Using an antenna analyzer as a network analyzer...

[cillcalley]

I've searched the Web, but I can find no mention of using a common USB antenna analyzer (such as the Mini60/SARK-100 with software) to graph the band-pass of an HF filter. To me, it seems like a no-brainer, but perhaps I'm missing something critical here:
Antenna analyzer-->LPF/HPF/BPF circuit-->50 Ohm termination.

Sure is cheaper than a network analyzer, vector OR scalar!!!! Thoughts?

-Bill

 

[JimB]

I am familiar with antenna analysers, I have an AIM4170 myself.

Until I saw your post, I had never heard of the Mini60 or Sark100 so I did a bit of googling and found various comments both good and bad.
There seem to be some variations which originate in China and come without documentation and do not work correctly.

As for using an antenna analyser to measure the response of a filter, that would not be possible using the Mini60 or the AIM4170.
An antenna analyser has only one measurement port and so it can only measure the impedance. Looking at a standard 50ohm termination via a filter will not give you the frequency response of the filter. Yes it will measure the impedance seen through the filter, but that is not the same as measuring the frequency response.

To measure the frequency response you will need an analyser with two ports, a source and a detector such as the MiniVNA.
Look here:
http://miniradiosolutions.com/

The downside of course is that the MiniVNA costs a lot more than the Mini60/Sark100.

JimB

 

[cillcalley]

Thanks JimB -- I was afraid of that (but hope springs eternal!). But with a vector antenna analyzer (like the Mini60), couldn't you get a decent indication of the filter's passband via the return loss vs. frequency measurement function?

 

[JimB]

To put your hypothesis to the test I did some tests on a transmitter output filter.
The filter is a low pass filter for the 7MHz amateur band, the cut-off frequency is around 10MHz.

I tested the insertion loss of the filter, using a spectrum analyser and tracking generator.
The results are shown here:

The horizontal scale is 10MHz/div, and the vertical scale is 10dB/div.

I then terminated the output of the filter with a 50Ohm load, and measured the return loss at the input of the filter using an antenna analyser.
The results are shown here:

I would not like to use the return loss as an estimation of filter insertion loss.

JimB

 

[cillcalley]

Wow JimB, thanks for doing that!! I'm trying to do the same on my end, but am having interface problems with my antenna analyzer. However, since I have no real test equipment, I could definitely live with the insertion loss giving me a decent indication of the filter's passband!

 

[JimB]

I could definitely live with the insertion loss giving me a decent indication of the filter's passband!

Mmm.. the insertion loss IS the passband, ie the response to a signal passing through the filter. Just different terminology.

JimB

 

[cillcalley]

Mmm.. the insertion loss IS the passband, ie the response to a signal passing through the filter. Just different terminology.

JimB

Sorry; I meant return loss. Senior moment.

 

[cillcalley]

Also, since in an ideal filter the square root of 1-s11^2 equals s21, it would mean that it would be quite easy to write an Excel formula to convert an entire table of s11 values directly into an s21 graph. This would most certainly give me an approximation of the bandpass of a real-life filter...

 

[Mikebits]

Jim, I would love to see your lab, I bet it looks pretty cool

 

[Mosaic]

Over the last couple of months I've been researching VNAs etc.
For HAM use and such I'd suggest the new VNWA kit.
**broken link removed**

Because I am interested in blue tooth and wireless comms I needed something a bit more capable:
I eventually purchased a HP8753D (0.03Mhz -3.00Ghz) with the built in dual S ports with a valid cal. certificate for $2100 on Ebay. This also requires about $400 worth of SOL cal. gear with a custom cal. constant disc., $100 of APC7 adapters etc.
The VNA is (28c64EPROM) 'hackable' to open up several 'options' such as Time domain (For TDR) and swept 2nd & 3rd harmonics. Likewise the OS firmware is upgradeable to the latest available via programming the 4, 27c020 EPROMs, which are socketed.

There is even an outside possibility of a 6Ghz option hack as the stock ALC oscillator down converts to 3Ghz from a variable 3.0- 6.8Ghz onboard source and a 3.8Ghz fixed cavity source.

With the EPROM feature unlocking, the VNA 'value' jumps by about 50%. One wonders why EBAY sellers have not figured this out...and 'upgraded' the units they sell.

 

[cillcalley]

Thanks Mosaic. The VNWA3 looks great, but is above my pain threshold, price wise. (Their SOL Cal kit are a real steal -- I forget what my former employers paid when they bought the HP/Agilent ones, but it was BIG money!).

 

[Mosaic]

Well a DIY Spect Analyzer can give u the scalar results.
https://www.robkalmeijer.nl/technie...hniek/hambladen/qst/1985/11/page23/index.html
or
https://www.qsl.net/n9zia/spec/

 

[Mosaic]

Hmm...this also looks pretty good:
https://www.scottyspectrumanalyzer.com/

 

[Mosaic]

There's a 'group buy' of the PCBs etc. for the scotty SA going on now in the yahoo group. Somewhere around $35 for the PCB panel with the sub boards.
Perhaps you can jump in as it is opportune, plus any build problems will be able to be supported as lots of those folks have DIY experience with this SA.
You can even go for the VNA version as its all 'additive' features.

https://groups.yahoo.com/neo/groups/spectrumanalyzer/conversations/messages/11524

 

[dr pepper]

I have a arduino based analyser, the 50r comparison resistor is in a socket, I just change it out for a 10k or whatever to check the response of a filter or xtal, its probably not as good as a 500 notes version, but it cost next to nothing.