Posts by AB6RF

    To visualize the radiation patter, and specifically the -10 dB beamwidth, I have two more plots to share.
    Here again the Theta angle is from the Z-axiz (zero degrees is straight up), and Phi angle is from the X-axis.

    Each plot shows four "slices", at 0°, 45°, 90° and 135° Phi.


    4-turn Helix


    Let's see how bad job Google translate will do...

    Können Sie ein Relais oder einen FET mit Zeitverzögerung hinzufügen, sodass die PA erst dann mit Strom versorgt wird, wenn sich die Spannung stabilisiert hat?

    Can you add a relay or a FET with time delay, so that power to the PA is connected only after the voltage has stabilized?


    I'm not an antenna expert either, just an amateur :)

    Ideally, yes a perfect circular polarization has an axial ratio of zero.
    Anything greater than zero, it's an elliptical polarization.

    In some ways, pure linear polarization and perfect circular polarization are special cases of the elliptical polarization.
    (axial ratio becomes either infinity, or zero)

    But just like with my "obsession" with the phase noise, it's very interesting to measure and try to optimize, but in reality it doesn't matter that much. It just needs to be "good enough". We're not doing EME here.

    Let's look at the magnitude of the polarization loss vs. axial ratio.

    This table shows the power loss (second line from the left), as a function of the TX axial ration (left), and RX antenna axial ration (third from the left).

    I added three examples in there.

    Green line: 3 dB TX antenna axial ratio (my POTY), 1 dB RX antenna axial ratio -> loss = 0.2 dB

    Purple line: 3 dB TX antenna axial ratio (my POTY), 0 dB RX antenna axial ratio -> loss = 0.1 dB

    Red line: 20 dB TX antenna axial ratio (= almost perfect linear polarization), 0 dB RX antenna axial ratio -> loss = 2.2 dB

    So a 3 dB axial ratio TX antenna causes a negligible loss (would be a concern for a EME station, but not here).

    But yes, I also agree, based on the S11 I have with this POTY, the "double dip" is there, and the hump is exactly at 2.4GHz, I was expecting, or guessing, that the resulting polarization would be more ideal.

    Last thing I want to point out, the axial ration is better in the bore axis, and becomes higher when you go off-axis. The 3 dB result is integral over the (almost)entire radiation pattern.


    I analyzed the raw antenna test data in Excel. The Theta angle measurement was done every 10°, so for a -10dB beamwidth I had to interpolate between the measurement points. So these are not absolute accurate.

    POTY -10 dB beamwidth = 110°

    4-turn Helix -10 dB beamwidth = 86°

    (if you include the impedance matching vane, it's closer to a 4.5 turns total)

    This is calculated for the total power. (Sum of each linear polarization)

    And also the calculated axial ratio integrated over the entire pattern.

    POTY axial ratio 3 dB

    Helix axial ratio 1.4 dB

    Hello Robert,

    The test chamber software does not report the -10dB beamwidth, but I can calculate it from the raw data.

    Give me few days, I need to get access again to the test chamber, I saved the raw data but it's in binary format, I need export it into some other format for analysis.

    Yes, run out of time yesterday, but here are the Helix numbers.

    About 4.2 turns, 42mm diameter, 2mm copper wire. S11 @ 2400MHz = -30dB.

    Efficiency = -0.5 dB (= 89%)

    Directivity = 11.2 dB

    Gain = 10.7 dBi

    -3 dB Beamwidth (E plane) = 47°

    -3 dB Beamwidth (H plane) = 50°

    2D radiation pattern in X- and Y-plane.


    I finally had an opportunity test my POTY and Helix antennas in an antenna test chamber, here are some of the results, first the POTY results.

    I tuned this POTY just using the S11, no further tuning was done to optimize the radiation pattern.

    For a reference, here's my final POTY S11.

    Here's a picture of the POTY in the antenna chamber, and the coordinates for the plots.

    Efficiency = -0.5 dB

    Directivity = 10.3 dB

    Gain = 9.8 dBi

    -3 dB Beamwidth (E plane) = 56°

    -3 dB Beamwidth (H plane) = 64°

    Here are two 2D plots of the radiation pattern, one is in the X-plane and other is in Y-plane.

    (Total power, sum of "vertical" and "horizontal" polarizations)

    And here's an animation of the 3D pattern.

    POTY 3D Animation

    Another very interesting thing is to take a look at the two linear polarization fields created by the POTY.

    The theory is that it should create two fields that are equal in magnitude, but 90° out of phase.

    Here's a polarization plot of the main beam. Each plot is one polarization.

    (Theta and Phi, in ham radio we would call them Vertical and Horizontal)

    Very equal, and almost exactly 90° out of phase!

    In other words, on the bore axis of the main beam, the axial ratio is very low.

    But further off-axis, the pattern gets more messy, and the axial ratio gets larger.

    Integrated over the (almost)entire pattern, the average axial ratio is about 3 dB.


    I should explain the coordinate system.

    Phi angle is in the X-Y plane, with 0° being towards the positive X-axis. Similar to what’s usually called the Azimuth.

    Theta is in Z-Y plane, with 0° towards the positive Z axis (= straight up), could be called the Elevation.


    Today I modified my "NXP TFF1014" LNB to use an external 25MHz reference.

    I removed the internal crystal, and connected a small 1:1 transformer to the reference pins of the PLL chip.

    The "upper" PLL chip in this picture is now disabled, and the "lower" one is the active, and the crystal in the middle is replaced by a transformer.

    Looks like the modification was successful.

    1) With the internal crystal the frequency error was about 177kHz, with the external OCXO the frequency error is about 20kHz.

    (temperature in my garage is currently about 15°C)

    Also, with the OCXO there's no detectable frequency drift after about 10 minutes of initial warm-up time.

    2) With the external OCXO, also the phase noise is slightly better than with the internal crystal.

    PN with the internal crystal, measured at the 739MHz IF output.

    PN with the external OCXO

    Again, the green line is the PN of the 10.489GHz signal generator, measured directly with a straight connection from the generator into the analyzer.

    The good phase noise of the OCXO is one of the reasons why I didn't want to use a PLL based reference oscillator.

    If locked to the GPS, they can have a better long-term frequency stability, but often the phase noise is much worse.

    On a related note, I noticed that I had made a mistake when writing down the part numbers in this LNB.

    I thought the NXP chip is TFF1014. It's not, it's actually a TFF1013.

    Looking at the TFF1013 and TFF1014 data sheets, the newer TFF1014 has 2dB lower noise figure.

    But it doesn't really matter much because there are two LNA stages in front of the PLL chip.

    Doing some quick math, with TFF1013 (NF = 9dB) the cascade NF of the LNB is 1.4dB.

    With the TFF1014 (NF = 7dB) the cascade NF would be only marginally better at 1.3dB.


    Continued with my science project. (I love science projects!)

    Today I attempted to compare the performance of the two LNBs that I have.

    I don't have suitable dish, and I certainly don't have "visibility" to the QO-100, so my testing is done just using test equipment in my garage.

    I set up each LNB one at a time on a tripod (no dish) pointing directly to a test antenna that was connected to a 10GHz signal generator.

    The exact test frequency was 10.489GHz.

    Then I used the spectrum analyzer to measure the phase noise, and a crude S/N of the down converted 739MHz signal at the LNB output.

    One of my LNBs is using the Rafael RT320M chip, and second one is using NXP TFF1014.

    The short summary is:

    1)My TFF1014 LNB provides about 2dB better SNR in this setup.

    2)The TFF1014 LNB has significantly more gain.

    (I almost wonder if there's something wrong with the RT320M LNB, but since the S/N is OK, I do think the LNB is undamaged)

    3) The RT320M based LNB has slightly better phase noise.

    First plot, RT320M S/N.
    In the plot you can see the noise floor of the spectrum analyzer, the pass-band of my 740MHz SAW filter, and the in-band noise floor of the LNB. S/N = 40.5dB (specific to these spectrum analyzer settings_)

    Next plot, same setup, same settings, TFF1014.

    S/N = 43.3dB

    Then the phase noise plots.

    The green "reference" in both PN plots is the PN of the signal generator measured directly at 10.489GHz.



    Happy New Year, from AB6RF

    No need for exclamation marks.

    Why? Because I like building stuff, and testing how it works.

    If YOU don't like reading about it... Please turn the VFO, as they say.

    Merry Christmas from AB6RF

    More phase noise testing.

    Today I measured the PN of the Pluto output signal at 2.4GHz.

    Looks like my used 38.8800MHz OCXO (= unknown history, could be damaged etc.) produces a slightly worse PN than the internal TCXO in Pluto.

    In the plot there are three lines.

    Black = measure PN with the OCXO

    Blue = smoothed PN with OCXO

    Green = smoothed PN with internal TCXO

    Below 300Hz, and above 100kHz offset the TCXO PN is better.

    Between 300Hz and 100kHz the PN is equal.

    It's also possible that the LTC6957 ext clock buffer in Pluto causes the PN degradation.

    It does certainly add little bit of clock jitter.

    Now I'm tempted to purchase a different 38.88 or 40.00MHz OCXO to make a comparison with the one I have.


    It's interesting.

    I'm using a Pluto HW version C, FW version 0.35

    And my external reference clock frequency is 38.880MHz.

    In my case, entering these variables does not work.
    fw_setenv refclk_source external

    fw_setenv ad936x_ext_refclk "<38880000>"

    The Pluto DOES switch to using the external clock, but the default setting for ad936x_ext_refclk persists after a reset.

    (In my case 40000115)

    So the Pluto is off frequency.

    To make it work, I have to enter:

    fw_setenv refclk_source external

    fw_setenv ad936x_ext_refclk_override "<38880000>"

    Then all is good, and the correct ref frequency setting is used by the Pluto.


    I have few used OCXO modules, and I made boards to use them for a 25MHz reference for the LNB, and a 38.88MHz reference for the Pluto.

    But like it has been discussed, the reference must have very low phase noise, so I tried to measure the phase noise of these OCXOs.

    I ran into a problem.

    I'm using a very nice R&S FSV Spectrum Analyzer to do the PN measurement, but even it has too high internal PN.

    Every module I measure shows the the same PN plot, it's the noise floor of the Spectrum Analyzer.

    So for the Pluto, what I need to do next is make a 2.4GHz test tone out of the Pluto, and then measure the PN of that with the internal TCXO and with my OCXO.

    Hopefully at the output frequency I can actually see the PN effect of the reference clock.


    Fabiano, could you share here what you had to do to make it work with the external 50 MHZ reference?

    I'm sure others would benefit from the information.

    Initially you added these two settings, but it didn't work.

    fw_setenv refclk_source external

    fw_setenv ad936x_ext_refclk ‘<50000000>’

    What was done to make it work?



    While I was planning the transceiver system, I realized I will need several different power supply rails.

    I also wanted to keep it pretty compact and efficient, so I made a custom board that fits my needs.

    The board takes a 24V input from a AC/DC power supply, adds some filtering and protection circuits.

    It provides the following outputs:

    - Relay switched 24V output for PA. Can be "always ON" or controlled by a GPIO

    - 5V (SMPS) Out on USB connector for Raspberry Pi

    - 5V (LDO) Out on USB for Pluto

    - 5V (LDO) Out for 25MHz and 38.8MHz OCXO

    - 5V (LDO) Out for PA Bias

    - 12V (LDO) Out for the LNB

    I used separate LDOs for the "sensitive" blocks to provide clean voltage and have isolation between the different blocks.


    Today redid the helix.

    I have to say the tuning of the matching strip is incredibly sensitive. I'm glad I have a nice VNA, without one it would be simply impossible to tune one of these.

    I used 2mm diameter wire in an effort to try to minimize the 10GHz shadowing.

    4 turns, 42mm helix inside diameter.

    The support arm is made of 0.25 inch Polypropylene.

    S11 = -30dB

    At some point in the coming weeks I'm hoping to measure the radiation pattern (and ellipticity) of my POTY and the helix.


    I ordered some Polypropylene (PP) for a better support for the helix.

    But in the mean time, today I made the POTY feed.

    (I think it's a right of passage for all QO-100 users to make at least one POTY, hi)

    I had the parts made out of 1,5mm copper.

    After testing it, I think it would be better to use thinner material for the feed element.

    With 1,5mm material, the resonances are not very pronounced. Maybe a thinner element would give sharper resonances.

    I did have to use some dielectric (foam PVC) to tune my POTY, but the end result seems OK.

    The two symmetrical resonances are there, on both sides of 2,4 GHz.

    And on 2,4 GHz the S11 is about -15dB. Not great but OK.

    Interestingly, by loading the antenna with a Plexiglas sheet directly in front of it, it gets even better.