TIME-INTERLEAVED ANALOG-TO-DIGITAL CONVERTER SYSTEM

• US 20100049483A1
• Filed: 06/30/2006
• Published: 02/25/2010
• Est. Priority Date: 06/30/2006
• Status: Active Grant

• Current Assignee: Teledyne Signal Processing Devices Sweden AB (Teledyne Technologies Incorporated)
• Original Assignee: Signal Processing Devices Sweden AB (Teledyne Technologies Incorporated)
• Inventors: Johansson, Håkan, Löwenborg, Per
• Granted Patent: US 8,307,248 B2
• Time in Patent Office: Days
• US Class Current: 703/2
• CPC Class Codes: H03M 1/1009 Calibration
• CPC Class Codes: H03M 1/1215 using time-division multiplexing

Abstract

A method and a module for estimating a plurality of relative channel-error for at least one signal with respect to a reference signal. The signals and are produced by an analog-to-digital module including parallel and time interleaved analog-to-digital converters and are received by an estimation module. The method is performed by the estimation module and includes defining a function representing a relationship between the reference signal and an arbitrary signal in the group of signals, selecting a first reference signal in the group of signals, selecting a second signal from the remaining signals in the group, optimizing the function so as to obtain an estimate of the plurality of relative channel-error, and repeating the selecting a second signal and optimizing the function for each remaining signal.

Claims

1. A method for estimating a plurality of relative channel-errors (d_k, G_k,C_k) for at least one signal (X_k) with respect to a reference signal (X₀), which signals (X₀, X_k) belong to a group of digitized signals (X₀-X_N-1) produced by an analog-to-digital module comprising a plurality of parallel and time interleaved analog-to-digital converters and received by an estimation module;

wherein said method is performed by said estimation module, the method comprising;

defining a function (F(d_k,G_k,C_k)) representing a relationship between said reference signal (X₀) and an arbitrary signal (X_k) in said group of signals (X₀-X_N-1);

selecting a first reference signal (X₀) in said group of signals (X₀-X_N-1);

selecting a second signal (X_k) from the remaining signals (X₁-X_N-1) in said group;

optimizing the function (F(d_k,G_k,C_k)) so as to obtain an estimate ({circumflex over (d)}_k, Ĝ

_k, Ĉ

_k) or a representation thereof of said plurality of relative channel-errors (d_k,G_k,C_k); and

repeating selecting a second signal and optimizing the function for each remaining signal (X₁-X_N-1).

2. The method according to claim 1, wherein the optimizing is preformed by an iterative process comprising:

feedback of the channel-error or channel-errors estimated in the previous iteration so as to adjust the channel-error in the sequence of the second signal (X_k); and

updating the function (F(d_k,G_k,C_k)) with a sequence of the second signal (X_k) having channel-errors adjusted by said feedback; and

repeating said feedback and said update until a satisfactory estimation is obtained.

3. The method according to claim 1, wherein said plurality of estimated channel-errors ({circumflex over (d)}_k, Ĝ

_k, Ĉ

_k) is calculated by defining said function (F(d_k,G_k,C_k)) as a measure of the difference between said sequences with respect to said plurality of channel-errors (d_k,G_k,C_k), and by performing an optimization by minimizing the defined function (F(d_k,G_k,C_k)).

4. The method according to claim 2, wherein at least one of the estimated channel-errors ({circumflex over (d)}_k,Ĝ

_k, Ĉ

_k) is an estimated delay ({circumflex over (d)}_k) corresponding to a delay mismatch (d_k), which estimated delay ({circumflex over (d)}_k) is applied to the selected signal (X_k) using a fractional delay filter.

5. The method according to claim 4, wherein said fractional delay filter is implemented using a Lagrange, Farrow Structure or Thiran Allpass.

6. The method according to claim 1, wherein the function is defined as

$F$

( d k , G k , C k ) = ∑

n = n 0 n 0 + N - 1 

( y k 

( n , d k , G k , C k ) - x 0 

( n ) ) 2 .

7. An estimation module for estimating a plurality of relative channel-errors (d_k,G_k, C_k) or a representation thereof for at least one signal (X_k) with respect to a reference signal (X₀), which signals (X₀, X_k) belong to a group of digitized signals (X₀-X_N-1) produced by an analog-to-digital module comprising a plurality of parallel and time interleaved analog-to-digital converters, where said estimation module is configured to operatively receive said digitized signals (X₀-X_N-1) and to operatively select a first reference signal (X₀) in said group of signals (X₀-X_N-1), and wherein the estimation module is further configured to operatively:

select a second signal (X_k) from the remaining signals (X₁-X_N-1) in said group of signals;

optimize a function (F(d_k,G_k,C_k)), representing a relationship between the first signal (X₀) and the second signal (X_k) so as to obtain an estimate ({circumflex over (d)}_k, Ĝ

_k, Ĉ

_k) or a representation thereof of said plurality of relative channel-errors (d_k,G_k,C_k); and

repeat said selection and optimizing for each remaining signal (X₁-X_N-1) in said group of signals.

8. The estimation module according to claim 7, wherein the module performs said optimizing by being further configured to operatively:

feedback the channel-error or channel-errors estimated in the previous iteration so as to adjust the channel-error in the sequence of the second signal (X_k); and

update the function (F(d_k,G_k,C_k)) with a sequence of the second signal (X_k) having channel-errors adjusted by said feedback; and

repeat said feedback and said update until a satisfactory estimation is obtained.

9. The estimation module according to claim 7, wherein the estimation module is further configured

to operatively calculate said plurality of estimated channel-errors ({circumflex over (d)}_k, Ĝ

_k, Ĉ

_k) by utilizing a function (F(d_k,G_k,C_k)) defining a measure of the difference between said sequences with respect to said plurality of channel-errors (d_k,G_k,C_k), and by performing an optimization by minimizing the defined function (F(d_k,G_k,C_k)).

10. The estimation module according to claim 9, wherein the estimation module is further configured

to operatively estimate a plurality of channel-errors, wherein least one of the estimated channel-errors ({circumflex over (d)}_k,Ĝ

_k, Ĉ

_k) is an estimated delay ({circumflex over (d)}_k) corresponding to a delay mismatch (d_k), and to apply the estimated delay ({circumflex over (d)}_k) to the selected signal (X_k) using a fractional delay filter.

11. The estimation module according to claim 10, wherein said fractional delay filter is implemented by using a Lagrange, Farrow Structure or Thiran Allpass.

12. The estimation module according to claim 7, wherein the estimation module is further configured to operatively use the function:

$F<br><br>\left(d_k,G_k,C_k\right)=\underset{n=n_0}{\overset{n_0+N-1}{\sum <br><br>}}<br><br>{\left(y_k<br><br>\left(n,d_k,G_k,C_k\right)-x_0<br><br>\left(n\right)\right)}^2.$

13. A system for reducing or eliminating a plurality of relative channel-errors (d_k,G_k,C_k) for at least one signal (X_k) with respect to a reference signal (X₀), which signals (X₀, X_k) belong to a group of digitized signals (X₀-X_N-1) produced by an analog-to-digital module comprising a plurality of parallel and time interleaved analog-to-digital converters, the system comprising:

an estimation module configured to operatively receive said digitized signals (X₀-X_N-1) and to estimate a plurality of relative channel-errors (d_k, G_k, C_k) or a representation thereof, and

a reconstruction module,

being arranged to operatively apply said plurality of relative channel-errors (d_k,G_k,C_k) or a representation thereof to said selected signal (X_k), belonging to a group of digitized signals (X₀-X_N-1) produced by the time interleaved analog-to-digital module, so as to reduce or eliminate the relative channel-error (d_k,G_k,C_k).