Calibration is a procedure of configuring an instrument to ensure the measurement results of a sample is within the acceptance range. In instrumentation design, removing or minimizing the factors that cause the inaccurate measurements is an important process used to maintain instrument accuracy.
The calibration procedures may vary in different instruments. It can be as simple as calibrating our home weighting scale or as complex as calibrating the high-end vector network analyzer (VNA) which requires a calibration kit and error correction mathematical models. In this article, I will share the basic knowledge about RF instrument calibration which is a fundamental knowledge needed for any RF test engineers.
What is Error?
Before we dive into the error correction method used in RF calibration, let me explain first of what defines an error. In all real life scenarios, measurements are influenced by the environment and limited by the capability of the measuring instrument used. The results is a reduction of the accuracy of the measured results, making “what we see” questionable. The difference between the true and the measured value is called error. Generally, errors can be classified as systematic errors and random errors.
Random errors are caused by random effects in the instrument’s environment such as component noise or temperature. These errors are almost impossible to trace or quantify. Thus, it is impossible to correct or compensate for it in measurement process. On the other hand, systematic errors are the consistent or repeated errors that exist in the same measurement. Hence, these errors can be identified from the measurements.
Although systematic errors cannot be removed completely, it can be minimized to a sufficient degree based on the error correction method which we will explain in a later paragraph. An example of systematic error of VNA S-parameters is the signal reflection which affects the source and load matching.
General 1-Port Error Model for Calibration
Now that we understand what kind of errors are there, let’s proceed to the error computation for calibration. Although there are many calibration methods that had been introduced since the 80s, they all share some common workflow in performing one-port calibration.
Take the figure below, it shows the signal flow image for a one port device under test (DUT) with an uncalibrated VNA. ΓDUT and ΓM are the reflection coefficient of the DUT and measured respectively. The between this two values is the systematic error that we want to remove.
To extract the error terms, an error box is modeled as shown above. The error box consist of directivity (D), reverse (R), and source match (S). Taking short, open, load (SOL) calibration as an example, the ΓDUT should be known based on the calibration kit characteristic, and ΓM is known based on the measured results. By forming three equations based on eq. 1 below, the three error terms can be solved by the three SOL equations. Now, we can perform the error correction for any DUT based on the extracted error terms.
Other than SOL 1-port calibration, short, open, load, thru (SOLT) and thru, reflect, load (TRL) are some of the most commonly used 2-ports calibration techniques. These two methods dominate in the testing of coaxial and waveguide components. Perhaps I can talk more of these in future articles.
When Does Instruments Require Calibration?
Not all instruments need to be constantly calibrated at every use. But in general, calibration has to be done in these circumstances:
- when an instrument is newly bought
- after being repaired
- after a specified time period has elapsed without usage
- before or after a critical measurement
As you can see, the calibration has to be done quite frequently to ensure accuracy. For complex instruments, calibration can be tedious work. Therefore, some manufacturers have designed electronic calibration (Ecal) kits which automate certain calibration processes.
In short, we covered fundamental knowledge involved in calibration, errors, and error corrections model in this article. Calibration is a big scope. There are more topics that can be discussed such as the 2-port error model, different error correction techniques, and their advantages and disadvantages. Perhaps I can talk more of these in future articles.
In FILPAL, test and measurement is a very common occurrence. Check out our website at http://www.filpal.com for more info of what we do!
Originally published at http://filpal.wordpress.com on May 15, 2022.
