Coded data generation or conversion – Analog to or from digital conversion – Analog to digital conversion
Reexamination Certificate
2003-04-09
2004-08-31
Jeanglaude, Jean Bruner (Department: 2819)
Coded data generation or conversion
Analog to or from digital conversion
Analog to digital conversion
C341S156000
Reexamination Certificate
active
06784820
ABSTRACT:
CROSS-REFERENCE TO RELATED APPLICATIONS
Not applicable.
STATEMENTS REGARDING FEDERALLY SPONSORED RESEARCH
Not applicable.
FIELD OF THE INVENTION
This invention relates generally to analog-to-digital converters and more particularly to a high resolution, expanded dynamic range analog-to-digital converter.
BACKGROUND OF THE INVENTION
In signal processing applications, it is often desirable to detect small variations in a signal having relatively large background variation. Many digital signal processing applications require signal sensitivities well beyond the capabilities of existing analog-to-digital (A/D) converters. An example is the detection and exploitation of minute magnetic perturbations from objects of interest in the presence of the earth's substantially larger magnetic field. One method of extending the signal resolution is to subtract an offset and amplify the result so that only the amplified difference is presented to the A/D converter. Small signal variations due to the object of interest can then be observed successfully as long as the variations do not exceed the signal amplitude capabilities of the A/D device. This technique generally works well when the sensor measuring the signal is stationary relative to the large undesired signal component, but fails when relative motion of the sensor and undesired signal source cause large signal variations, such as when a vector magnetometer is mounted on a moving platform.
Other conventional approaches to measuring small signal variations in the presence of sensor instability have either been to stabilize the sensor or increase the resolution of the A/D device. Stabilizing the sensor is relatively expensive, requiring mechanical fixturing, actuators and a control loop to continuously and instantaneously adjust the sensor as needed to maintain a fixed orientation. Such a stabilized system has a finite response time and orientation errors that effectively introduce noise onto the signal being measured and fundamentally limit the achievable sensor sensitivity. Increasing the resolution of the A/D is generally not a cost effective or viable option in certain applications. Furthermore, the greater the resolution, the slower the sensor operates, thus preventing an A/D with adequate resolution from having a sufficient respone time.
A/D converters are in widespread use in electronics and computer applications for converting real-world signals into a digital domain for further processing. A/D converters are characterized by the number of bits of resolution and operating speed (i.e., number of samples per second). Typical high speed A/D converters have low resolution (i.e., a small number of bits in data word), and high resolution A/D converters are generally slow. For some applications, the performance of available A/D converters is inadequate. One means of extending the precision of an A/D converter is to offset the analog signal by the signal's average direct current (DC) value and then to amplify the analog signal level prior to A/D conversion. This technique can be used when the signal of interest includes small signal excursions around a DC value which is less important. The limitation of this approach occurs when the analog signal moves outside a valid range, causing the A/D converter and/or amplifier to saturate and invalidating the data.
The performance of a conventional magnetic sensor is limited by available analog to digital converters. The earth's magnetic field varies depending on where the sensor is located, but is typically in the range of ±45,000 nano Tesla (nT). A typical signal has a sub nT value on top of a 30,000 nT pedestal representing the earth's background field. The flux gate sensors are capable of handling the dynamic range of ±70,000 nT, and have noise floors of below 0.010 nT per square root Hz at 1 Hz. To digitize this signal with an LSB of 0.008 nT requires an A/D of at least 24 bits. While there are twenty-four bit A/Ds available, a careful inspection of the twenty-four bit A/D's specifications indicates that for the frequency band of interest (generally 4 milliHertz to 70 Hertz for magnetic sensors), the A/D performance is effectively well below twenty-four bits. For example, an Analog Devices 7716 sigma delta device nominally provides 22 bits of data, but at the lowest update rate the digitization noise is such that the device is equivalent to a nineteen or twenty bit A/D.
One solution to this problem uses a very stable digital-to-analog (D/A) converter to provide a voltage for subtracting the majority of the earth field pedestal before the input signal is digitized. Then the signal having the offset removed can be amplified before it is digitized. This technique provides an improvement of approximately 3 to 6 bits in the magnetometer example above resulting in a least significant bit (LSB) corresponding to about 0.002 nT. This method relies of the fact that sensor is stationary, so the underlying earth field is fairly constant; and also, that the removal of the DC pedestal does not adversely affect the system's ability to process the data and localize targets.
However, in a synthetic aperture application, the sensor is moved through the earth's field, thus the background is changing due to earth field gradients and due to the changing attitude of the vector sensor in the field. The conventional solutions do not work under these conditions because the offset needed to null the background is constantly changing.
U.S. Pat. No. 6,441,767, “Method and System for Adjusting a Threshold Control in an Analog-to-Digital Converter” describes a differential amplifier, an adjustable offset voltage, and a feedback mechanism for adjusting the offset for reducing unwanted offset voltages internal to the A/D converter that drift causing overvoltage clipping to occur at unpredictable voltage levels. U.S. Pat. No. 6,441,767 teaches amplifying an offset input signal to correct an internal error mechanism within the A/D converter, but does not describe extending the resolution or dynamic range of the A/D converter. U.S. Pat. Nos. 6,445,329 and 6,437,717 describe multi-stage A/D converters having multiple stages. U.S. Pat. No. 6,445,329 teaches the use of a reference voltage and offset voltage selection blocks to optimize subsequent subrange analysis blocks for linearity and operating speed. U.S. Pat. No. 6,437,717 teaches using the difference between the analog input signal and threshold values to correct offset errors. The accuracy of these multistage converters is limited by the ability to correctly track the input signal. Noise levels limit the converter performance to that of a lower resolution A/D converter. The noise level is relatively high when the input signal hovers near a value that causes the offsets to change state.
Subtracting an offset from the input signal generally adds noise to the output signal because the offset is never completely noise free. Obtaining a variable offset that is as accurate as the desired precision for the measurement is relatively difficult. Typically, the offset circuit is made up of a precision voltage reference and a voltage divider circuit. The precision reference can be relatively low noise (on the order of the desired measurement precision), but the voltage divider precision is a problem because the voltage divider component tolerances limit the accuracy of the voltage reference. This is true whether the voltage divider is made from discrete passive resistors or the resistors are fabricated in the silicon of an integrated circuit D/A. For example, 0.1% tolerance on resistors used in a voltage divider cause a 0.1% error. The 0.1% error is about 1000 times larger than one part per million accuracy which is desired in certain applications.
U.S. Pat. Nos. 6,445,329 and 6,437,717 teach computing a new offset each cycle and subtracting the offset from the input signal during the process of generating high resolution outputs. The precision in these conventional systems resulting from extra bits of resolution is negated by the adde
Casalegno James William
Kohnen Kirk K.
Monte Frank Philip
Slater Eric Kent
Daly, Crowley & Mofford, LLP.
Jeanglaude Jean Bruner
Raytheon Company
LandOfFree
High resolution, high dynamic range analog-to-digital... 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 resolution, high dynamic range analog-to-digital..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and High resolution, high dynamic range analog-to-digital... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3298041