High-speed sample-and-hold circuit with gain

Details High-speed sample-and-hold circuit with gain High-speed sample-and-hold circuit with gain

2001-05-25

2002-11-12

Williams, Howard L. (Department: 2819)

Coded data generation or conversion

Sample and hold

C341S155000, C327S096000

Reexamination Certificate

active

06480128

ABSTRACT:

BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to the field of sample-and-hold circuits. In particular, the present invention relates to sample-and-hold circuits for use in analog-to-digital conversion systems with a very high sampling rate.
2. Discussion of Related Art
As is well known in the art, settling of the input signal of an analog-to-digital converter during the sampling phase is critical to the performance of the analog-to-digital conversion. Preferably, such settling should be performed below one quarter of a least significant bit voltage of an analog-to-digital converter to provide a margin for the settling of the analog-to-digital converter input circuitry, e.g. the analog latch within a comparator bank.
An example of a known analog-to-digital conversion system is shown in FIG.
1
. The analog-to-digital conversion system
100
includes a closed-loop driving amplifier
102
, which has a gain, and a sample-and-hold switch
104
. The gain of the driving amplifier
102
is defined by the ratio of the feedback resistors R
2
and R
1
and the sample-and-hold switch
104
can be built using a single NMOS transistor that is controlled by a sample clock
105
. The input capacitance of the analog-to-digital converter
106
can perform as a storage capacitor during the hold phase. If the input signal of the sample-and-hold switch
104
changes within the hold period, settling to a new value is required in the sampling period. The voltage step that has to be performed between two adjacent samples can be as high as the full-scale analog-to-digital converter voltage. At very high sampling rates, it is not possible to settle such a voltage step using the closed-loop driving amplifier
102
driving the sample-and-hold switch
104
because of the limited unity gain bandwidth of the amplifier
102
. The unity gain bandwidth has to be approximately an order of magnitude higher than the sample clock frequency if settling to one quarter of the least significant bit voltage of the analog-to-digital converter is to be obtained. The situation becomes even worse if the amplifier
102
has to provide gain because the closed-loop corner frequency, which is defining the settling time constant, decreases linearly with the gain compared to the unity gain bandwidth of the amplifier. Thereby, the maximum possible sampling rate using a sample-and-hold driver with an amplifier in a closed-loop configuration is limited.
Another example of an analog-to-digital conversion system that uses a closed-loop amplifier is shown in FIG.
2
. The analog-to-digital conversion system
200
includes a closed-loop driving amplifier
202
that has an open-loop buffer circuit, such as an NMOS source follower
204
, positioned at the output of the amplifier
202
to perform high-speed settling. The open-loop buffer circuit has gain-variations caused by technology variations during processing. Gain variations can be controlled by incorporation of a replica buffer, such as an additional NMOS source follower
206
, in the feedback branch of the closed-loop amplifier
202
. The replica buffer must be operated at the same current density as the output buffer. An implementation using a scaling factor N for both the (W/L)-ratio and the current-source currents can be chosen. Harmonic distortion in the output buffer, which can be due to variations of the threshold voltage of the source follower having a body effect, can be corrected for by the same replica buffer approach.
During operation of the analog-to-digital conversion system
200
, a change of the output signal at nodes vom and vop up to the full-scale value of the analog-to-digital converter is possible between two adjacent samples at input signal frequencies close to half the sampling frequency. Therefore, during switching of the sample and hold switch
208
from hold to sampling phase, large voltage transients occur at voutm and voutp that have to be settled within half the clock phase. These transients cause the gate potential of the source-follower
204
to change due to a kickback effect. Because the gate of the source-follower buffer
204
is tied to the output gain nodes vgainm and vgainp of the gain-setting amplifier
202
, a settling process of the closed loop amplifier
202
is triggered. During this settling process the output voltage at the gain nodes vgainm and vgainp of the core amplifier and hence the gate voltage of the source-follower
204
driving the sample and hold switch
208
will show ringing. This will alter the output voltage voutm and voutp of the sample and hold switch
208
. Hence, at very high sampling rates settling to a quarter of a least significant bit voltage within the sampling period will be distorted and may not be accurately performed if the kickback can not be reduced and/or the ringing can be avoided. One way to avoid ringing during the mentioned settling process is to design the gain-setting amplifier for a high phase-margin. However, this limits the maximum signal bandwidth that can be processed by the gain-setting amplifier because the unity gain bandwidth has to be reduced.
Due to the level-shift performed by the NMOS source follower the common-mode voltage at the gain nodes vgainm and vgainp is very high. Hence it is not possible to achieve a high DC gain of the amplifier at these nodes at low supply voltages below 2 V because the required drain-source voltages for the pull-up current-source, which is part of the amplifier
202
and can be a DC current source or the output of a current mirror controlled e.g. by the amplifier's input stage, is not high enough to use cascode transistors in order to improve the output impedance of the source. The high gain is required in order to set the gain of the total circuit by the ratio of the feedback resistors R
2
and R
1
and to achieve low harmonic distortion. If the amplifier's gain is not sufficiently high, i.e. >50 dB, gain variations due to process variations will alter the total gain of the system.
A high gain might be obtained in case both the output voltage range and the output common-mode voltage are significantly reduced. However, a limited output range results in a limited analog-to-digital converter full-scale range, which immediately reduces the analog-to-digital converter's performance due to analog noise.
In summary, prior analog-to-digital conversion systems that have a sample-and-hold switch driven by closed-loop amplifiers have several shortcomings. For example, such analog-to-digital conversion systems have longer settling times at maximum input voltage steps that occur at input signal frequencies close to the Nyquist frequency, i.e. half the sampling frequency. During the sampling phase, at very high sampling rates a closed-loop amplifier circuit is not capable of settling a full-scale voltage step at the output of the sample-and-hold switch to below one quarter of an analog-to-digital converter least significant bit voltage at typical analog-to-digital converter resolutions up to 6 bit and a differential full-scale voltage of up to 1V peak-to-peak. Such large signal voltage steps occur at input signal frequencies close to the Nyquist frequency. Accordingly, if settling to a quarter of an analog-to-digital converter least significant bit at typical analog-to-digital converter resolutions up to 6 bit and a differential full-scale voltage of up to 1V peak-to-peak is required to provide headroom for the settling of the first analog latch stage of the analog-to-digital converter, the maximum sampling rates are limited.
Another problem of analog-to-digital conversion systems that use a closed-loop amplifier and single open-loop source-followers to drive the sample-and-hold switch is that for signal frequencies near the Nyquist frequency, incomplete settling of maximum input voltage steps can occur due to ringing of the gain-setting amplifier caused by the kickback onto the amplifier's gain-node. Incomplete settling introduces sampling errors and can be the cause of harmonic distortion of the analog-to-digit

Affiliated with

Also associated with

Rating

High-speed sample-and-hold circuit with gain 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-speed sample-and-hold circuit with gain, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and High-speed sample-and-hold circuit with gain will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2964089