antenna-theory.com

[Sekutma]

Hello, I am currently working on a product that is essentially 2x2 MIMO in the cell bands (698mhz to 960mhz, and 1700mhz to 2700mhz). We are trying to figure out the best spacing and I have been tasked to find this ECC number. So after several days of trying to figure this out I am back at square one.

I have been trying to use your simplified formula (using the S12 measurements). The problem is that my ECC at the lower frequencies is lower then the ECC at the higher frequencies which doesn't make sense. At higher frequencies the spacing should be more then enough.

If anybody could help I would greatly appreciate it. Starting with what measurements I need to take and what output from the VNA (smith? polar? log mag?). It has been several years since I took complex math and learned about S-parameters but I don't need a full explanation of either. Just a nudge in the right direction should get me back on track.

In case this does matter we use an HP 8753D for our VNA. We also have an Anritsu MS2026A for our anechoic chamber but I can use that for this if need be.

[admin]

Why do you expect ECC to be the same at the low and high frequencies?

[Sekutma]

I don't expect it to be the same. I would expect it to be lower at higher frequencies. The wavelength of 698 mhz is larger then 2700 mhz. Say I'm getting a ECC of 0.472 at 698 mhz, then the ECC at 2700 mhz shouldn't be 0.982. I would think closer to 0.1 or even lower. That is just a guess though, I would much rather just calculate it correctly.

[admin]

OK, I misunderstood. You are correct, ECC should decrease with higher frequency.

Have you confirmed that if you put the same antenna pattern into your program you get an ECC of 1.0?

If you use the isolation (s-parameter) version for Ecc (valid for efficient antennas), then you just save the 2-port s-parameter file. This will give you s11, s12, s21, s22 (all complex numbers, depending on the file type they will give you either real+imag or mag/phase). Then you should just be able to plug in the values into equation [3] on here:
http://www.antenna-theory.com/definitions/envelope-correlation-coefficient-ecc.php

Note that smith/polar/logmag are all different representations of the same s-parameters, but all use the same underlying data.

[Sekutma]

When you say save are you talking about from the analyzer to an output file? The 8753D uses a floppy drive so that may not be possible. I was basically printing off the screen captures and then running the numbers by hand. Of all the output formats, I used "real" and "imaginary" to build my complex numbers. Which are different from the smith chart numbers (should those be the same?).

Let's back off this specific calculation for the moment and let me ask about isolation. Is there a magic number for isolation? Essentially I have been told to get my isolation to under -15 dbi. Even with my elements 5 inches apart I can't get my isolation under -10 dbi. However in my hand I hold another antenna with only 4 inches of separation and it gets -15 dbi! At 700 mhz a quarter wavelength is approximately 4.22 inches so somehow they get lower isolation under a quarter wavelength of separation.

I was hoping to use ECC as a way to help me figure out how to get my isolation better but it's feeling like an unnecessary headache at this point. The goal is to build something that works for all the cell bands with a low enough isolation to compete. Despite my SWR being lower and gain being higher I am being told my isolation will keep it from performing better.

[admin]

Yes, you're complex numbers read off the smith chart should be the same as what is exported in the s-parameter file. There is only one value for s11, it isn't open to interpretation.

Isolation is pretty important for MIMO operation. Basically if the isolation was 0, it would be the same as having one antenna. From data I've seen, >13 dB is good. Everyone comes up with their own metrics though, adn there's no absolute answer.

ECC is much less useful than isolation. You won't learn anything from it. So I would recommend focusing your attention on s12 (isolation) measurements only if that is what you are after. Don't make stuff harder than it needs to be.

I can't debug the setup on isolation without lookjing at exactly how things are grounded and relative orientations. I would recommend just trying new experimetns to further the grasp on the situation.

[Sekutma]

Thank you very much! I'm making an assumption that because you are admin that you are the site owner. I have been using this information on and off for two years now so thank you very much!

[admin]

No, my name as Avery D. Miniwood, I just concatenated it to admin. Sorry for the confusion