This tech note serves to explain how to interpret the spurious data we provide on our datasheets, how we obtain this data, and things to keep in mind about spurious products when designing and selecting components for a system. If you are interested in what spurs are, how to measure them, or which spurs are important, see here, here, and here.
How Do We Measure Spurs?
At Marki Microwave, we test mixer spurs in two ways: on a spectrum analyzer and using a nonlinear vector network analyzer. For datasheet characterization, measurements are typically performed on a network analyzer and confirmed with ‘spot checks’ on a spectrum analyzer. For example, down-conversion spurs are tested by injecting a swept fundamental RF tone and mixing it with a swept fundamental LO tone to produce a fixed IF tone. To measure the down conversion spurs at the IF output, we tune the test receiver to sample the spur of interest (i.e. 2RF x 2LO, 3RF x 1LO, etc.) within the usable IF bandwidth since out of band spurs are assumed to be filtered out. Then, using all of these discrete data points, we convert to dBc, take an average, and report that average in the spur table. Since spur table values are averages, the spur level for a given frequency plan can vary quite a bit from the listed number. Looking at a sample spur table against the spurious suppression plot from any of our mixer datasheets displays this fact.
Below is the 2IFx1LO spurious suppression plot of the MM1-0222H as well as a truncated spur table next to it taken off of the datasheet. To create this up-conversion spurious suppression plot, a swept fundamental LO and a fixed IF were injected to the mixer and the test receiver tuned to sample the 2IFx1LO spur. The specific frequency plan measured is:
*Note that although the RF/LO sweep goes outside the mixer’s guaranteed spec. of 22 GHz we only consider the spurs that are generated within the mixer band when we average the values for the spur chart
We follow this same procedure to obtain data on every spur up to the 5IF x 5LO spur.
Notice that the spurious suppression varies by almost 30 dB across the band and is far from a constant value, yet the spur table shows just 69 dBc, the average across the band. These perturbations show up due to the sensitivity of non-linear devices to small changes and thus, the spur table measurements should be taken with a grain of salt, as the results across the band will vary.
In addition to RF/LO frequency variation, spur levels can vary due to IF frequency, LO power, waveform, port impedances, and other packaging concerns. Depending on the mixer technology (Legacy/T3 vs MMIC/MT3) there may be unit to unit variation that can be minor to substantial as well.
How Can I Get Information on Spurs?
At this point it is becoming clear that the spur table is a tool, and just like any tool in an RF designers tool kit, it is only useful if the engineer knows how to use it. If a closer approximation is needed, one could download the simulation files for our non-linear devices and run a simulation in ADS or MWO. To get a rough idea of the simulation accuracy, below are the spur levels of an ADS simulation of the MM1-0312H compared with the same frequency plan tested on a spectrum analyzer. Here we see that actual values deviate slightly, but the simulation can give you a very good idea of approximate spur levels using your frequency plan. This significant increase in accuracy compared to the spur table is due to the simulation taking into account your unique frequency and power parameters.
When it comes to accurate results however, there is no substitute for measured data. The best way we can recommend to characterize the spur levels of your design is to run the test yourself, or email [email protected] with your frequency plan and desired spurious data and we will run the tests for you!
- Marki Microwave only measures the spurs that show up in band of the mixer
- Spur tables show the average of the varying spur level across the band which will not be equivalent to the spur levels of a CW test
- Many factors can affect spur levels across band so spur tables, datasheet plots, and simulations should be used as design tools that have limits and cannot fully characterize the performance
- There is no substitute for measured data when accuracy is imperative and Marki Microwave is happy to help you by running tests for you or sharing our VNA template with you to use on your own VNA
I have some questions to ask.
Let’s say that my wanted RF is at 850MHz and LO is at 800MHz. IF is at 50MHz. Therefore image is at 750MHz (low side injection).
But some says that the image is at 1650MHz (800+850). I am confuse as 1650 is 1,2 response where w relates to 2xLO. Why is it some people says image is the sum of wanted RF and LO?
Regarding half-IF, is it fair to say that IF a spur is at half-IF (RF+IF/2) it would fall into IF.
Finally if the LO to IF is less, are there techniques to improve isolation or is it justcreduce LO drive