Playing with stubs

I cut a couple of stubs for 1090 Mhz and did a little testing using RTLSDR Scanner to see what effect they’d have on reception. It seems the shorted stub does a decent job of isolating the desired frequency and the open one attenuates it pretty well. Can anyone comment on the difference in the “base” noise level, for lack of a better term? The bottom line that seems much lower with the open stub than in the scan with the shorted stub.

The stubs

No stub

Shorted stub

Open stub

Congratulations, this is good stuff!

Beware in making interpretations – the vertical scales among these plots are different! Easier to compare for small differences if the vertical scales are the same.

Comparing the shorted stub and open stub plots – the shorted stub is a bandpass filter and the open stub is a notch (band reject) filter, both looking to be centered around 1090.

In both plots the amplitude of the spike around 1036 is about the same, -36 dB.

Looking at the spikes around 1066 and 1080 in the open and shorted stub plots, as well as the range around 1090, the shorted stub plot shows more energy around 1090 and attenuation of the spikes at 1066 and 1080, as we’d expect for a bandpass filter. The open stub plot shows significant attenuation of the energy around 1090 (around 12 dB) and slightly higher energy around 1066 and 1080 – showing the bandwidth of the notch around 1090.

As to the noise floor, given the difference in vertical scales, it looks close, with the software reporting +/- 2dB among the three plots (also shows a noisy receiver).

Reading differences from these kinds of plots produced with live signals can be difficult. Longer durations (seconds) with averaging - max hold can help. When I do these kinds of captures on my spectrum analyzer I do a max hold for a few seconds (using the same interval for subsequent captures). My spectrum analyzer (and when I use it with the scalar network analyzer) don’t “help” by auto-ranging the amplitude (vertical) scale… In the absence of lab instruments such as a vector or scalar network analyzer, a wideband noise source or comb generator can be used to provide reference/calibration spectra.

I like hard data!

bob k6rtm

Very interesting. I’m going to try these scans again with longer durations (will take a long time to do the scans). But - this is what’s weird - I didn’t make any changes to the scan parameters between scans. All I did was swap out the stubs. I have no idea why the scales are different.

I’d bet it’s the software automatically “being helpful” and adjusting the scale to make best use of the screen. Should be a way to override the vert axis settings, or I hope there is!

bob k6rtm

@jepolch Good work!
What is the frequency you are trying to get rid of and what is the length of your stubs? I wasn’t sure if the squares were inches or centimeters.

You’re right. It was set to “auto range level”. I turned that off and will re-do the scans. Also, you’ll notice the stubs are different. I made the white one recently but made the black one some time ago. I used the black one for convenience, rather than make a new one (and use a $1 coax connector). But, even though both stubs have a VF of .82, they don’t have identical shielding or connectors, so I’m going to make another stub identical to the white one, except it will not be shorted. That should produce more consistent results.

1090 Mhz. The squares are inches. The stubs are cut as close to 56.58 mm as I could make them.

I re-did the open stub to make it identical to the shorted stub (except not shorted). I set RTLSDR Scanner to not auto level. I ran 8 sweeps with each configuration with a 1 second dwell time.

No stub

Shorted stub

Open stub

for most of us, is tuning the [shorted] stub likely to be a mater of trial and error?

Very much so – as the length includes the connecting tee, and is a function of the velocity factor of the cables involved.

The good news, though, you can leave the stub a bit long, make a measurement, trim and repeat (getting as close as you dare).

I’m thinking of looking at the overall response of a shorted stub centered on 1090 (a band-pass filter). What’s the insertion loss, and how well does it do, particularly in that difficult region from 800 to 960 or so? Some fun for this wekeend…

bob k6rtm

yeah, it’s a pain, but you only do it once…

I did some more testing today using a cantenna with and without a shorted stub. The results were underwhelming. I’m using an older DVB-T dongle with (R820T) at the end of a 10 meter active USB cable. I’m thinking of moving the dongle to the roof using this cable.

Cantenna with no stub

Cantenna with shorted stub

Rant: It sure is a pain having to upload pictures to another web site and copy the links here. Will we ever be able to upload pictures directly to FlightAware?

I’m thinking of doing surgery on a power inserter for this and removing the choke, this is because I have an antenna amplifier under the antenna - putting a shorted stub there would need a blocker for the DC that the amplifier would pass to the Satellite dish LNB (if it were used as intended).

Or would I be better to use something like a 100pF capacitor as the short on the stub?

I have the same dilemma. My antenna is on the roof (now covered with 6 inches of snow). The amp is attached directly to the antenna, so I can’t insert the shorted stub into the line. The power injector is inside the house though, and I have a spare to experiment with. Can you give more detail on the modification to your inserter/injector for folks like me who are kind of clueless on the electrical theory side of things but who are handy with a soldering iron? With your mod would it be possible to use a shorted antenna without the need for a DC blocker? Details, man! :laughing: Thanks.

Google for Type F dc block – that will work if you are using a satellite-style LNA and want to put the shorted stub between the antenna and the LNA (antenna → dc block → LNA → power injector → SDR). Lots of them for $3 or thereabouts.

If you’re not using type F, but instead using SMA, your DC blocks are going to be more expensive; check eBay.

And if you want to put the shorted stub between the LNA and its power injector, then you’re looking for a different kind of beast. I can draw one out if someone is interested; you could do one with a pair of power injectors connected more or less back to back and putting the tee and stub between them (this is probably illegal in a number of Southern states in the U.S. Well not really, but any RF engineer seeing such will run away from it wringing his hands and crying out “Unclean! Unclean!”). Stub tuning will undoubtably be changed, so the stub will have to be measured and cut for this setup.

Another way to get power to an entire system is through inexpensive POE adapters. I picked up a TP-LINK TL-POE200 adapter kit:

Lots of places have them for around $25. This one gets bad reviews from folks who expect it to do Gigabit. We don’t have that problem with the Pi… The kit contains a wall wart, POW injector, POE splitter, and some short cables. You will have to do something about going from the barrel dc output connector to your Pi.

I’ve got a Pi happily running on the end of a 50 foot Ethernet cable using a pair of these, with the goal of putting a system up in the attic and greatly shortening an antenna run.

bob k6rtm

Hi Bob. I’m familiar with F type DC blockers. I have a couple, but I’m not using them now because neither of my antennas is shorted. But maybe I misunderstood Peter saying he had a mod for a power injector that would allow you to put a shorted stub on the line without the need of a blocker. I can’t get to my antenna now, so I’m not able to go up and put the stub and DC blocker in place. I was thinking there was something I could do inside the house. I have one power injector on my line to the antenna now. I have a spare power injector.