Error Vector Magnitude (EVM) is a measure used to quantify the quality or performance of a modulated signal from a transmitter or receiver. In simple terms, if we consider a constellation diagram the EVM is the magnitude of the difference between the measured vector and the ideal (reference) vector. This can be visualized as below.
It can be seen from this simple diagram that EVM is influenced by a number of parameters such as below:
- Phase Error
- Frequency Error
- Magnitude Error
- Noise that contributes to all of the above
Each of these areas are contributed to not only by the signal being measured, but also by the test instrument itself which has an effect on how well it can capture the signal, but also how it is able to generate an “ideal” reference signal to use for the calculation. If we take a look at this block diagram which shows a model of transmitter EVM contributions such errors also exist during the demodulation process. Therefor the limit of EVM demodulation performance can only be as good as the error contributions added during the demodulation process in the signal analyzer.
Many of the effects we are able to correct for in DSP as part of signal synchronization etc. Although phase noise is not so easy to correct for and has a direct impact on performance.
Below are some graphics showing a visual representation of the effects of the distortions described in the model on the constellation diagram.
EVM Performance when measuring the new 802.11ac 80MHz standard 256QAM
Below are two graphics showing screenshots taken from an R&S Vector Signal Analyzer on an 802.11ac signal with a specific EVM of the transmit signal. The good signal has 33dB of EVM and the bad signal only 30dB of EVM performance. It is clear to see on the bad signal that some symbols (constellation points) are much further away from the ideal point which would result in more error and a poorer EVM result.
|
Constellation with 30dB of EVM Performance
|
Constellation with 33dB of EVM Performance
|
Instruments with a worse EVM performance will contribute directly to this error and on signals over a wide bandwidth and such high order modulation schemes (256QAM / 1024QAM etc) EVM performance and phase noise performance of the test instrument becomes critical to measuring such parameters.
Graphics courtesy Rohde & Schwarz FSV-K70 user manual and some very helpful colleagues relating to the EVM plots above.
Posted in Test and Measurement, Tutorials, Wireless Technology
Post Tags: 802.11ac »
Error Vector Magnitude »
EVM »
FSV »
FSW »
Spectrum Analysis »
Vector Signal Analyzer
Interview with EngineeringTV (Microwaves&RF magazine) at European Microwave 2011 for our new R&S FSW high end spectrum analyzer. Serious receding hair line and slight scouse accent (brought up on the border between Manchester and Liverpool).
Posted in Events & Conferences, Test and Measurement
Post Tags: EuMW2011 »
European Microwave »
FSW »
Spectrum Analysis
When we talk about spectrum analysis what do we consider the most important aspects? Normally it depends on whether your background is based on time domain (scopes) or frequency domain measurements.
The scope guys will tell you bandwidth is everything so you can see the smallest glitch in a time domain signal. If you want to inspect all the detail possible in the frequency domain a spectrum analyzer is really the only tool you really should be using. The frequency domain guys will tell you its all about dynamic range (spurious free dynamic range) and phase noise performance.
Without these aspects it is not possible to get the detail required on the smallest spurs that may be created in a device under test. Or to characterise distortion performance of amplifiers with precision. Scope A/D converters have very wide bandwidth but do not have the resolution or spurious free dynamic range required to create an accurate picture of spurious and other phenomena in the frequency domain.
With this in mind, R&S has released what is seen as the highest performance spectrum analyzer ever to the market and below is an overview of what makes it so high performance, and why it is important for measurement tasks in the frequency domain.
R&S FSW Performance Overview – all the important points that make it the highest performing spectrum analyzer ever
Continue reading » » The highest performance spectrum analyzer ever – R&S FSW
Posted in RF and Microwave News, Test and Measurement
Post Tags: FSW »
R&S »
Spectrum Analysis »
Spectrum Analyzer »
Vector Signal Analyzer
The latest 1.61 SP2 firmware for the R&S FSV Signal Analyzer sneakily included noise correction as a free of charge upgrade. This functionality is normally reserved for high end instruments as it gives an improvement in dynamic range for any given resolution bandwidth and allows measurements of signals close to the noise floor of an analyzer. Our previous post on this topic explains how this works and also gives information on R&S application note 1EF76: Improved Dynamic Range with Noise Correction. If I get time and happen to grab some screen shots in the coming weeks I’ll make sure I post them to give you an idea of what it looks like on the FSV.
Posted in RF and Microwave News, Test and Measurement
Post Tags: noise correction »
Spectrum Analysis »
Spectrum Analyzer
In a previous post we talked about noise figure and how ultimately thermal noise limits the performance of a spectrum analyzer. This limits our ability to see signals that would be in the noise floor of the analyzer. For quite some time it has been possible to “remove” the noise of a spectrum analyzer from the measurement to give an improved dynamic range. This can be extremely useful when trying to view spurs or improve channel power measurements and can give the user a significant performance increase (depending on settings) and allow us to measure almost to the thermal limit of -174dBm/Hz at lower frequencies (1GHz).
Sometimes it may be difficult to measure spurs due to them being such low power and also the power of a carrier being fed into the analyzer which may have the effect of over driving the input. The input signal may also be pulsed (as in this case) which means a longer sweep time is needed to see the whole spectrum. The challenge here is the trade off in resolution bandwidth, sweep speed and dynamic range.
The plot below shows the noise floor of spectrum analyzer with noise correction disabled and a normalised noise floor (1Hz RBW) of around -135dBm/Hz at the 11.5GHz center frequency. In this application we still want to see the peak of our input signal and do a relative measurement between the peak and any spurs or the noise floor, hence the use of delta markers on the screen.
Continue reading » » Spectrum Analyzer Noise Correction for Increased Dynamic Range
Posted in Test and Measurement, Tutorials
Post Tags: Noise Figure »
Spectrum Analysis »
Spectrum Analyzer »
spur search
