Click here to go to our page on measuring noise figure using Y-factor (might not be live yet!)
Click here to learn how to interpolate using Excel
New for December 2024. Long ago we provided an explanation of how to use Excel for interpolating data. It is one of our more popular pages, Besides engineering, as it holds appeal for people in financial industries. We hate to say it, but using Excel for this mundane task is pretty painful.
Recently we filmed a video on how to measure noise figure using Y-factor. It is explained on this page. We took 100 points of data from 10 MHz to 5 GHz, which works out to a very weird frequency step, and a ton of frequency points. Once we saved the Y-factor data, we needed ENR data over the same frequency set.
A light-bulb went off... why no use Microwave Office (or any other EDA software that deals with SNP files) to do the heavy lifting for interpolating the data? Trust us, this is way easier than trying to use Excel.
The first step is you have to enter the ENR points from the noise source into an S1P file, we did this in Excel before we stored it as a text (S1P) file. As you will be hand-typing the ENR data be sure to check for accuracy. The "angle" is meaningless but required. We filled the angle column with zeros for all the points. And don't forget the line at the top that tells MWO that units are GHz, data is in magnitude (not dB) and system impedance is 50 ohms. None of this is important, it is just needed to fool the squirrels inside the Microwave Office solver.
# GHz S MA R 50.0
1 19.04 0
2 19.02 0
3 18.94 0
4 18.75 0
5 18.58 0
6 18.39 0
7 18.21 0
8 18.04 0
9 17.82 0
10 17.73 0
11 17.47 0
12 17.33 0
13 17.09 0
14 16.96 0
15 16.8 0
16 16.59 0
17 16.46 0
18 16.3 0
19 16.2 0
20 16.11 0
21 15.98 0
22 15.95 0
23 15.92 0
24 15.75 0
25 15.76 0
26 15.75 0
26.5 15.73 0
27 15.81 0
28 15.72 0
29 15.75 0
30 15.82 0
31 15.85 0
32 15.97 0
33 16.04 0
34 16.13 0
35 16.29 0
36 16.32 0
37 16.47 0
38 16.51 0
39 16.59 0
40 16.54 0
41 16.37 0
42 16.10 0
43 15.69 0
44 15.07 0
45 14.19 0
46 13.60 0
47 13.46 0
48 13.14 0
49 12.53 0
50 11.63 0
Our frequency step works out to
50.40404040404040000000 MHz
That's ugly. But it will all come out in the wash. You just need to set up a frequency list in MWO that hits the correct points, like this:
Now let's drop the S1P file into a schematic and add a port:
Then just run a simulation and "graph" S11 magnitude into a table. These data can be selected and copied, for export into Excel. Here is the table that MWO generated.
|S(1,1)| (GHz) |S(1,1)|
ENR data ENR data
Frequency
0.01 19.06
0.0604040404 19.059
0.1108080808 19.058
0.1612121212 19.057
0.2116161616 19.056
0.262020202 19.055
0.3124242424 19.054
0.3628282828 19.053
0.4132323232 19.052
0.4636363636 19.051
0.514040404 19.05
0.5644444444 19.049
0.6148484848 19.048
0.6652525252 19.047
0.7156565656 19.046
0.766060606 19.045
0.8164646464 19.044
0.8668686868 19.043
0.9172727272 19.042
0.9676767676 19.041
1.018080808 19.04
1.068484848 19.039
1.118888889 19.038
1.169292929 19.037
1.21969697 19.036
1.27010101 19.035
1.32050505 19.034
1.370909091 19.033
1.421313131 19.032
1.471717172 19.031
1.522121212 19.03
1.572525252 19.029
1.622929293 19.028
1.673333333 19.027
1.723737374 19.026
1.774141414 19.025
1.824545454 19.024
1.874949495 19.023
1.925353535 19.021
1.975757576 19.02
2.026161616 19.018
2.076565656 19.014
2.126969697 19.01
2.177373737 19.006
2.227777778 19.002
2.278181818 18.998
2.328585858 18.994
2.378989899 18.99
2.429393939 18.986
2.47979798 18.982
2.53020202 18.978
2.58060606 18.974
2.631010101 18.97
2.681414141 18.965
2.731818182 18.961
2.782222222 18.957
2.832626262 18.953
2.883030303 18.949
2.933434343 18.945
2.983838384 18.941
3.034242424 18.933
3.084646464 18.924
3.135050505 18.914
3.185454545 18.905
3.235858586 18.895
3.286262626 18.886
3.336666666 18.876
3.387070707 18.866
3.437474747 18.857
3.487878788 18.847
3.538282828 18.838
3.588686868 18.828
3.639090909 18.819
3.689494949 18.809
3.73989899 18.799
3.79030303 18.79
3.84070707 18.78
3.891111111 18.771
3.941515151 18.761
3.991919192 18.752
4.042323232 18.743
4.092727272 18.734
4.143131313 18.726
4.193535353 18.717
4.243939394 18.709
4.294343434 18.7
4.344747474 18.691
4.395151515 18.683
4.445555555 18.674
4.495959596 18.666
4.546363636 18.657
4.596767676 18.649
4.647171717 18.64
4.697575757 18.631
4.747979798 18.623
4.798383838 18.614
4.848787878 18.606
4.899191919 18.597
4.949595959 18.589
5 18.58
Below, we plotted the original ENR table (purple) and the interpolated data (blue). Note that we extrapolated outside of the known data (10 MHz data needed but ENR table only goes down to 1 GHz). It appears that MWO didn't do anything stupid and blow up. Impressive!
We realize it would be just as easy to use MATLAB to do interpolation. What you use to do interpolation depends on what applications you have on hand. In our case we have MWO but not MATLAB. Use whatever you have and move on to analyzing the data rather than crunching it. That's the hard part!