dB loss is caused by the structure because the subsequent two paths compensate the 3dB gain after vector synthesis is carried out.
The broadband impedance transformation characteristic of QHC is shown in fig. 6, and when f is 30GHz during the characteristic impedance changes from 10 ohms to 70 ohms, the return loss is always kept below-17 dB, which proves that the characteristic impedance can be adjusted according to the requirement.
The noise figure of the LNA-PS system before and after noise matching is shown in FIG. 7, k1=0.79、n=1.71、L1=320pH、L2The 3-dB bandwidth of the LNA-PS system is 26-34GHz at 110, and a minimum noise figure of 5.2dB is achieved at 32 GHz. The minimum noise coefficient of the structure is 5.2dB, and 6.2dB is reduced compared with the condition that LNA and QHC are independently designed and directly connected without noise matching.
In the embodiment, for a specified characteristic impedance, in combination with the structure of fig. 3, the metal layer with the closest total thickness is selected by using the current process, so that the equivalent inductances of the two coils are the same and are as low as possible, and a coupling coefficient as high as possible is realized; iterative optimization is used to achieve simultaneous matching of QHC and LNA for amplitude and noise.
The foregoing embodiments may be modified in many different ways by those skilled in the art without departing from the spirit and scope of the invention, which is defined by the appended claims and all changes that come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.