U.S. patent application number 16/375783 was filed with the patent office on 2020-01-02 for iq mismatch correction for zero-if/low-if tx/rx.
The applicant listed for this patent is TEXAS INSTRUMENTS INCORPORATED. Invention is credited to Jaiganesh Balakrishnan, Sthanunathan Ramakrishnan, Jawaharlal Tangudu.
Application Number | 20200007365 16/375783 |
Document ID | / |
Family ID | 69055480 |
Filed Date | 2020-01-02 |
![](/patent/app/20200007365/US20200007365A1-20200102-D00000.png)
![](/patent/app/20200007365/US20200007365A1-20200102-D00001.png)
![](/patent/app/20200007365/US20200007365A1-20200102-D00002.png)
![](/patent/app/20200007365/US20200007365A1-20200102-D00003.png)
![](/patent/app/20200007365/US20200007365A1-20200102-D00004.png)
![](/patent/app/20200007365/US20200007365A1-20200102-D00005.png)
![](/patent/app/20200007365/US20200007365A1-20200102-M00001.png)
![](/patent/app/20200007365/US20200007365A1-20200102-M00002.png)
![](/patent/app/20200007365/US20200007365A1-20200102-M00003.png)
United States Patent
Application |
20200007365 |
Kind Code |
A1 |
Tangudu; Jawaharlal ; et
al. |
January 2, 2020 |
IQ MISMATCH CORRECTION FOR ZERO-IF/LOW-IF TX/RX
Abstract
IQ mismatch correction for analog chain IQ mismatch impairments
is based on a two-filter architecture. In either RX or TX, an IQmc
mismatch corrector (digital chain) filters I and Q digital signals,
and includes an I-path to receive the I signal, and a Q-path to
receive the Q signal, and is configured with two filters: an
in-path filter to filter either the I signal or the Q signal
received in the same path; and a cross-path filter to filter either
the I signal or the Q signal received in the other path. The IQmc
mismatch corrector can include: an I-path delay element to provide
a delay to the I signal corresponding to a delay through either the
in-path filter or the cross-path filter; and a Q-path delay element
to provide a delay to the Q signal corresponding to a delay through
either the in-path filter or the cross-path filter.
Inventors: |
Tangudu; Jawaharlal;
(Bangalore, IN) ; Ramakrishnan; Sthanunathan;
(Bangalore, IN) ; Balakrishnan; Jaiganesh;
(Bangalore, IN) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
TEXAS INSTRUMENTS INCORPORATED |
Dallas |
TX |
US |
|
|
Family ID: |
69055480 |
Appl. No.: |
16/375783 |
Filed: |
April 4, 2019 |
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
H04L 27/2064 20130101;
H04B 1/30 20130101; H04B 1/0007 20130101; H03D 3/009 20130101; H04B
1/40 20130101; H04L 27/364 20130101; H04L 27/22 20130101; H04L
27/3863 20130101 |
International
Class: |
H04L 27/20 20060101
H04L027/20; H04L 27/22 20060101 H04L027/22; H04B 1/40 20060101
H04B001/40 |
Foreign Application Data
Date |
Code |
Application Number |
Apr 4, 2018 |
IN |
201841012763 |
Claims
1. A circuit to provide IQ mismatch correction for use in a system
for radio frequency (RF) communication including a transmit (TX)
end, and/or a receive (RX) end, the circuit comprising: at one of
the TX end or the RX end, an analog signal chain and a digital
signal chain: the analog signal chain including analog circuitry
that introduces IQ mismatch signal impairments, the digital signal
chain including an IQmc mismatch corrector to filter In-phase and
Quadrature digital signals to provide IQmc correction to correct
the IQ mismatch impairments, and an interface between the digital
signal chain and the analog signal chain; and the IQmc mismatch
corrector, including: an I-path coupled to receive the In-phase
digital signals (I signal), and a Q-path coupled to receive the
Quadrature digital signals (Q signal); and an in-path filter
element to filter either the I signal or the Q signal received in
the same path, and a cross-path filter element to filter either the
I signal or the Q signal received in the other path.
2. The circuit of claim 1, the IQmc mismatch corrector further
including: an I-path delay element included in the I-path to
provide a delay to the I signal corresponding to a delay through
either the in-path filter element or the cross-path filter element;
and a Q-path delay element included in the Q-path to provide a
delay to the Q signal corresponding to a delay through either the
in-path filter element or the cross-path filter element.
3. The circuit of claim 1, wherein the cross-path correction filter
is connected from the Q-path to the I-path; and the in-path
correction filter is connected in the Q-path; so that, the Q signal
input is applied to the in-path and cross-path correction filters,
and IQmc correction is provided for the I-path (Icorr) and the
Q-path (Qcorr).
4. The circuit of claim 1, wherein the cross-path correction filter
is connected from the I-path to the Q-path; and the in-path
correction filter is connected in the I-path; so that, the I signal
input is applied to the in-path and cross-path correction filters,
and IQmc correction is provided for the I-path (Icorr) and the
Q-path (Qcorr).
5. The circuit of claim 1, wherein the cross-path correction filter
is connected from the I-path to the Q-path; and the in-path
correction filter is connected in the Q-path; so that, the I signal
input is applied to the cross-path correction filter, and the Q
signal input is applied to the in-path correction filter, and IQmc
correction is provided for the Q-path (Qcorr).
6. The circuit of claim 1, wherein the cross-path correction filter
is connected from the Q-path to the I-path; and the in-path
correction filter is connected in the I-path; so that, the Q signal
input is applied to the cross-path correction filter, and the I
signal is applied to the in-path correction filter, and IQmc
correction is provided for the I-path (Icorr).
7. The circuit of claim 1, the IQmc mismatch corrector to provide
IQmc correction that results in a change in linear response through
the analog and digital signal chains, the circuit further
comprising: in the analog signal chain, a digital step attenuator
(DSA) to provide a selected DSA step attenuation; in the digital
signal chain, a DSA gain/phase correction block to provide a DSA
correction signal to the DSA to correct the selected DSA step
attenuation, and a DSA gain/phase correction update module to
adjust the DSA correction signal to compensate for the change in
linear response.
8. The circuit of claim 1, wherein the in-path and cross-path
filter elements include filter coefficients selectively updated
based on selectively estimated IQmc mismatch impairment; and the
in-path and cross-path filter elements are implemented in hardware,
and the filter coefficients are selectively updated by computation
in firmware based on estimated IQmc mismatch impairment.
9. A transceiver circuit for transmitting and receiving radio
frequency (RF) communication signals, including a transmit (TX)
end, and a receive (RX) end, at the TX end and/or the RX end the
circuit comprising: an analog signal chain, and including analog
circuitry that introduces IQ mismatch signal impairments; a digital
signal chain including an IQmc mismatch corrector to filter
In-phase and Quadrature digital signals to provide IQmc correction
to correct the IQ mismatch impairments; and an ADC
(analog-to-digital conversion) circuitry to provide an interface
between the analog signal chain and the digital signal chain; the
IQmc mismatch corrector including: an I-path coupled to receive the
In-phase digital signals (I signal), and a Q-path coupled to
receive the Quadrature digital signals (Q signal); and an in-path
filter element to filter either the I signal or the Q signal
received in the same path, and a cross-path filter element to
filter either the I signal or the Q signal received in the other
path.
10. The circuit of claim 9, the IQmc mismatch corrector further
including: an I-path delay element included in the I-path to
provide a delay to the I signal corresponding to a delay through
either the in-path filter element or the cross-path filter element;
and a Q-path delay element included in the Q-path to provide a
delay to the Q signal corresponding to a delay through either the
in-path filter element or the cross-path filter element.
11. The circuit of claim 9, wherein the cross-path correction
filter is connected from the Q-path to the I-path; and the in-path
correction filter is connected in the Q-path; so that, the Q signal
input is applied to the in-path and cross-path correction filters,
and IQmc correction is provided for the I-path (Icorr) and the
Q-path (Qcorr).
12. The circuit of claim 9, wherein the cross-path correction
filter is connected from the I-path to the Q-path; and the in-path
correction filter is connected in the I-path; so that, the I signal
input is applied to the in-path and cross-path correction filters,
and IQmc correction is provided for the I-path (Icorr) and the
Q-path (Qcorr).
13. The circuit of claim 9, wherein the cross-path correction
filter is connected from the I-path to the Q-path; and the in-path
correction filter is connected in the Q-path; so that, the I signal
input is applied to the cross-path correction filter, and the Q
signal input is applied to the in-path correction filter, and IQmc
correction is provided for the Q-path (Qcorr).
14. The circuit of claim 9, wherein the cross-path correction
filter is connected from the Q-path to the I-path; and the in-path
correction filter is connected in the I-path; so that, the Q signal
input is applied to the cross-path correction filter, and the I
signal is applied to the in-path correction filter, and IQmc
correction is provided for the I-path (Icorr).
15. The circuit of claim 19, the IQmc mismatch corrector to provide
IQmc correction that results in a change in linear response through
the analog and digital signal chains, the circuit further
comprising: in the analog signal chain, a digital step attenuator
(DSA) to provide a selected DSA step attenuation; in the digital
signal chain, a DSA gain/phase correction block to provide a DSA
correction signal to the DSA to correct the selected DSA step
attenuation, and a DSA gain/phase correction update module to
adjust the DSA correction signal to compensate for the change in
linear response.
16. The circuit of claim 9, wherein the in-path and cross-path
filter elements include filter coefficients selectively updated
based on selectively estimated IQmc mismatch impairment; and the
in-path and cross-path filter elements are implemented in hardware,
and the filter coefficients are selectively updated by computation
in firmware based on estimated IQmc mismatch impairment.
17. A method to provide IQ mismatch correction, for use in a system
for radio frequency (RF) communication, the method useable at a
transmit (TX) end, and/or at a receive (RX) end in which an analog
signal chain includes analog circuitry that introduces IQ mismatch
signal impairments, the method comprising: digital filtering, in an
IQmc corrector, In-phase and Quadrature digital signals to provide
IQmc correction to correct the IQ mismatch impairments; the digital
filtering for IQmc correction accomplished by: receiving, in an
I-path, the In-phase digital signals (I signal), and receiving, in
a Q-path, the Quadrature digital signals (Q signal), and filtering
in an in-path filter element either the I signal or the Q signal
received in the same path, and filtering in a cross-path filter
element either the I signal or the Q signal received in the other
path.
18. The method of claim 17, the IQmc mismatch corrector further
comprising: introducing a delay in the I-signal, with an I-path
delay element included in the I-path, corresponding to a delay
through either the in-path filter element or the cross-path filter
element; and introducing a delay in the Q-signal, with a delay
element included in the Q-path, corresponding to a delay through
either the in-path filter element or the cross-path filter
element.
19. The method of claim 17, wherein the IQmc mismatch corrector is
configured with the in-path filter element and the cross-path
filter element arranged so that either: the IQmc correction is
provided for the I-path (Icorr) and the Q-path (Qcorr); or the IQmc
correction is provided for one of the I-path (Icorr) or the Q-path
(Qcorr).
20. The method of claim 17, wherein the analog signal chain
includes a digital step attenuator (DSA) to provide a selected DSA
step attenuation, the method further comprising; determining a
change in linear response through the analog and digital signal
chains that results from the digital filtering for IQmc correction;
and providing a correction signal to the DSA to adjust the selected
DSA step attenuation to compensate for the change in linear
response.
21. The method of claim 17, wherein: the in-path and cross-path
filter elements include filter coefficients selectively updated
based on selectively estimated IQmc mismatch impairment; and the
in-path and cross-path filter elements are implemented in hardware,
and the filter coefficients are selectively updated by computation
in firmware based on estimated IQmc mismatch impairment.
Description
CROSS-REFERENCE TO RELATED APPLICATIONS
[0001] Priority is claimed under 37 CFR 1.78 and 35 USC 119(e) to
INDIA Provisional Application 2018/41012763 (Docket TI-78469P5),
filed 2018 Apr. 4, which is incorporated by reference in its
entirety.
BACKGROUND
[0002] Wireless infrastructure can employ zero-IF and
low-intermediate frequency (zero/low-IF) architectures for TX and
RX. An analog signal chain provides RF TX/RX, and a digital signal
chain operates at baseband, commonly with DUC/DDC (digital
upconversion/downconversion). The analog chain implements complex,
quadrature (IQ) modulation/demodulation. The analog and digital
signal chains are interfaced with a TX DAC or RX ADC.
[0003] The analog IQ chain includes a complex IQ modulator (TX), or
demodulator (RX). Mismatches between the I and Q signal chains (IQ
errors) include frequency independent gain/phase mismatch in the
local oscillators, and frequency dependent mismatch errors in
mixers and filter transfer functions, and delay errors through the
I and Q signal paths. These I/Q mismatch errors lead to side-band
leakage (I/Q imbalance).
[0004] TX/RX IQ mismatch in the analog chain can be corrected in
the digital chain with digital TX/RX IQmc (IQ mismatch
correction/compensation): at the TX end, the digital TX IQmc
corrector essentially pre-distorts the baseband signal to
compensate for IQ mismatch in the downstream analog IQ chain; at
the RX end, the digital RX IQmc corrector corrects for IQ mismatch
introduced by the upstream analog IQ chain.
[0005] IQmc corrector architectures commonly use filters to correct
for IQ mismatch errors. The filter is a construct of a specified
number of filter taps, with selectively assigned (estimated)
coefficients, and associated MAC (multiply-accumulate) elements,
and delay (Z transform) elements.
BRIEF SUMMARY
[0006] This Brief Summary is provided as a general introduction to
the Disclosure provided by the Detailed Description and Drawings,
summarizing aspects and features of the Disclosure. It is not a
complete overview of the Disclosure, and should not be interpreted
as identifying key elements or features of, or otherwise
characterizing or delimiting the scope of, the disclosed
invention.
[0007] The Disclosure describes apparatus and methods for IQ
mismatch correction based on a two filter architecture, such as for
use in an RF communications system with zero/low-IF TX or RX.
[0008] According to aspects of the Disclosure, a circuit to provide
IQ mismatch correction can be used in a system for radio frequency
(RF) communication including a transmit (TX) end, and/or a receive
(RX) end. At either the TX end or the RX end, an analog signal
chain includes analog circuitry that introduces IQ mismatch signal
impairments, and a digital signal chain includes an IQ mismatch
corrector to filter In-phase and Quadrature digital signals to
provide IQmc correction to correct the IQ mismatch impairments. The
IQ mismatch corrector, includes an I-path coupled to receive the
In-phase digital signals (I signal), and a Q-path coupled to
receive the Quadrature digital signals (Q signal). The IQmc
mismatch corrector further includes two filter elements: an in-path
filter element to filter either the I signal or the Q signal
received in the same path; and a cross-path filter element to
filter either the I signal or the Q signal received in the other
path. The IQmc mismatch corrector can include: an I-path delay
element included in the I-path to provide a delay to the I signal
corresponding to a delay through either the in-path filter element
or the cross-path filter element; and a Q-path delay element
included in the Q-path to provide a delay to the Q signal
corresponding to a delay through either the in-path filter element
or the cross-path filter element.
[0009] According to other aspects of the Disclosure, a transceiver
circuit for transmitting and receiving radio frequency (RF)
communication signals, includes a transmit (TX) end, and a receive
(RX) end. At the TX end and/or the RX end the circuit includes: an
analog signal chain with analog circuitry that introduces IQ
mismatch signal impairments; and a digital signal chain including
an IQmc mismatch corrector to filter In-phase and Quadrature
digital signals to provide IQmc correction to correct the IQ
mismatch impairments. The IQmc mismatch corrector includes: an
I-path coupled to receive the In-phase digital signals (I signal);
and a Q-path coupled to receive the Quadrature digital signals (Q
signal). the IQmc mismatch corrector further includes two filter
elements: an in-path filter element to filter either the I signal
or the Q signal received in the same path; and a cross-path filter
element to filter either the I signal or the Q signal received in
the other path. The IQmc mismatch corrector can include: an I-path
delay element included in the I-path to provide a delay to the I
signal corresponding to a delay through either the in-path filter
element or the cross-path filter element; and a Q-path delay
element included in the Q-path to provide a delay to the Q signal
corresponding to a delay through either the in-path filter element
or the cross-path filter element.
[0010] According to other aspects of the Disclosure, a method to
provide IQ mismatch correction, for use in a system for radio
frequency (RF) communication. The method is useable at a transmit
(TX) end, and/or at a receive (RX) end in which an analog signal
chain includes analog circuitry that introduces IQ mismatch signal
impairments. The method comprises digital filtering, in an IQmc
corrector, In-phase and Quadrature digital signals to provide IQmc
correction to correct the IQ mismatch impairments. The digital
filtering for IQmc correction is accomplished by: receiving, in an
I-path, the In-phase digital signals (I signal); and receiving, in
a Q-path, the Quadrature digital signals (Q signal); and filtering
in an in-path filter element either the I signal or the Q signal
received in the same path; and filtering in a cross-path filter
element either the I signal or the Q signal received in the other
path. The method can include: introducing a delay in the I-signal,
with an I-path delay element included in the I-path, corresponding
to a delay through either the in-path filter element or the
cross-path filter element; an introducing a delay in the Q-signal,
with a delay element included in the Q-path, corresponding to a
delay through either the in-path filter element or the cross-path
filter element.
[0011] Other aspects and features of the invention claimed in this
Patent Document will be apparent to those skilled in the art from
the following Disclosure.
BRIEF DESCRIPTION OF THE DRAWINGS
[0012] FIGS. 1A-1B illustrate example RF zero-IF receivers 10A/10B,
each including an analog signal chain 11 with a complex RF IQ
Demodulator 30, and a digital signal chain 12A/12B with an IQmc
Corrector 100, configured with an IQmc two-filter architecture
according to the Disclosure.
[0013] FIG. 2 provides example waveforms illustrating IQ mismatch,
including an IQ mismatch image signal 220.
[0014] FIGS. 3A and 3B illustrate an example implementation of an
IQmc corrector based on an IQmc two-filter architecture according
to the Disclosure, with in-path (g.sub.i) and cross-path (g.sub.q)
filter elements: FIG. 3A illustrates an example two-filter IQmc
corrector 300, with Q input to in-path and cross-path correction
filters 311/312, providing IQ mismatch correction for both paths
Icorr and Qcorr; and FIG. 3B illustrates an equivalent model for
the two-filter IQmc corrector architecture 300, including a complex
filter g.sub.i+jg.sub.q, with g.sub.i and g.sub.q filters
(g.sub.i+jg.sub.q) to correct IQ mismatch.
[0015] FIGS. 4A and 4B illustrates an example four-filter
architecture for an IQ mismatch corrector 400: FIG. 4A illustrates
and example (HW) implementation of a four-filter IQmc corrector
400; and FIG. 4B illustrates an equivalent model for the
four-filter IQmc corrector architecture 400; and.
[0016] FIGS. 5A-5C illustrate example alternate implementations of
an IQmc corrector based on an IQmc two-filter architecture
according to the Disclosure, with in-path (g.sub.i) and cross-path
(g.sub.q) filter elements: FIG. 5A illustrates an alternate IQmc
corrector 510 (symmetrical with the IQmc corrector 300 in FIG. 3A),
with I input to in-path and cross-path correction filters 511/512,
and providing IQmc correction in both paths Icorr and Qcorr; FIG.
5B illustrates an alternate IQ corrector 520 with I and Q input to
in-path and cross-path correction filters 521/522, but with IQmc
correction applied only to the Q-path Qcorr; and FIG. 5C
illustrates an alternate IQ corrector 530 (symmetrical with the
IQmc corrector 520 in FIG. 5B), with I and Q input to in-path and
cross-path correction filters 531/532, and IQmc correction applied
only to the I-path Icorr.
DETAILED DESCRIPTION
[0017] This Description and the Drawings constitute a Disclosure,
including design examples and implementations, and including
illustrating various technical features and advantages for: IQ
mismatch correction based on a two-filter architecture, with
in-path (g.sub.i) and cross-path (g.sub.q) filter elements.
[0018] This Disclosure uses the following nomenclature. IQmc (IQ
mismatch correction/compensation) means digital
correction/compensation for IQ mismatch (imbalance) introduced in
an analog IQ chain, RX or TX: in the TX chain, the TX IQmc
corrector essentially pre-distorts the baseband signal to
compensate for IQ mismatch in the downstream analog IQ chain; and
in the RX chain, the RX IQmc corrector compensates for IQ mismatch
in the upstream analog IQ chain.
[0019] The IQmc corrector filters can be implemented in hardware or
firmware, using coefficients that can be updated using on-line or
off-line estimation. For the filters the HW/FW trade-off is
typically based on operating frequency. For the example, IQmc
correctors in this Disclosure, the IQmc filters are implemented in
hardware, and updated with coefficients estimated in firmware.
[0020] The Disclosed example application for the IQmc two-filter
architecture is an RF zero-IF receiver. The IQmc two-filter
architecture according to the Disclosure is also applicable to TX
IQmc correction, and to low-IF RX/TX architectures.
[0021] In brief overview, IQ mismatch correction for analog chain
IQ mismatch impairments is based on a two-filter architecture. In
either RX or TX, an IQmc mismatch corrector (digital chain) filters
I and Q digital signals, and includes an I-path to receive the I
signal, and a Q-path to receive the Q signal, and is configured
with two filters: an in-path filter to filter either the I signal
or the Q signal received in the same path; and a cross-path filter
to filter either the I signal or the Q signal received in the other
path. The IQmc mismatch corrector can include: an I-path delay
element to provide a delay to the I signal corresponding to a delay
through either the in-path filter or the cross-path filter; and a
Q-path delay element to provide a delay to the Q signal
corresponding to a delay through either the in-path filter or the
cross-path filter.
[0022] FIGS. 1A-1B illustrate example RF zero-IF receivers 10A/10B.
The RX chain includes an analog receiver chain 11 with IQ
demodulation, and digital IQ chains 12A/12B. The receivers include
analog/digital IQ chains 15A/15B.
[0023] The analog chains 11 are identical, and include an RF input
stage with LNA 22 and DSA 24, providing RF amplification and
digital step attenuation, with RF input to a quadrature (IQ)
Demodulator 30, driven by a local oscillator LO. The IQ Demodulator
30 and Filter 40 provide I and Q baseband signal inputs to a
delta-sigma ADC 50, for conversion to digital I and Q streams input
to the digital signal chain.
[0024] The digital chains 12A/12B each include a Decimation Filter
60, followed by an IQmc Corrector 100. IQmc Corrector 100 is
implemented with a IQmc two-filter architecture according to the
Disclosure.
[0025] IQmc Corrector 100 is followed by DSA Gain/Phase Corrector
70. As described below, the IQmc correction can affect linear
response dependent upon the IQmc filter response. In FIG. 1B, the
digital chain 15B includes an additional Gain/Phase Corrector
Update module 75 to update the DSA Gain/Phase Corrector 70 to
compensate for changes in linear response resulting from changes in
IQmc correction.
[0026] Gain, phase, delay mismatch in the IQ Demodulator 30
(including the LO), and/or baseband Filters 40 operating on IQ
signals results in IQ mismatch errors (imbalance or leakage), which
limits SFDR (Spurious Free Dynamic Range) unless corrected.
[0027] Gain/phase mismatch in the LO results in
frequency-independent IQ mismatch. Mismatch in IQ chain components,
filter transfer functions, and LO delay, results in
frequency-dependent IQ mismatch.
[0028] FIG. 2 provides example waveforms illustrating IQ mismatch,
including an example RX signal 210, and its IQ mismatch image
signal 220. Frequency dependent IQ mismatch generates images at a
negative frequency reflected about DC. For example, an input tone
at frequency f0 results in a IQ mismatch image at -f0.
X(f)=X.sub.orig(f)+H(f)X.sub.orig*(-f), where Xorig(f) is the
baseband equivalent input to the system, and X(f) is the output of
the analog chain affected by IQ mismatch. H(f) is the IQ mismatch
of the analog chain, which can be estimated by IQ mismatch
estimator logic (on-line or off-line). The time domain equivalent
of H(f), h(n), can also be computed by an IQ mismatch estimator, to
enable IQ mismatch filter coefficient update. In general, h(n) is a
complex filter, because the IQ mismatch H(f) is in general not
conjugate symmetric about 0 Hz.
[0029] FIGS. 3A and 3B illustrate an example implementation of an
IQmc corrector based on an IQmc two-filter architecture according
to the Disclosure. The IQmc two-filter architecture includes
filters g.sub.i and g.sub.q (g.sub.i+jg.sub.q), and specifically,
in-path (g.sub.i) and cross-path (g.sub.q) filter elements,
configured to correct IQ mismatch.
[0030] FIG. 3A illustrates an example IQmc corrector 300 with
Q-path input to in-path and cross-path correction filters 311/312
(g.sub.i and g.sub.q). This example two-filter IQmc corrector
architecture implements IQmc correction in both I and Q paths,
outputting Icorr and Qcorr. The IQmc filter coefficients g(n), can
be determined from G(f), which can be computed from H(f) according
to:
G ( f ) = 2 H ( f ) 1 - H ( f ) ##EQU00001##
where H(f) is the IQ mismatch. The example IQ mismatch filters are
implemented in hardware, with filter computation/update implemented
in firmware. The output of the IQmc corrector is:
X.sub.corr(f).apprxeq.X.sub.orig(f)(1+G(f)/2)
[0031] FIG. 3B illustrates an equivalent model for the IQmc
corrector 300 including a complex filter (g=g.sub.i+jg.sub.q). The
IQmc corrector effectively removes the IQ mismatch seen in the
analog chain output due to its structure.
[0032] To reduce complexity of the computation of G(f) from H(f),
an approximate filter can be used based on
G(f).apprxeq.2H(f)(1+H(f)), or for further simplification,
G(f).about.=2H(f). Note that these relations in frequency domain
can also be implemented directly in time domain if h(n) (the IQ
mismatch estimate in time domain) is directly available. For
example, G(f).apprxeq.2H(f)(1+H(f)) can be implemented as
g(n).apprxeq.2h(n)+2h(n)*h(n), where * represents a convolution
operation.
[0033] For comparison, FIGS. 4A and 4B illustrates an example
four-filter architecture for an IQ mismatch corrector 400.
[0034] FIG. 4A illustrates an example implementation of a
four-filter IQmc corrector 400, using four real filters (two
in-path, and two cross-path). Each filter commonly uses a multi-tap
FIR with tens of taps running at hundreds of MHz in RX, and at even
higher rates in wideband TX. FIG. 4B illustrates an equivalent
model for the four-filter IQmc corrector architecture 400
[0035] The four filter architecture can be reconfigured as a three
filter architecture, optimizing a complex multiplication to use
three multiplies instead of four, with an extra addition. The
higher complexity of such a 4 or 3 filter architecture is due to
the number of coefficients running at high data rates.
[0036] Referring back to FIG. 3A, the Disclosed IQmc corrector
architecture uses only two real filters, one in-path and one
cross-path, while maintaining IQmc correction. The disclosed IQmc
two-filter correction architecture enables filter design area/power
that is 2/3 of a 3 filter structure, or 2/4 of a 4 filter
structure.
[0037] FIGS. 5A-5C illustrate example alternate implementations of
an IQmc corrector based on a two-filter architecture according to
the Disclosure, with different arrangements of in-path (g.sub.i)
and cross-path (g.sub.q) filter elements, and IQmc correction.
[0038] FIG. 5A illustrates an alternate example two-filter IQmc
corrector configuration 510, with a configuration symmetrical to
the two-filter IQmc corrector 300 of FIG. 3A. IQmc corrector 510
includes I-path input to in-path and cross-path correction filters
511/512 (g.sub.i and g.sub.q), and provides IQmc correction to both
paths Icorr and Qcorr. Note, however, that the filter coefficients
(G(f)) for the IQmc corrector 510 of FIG. 5A are not the same as
the filter coefficients for FIG. 3A. The filter coefficients g(n)
for the IQmc corrector architecture 5A can be computed based
on:
G ( f ) = - 2 H * ( - f ) 1 + H * ( - f ) ##EQU00002##
which is similar to the equation for the two-filter IQmc corrector
300 in FIG. 3A.
[0039] FIGS. 5B-5C illustrate alternate two-filter IQmc corrector
configurations with both I-path and Q-path inputs through in-path
and cross-path filter elements, but with IQmc correction applied in
only one path (either Qcorr or Icorr).
[0040] FIG. 5B illustrates an alternate IQ corrector 520 with
I-path and Q-path input to cross-path and in-path correction
filters 522/521 (g.sub.i and g.sub.q), providing IQmc correction on
the Q path Qcorr.
[0041] The IQmc filter coefficients can be determined according
to:
G ( f ) = 2 H ( f ) 1 + H * ( - f ) 1 - H ( f ) H * ( - f )
##EQU00003##
where H(f) is the IQ mismatch estimate. That is, the IQmc filter
coefficients G(f) are computed from IQmc mismatch estimates H(f)
and H*(-f), with H* representing the conjugate of the IQ mismatch
estimate.
[0042] IQmc filter computation can be solved if mismatch estimates
H(f) are available for both f.sub.0 and -f.sub.0. If not,
assumptions can be made about the missing frequency to enable
filter coefficient computation. For example, it can be assumed that
the missing estimate is the same as a nearby estimate, or if no
nearby estimates are present, then it can be assumed that
H(f.sub.0)=H*(-f.sub.0) (i.e., estimates are assumed to be
conjugate symmetric). Or, if H(f) is small, higher order terms for
computing G can be neglected. The output of the IQmc corrector
is:
X.sub.corr(f).apprxeq.X.sub.orig(f)(1+G*(-f)/2)
[0043] FIG. 5C illustrates an example alternate IQmc corrector 530,
that is symmetrical with the IQmc corrector 520 in FIG. 5B. IQmc
corrector 530 uses I and Q input to in-path and cross-path
correction filters 531/532 (g.sub.i and g.sub.q), providing IQmc
correction on the I path Icorr.
[0044] IQmc filter updates can cause changes in linear response of
the signal based on the level of the IQmc correction (IQ
imbalance). If IQmc correction is fixed, then the impact of IQmc
correction on linear response of the signal is also fixed, and can
be absorbed into the channel estimation, and therefore causes no
decoding errors. However, a change in IQmc correction can cause a
gain-step error (DSA step attenuation error) in the signal
gain/phase, which can cause decoding errors. Such a change in IQmc
correction can occur when IQ mismatch H(f) changes over time due
to, for example, temperature changes.
[0045] For two-filter IQmc correction architectures according to
the Disclosure, the impact on linear response can be on the order
of the correction filter level, but the exact form varies with the
type of two-filter architecture. For example, the two-filter
architecture of FIG. 3A has an output which is
Xcorr(f)=Xorig(f)(1+G(f)/2). But the two-filter architecture of
FIG. 5B (and 5C) has an output which is
Xcorr(f)=Xorig(f)(1+G*(-f)/2). If the G(f) is not changing
significantly with time, then this is equivalent to a small change
in the linear response which can be estimated and corrected by the
baseband receiver as part of channel estimation. A significant G(f)
change, as can happen with periodic coefficient updates that can
cause a step change in correction (i.e., a step-change coefficient
update), can cause a step change in the linear response that can be
difficult for the baseband receiver to adjust to. IQmc filter
update affects the linear response by an amount dependent on the
IQmc filter response.
[0046] Gain-step errors, such as resulting from changes in IQmc
filter update, can be counteracted by updating DSA gain/phase error
correction based on the linear error introduced by IQmc filter
update, without affecting overall linear response. The DSA
gain/phase error correction module corrects a gain and phase error
introduced in the DSA. If a is the gain mismatch and .theta. is the
phase mismatch introduced, then the analog signal is multiplied by
(1+.alpha.)e.sup.j.theta., where .alpha. and .theta. are typically
small and are also dependent on the current DSA setting used. To
correct this the DSA gain phase corrector module multiplies the
signal by (1-.alpha.)e.sup.-j.theta. to remove this error.
[0047] Referring to FIG. 1B, the digital chain 15B includes a DSA
Gain/Phase Correction block 70 to correct DSA settings for
gain-step errors and phase changes with DSA settings. The digital
chain 15B includes an additional Gain/Phase Correction Update
module 75 to update the DSA Gain/Phase Correction block 70 to
compensate for the change in linear response according to aspects
of the Disclosure.
[0048] For the example two-filter IQmc corrector architecture of
FIG. 3A, the example Gain/Phase Correction Update module 75 updates
the DSA Gain/Phase Corrector 70 to compensate for the change to the
linear response resulting from a change/update in IQmc filtering.
For example, the center tap of the IQmc filter, g(d)/2, can be
added to the DSA Gain/Phase Corrector 70. In other words, the
correction can be modified as
(1-.alpha.-g.sub.ig(d)/2)e.sup.-j(.theta.+g.sup.Q.sup.(d)/2), where
g.sub.ig(d) and g.sub.Q(d) are the real and imaginary parts of
center tap g(d). If DSA Gain/Phase Corrector 70 is multi-tap, then
g(d)/2 for the sub-set of available taps can be provided as the
update to remove G(f)/2. This update to DSA gain/phase correction
can be computed in firmware whenever the IQ mismatch estimation is
updated by the IQ Imbalance Estimator 101.
[0049] An IQmc Imbalance Estimator 101 estimates the latest IQ
mismatch estimate H(f), and sends it to the Corrector Filter
Generator block 105. The Corrector Filter Generator 105 computes
IQmc filter coefficients (g) from mismatch estimates H(f).
Gain/Phase Correction Update module 75 computes the update needed
for the DSA Gain/Phase Corrector 75, and also updates the DSA
gain/phase coefficients. This update is applied with a delay to the
DSA Gain/Phase Corrector 70 so that the samples for which the IQmc
filter coefficients are updated, are also used to provide modified
DSA gain/phase compensation.
[0050] The Disclosure provided by this Description and the Figures
sets forth example designs and applications illustrating aspects
and features of the invention, and does not limit the scope of the
invention, which is defined by the claims. Known circuits,
connections, functions and operations are not described in detail
to avoid obscuring the principles and features of the Disclosed
example designs and applications. This Disclosure can be used by
ordinarily skilled artisans as a basis for modifications,
substitutions and alternatives, including adaptations for other
applications.
* * * * *