making my own antenna analyzer

[djacobow]

I'm thinking of building my own simple antenna analyzer. There are some examples out there, like the K6BEZ example and its variations. They use an AD9850 module or similar as a signal source and a  SWR bridge and a diode detector. It sounds, though, like the diodes are operating right at the beginning of their conduction, so the response is nonlinear and not so good. One variation uses an AD8307 or similar to function as a detector instead. The AD I would use would be the standard 8-10-12 bitter available on a cheap MCU.

I'm pretty new to RF. What are the gotchas I need to worry about in making a half-decent instrument? Do I have to do anything special with the impedance of the antenna traces? I'm guessing no since the distances on the board are small fractions of the HF wavelengths I want to work with.

What else can go wrong?

I believe the AD9850 has a sine wave output, but some other parts that can go faster (SiLabs 54xx) generate a square wave. I'm correct in assuming that the harmonics will have different SWRs and so they will just mess up a the reading. In that case, I'll need an array of filters and the ability to switch between them depending on the band. That seems like a fair bit of extra complexity I'd like to avoid if I can.

-- dave j

[Theboel]

I'm thinking of building my own simple antenna analyzer. There are some examples out there, like the K6BEZ example and its variations. They use an AD9850 module or similar as a signal source and a  SWR bridge and a diode detector. It sounds, though, like the diodes are operating right at the beginning of their conduction, so the response is nonlinear and not so good. One variation uses an AD8307 or similar to function as a detector instead. The AD I would use would be the standard 8-10-12 bitter available on a cheap MCU.

I'm pretty new to RF. What are the gotchas I need to worry about in making a half-decent instrument? Do I have to do anything special with the impedance of the antenna traces? I'm guessing no since the distances on the board are small fractions of the HF wavelengths I want to work with.

What else can go wrong?

I believe the AD9850 has a sine wave output, but some other parts that can go faster (SiLabs 54xx) generate a square wave. I'm correct in assuming that the harmonics will have different SWRs and so they will just mess up a the reading. In that case, I'll need an array of filters and the ability to switch between them depending on the band. That seems like a fair bit of extra complexity I'd like to avoid if I can.

-- dave j

Hi Dave,

Nothing wrong with Your plan, depent with your requirement. but there is a "mature" project in HF range (0-60Mhz) in "kits" version or DIY if You insist
or ready for use up to UHF range.
N2PK
www.makarov.ca
VNWA
www.sdr-kits.net
"something more conventional"
http://www.changpuak.ch/electronics/Arduino-Shield-TOBI.php
VNWA

[VK5RC]

Have you seen the VK5JST analyser?  http://www.ahars.com.au/about/kits/
I think the construction notes also contain a theory of operation. If I recall correctly a lot went into getting the oscillator to be very constant output across the frequencies.

[Towger]

K6BEZ has designed a new version in conjunction with the Ham Radio 360 podcast.  He was on it a few weeks ago talking about the design and the possibility of increasing its range up to cover the 2m band.   I have also seen a 3rd party version which used a graphical display rather than the two line lcd which would be an interesting improvement.

[Bud]

You need to understand that power detector ICs are wide band, and will sum together all of the power at their inputs, so the output will have a sum of the signals you measure (your DDS) and everything else that makes its way to the detector input. Basically that means your measurements will be off depending on the ratio of the useful and stray power. It is one''s guess by how much and will depend on a particular environment. You could mitigate the issue by using an LPF at the detector input. This is often overlooked in simple designs. Tuned detectors do not have that problem.

[djacobow]

Hi Dave,

Nothing wrong with Your plan, depent with your requirement. but there is a "mature" project in HF range (0-60Mhz) in "kits" version or DIY if You insist
or ready for use up to UHF range.
N2PK
www.makarov.ca
VNWA
www.sdr-kits.net
"something more conventional"
http://www.changpuak.ch/electronics/Arduino-Shield-TOBI.php
VNWA

Wow, I was not aware of any of these. The TOBI is along the lines of where I was planning to go, though I was going to put the uC on the board, along with an LCD. I definitely want my final result to be a standalone handheld design with an optional computer interface.

The N2PK design has higher range and a more sophisticated approach with the quadrature DDS generators. OTOH, it seems to maybe guild the lily a bit with the battery charger, switching power supply, etc.

Anyway, this is great for me, since they both have schematics I can crib from.

Thanks for the pointers!

[djacobow]

Have you seen the VK5JST analyser?  http://www.ahars.com.au/about/kits/
I think the construction notes also contain a theory of operation. If I recall correctly a lot went into getting the oscillator to be very constant output across the frequencies.

Another one I was not aware of. Thanks for the tip!

[djacobow]

You need to understand that power detector ICs are wide band, and will sum together all of the power at their inputs, so the output will have a sum of the signals you measure (your DDS) and everything else that makes its way to the detector input. Basically that means your measurements will be off depending on the ratio of the useful and stray power. It is one''s guess by how much and will depend on a particular environment. You could mitigate the issue by using an LPF at the detector input. This is often overlooked in simple designs. Tuned detectors do not have that problem.

Just to make sure I understand what you are saying, the issue is that "actual" received signals picked up by my antenna will be measured along with the injected signal, and if they are of comparable power, I'll get erroneous readings.

But the analyzer will be injecting a +n dBm signal, where n is 0-10. I've never measured the signal coming from my antenna. Is there typically anything coming in that is remotely that large, unless there's somebody keying up next door or something?

[Bud]

That is right, i once lived in an appartment building that had cell station antennas on the roof, which rendered my wireless headset useless. Same here. You can hook up you antenna to a scope (loaded to 50 Ohm to emulate the analyzer input) and see how much RF voltage you get within the scope's bandwidth. In one instance my friend and i measured a few volts of RF, he lived in a high rise appartment building with a long wire antenna from the appartment window down to a tree. So to mitigate the problem you would need to increase the generator signal strength or put an LPF in front of the receiver, unless you use a tuned receiver as some of those analyzers do. The AD power detector ICs only look sweet for this application but when you start digging they are not. Again, it may be ok for an entry level analyzer or for one used in a fairly quiet RF environment. Otherwise readings may be jumpy.

[sever2k6]

http://ra4nal.qrz.ru/vna.shtml
https://bitbucket.org/kuchura/antennaanalyzer/wiki/Home

[C]

djacobow,
KISS can be an idea that creates problems.

It has been mentioned the need to keep signal source at fixed level for example.
The need to know the test signal vs the returned test signal and other signals.

A not quite so kiss circuit can make some of this easer.
If you run your signal generator to a splitter then you can measure the level of test signal during the test. You can swap the splitter outputs to calibrate one output vs the other output.
You now have two outputs from your signal generator, go the next step and test two antennas at same time. You can now compare one antenna to a second. If the second is just a load replacing the antenna, then you have more to work with.
Just try to match as best as you can all circuits after above splitter. Again you can swap connection to load and antenna to calibrate. You can even swap two loads.
This costs a little more, but also lets you test & improve one path with respect to other. Just match up both paths to contain the improvements.

So a kiss simple splitter can make things easer with addition of some matched circuits which can also help in long run.
 

[cncjerry]

I built and designed a bunch of them as in theory it is easy but in practice not so easy.  I ended up building two N2PK VNAs because I had two different master oscillators and had a lot of time on my hands.   The biggest issue with a lot of the simple designs as someone mentioned above is the power summation and most of that comes from broadcast bands.  You really need a way to limit the power to that wanted and the N2PK detects at near DC which eliminates a lot of the problems.

[awallin]

we have a rigexpert 600MHz at work and it is really convenient and easy to use.
they are nice enough to provide block diagrams on their website.
https://www.rigexpert.com/index?s=aa1000&f=inside
https://www.rigexpert.com/index?s=aa54&f=inside