CLC935 datasheet(6/12 Pages) NSC

6 / 12 page

http://www.national.com

Overdrive Recovery Plots

These plots indicate ADC time domain settling from a

50% overdrive condition. A very fast, +1.5V or -1.5V to

0.00V pulse, with a period slightly shorter (100ps) than

that of the CONVERT clock, is used as the input source.

The ADC is therefore “slipped” through the input wave

form and the output data is plotted after being smoothed

using a 5 point sliding average. The slip rate (period

difference between clock and input) and data point

number are used to generate the time axis. For the sake

of plot resolution, only fine settling is shown.

Analog-to-Digital converters are specified in many ways.

As a component achieves higher performance, its

specifications and their definitions can become more

critical.

Fortunately, the vast number of converter

applications can generally be placed into one of two

classes. These are processed data and non-processed

data applications. The distinction seems quite simple but

the split implies a completely different approach in

specifying A/D converters for a given application.

The processed data area includes the frequency

domain applications which employ Fourier processing

(FFT).

Also in this category are the highly averaged

applications, usually concerned with low noise. In each

case, the converter’s data is averaged or convolved

mathematically. This processing reduces the apparent

noise level in the output data. For FFTs, the noise is

simply spread over a large number of frequency

bins. For simple averaging approaches, the Gaussian

distribution of noise is greatly reduced, appearing

to increase the converter’s resolution. Processed

applications include radar, network and spectrum

analyzers, communications receivers, etc.

The non-processed applications tend to take the convert-

er’s data in its original form with very little processing.

This means that the noise reduction benefits of the

processed applications are not seen.

The non-

processed area is composed primarily of time domain

applications like imaging, DSO’s, ultrasound, etc.

The processed vs. non-processed issue has several

implications in terms of converter specifications. For the

non-processed (time domain) systems

the dominant

converter specifications deal with noise (SNR) and

converter accuracy (DNL). The converter’s quantization

noise and input stage noise dominate converter

accuracy . The harmonic distortion (primarily INL) of the

converter is generally of little interest given that most time

domain applications present data for visual analysis and

tend to focus on “local” accuracy rather than over the

full input range.

“Local” accuracy is best described

through the standard noise measurements, such as SNR

and DNL.

In the frequency domain application areas, the noise of

the converter is processed to the point where, for almost

all systems, it is no longer of issue. This is manifested as

a reduction in the apparent noise floor. The actual RMS

noise is not reduced, but is spread over more and more

frequency bins as processing levels are increased.

Unfortunately, the harmonic distortion performance of the

converter is not affected by increased processing. This

makes the harmonic performance, or more specifically

the spurious performance, the dominant error source

for frequency domain applications. SFDR becomes

the dominant specification for determining converter

performance in the frequency domain.

Signal-to-Noise Ratio (SNR) is the ratio of the power

contained in the fundamental signal compared to the

power contained in the entire noise floor. That is to say

all individual noise components are added together to

arrive at an integrated noise power. For SNR, harmonic

power is excluded from the noise measurement. SNR is

particularly important in time domain applications like

digital image processing and infrared imaging, where

conversion accuracy can be heavily degraded by

integrated noise.

Signal-to-Noise-and-Distortion (SINAD) is the ratio of

the fundamental signal power to the power at all other

frequencies. This includes all noise as well as all

harmonics. SINAD is a worst case specification for A/D

converters, combining variables from both frequency and

time domains. The value of SINAD in high-performance

converter applications is not clear since it does not

accurately predict the best converter for a given

application. Because data converter applications tend to

fall into either noise-sensitive time-domain applications

or distortion-sensitive frequency-domain applications,

SINAD is not specified for the CLC935 data converter.

Total Harmonic Distortion (THD) is the combined

power of a specified number of harmonics, compared to

the power of the fundamental signal. Harmonics are

located at predictable frequencies, spaced at integer

multiples of the fundamental signal.

For example, a

1MHz fundamental would generate harmonics at 2MHz,

3MHz, 4MHz, ... and so on. In practice, only the first five

harmonics contribute significantly to THD, although more

may be included in the measurement. THD does not

tend to apply well in frequency domain applications which

are by their nature very SFDR oriented. In time domain

applications, THD is indicative of full-scale input range

distortion, however the high-performance time domain

applications are generally most interested in local

distortion performance. Local distortion and accuracy is

dominated by DNL.

The use of THD for applications

requiring local performance is not likely to yield accurate

or repeatable results and therefore THD does not appear

in the CLC935 specifications.

Understanding A/D Dynamic Specifications

CLC935 bảng dữ liệu(PDF) 6 Page - National Semiconductor (TI)

CLC935 bảng dữ liệu(HTML) 6 Page - National Semiconductor (TI)

Html Pages

1 2 3 4 5 6 7 8 9 10 11 12

bảng dữ liệu tải về

Link URL