What is Data Acquisition?
What Makes Up a PC-based Data Acquisition System (DAS)?
What Are the Main Requirements to Consider When Choosing a DAS?
Your Selection Process
Glossary
What is Data Acquisition?
Data acquisition is the science of measurement (acquisition) of electronic signals. Modern data acquisition systems (DAS) convert these signals to digital form to be transmitted to a computer and stored, normally on a hard disk. Most DAS systems are designed around a PC.
What Makes Up a PC-based Data Acquisition System (DAS)?
These are the typical main components of a PC-based DAS:
signal conditioning hardware (may include isolation) analog to digital conversion hardware buffering/storage hardware and hardware for transmission to a PC PC software to convert to a meaningful form, store it to a hard disk, display it, analyse it etc.
What Are the Main Requirements to Consider When Choosing a DAS?
The DAS should:- sample fast enough for your application; - provide an anti-aliasing hardware filter to remove high frequency noise; - measure the type of signals generated by your measurement sensors; - provide signal excitation if they are required by the sensors and not already provided; - measure the entire range of signal of interest (eg. high or low enough voltage); - provide the necessary signal conditioning and isolation as required; - provide the necessary portability (if required); - have the necessary resolution; - have the necessary accuracy; - ideally, provide simultaneous sampling; (all samples corresponding to the same moment in time); check if your application is sensitive to skewed sampling; - have a suitably low noise level; - have the necessary software to adjust collection variables, save the collected data to disk, save the collection documentation, plot and analyse the data and export it to other formats you may need; The hardware should have few switches and physical adjustments - this simplifies operation. The software should be intuitve, easy to learn and versatile.
Your Selection Process
1) determine the signal types to be measured (general purpose voltage, current, frequency, digital (binary on/off),thermocouple, RTD, strain gauge etc.);
2) determine the minimum (may be negative) and maximum signals to be measured (eg. -2 to 5 VDC). This will determine the signal ranges you require (eg. -5 to 5 V);
3) determine the frequency of the signals you want to measure. Ideally, you should measure 10 times faster;
4) check if your sensors require excitation (controlled power) not already provided;
5) estimate the smallest change in signal you need to measure - then relate it to the smallest signal range available in the hardware that spans the signal of interest. For example, if you need to measure -2 to +5 V, and the hardware has a -5 to +5 V range, and you need to see changes of 1 mV, then you need a resolution of at least 0.01 %. This would imply that a 12 bit system is insufficient. This normally means going to a 16 bit system;
6) ask the vendor the noise level in the digital measurements for a constant analog source; (an answer of "none" is not believable). Verify that the noise will not reduce the effective resolution below the minium you require;
7) determine if you need portability and on-board memory for unattended operation (short term or long term);
8) determine if you need single ended (common ground at inputs) or differential (separate grounds);
9) determine if you need isolation. This is an important safety feature to protect you, your computer, your sensors and any other connected devices (eg. an industrial computer system);
10) determine if it is important to sample the signals at the same moment;
11) determine how ranges and inputs types are set. Hardware with DIP switches and pots (potentiometers) should be avoided unless input ranges are changed rarely. An additional issue with pots is their temperature sensitivity;
12) determine the software requirements; what type of plots and plot tools (zooming etc.) are required; what analysis tools (eg. frequency spectrum Fourier analysis) are required; what methods of bringing data in and out of the software (import, export, DDE etc.) are required;
13) determine if you need any software drivers to create your own custom application.
Glossary
ADC - analog to digital converter - converts the analog signal to digital form. Aliasing - a misrepresentation of the measurement caused by sampling more slowly than the signal being measured. For example, if your signal had a 2500 Hz component and you sampled at 1000 samples per second, aliasing would occur. The waveform that you observe in your sampled data may have a "ghost" wave at, for example, 100 Hz. This wave actually never existed and is a result of aliasing. Aliasing cannot be easily removed once it occurs and often, users are unaware that it has occurred, resulting in incorrect conclusions being drawn. To avoid aliasing, the sample period must be increased or a hardware filter be implemented before sampling occurs to remove high frequencies. DAC - digital to analog converter - converts a digital signal to analog form (for output) anti-aliasing filter - a hardware filter selected to remove high frequencies (usually noise); A software numeric filter (also called "digital" filters) CANNOT prevent aliasing during hardware sampling. The filter should have a cutoff frequency of half the sample rate. For example, for a sample rate of 1000 samples/sec. DAS - data acquisition systems. Msec - milliseconds - 1000 msec equals 1 second. Nyquist frequency - theoretically, the minimum sampling frequency for a given signal frequency of interest. This is double the signal frequency. For example, to measure a 10 Hz signal, theoretically you would have to sample at 20 samples/sec. In practice, a ratio of 10:1 is advisable (in this example, 100 Hz). Resolution - the smallest change in input signal measureable by the DAS. This is normally expressed in terms of the number of bits in the resultant digital value generated by the ADC resolution. The vast majority of ADC's provide 12 bits (1 part in 4096 or 0.024 %) or 16 bits (1 part in 65536 or 0.0015 %). Sampling Frequency or Rate - the number of samples that can be measured per second - normally expressed in samples/sec or Hz. Check how this relates to the sampling rate per channel. Sampling Period - the inverse (reciprocal) of Sampling Frequency - normally expressed in msec/sample or sec/sample. For example, 10 samples/sec is equivalent to 100 msec/sample.