Dynamic Signal Measurement Basics Hardware Basics

Download 18.26 Kb.
Date conversion13.01.2017
Size18.26 Kb.

Quick Look

Dynamic Signal Measurement Basics

Hardware Basics

Filters and Delta Sigma Converters

A vital part of the dynamic signal analysis is the proper acquisition of these signals for their posterior analysis. High quality hardware should be used to provide the necessary signal conditioning to prevent aliasing and distortion in general of the signal for its processing on the frequency domain.
Amplifiers, filters and converters play an essential role on this digitalization process. We will go over some of the basics of these components and their use on dynamic signals acquisition.

We’ll talk about the architecture of both the 44’s and 45’s series. We’ll talk about the specifics of the digital anti-aliasing and anti-imaging filtering capabilities as well as the Delta Sigma converters technology and advantages.

The simpified architecture shown here is a two inputs / two outputs such as the one used on the 4551, 4453 and 4451. The 4452,4454 and 4552 have four inputs and they have the same architecture than the two (upper part ) inputs shown here.
The PCI-4451/52 and NI-4551/52 boards have AC / DC coupling capabilities that you can change from you software, you have also a minimum range of +-10mv and a maximum of +-42v on the inputs with several ranges in between (gain selections between -20dB and 60dB in 10dB increments).
(The PCI-4453/54 have AC/DC coupling but only a +-10V input range)
All channels are simultaneous and coupling and gain can be set on a per channel basis, there is a delta sigma converter per channel and digital and analog filter per channel as well.
The 4551 and 4552 instruments have DSA Measurement Engine Processor on board and the TIO digital I/O chip . The 4451 and 4452 boards will not have the processor and will use the DAQ-STC counter timer chip. Besides that the architecture is basically the same.

According to Shannon sampling theorem, the highest frequency (Nyquist frequency: fN) that can be analyzed is fN = fs/2, where fs is the sampling frequency. Any analog frequency greater than fN will, after sampling, appear as a frequency between 0 and fN. Such a frequency is known as an “alias” frequency. In the digital (sampled) domain, there is no way to distinguish these alias frequencies from the frequencies that actually lie between 0 and fN. Therefore, these alias frequencies need to be removed from the analog signal before sampling by the A/D converter.
To remove these components present at higher frequencies than the Nyquist frequency, an analog lowpass filter must be used. This anti-aliasing filter should exhibit a flat in-band frequency response with a good high frequency alias rejection and a fast roll-off in the transition band.
The delta sigma converters and digital filters with re-sampling, oversampling and noise shaping procedures will provide a second powerful software adjustable antialiasing fitering stage.

More or less the same principle applies to the outputs , analog and digital filtering and oversampling are needed as well to provide an image free output spectrum.
Specifics and detailed explanation of how filtering and converting is performed on the DSA boards for inputs and outputs as well as specs of filters and overall performance of the board can be found on the NI 4551/4552 and PCI 4451/4452 Users Manuals downloadable from our web page at www.natinst.com.

Delta sigma converters are used for analog to digital and digital to analog conversion on NI DSA boards. This state of the art technology provides excellent conversion performance when dealing with dynamic signals. Its outstanding linearity compared with regular resistor and capacitor network converters is not its only advantage, strong antialiasing levels are also reached by its very unique technology.
Delta Sigma converter in these boards provide 16 bits resolution but are hardware wise 1 bit , 128 times oversampling high speed converters, some of the basics behind these antialiasing features is explained here.
The results of the 1-bit oversampling digitizer are noise-shaped, filtered, and decimated to provide a high resolution output.
This type of digitizer exhibits excellent differential and integral linearity and no trimming for linearity is required. The key to this excellent linearity is that the transfer function contains only two points for the two possible states of a single bit (more on this later).

In any application that requires analog to digital conversion, understanding the application’s dynamic range requirements is essential. In dynamic signal acquisition (DSA) applications, the requirements is typically a minimum of 90 db SNR.
There are many vibration DSA applications where you must resolve a very low level signal from a large exciter signal. A failure analysis on a bearing would require resolving small frequency components indicative of an upcoming failure out of the almost overwhelming components because of the rotating machinery. A converter with less than 16-bits would lose the critical signal in its noise floor. DSA does not require a delta-sigma converter, but its features make it ideally suited for this type of application.
Users do not typically think of delta-sigma converters as being useful in control type applications. Careful understanding of the required loop times may actually show that these types of converters are indeed useful. Special attention should be paid to the group delay incurred in the analog anti-aliasing filter which you must when a non delta-sigma converter is being used. Sometimes this delay can actually be greater than the delay through a delta-sigma converter . delta-sigma converters can solve most control applications requiring greater than a 10 ms loop time. If the loop time is between 1 and 10 ms then you must carefully consider the delays in order to determine the usefulness of delta-sigma converters.

Oversampling simply redistributes the noise energy across a larger frequency range. The noise density remains the same.

Total noise energy = q2/12 and that energy is spread out evenly across from DC to fs/2. This assumes a non noise-shaped signal which is sufficiently busy through dithering or activity in the input signal.

Total noise energy = q2/12 and that energy is spread out evenly across from DC to fs/2. This assumes a non noise-shaped signal which is sufficiently busy through dithering or activity in the input signal.

Again, like oversampling, noise shaping only redistributes the noise. The total noise energy remains the same. Think of a tubular balloon and imagine squeezing one end of it. This is analogous to noise-shaping, where the air has been pushed towards the other end of the balloon.

In general this is what the Delta and Sigma heart of the converter are used for, and some of its advantages and disadvantages.

This shows the linear transfer function presented by the Delta Sigma converters and the simplified architecture of such 1 bit converters that will provide the 16 bits desired resolution and measurement quality.

The DSA boards can be programmed to have a desired Alias Free Frequency Span, this is, the cutoff frequency of the digital filters on the board can achieve filtering on user selectable frequency spans (DC -to->Alias Free Span).

Some of the advantages of digital filtering on board.

The filter roll-off of the actual filter will look almost like an ideal filter. We have exaggerated the slope to illustrate the effects of quantization noise and aliasing.

The database is protected by copyright ©dentisty.org 2016
send message

    Main page