POTY and return loss plot

  • I repeated the measurements with a short 20cm pigtail from the patch to the miniVNA Tiny and according recalibration. This is what i get now:

    This looks much more consistent with what I would have expected. Thanks for the hint DL9SEC. Those larger loops are not observed anymore (although they were to be expected as I've learned here) and the measurements are less noisy. I simply wasn't aware of the impact of the cable length. I put markers 1 and 2 where I would suspect the double dips to be located and marker three around 1.4GHz.

    G0MJW Indeed, the real test is what counts. Unfortunately my PA is not ready yet, so I can happily induldge in more or less meaningful measurements.

    To me everything looks okay-ish so far and I think I can leave my patch as it is and start doing some careful tests on the air as soon as my PA is finished.

    Just for my own understanding: taking the simplified two antenna model into account and assuming that in my measurement above marker 1 corresponds to the inductive and marker 2 to the capacitive antenna: their difference in phase, should be 90° in the optimal case. Is this correct?

    Thanks to everyone so far already

  • G0MJW we do have a link budget with is in very good agreement with the observed performance. But as you feared, this has too much NDA'ed information in it and for the sake of staying clean, I will not publish any numbers myself (sorry for being paranoid).

    However, if anybody else is generating a spreadsheet which can be verified by the QO100 community, there is nothing we can do about...

    OK - understood. Not everything in the spreadsheet will be under NDA though. I see the noise floor at ~10dB. Assuming antenna gain of 45 dB and a noise temperature of 400k I have a G/T of 19 dB. PSK Beacon SNR is 31.5 dB. This is according to SDR Console, I don't know if it is right but as long as it is consistent it does not matter.

    To equal that SNR in the same bandwidth with a similar signal, I need to transmit about 200mW.

    This is where errors start to creep in as I find if I TX at this level I am a lot weaker than some other stations who are apparently sending at the same level as the beacons.

    My dish gain on 2.4 GHz is (assuming Willi's simulation is correct) is 32.5 dB so an EIRP of 25.5 dBW is what I need to reach the beacon level. It should be mostly the same for everyone else.

    Now we could say at this point job done. To have a sensitivity such that the transponder noise is equal to the antenna noise one needs a dish with 7 dB less gain (not exactly, close enough though) other things being equal. -7 dB equates to 1/5th of the area, about 1.1m. The equal noise is an arbitrary target by the way but a fair one.

    There are some improvements to the ground station that could be made, I assumed 50% efficiency and a noise figure of 3 dB and about 120k dish temperature. Realistically, this could be improved to 200k with a lower noise figure and better side-lobes with 70% dish efficiency. That's probably the limit without going to extremes. That would allow another 3 dB and a 75cm dish.

    Part 2 after lunch

  • To get a specific SNR with known EIRP tells us what the satellite G/T is. Unfortunately I do not know what the SR bandwidth is in SDR Console but luckily I did some spectrum analyser measurements recently.

    The beauty of this is I can go back and do analysis on the data. This tells me the Resolution bandwith and other settings and allows me to measure stuff - the result of which is the PSK beacon is 18 dB above the transponder noise floor in 3 kHz. The beacon is 400 baud BPSK and from observation the majority of it's power lies within 1-1.5 kHz so we may have a correction to make.

    However, looking at the data the analyser was set to peak so the noise is peak detected too, and video averaged so it's level is somewhat uncertain. What I do know is it is at least 18 dB, probably 22 dB, possibly more. Too much uncertainty. Send contrary arguments to my server at /dev/null.

    If it were 22 dB, plugging this into a link budget gives a G/T of something like -6dB or -5.5 dB. That's not so unreasonable as a global satellite beam is about 22 dBi. However the errors in guessing the beacon EIRP based on my 200mW to equal it and not being quite sure where the noise floor is are too much.

    Time to try a measurement I think -

  • To get a specific SNR with known EIRP tells us what the satellite G/T is. Unfortunately I do not know what the SR bandwidth is in SDR Console but luckily I did some spectrum analyser measurements recently.

    The beauty of this is I can go back and do analysis on the data.

    Hi Mike & all,

    Maybe we can ask Simon G4ELI, how does the software work in this regard.
    On the latest release 3.0.13


    he added a few niceties, notably an analog S-Meter, wich he describes here:

    73 de CT2IRJ

    Salomão Fresco

    Callsign: CT2IRJ

    GRID Locator: IM59re -- CQ Zone 14 / ITU Zone 37

    QTH: Fazendas de Almeirim, Portugal

  • Quick test with G4JNTs Signal-Noise monitoring package - 1W CW TX, SNR 58 dB Hz which suggests an uplink G/T of -11 dB/K. That is much worse than I expected so either my dish is not good or it's not very sensitive. I do recall much better results earlier on before the gain was reduced, but I don't think it could be that bad so perhaps a combination of both. It would be nice to have some other measurement datapoints.

  • Yesterday I made some not so scientific measurements. I compared Bamatech DJ7GP and PE1CKK POTY 2.4GHz patch antennas. Already known that DJ7GP design is not circular polarized but POTY should be.

    I measured SWR for both antennas and transmission over short (less than 1m) distance. RX antenna was 2.4GHz quad.

    DJ7GP patch and RX quad.

    DJ7GP H / V polarization difference is more than 10dB in room environment with all of the room reflections. Anyway you can see clear H / V difference when rotatin RX antenna.

    One clear resonance. No sign of double peak. Same also in the Smith chart (no picture).

    Just assembled PE1CKK POTY in the test bench.

    Transmission loss vary only dB or two at 2400.175 test frequency. When RX antenna is rotated transmission under or above the test frequency is warying even up to 20dB but transmission at the test frequency stays almost constant. What I understand this is a good sign of circular polarization.

    However there are no clear sign of the double resonance around the test frequency. I did not find good explanation for this.

    I even did new calibration for the 8753C with no cable and conected the antenna directly to the test port. No noticeable change. Do anyone have good explanations for this. Also I am interesting to hear / see results if someone have done H / V transmission tests for this type antennas. (click picture to see full picture. For unknown reason full picture is not shown here)

  • However there are no clear sign of the double resonance around the test frequency. I did not find good explanation for this.

    Can you look at it with a smith chart? This will hopefully show a loop. There is not always a double dip visible on a RL plot depending on how it is matching. Here is what mine looked lie on an xavna before and after tweaking.

  • Yesterday I made some not so scientific measurements. I compared Bamatech DJ7GP and PE1CKK POTY 2.4GHz patch antennas. Already known that DJ7GP design is not circular polarized but POTY should be.

    Can I ask why you say the Bamatech is not circular polarized the web text says it is, I am curious as I know a few that have just bought a couple of these????


  • Again sorry I do not know why pictures are not full size. Click the pictures to see them.

    Unfortunately DJ7GP patch does show only one resonance. Transmission tests supports this finding. If you listen the antenna with dipole or quad and rotate it very clear minimum and maximum are present at the transmission attenuation. In some other mail I have seen note: "First generation DJ7GP antennas are linear polarized". Are there already other designs available?

    POTY tuned with some dielectric material under "long resonator". You can sind two points where j=0 (resonance points). Transmission tests shows quite little variation in the transmission attenuation when RX antenna is totated. This is a sign of circular polarization.

    At 2.4GHz measurement setup is quite critical. A lot depends where you set (or can set) your calibration reference plane. This picture demonstrates what happens if for example one SMA I connector is inserted. This means about 160ps delay = some tens of degrees rotation at the Smith chart. Same tuning as in the previous picture looks very different.

  • Another old example of 1300MHz +/-45deg cross dipole. Here we have same principle as in the POTY patch plate. One leg is longer and other shorter than required resonance length. One leg is inductive (+j) and other capacitive (-j). In optimal case these are +/-45 degrees phase at the design frequency. In this example not exactly there but antenna is working as expected and SWR plot also shows clear double peak.

  • Can I ask why you say the Bamatech is not circular polarized the web text says it is, I am curious as I know a few that have just bought a couple of these????


    The Bamatech patch works on the same principle as mine - two resonances in the right phase and amplitude relationship . This is achieved in that design through the offset tuning capacitor (screw) but for whatever reason it isn't showing them here. I built and then simulated a similar solution but I could not get it to work properly. Maybe with more tweaking but It was just one of several possibilities. James Miller, G3RUH designed a very good antenna for AO40 on exactly that principle and with a choke ring. It's well suited to QO100 too, but it is single band. Adding a choke ring to the POTY improves the backward sidelobes greatly, but it's not needed for TX only use.

    A critical dimension with the POTY turned out to be the spacing, 3mm and less worked, by 6mm it didn't. I put this down to the effect of the waveguide that penetrates it. Maybe if a Bamatech like design used a smaller spacing it would also work. It would have the advantage of an easier to adjust tuning element. Tuning is very critical, so if you have one don't go adjusting the tuning unless you have the test gear to evaluate the effect. Otherwise you tend to get a well matched linear polarised feed.

    If I can find an old simulation file I will add an image to show what I mean.

  • Yes - I found my original simulation 20/12/2018 - no changes at all it surprised my and ran in the updated student CST.

    Here is the 3D model - note the piston like capacitor - I probably need to adjust the disc dimension to have a stronger tuning effect but this was all a year ago.

    The match is really good

    So is the pattern

    Smith chart says it all though

    And that's confirmed by the axial ratio - this antenna is not circularly polarised.

    What I will investigate is can it be made circular with more capacitance and a smaller disc. I had no success at the time but it was a year ago and I had only a day's of experience at the time.

  • Personally I am using the POTY, but I know of two that have just bought what I believe is the DJ7GP design from Passion-Radio, they are not in use yet, but they do have a tuning screw that has not been tweaked by them!

    So I am hoping for them it to be OK. I did not realise there was two versions of the antenna before I saw this post and I was trying to determine the physical difference between the two to check.


  • I don't suppose anyone has accurate dimensions for the feed? I had a go at re-simulating and I am getting an excellent, linearly polarised response. I am assuming the placement of the feed and tuning screw are the magic I am missing.

  • G0MJW: There are a few articles on DJ7GP's feed (iirc in Funkschau) by DJ7GP himself, where he also states that its CP isn't ideal.

    I made one almost exactly one year ago with a 'FYM-mod' (additional screw) but at that time Mike and I found each other to converge to something more simpler which was baptized POTY on 3 March 2018 ; -)

    G0MJW: for the DJ7GP dimensions (omit the FYM screw) see: https://pbs.twimg.com/media/Du…lzH?format=jpg&name=large

    And also these Twitter posts:



    A year has gone by .... ; -)

  • A year has gone by .... ; -)

    Time has a habit of catching up with you and then just keeps on going.


    John is to blame. Along with Robert Watson who suggested I tried CST student.


    Started with an ellipse which was the best, then tried circular with tabs/cutouts and eventually moved to the square version for ease of making. Most things I turn on my lathe are elliptical but that's not deliberate of controlled. This design progressed 19th-24th December ( I had to learn about EM simulation in CST first and then get past the limitations of the free version).

    By the time we got to Christmas Paul had flattened quite a lot of old 35mm copper pipe and CNCd several prototypes. I am still using the elliptical version and Paul is still using the original that Remco later christened POTY.


    Here is the original RemcoCADTM

    [Blocked Image: https://pbs.twimg.com/media/DvMy5GZXgAAcPNZ?format=jpg&name=4096x4096]

  • Trying to adhere to the topic:

    From the TX viewpoint, for simple circularly polarized feeds one still needs to achieve splitting of the signal into two equal amplitude components with +-45 degrees shifts, i.e. two orthogonal fields from two virtual elements with impedances of 50+j50 ohms and 50-j50 ohms. Parallelly fed this results in a total feed impedance of 50+j0 ohms purely resistive.

    These two resonances should be evident from the hardware when measuring with a VNA: there should clearly be both impedances with their relevant resonances, ideally symmetrically about the nominal center frequencies. Accomplishing all this with a simple feed plate, a single feed point and a single disturbance (the screw) is somewhat challenging and tough to replicate in a hardware duplicate. But with very tight manufactureing tolerances it IS possible over a very narrow frequency range (obviously). This is why it is so very difficult to copy mechanical drawings with good overall results. You really need to know what you are doing and also have the instrumentation. Pushing limits, I routinely measure ellipticity on HF in the confines of my small basement - figure that one out ;-)

    Once you have determined the apparent crossed polarized fields of your particular CP feed, you should be able to clearly affect each resonance individually (might take a piece of dielectric on a very thin dielectric probe). I usually use small pieces of PTFE, Rexolite, plastic or even just my finger tips for this type of work.

    After decades of work with CP polarization for a very wide range of applications and with quite a few years on HF too, last year I finally put together a (quadrature) dual channel polarimeter demonstrator for HF and this spring I also added a dual channel downconverter for 2.4 GHz for demonstration purposes:


    Now, for my tripole antenna experiments I obviously need a three channel polarimeter. Think of this as an analogue oscilloscope CRT with magnetic deflection using external coils (e.g. Cossor 1039 or vintage radar CRT): instead of two orthogonal coil pair, only three deflection coils are needed in trigature, i.e. spaced at 120 degrees. Obviously magnetic deflection if bandwidth limited, but the display is on an arbitrary IF anyway. And I just love the dual time constant phosfors of my radar CRT :-)

    Until now I have been using vintage HP Vector and Constellation Displays with various homebrew 3 phase/2 phase RF transformers and matching three channel coherent converters depeding the on frequency band. This is all very fine for displaying CP ellipticity directly and in real time, but all the way, it is imperative to see also the complex impedances, such as (plot of the three resonances, 0 degrees and +-120 degrees with the total return loss of the trigature hybrid with imbalance load, 5dB/div):


    The overall ellipticity (linear deflection) of this contraption may be seen in the last plot of the first link.

    Michael, oh2aue

    P.S. winding of CRT deflection coils is described in detail in the Bell Labs microwave series... :-)

    P.P.S. how many multipole antennas can you count in the AO-40 photo on the left?


    "If you have data, you have something, if you do not, you have nothing." (Bengt Hultqvist, SK 24.02.2019)

  • PA3FYM,

    well, here in Finland we have this thing called "nollatutkimus", or zero-research, quite literally referring to putting effort into restudying something that has been evident for eons ;-)

    Something like this, only this is on 7 MHz:

    We all know that the polarisation ellipse of a correctly aligned POTY would on the polarimeter CRT be a perfect circle (when measured in a reflection free environment - here at the 62nd parallel north, snow melts and the garden dries to boot-comfort usually around April/May...).

    Meanwhile, I am integrating a Trio CO-1303D oscilloscope into the system, more pleasing for the esthetic eye with it's vintage round CRT and all that :)



    "If you have data, you have something, if you do not, you have nothing." (Bengt Hultqvist, SK 24.02.2019)

    Edited once, last by oh2aue ().