Miscellaneous active electrical nonlinear devices – circuits – and – Specific identifiable device – circuit – or system – Unwanted signal suppression
Reexamination Certificate
2000-03-29
2001-10-16
Zweizig, Jeffrey (Department: 2816)
Miscellaneous active electrical nonlinear devices, circuits, and
Specific identifiable device, circuit, or system
Unwanted signal suppression
C327S558000
Reexamination Certificate
active
06304134
ABSTRACT:
FIELD OF THE INVENTION
The present invention pertains in general to continuous time filters, and more particularly to a continuous time filter having an arrangement of transconductor elements that provides programmable high-frequency boost to the continuous time filter. The present invention relates to continuous time filters for use in a read channel of a disk drive.
BACKGROUND OF THE INVENTION
Filters are input/output devices that reject or pass signals based on their frequency content; that is, the distribution of their energy across a range of frequencies. Every filter has a pass band that defines its effect on the frequency content of the input signals. For example, low-pass filters pass signal frequencies below a certain frequency known as a cut-off frequency and reject frequencies above the cut-off frequency, whereas high-pass filters pass frequencies above a cut-off frequency. In contrast, band-pass filters pass frequencies that are between upper and lower cut-off frequencies.
In addition to their pass-band characteristics, filters are also characterized according to their construction and temporal operation. Filters typically include a network of resistors and capacitors and transistors which define the pass band or frequency response. When most of these components are fabricated on a common substrate or foundation, the filter is known as an integrated filter. As for temporal operation, filters operate either in continuous time, which allows their inputs and outputs to change at any time, or in discrete time, which allows changes at only specific time increments.
The use of continuous time filters that utilize one or more transconductor stages and having a bandwidth that is a function of the transconductance G
m
of each of the stages is becoming more widespread. This G
m
value can vary as a result of process variations, temperature variations, etc.
Particularly at high frequencies, the transconductance-capacitor (G
m
-C) technique has emerged as a design approach based on biquad building blocks as well as LC ladder simulations. Compared to operational amplifiers, transconductances generally result in simpler circuitry with fewer undesirable and hard-to-model parasitics which allows for superior high-frequency performance. Although some bipolar circuits have been used, most recent attention has focused on CMOS design.
Typically, when the continuous time filter is used in connection with a read channel of a disk drive, the filter is placed inside the automatic gain control (AGC) loop where it can be used to perform a dual role. It can cut off high-frequency noise in an amplified read signal resulting in lower bit error rates. A second filter objective is to equalize the bit stream (i.e., to slim the data pulses), allowing higher bit densities. Additionally, to minimize pulse peak shifts in time, a filter with a linear phase behavior (or constant group delay) is desired. Furthermore, the filter's group delay should be independent of the amount of equalization. This equalization can be accomplished typically with a feed-forward design.
In many communication systems, such as hard disk drive, tape and optical drive, analog adaptive filters are used to equalize the channel response or, in other words, the bit stream. As a result of the limited programmability of analog systems, the “adaptiveness” of the continuous time filter (CTF) is restricted to programming bandwidth and the addition of programmable high-frequency boost at every bandwidth setting.
FIG. 1
illustrates the effects of the programming of bandwidth without boost. In contrast,
FIG. 2
illustrates the effect of boost programming with one bandwidth setting. The boost programming should not be dependent on the bandwidth setting. The boost programming should alter only the magnitude response of the filter. The phase and group delay response should not change.
The DC gain of the filter must be reduced such that the maximum value of the magnitude response remains relatively constant irrespective of the boost value. This aspect is shown in FIG.
3
. However, turning back to
FIG. 2
, it can be seen that with increasing boost, the cutoff frequency increases.
In
FIG. 3
, it can be seen that the gain increases with increasing the boost. Typically, the transfer function of the filter is an equiripple group delay approximation, usually fifth or seventh order. Butterworth responses have also been used.
FIG. 4
illustrates a schematic of conventional boost circuit. To implement boost, a differentiated version (k
1
sV
in
) The present invention relates to continuous time filters for use in a read channel of a disk drive.) of the input signal V
in
is injected into the internal node of the second order section. The response of the biquad in the absence of boost, where k
1
=0, is denoted as H(s), where
H
(
s
)
=
1
D
(
s
)
(
1
)
If k
1
≠0, it can be shown that Equation 2 holds.
H
′
(
s
)
=
1
-
k
1
s
2
D
(
s
)
(
2
)
Most high-speed filter designs use the G
m
-C technique to realize integrators. The schematic of a G
m
-C biquad is shown in FIG.
5
. In this figure, the transconductors are digitally tunable in order to realize bandwidth programming.
In order to implement boost, a differentiated version of the input signal V
in
needs to be injected into capacitors
502
and
504
. This can be implemented in two ways. The schematic of the circuit is illustrated in FIG.
6
. The voltage amplifier with a programmable gain k
1
is coupled by means of floating capacitors
602
and
604
to capacitors
502
and
504
, respectively. However, this approach has various problems. The finite output impedance of amplifier k
1
causes aberrations in the frequency response of the filter as the bandwidth is programmed. In other words, the response of the filter degrades progressively as the bandwidth is increased. Additionally, the parasitic capacitance of the bottom plate of capacitors
602
and
604
is not accurately determinable, and this results in uncertainty in the frequency response. The circuit of
FIG. 6
results in a noisy differentiation process.
Reducing DC gain in the presence of programmed boost is accomplished by reducing the level of the input signal to the filter. A reduced input signal results in the filter being susceptible to noise.
Another technique is illustrated in FIG.
7
. Here, the differentiated version of the input signal is tapped off from a node of a previous biquad. This input signal is converted into a current by the transconductance (k
1
G
m
) and injected into capacitors
502
and
504
. However, this circuit has problems including the output of the transconductor changes as the boost is programmed at a fixed bandwidth setting. This problem is especially serious in CMOS designs where output impedances are already low. Due to this change in output impedance, the phase and group delay responses change as boost is varied. Additionally, the programmable transconductor (k
1
G
m
) needs to be programmable by a much larger range because boost programming, which is determined by k
1
, and bandwidth programming, which is determined by G
m
, are implemented in the same transconductor. This is possible in bipolar designs because of the exponential nature of the devices but difficult to perform in CMOS technology without serious power penalties. Furthermore, it is still difficult to reduce the DC gain as boost is programmed.
SUMMARY OF THE INVENTION
The present invention employs a follow-the-leader (FLF) approach. A third order filter section is implemented. The boost is implemented by adding the outputs of the first and third integrators (i.e., the voltages across the capacitors of the respective integrators) in a weighted fashion. The present invention uses a digital programmable transconductor with constant input capacitance.
REFERENCES:
patent: 4771196 (1988-09-01), Mead et al.
patent: 5293086 (1994-03-01), Tanigawa et al.
patent: 5625317 (1997-04-01), Deveirman
patent: 5886586 (1999-03-01), Lai et al.
patent: 5912583 (1999-06-01), Pierson et al.
patent: 5994951 (19
Brady W. James
Swayze, Jr. W. Daniel
Telecky , Jr. Frederick J.
Texas Instruments Incorporated
Zweizig Jeffrey
LandOfFree
High-frequency boost technique does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with High-frequency boost technique, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and High-frequency boost technique will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2612753