(19)
(11)EP 1 330 658 B1

(12)EUROPEAN PATENT SPECIFICATION

(45)Mention of the grant of the patent:
21.12.2005 Bulletin 2005/51

(21)Application number: 01975435.7

(22)Date of filing:  25.09.2001
(51)International Patent Classification (IPC)7G01R 31/00
(86)International application number:
PCT/US2001/030158
(87)International publication number:
WO 2002/027336 (04.04.2002 Gazette  2002/12)

(54)

BIST METHOD FOR TESTING CUT-OFF FREQUENCY OF LOW-PASS FILTERS

BIST VERFAHREN ZUR PRUFUNG DER GRENZFREQUENZ EINES TIEFPASSFILTERS

PROCEDE BIST DE VERIFICATION DE LA FREQUENCE DE COUPURE DE FILTRES PASSE-BAS


(84)Designated Contracting States:
AT BE CH CY DE DK ES FI FR GB GR IE IT LI LU MC NL PT SE TR

(30)Priority: 26.09.2000 US 670306

(43)Date of publication of application:
30.07.2003 Bulletin 2003/31

(73)Proprietor: INTEL CORPORATION
Santa Clara, CA 95052 (US)

(72)Inventors:
  • SUSUMA, Hara
    Gold River, CA 95670 (US)
  • IHS, Hassan
    Sacramento, CA 95827 (US)

(74)Representative: Molyneaux, Martyn William et al
Harrison Goddard Foote 40-43 Chancery Lane
London WC2A 1JA
London WC2A 1JA (GB)


(56)References cited: : 
US-A- 5 625 317
US-A- 5 748 001
  
  • PATENT ABSTRACTS OF JAPAN vol. 1999, no. 12, 29 October 1999 (1999-10-29) -& JP 11 177379 A (NEC CORP), 2 July 1999 (1999-07-02) -& US 6 366 161 B1 (KOAZECHI SHINICHI) 2 April 2002 (2002-04-02)
  
Note: Within nine months from the publication of the mention of the grant of the European patent, any person may give notice to the European Patent Office of opposition to the European patent granted. Notice of opposition shall be filed in a written reasoned statement. It shall not be deemed to have been filed until the opposition fee has been paid. (Art. 99(1) European Patent Convention).


Description

BACKGROUND OF THE INVENTION


Field of the Invention



[0001] The present invention generally concerns measuring circuit characteristics of filters, and in more particular concerns built-in testing schemes for measuring the cut-off frequency of low-pass filters.

Background Information



[0002] Low pass filters are used in many applications where out of band components of a signal or noise have to be rejected. In many instances, they are implemented in mixed signal integrated circuits (ICs); such ICs are used extensively in telecommunications for pulse templating. For example, the transmitter in typical HDSL transceivers comprises a sigma delta digital-to-analog modulator followed by a large switched capacitor noise-shaping filter. In many cases, the bandwidth of the filter has to be measured to comply with template specifications. This requires the cut-off frequency of the filter to be measured.

[0003] Testing mixed signal ICs is known to be a difficult and time consuming task. A conventional test that is often used for testing the cut-off frequency of filters in mixed IC devices uses a multi-tone signal with a tone at the cut-off frequency and other tones at each side of cut-off frequency. During testing, the output signal of the device is digitized, and data processing is performed to extract the gains at each tone. In many cases, to ensure the compliance with different standards, the filter bandwidth of such ICs is programmable. Accordingly, when using the foregoing conventional testing method, a separate test must be performed for each bandwidth setting, leading to long production test times and expense. In addition, conventional test schemes often require connecting the circuit under test (CUT) to complex and expensive mixed-signal automated test equipment.

[0004] PATENT ABSTRACTS OF JAPAN voL 1999, no. 12,29 October 1999 (1999-10-29) -& JP 11 177379 A, 2 July 1999 (1999-07-02) -& US 6 366 161 B1 2 April 2002 (2002-04-02) discloses a technique for measuring and adjusting the cut-off frequency of an active filter circuit using on-chip circuitry. A conventional active filter circuit composed of adjustable resistors and/or capacitors and an operational amplifier is formed in a semicondutor substrate. The circuitry also includes mutiple switching elements connected between the filter circuit elements in a that enables the circuitry to be configured as two alternative circuits (one at a time): a low-pass filter or a Wien-Bridge oscillator. The circuit elements further include a counter and a decoder, which are coupled in series to the output of the operational amplifier. To set the cut-off frequency, the circuit is configured in the Wien-Bridge oscillator configuration by setting the switches in appropriate positions. The frequency of oscillation of the Wien-Bridge oscillator corresponds to the cut-off frequency of the low-pass filter when the same circuit elements are reconfigured via the switches to form the low-pass filter. The oscillation frequency is measured with the counter, which produces a count value that is fed into the decoder. The decoder compares the fount value with values stored in a lookup table, and adjusts one or more of the adjustable resistors and/or capacitors to change their values. This, in turn, changes the oscillation frequency. The process is repeated until the oscillation frequency meets the design frequency requirements. At this point, the switches are set to reconfigure the circuit to function as the low-pass filter.

[0005] The foregoing technique has some significant limitations. Firstly, the circuit elements must be configured so as to enable the formation of an active filter and a Wien-Bridge oscillator via the various switches. This limits the active filter circuit complexity, and thus filter characteristics. Secondly, the circuitry includes multiple analog switches, which increases circuit complexity and may degrade filter performance. Thirdly, the resistive and/or capacitive elements are adjustable, increasing the circuit complexity and cost.

[0006] Ideally, it would be desirable to provide a test scheme for measuring the cut-off frequency of a low pass filter that is both fast and simple. Additionally, such a scheme should be implemented on-chip such that only a limited amount of automated testing equipment (ATE) is required, preferably requiring only digital ATE.

SUMMARY OF THE INVENTION



[0007] According to a first aspect of this invention there is provided a method as claimed in claim 1 herein.

[0008] According to a second aspect of this invention there is provided a mixed-signal integrated circuit as claimed in claim 6 herein.

[0009] Preferred features of the invention are defined by the dependent claims.

BRIEF DESCRIPTION OF THE DRAWINGS



[0010] The foregoing aspects and many of the attendant advantages of this invention will become more readily appreciated as the same becomes better understood by reference to the following detailed description, when taken in conjunction with the accompanying drawings, wherein:

Figure 1 is a block schematic diagram of exemplary implementation of the present invention; and

Figure 2 is a block schematic diagram of an alternate implementation of the present invention.


DETAILED DESCRIPTION



[0011] The present invention comprises a method and circuitry for implementing a built-in self test (BIST) for determining the cut-off frequency of low-pass filters. The method comprises inserting a Circuit Under Test (CUT) into a feedback loop that looks like a sigma delta modulation loop and adjust the feedback loop so that it oscillates at the cut-off frequency of the filter. The frequency of oscillation can then easily be measured using either on on-chip counter or digital automated testing equipment.

[0012] A block diagram 10 corresponding to an exemplary implementation of such a feedback loop is depicted in FIGURE 1, comprising a CUT 12, a comparator 14, which can be as simple as a CMOS inverter, a digital delay line 16 that delays the comparator output signal with delay units corresponding to a clock period produced by a CLK signal 18, and a one-bit Digital-to-Analog Converter (DAC) 20 that limits the signal not to exceed the input voltage range of the CUT. The output from one-bit DAC 20 is fed back into the loop at a summing block 22, which also receives a X(z) input signal 24 that preferably should be zero (i.e. the middle range voltage in between the two voltage levels of one-bit DAC 20). Note that in FIGURE 2, this input signal is omitted, which is equivalent to a zero input. The feedback loop is designed so as to make the whole loop oscillate at the same frequency as the cut-off frequency of the CUT. The oscillation frequency or period constitutes the CUT test signature, which preferably is measured using an on-chip counter 26. Optionally, the oscillation frequency can be measured by external digital ATE device. Note that the CUT may be a discrete time circuit, as well as a continuous time circuit.

[0013] An optional configuration 10' for the circuit is shown in FIGURE 2, wherein components with like-named references perform similar functions to those shown in FIGURE 1 and discussed above. In configuration 10', the output of one-bit DAC 20 is simply feed directly into CUT 12 to form the feedback loop. It is noted that care must be taken when implementing this configuration so that a positive feedback is provided, otherwise the loop may become unstable.

[0014] The condition of oscillation in any linear circuit is given by Barkausen criterion:



   wherein H is the gain of the CUT, F, the gain of the feedback loop, and ϕHF the total phase of the loop. For the case of the Figure-1 scheme, the loop is non-linear, and the Barkausen criterion can no longer be applied directly. However, as with sigma delta modulation, the comparator acts as a voltage-controlled gain circuit that adapts the gain of the loop such that it is maintained at unity at all times. Although no well-established theory expressing the gain of the comparator exists, the Barkausen condition on gain can be written using a conceptual representation of the comparator gain, as follows:

wherein Gcomp is the voltage dependant gain of the comparator. The comparator (a sine function) can be seen as a square wave regenerative circuit. This means that, in an open loop, applying a square wave at the CUT input will result in a distorted signal at its output, but regenerated with a phase shift at the comparator output. Since the comparator regenerates the input square wave, the condition on the gain is always satisfied. As a result, the oscillation of the loop depends only on the Barkausen condition on the phase.

[0015] Assume F(z) is a constant. To satisfy the Barkausen criterion on phase, the phase in the whole loop should be -2π. As the signal inversion after the output of one-bit DAC 20 introduces -π phase shift, the CUT is required to contribute -π phase. Since every pole of the CUT is contributing a phase maximum of -π/2, the oscillation frequency should be larger than the second pole of the CUT, which in turn is larger than its cut-off frequency. The role of the F(z) delay line is to add more phase shift into the loop so as to adjust the oscillation frequency to the cut-off frequency. As the loop oscillates at the attenuated region of the CUT transfer function, the odd harmonics of the square wave present at its input are filtered out. This means that the effect of signal non-linearity on the phase can be neglected. The Barkausen condition on the phase can then be written:

wherein ϕHF is the phase transfer function of the CUT, T is the clock frequency, and nT is the delay in the delay line. The role of this equation is to extract the number of delay units n in the delay line that will make the loop oscillate at the CUT's cut-off frequency. For large discrete filters, this equation may be difficult to solve analytically. Simulation using a simple behavioral modeling of the CUT allows to tune the oscillation frequency. Generally, it takes no more than 10 delay units to set the oscillation frequency to the desired value.

[0016] The method works in the following manner. The delay line is adjusted until either a predetermined frequency of oscillation should be produced or a frequency of oscillation corresponding to the cut-off frequency of the CUT is reached through tuning the phase delay of the feedback loop. In the both cases, the on-chip counter can be used to determined if the frequency of oscillation does not correspond to the cut-off frequency of the CUT. Accordingly, in the second case the phase delay in the delay line can be adjusted until the loop is tuned, whereupon the cut-off frequency will then be known. The first case would be used in situations where a predetermined frequency characteristic of the CUT is desired, whereby if the difference between the oscillation frequency and the CUT frequency characteristic exceeds a threshold, the CUT is deemed to not meet the predetermined frequency characteristic. For example, a particular mixed-signal IC may need to have a low-pass filter with a cut-off frequency of 10 KHz +/- 5%. If the difference between the oscillation frequency and the CUT frequency characteristic was determined to exceed 5%, then the CUT would not qualify for acceptance.

[0017] The foregoing scheme provides many advantages over the prior art. Notably, it primarily uses digital blocks for the feedback implementation. It is also substantially immune against process and temperature variations. As with sigma delta loops, the offset in the comparator input doesn't affect the oscillation frequency. As a result, the comparator can be as simple as a CMOS inverter. In addition, digital delay line 16 may comprise an n-length shift register or an n=2P counter.

[0018] As discussed above, one-bit DAC 20 limits the input voltage in the CUT so as not to exceed its linear input voltage range, thereby eliminating the generation of undesired signal distortions. One-bit DAC 20 may comprise a few switches and a voltage reference, which is readily available on most mixed-signal ICs. Similar implementations may be used with switched-capacitor and switched-current circuits.

[0019] The method of the present invention is applicable to an array of communication circuits, as well as other mixed-signal circuits that employ filter elements. The method can be used with both continuous and discrete time filters. It can be used in sigma-delta digital-to-analog filters to check the noise-shaping bandwidth. It may also be applied to the Automated Gain Controller (AGC)-Equalizer pairs used as the analog front end in many communication applications. In this case, the AGC can be configured using some minor modifications so as to play the role of one-bit DAC 20. In addition, the method is not restricted to low-pass filters. Other pole/zero related parameters can be determined in any type of filter using the same scheme.

[0020] The method is substantially faster than the traditional multi-tone test presently used to determine the cut-off frequency of filters. Under the method, circuit characteristics can be obtained by using a Test Access Port (TAP) controller, which is generally available for testing most ICs. This eliminates the time required to set up GPIB interfaces and external AWGs and digitizers, as well as reducing the processing time.

[0021] The above description of illustrated embodiments of the invention is not intended to be exhaustive or to limit the invention to the precise forms disclosed. While specific embodiments of, and examples for, the invention are described herein for illustrative purposes, various equivalent modifications are possible within the scope of the invention, as those skilled in the relevant art will recognize. Accordingly, it is not intended that the scope of the invention in any way be limited by the above description, but instead be determined entirely by reference to the claims that follow.


Claims

1. A method of determining a frequency characteristic of a filter circuit (12) embedded in an integrated circuit (10), comprising:

providing a plurality of digital components (14, 16, 20) embedded on the integrated circuit corresponding to digital blocks in a feedback loop;

connecting the filter circuit to the plurality of digital components embedded on the integrated circuit such that the filter circuit is inserted into the feedback loop;

adjusting a delay component (16) of the digital components to adjust a phase delay of the feedback loop after insertion of the filter circuit until the feedback loop including the filter circuit is tuned; and

determining a frequency of oscillation when the feedback loop including the filter circuit is tuned, said frequency corresponding to the frequency characteristic of the filter circuit.


 
2. The method of claim 1, further comprising measuring the oscillation frequency as the feedback loop is tuned to produce a test signature of the filter circuit.
 
3. The method of claim 1, wherein said plurality of digital components comprise:

a comparator (14);

an adjustable delay component (16); and

a signal limiting component (20).


 
4. The method of claim 3, wherein the delay component comprises an n=2p counter.
 
5. The method of claim 1, wherein the filter circuit (12) corresponds to a low-pass filter and the frequency characteristic is the cut-off frequency of the low-pass filter.
 
6. A mixed-signal integrated circuit (10) comprising:

filter circuitry (12) having an input and an output, the filter circuit inserted into a feedback loop including:

a comparator (14) having an output and an input connected to the output of the filter circuitry;

an adjustable phase-delay component (16) having an output and an input connected to the output of the comparator;

a one-bit digital-to-analog converter (20) having an input connected to the output of the phase-delay component (16) and an output connected to the input of the filter circuitry (12) so as to form the feedback loop; and

a counter (26) connected to the output comparator (14), said feedback loop enabling a frequency characteristic of the filter circuit to be determined through a built-in self-test of the integrated circuit.


 
7. The circuit of claim 6, wherein the adjustable phase-delay component comprises an n-length shift register.
 
8. The circuit of claim 6, wherein the adjustable phase-delay component comprises an n=2p counter.
 
9. The circuit of claim 6, wherein the comparator comprises a CMOS inverter.
 
10. The circuit of claim 6, wherein the filter circuitry corresponds to a low-pass filter and the frequency characteristic is the cut-off frequency of the low-pass filter.
 


Ansprüche

1. Verfahren zur Bestimmung der Frequenzcharakteristik einer Filterschaltung (12), in eine integrierte Schaltung (10) integriert ist, wobei das Verfahren folgendes umfasst:

das Bereitstellen einer Mehrzahl digitaler Komponenten (14, 16, 20), die in der integrierten Schaltung gemäß digitalen Blöcken in einer Rückkopplungsschleife integriert sind;

das Verbinden der Filterschaltung mit der Mehrzahl von in die integrierte Schaltung integrierten digitalen Komponenten, so dass die Filterschaltung in die Rückkopplungsschleife eingefügt wird;

das Anpassen einer Verzögerungskomponente (16) zur Anpassung einer Phasenverzögerung der Rückkopplungsschleife nach dem Einfügen der Filterschaltung, bis die Rückkopplungsschleife, einschließlich der Filterschaltung, abgestimmt ist; und

das Bestimmen einer Oszillationsfrequenz, wenn die Rückkopplungsschleife einschließlich der Filterschaltung abgestimmt ist, wobei die genannte Frequenz einer Frequenzcharakteristik der Filterschaltung entspricht.


 
2. Verfahren nach Anspruch 1, wobei dieses ferner das Messen der Oszillationsfrequenz umfasst, wenn die Rückkopplungsschleife abgestimmt wird, so dass eine Testsignatur der Filterschaltung erzeugt wird.
 
3. Verfahren nach Anspruch 1, wobei die genannte Mehrzahl digitaler Komponenten folgendes umfasst:

einen Komparator (14);

eine regelbare Verzögerungskomponente (16); und

eine Signalbegrenzungskomponente.


 
4. Verfahren nach Anspruch 3, wobei die Verzögerungskomponente einen n=2p Zähler umfasst.
 
5. Verfahren nach Anspruch 1, wobei die Filterschaltung (12) einem Tiefpassfilter entspricht, und wobei es sich bei der Frequenzcharakteristik um die Grenzfrequenz des Tiefpassfilters handelt.
 
6. Integrierte Schaltung (10) mit Mischsignal, wobei diese folgendes umfasst:

eine Filterschaltkreisanordnung (12) mit einem Eingang und einem Ausgang, wobei die Filterschaltung, die in eine Rückkopplungsschleife eingefügt wird, folgendes aufweist:

einen Komparator (14) mit einem Ausgang und einem Eingang, der mit dem Ausgang der Filterschaltkreisanordnung verbunden ist;

eine regelbare Phasenverzögerungskomponente (16) mit einem Ausgang und einem Eingang, der mit dem Ausgang des Komparators verbunden ist;

ein Ein-Bit-Digital-Analog-Umsetzer (20) mit einem Eingang, der mit dem Ausgang der Phasenverzögerungskomponente (16) verbunden ist, und mit einem Ausgang, der mit dem Eingang der Filterschaltkreisanordnung (12) verbunden ist, so dass die Rückkopplungsschleife gebildet wird; und

einen Zähler (26), der mit dem Ausgangskomparator (14) verbunden ist, wobei es die genannte Rückkopplungsschleife ermöglicht, dass eine Frequenzcharakteristik der Filterschaltung durch einen integrierten Selbsttest der integrierten Schaltung bestimmt wird.


 
7. Schaltung nach Anspruch 6, wobei die regelbare Phasenverzögerungskomponente ein n-Längen-Schieberegister umfasst.
 
8. Schaltung nach Anspruch 6, wobei die regelbare Phasenverzögerungskomponente einen n=2P Zähler umfasst.
 
9. Schaltung nach Anspruch 6, wobei der Komparator einen CMOS-Inverter umfasst.
 
10. Schaltung nach Anspruch 6, wobei die Filterschaltkreisanordnung einem Tiefpassfilter entspricht, und wobei es sich bei der Frequenzcharakteristik um die Grenzfrequenz des Tiefpassfilters handelt.
 


Revendications

1. Procédé de détermination d'une fréquence caractéristique d'un circuit à filtre (12) intégré dans un circuit intégré (10) comprenant les étapes consistant à :

fournir une pluralité de composants numériques (14, 16, 20) intégrés sur le circuit intégré correspondant aux blocs numériques dans une boucle à contre-réaction ;

connecter le circuit à filtre à la pluralité de composants numériques intégrés sur le circuit intégré afin que le circuit à filtre soit inséré dans la boucle à contre-réaction ;

ajuster un composant de temporisation (16) des composants numériques pour ajuster une temporisation de phase de la boucle à contre-réaction après l'insertion du circuit à filtre jusqu'à ce que la boucle à contre-réaction y compris le circuit à filtre soit réglée ; et

déterminer une fréquence d'oscillation lorsque la boucle à contre-réaction y compris le circuit à filtre est réglée, ladite fréquence correspondant à la fréquence caractéristique du circuit à filtre.


 
2. Procédé selon la revendication 1, comprenant en outre l'étape consistant à mesurer la fréquence d'oscillation alors que la boucle à contre-réaction est réglée pour produire une signature test du circuit à filtre.
 
3. Procédé selon la revendication 1, dans lequel ladite pluralité de composants numériques comprend :

un comparateur (14) ;

un composant de temporisation ajustable (16) ; et

un composant de limitation de signal (20).


 
4. Procédé selon la revendication 3, dans lequel le composant de temporisation comprend un compteur n=2p.
 
5. Procédé selon la revendication 1, dans lequel le circuit à filtre (12) correspond à un filtre passe-bas et la fréquence caractéristique est la fréquence de coupure du filtre passe-bas.
 
6. Circuit intégré à signal mélangé (10) comprenant :

des circuits à filtre (12) ayant une entrée et une sortie, le circuit à filtre inséré dans une boucle à contre-réaction comprenant :

un comparateur (14) ayant une sortie et une entrée connectées à la sortie des circuits à filtre ;

un composant de temporisation de phase ajustable (16) ayant une sortie et une entrée connectées à la sortie du comparateur ;

un convertisseur numérique/analogique d'un bit (20) ayant une entrée connectée à la sortie du composant de temporisation de phase (16) et une sortie connectée à l'entrée des circuits à filtre (12) afin de former la boucle à contre-réaction ; et

un compteur (26) connecté au comparateur de sortie (14), ladite boucle à contre-réaction permettant à une fréquence caractéristique du circuit à filtre d'être déterminée via un auto-test intégré du circuit intégré.


 
7. Circuit selon la revendication 6, dans lequel le composant de temporisation de phase ajustable comprend un registre de décalage de n longueur.
 
8. Circuit selon la revendication 6, dans lequel le composant de temporisation de phase comprend un compteur n=2p.
 
9. Circuit selon la revendication 6, dans lequel le comparateur comprend un inverseur CMOS.
 
10. Circuit selon la revendication 6, dans lequel les circuits à filtre correspondent à un filtre passe-bas et la fréquence caractéristique est la fréquence de coupure du filtre passe-bas.
 




Drawing