Second build sold out. Message will appear here when store is ready for third build ordering.

wsprdaemon noise graphs

1235»

Comments

  • Thank you Rob, for your great work and effort - I really enjoy playing with the kiwi with WD...
    Rob, do you use pre-amp at the feedpoint of your beverage in KH6 ? I recommend it, because the ant becomes very useful on high bands as well...

    From about 15h utc or so, I made a quick and dirty test with an old W7IUV pre-amp, which I had around...
    After 150m of cable from the feedpoint of the NW-wire (170m at only 3m height) I inserted the pre-amp with a 100pF series C (just to dampen the lower frequency somewhat), hoping I can test, if there are really more spots, due to more sensitivity...
    Although the pre-amp is really not properly installed and the homebrew PCB is without any housing on a sideboard in the gardenhouse, results are very positive. What a difference!
    The long wire gets really sensitive now even on the high bands...
    Tomorrow I will try to make a more proper installation.
    It will be much more difficult to use a pre-amp on the east Beverage, because the broadcast stations targeting Europe are so loud (for example 17.5Mhz) while at the same time I cannot hear the bandnoise on 18.1Mhz.... so proper filtering is essential...

    "Another advantage of the preamp, beyond simply improving the kiwi's noise floor, is that it pushes common-mode ingress occurring after the preamp further down by the amount of the gain."

    That is, why I will use pre-amps with low input impedance (50Ohm matched) and very low common-mode sensitivity...
    Any pre-amp made for short monopoles with high impedance will just amplify the common-mode current as well...

    Ulli, ON5KQ
  • Ulli writes : "Any pre-amp made for short monopoles with high impedance will just amplify the common-mode current as well... "

    Which is why I build active short dipoles instead of monopoles and use them with very high CM rejection, high input Z OpAmp preamplifiers and CAT5 line drivers&receivers. 70 dB of CM rejection at low frequencies and still 30 dB at 30 MHz.
  • Hi Ulli,
    I use a DXE/Clifton LNA at KPH to compensate for the 10 dB loss introduced by the 8-way splitter feeding Kiwis 73-77. The net gain of about 10 dB from TCI-530 feed to Kiwi SMA input which helps overcome the poor 20 dB NF of the Kiwi on the higher bands. KPH is so quiet that by adding an additional 10 dB in front of Kiwi78 for a net gain of 20 dB, I was able to improve WSPR spots by about 1 dB on 20-10M and pick up a few very weak signals.
    In Maui at AI6VN/KH6 I put a DXE/Clifton at the feed point of the 100M Beverage and get excellent reception up through 15M.

    However, as delivered from the factory both of those DXE/Cliftons suffered from the internal LM340 10V linear supply oscillating at 100 KHz. By shorting out the inductor on the output side of the LM340 I was able to eliminate those oscillations.

    But whatever your LNA, it is good practice to put a 50 terminator on its input and look at the 0-30 MHz spectrum on the Kiwi before attaching the antenna to the input of your LNA.

    Rob
  • I now have the Northwest wire installed with a pre-amp directly at the feedpoint (after the impedance transformer). Then 200m of coax (about 9db loss) to the shack follows...
    The 170m wire is at 2-3m height and points straight towards W6/W7. With opening angle of approximately only +/-15degrees on 20m at -3db) east coast stations on 20m are not better than the vertical dipole (off direction already 20deg), but west coast at least 10db better S/N ratio.

    The pre-amp is a very simple MMIC, which is not very good - I managed it with a 3pole highpass filter at 20Mhz bypassed with a 1kOhm resistor. The gain curve is about 19db gain on 10m, about 10db on 20m and at 30m and lower there is no gain at all so the 19db MMIC gets about 19db att. in front for net 0db gain... that is hopefully just enough to prevent excessive overdrive of the cheap MMIC. It tends to produce IM2 products from the mediumwave broadcasters...
    For next winter I will make something better...

    I changed the Noise graphs for 630/2200m band - so that= they are visible again on the chart:
    http://graphs.wsprdaemon.org/ON5KQ/
    But keep in mind, the real values for these bands are 30db higher !

    Ulli, ON5KQ
  • It isn't clear to me how you are instructing WD to adjust the noise levels on your various receivers/antennas/bands. Can you share the conf lines? I would expect that your 170M Beverage on the 10M band would be configured for a net of +9 dB gain ahead of the Kiwi. So the configuration for that band would be '10:-9 '
  • edited February 2020
    Rob,
    - Beverage wire gain (according to 4Nec2 with Nec2 engine) has indeed 9.5dbi gain at 2.5m height
    - From the beverage wire there is a imp-transfo to match to 50Ohm - it is compensated for small reactance, so SWR on 10m is now 1:1.2 roughly... att of the transfo is below 0.2db
    - Then the Highpass filter (3pole T) has only very small loss on 10m , roughly 0.4db
    - Then the MMIC has 19.3db gain on 10m
    - 200m coaxcable introduce loss about 9.2db
    - additional losses of 2 connectors 1db

    So for 10m we get total gain to compensate for:
    9.5db-0.2db-0.4db+19.2db-9.2db-1db = 17.9db net gain on 10m
    Indeed I forgot the noisefigure of the preamp - it is about 2db according to MiniCircuits - I should substract that...

    But for the moment I calculated with 17.9db gain - so for 10m, I put in the default line -17.9 for 10m

    Here is the total line:
    DEFAULT:0,2200:15,630:0,160:18.4,80:12,80eu:12,60:9,60eu:9,40:7.4,30:5.8,20:-1.1,17:-10.2,15:-15,12:-16.9,10:-17.9"

    Note, that I really don't know the values for 2200 and 630 at the moment - I guess they are approx 30db higher... out of graph...
    The other band values are heavily influenced by the highpass filter between imp transfo 50Ohm side and the input of the MMIC amp. On 30m it results in attenuation - that just prevents heavy overloading.
    However I think tomorrow I will introduce another highpass filter to cut of everything below 1700kHz to get rid of the IM2 products from broadcast stations in the evening/night.

    Conclusion: with this DEFAULT settings the 15m and higher bands noisefloor is almost at -168dbm !!!
    My guess is, that the long wires on the highband don't have the gain in reality, NEC2 is suggesting. The efficiency due to earth losses drops dramatically which eats all the gain, NEC2 shows us. But I believe it is not true...

    I will try to verify this assumption to get an idea how much wrong is NEC2 with Beverage wires on the high bands...

    Ulli, ON5KQ

    Edit: I also installed an old W7IUV pre-amp to the East Beverage (also 170m at 1.5m). It is not at the feedpoint but after 30m of Coax in the garden house. In front of the pre-amp I use a similar highpass filter to prevent overloading. I did not measure the gain of the pre-amp on 10m, but estimate it for 16db. On 10Mhz and lower the bare pre-amp (without filter) has 20db gain.
    Let's see what the noise graph results are. I would expect both antenna's should have similar noisefloor now. But the absolute level on the high bands seem too low for me, due to NEC2 mistakes, I think...
  • edited February 2020
    Your configuration looks reasonable, but I wonder about the accuracy of the NEC model of the Beverage.

    At KPH I have used the gain numbers reported by my NanoVNA https://www.amazon.com/NanoVNA-Network-Analyzer-50KHz-900MHz-Antenna/dp/B07TMV17TP for the path from the antenna feed point to the Kiwi RF SMA input and ignored the antenna gain.

    In Maui this inexpensive Distill AM blocking filter suppresses overloads from nighttime AM while passing the LF/MF signals from the Beverage. https://www.amazon.com/Distill-Barebones-Broadcast-Bandstop-Applications/dp/B076CVW6LC

    Finally, your 'RAW' versus "SPOTS' on Jim's site http://www.wa2zkd.net:8088/today_int.html suggests that you aren't using WD for all your spotting and thus double spotting to wsprnet.org. If that is true, what motivated that setup?
  • "Finally, your 'RAW' versus "SPOTS' on Jim's site http://www.wa2zkd.net:8088/today_int.html suggests that you aren't using WD for all your spotting and thus double spotting to wsprnet.org. If that is true, what motivated that setup?"

    That is true! For about 1h now I use a comparison antenna on 40m. It runs on a different receiver, which I don't know how to implement with WD (ELAD FDM-S2)
    Some month ago I refurbished my old 8ele vertical array (broadside endfire transmit antenna. see foto on qrz.com) - so sometimes I use this antenna for comparison. I know exactly the current distribution of the elements as well as their resonant frequency for each element. Currently the beam is about 3-4db better in S/N...

    I found that unfortunately the 170m LW is still suffering slightly from IM2 products.

    Last time when I made this comparison, the beverage LW was almost always exactly the same in S/N, but there was no IM2 issue, as I didn't use any pre-amp. So the comparison today, was to confirm, whether there are some problems to be solved with the new installation of the preamp at the feedpoint of the Beverage wire - and yes, there still is!

    So tomorrow I will install my 1.7Mhz filter (7pole Tschebycheff, with less than 1db loss in passband). It will help the cheap MMIC of the current long wire to keep clean, not producing IM products.

    Thanks for info on your network analyser.
    many years ago, I build the N2PK design.
    Later I bought the DG8SAQ unit from Tom. It has almost the same dynamic range, but extended frequency range.
    I can't work without VNA any more...hi

    Ulli, ON5KQ
  • Today, I quickly compared the noiselevels of my 2x3m vertical active dipole with a different but known dipole (2x1m).
    The antenna factor of the shorter dipole has been calculated with LTspice and though the gain can be calculated for each band.
    From the comparison of the two antennas, I have a good idea, how much more efficient the 2x3m longer version is. (it is really a lot!!) That increase in efficiency is now reflected in the new numbers of the DEFAULT parameter.
    Conclusion:
    The current Noise plots written from about 13h utc today should show rather realistic noise levels for the vertical dipole now.

    It is clear, that the numbers NEC2 is calculating for the Beverage wires at 2.5m height is totally wrong, especially from 17m and up.
    I therefore introduced some 'additional loss' for the Beverage wires from 40m and higher. The additional loss goes drastically up from 17m and reaches >20db on 10m. I think, this will give more realistic numbers on the high bands for the beverage wires.

    My location is not at all quit - I guess just a little below residential levels from the ITU recommendation... really not comparable with 'the end of the world'..... hi

    Ulli, ON5KQ
  • need to switch back to the Beverage wires! No other antennas left for the moment - Reason:
    We expect starting sunday (peak in late evening) a potentially desasterous storm with windspeeds well above 120km/h - local weather channel with active warnings!
    So I took down all verticals ... only low wires left... tomorrow need to "prepare" the house
    fingers crossed...
    UIli
  • Is there any problem with the Grafana site to view the Noise graphs in more detail ?
    Of is it my computer with a problem ?

    It shows 502 error = bad gateway.. with no data
  • I don't know the cause of that problem, but I will look into it.
    I am also working to migrate our use of Grafana to a wsprdaemon.org server where I will have direct control of it.
  • About the Noise graphs on 160m:
    I notice the big difference of noise level at the kph site on 160m at daytime and nighttime. However it can be seen, that the noise level difference is lower with the vertical marconi antenna.

    So yesterday I kept both rx running all day with both Beverage antennas (east and north west) to watch, what the noise level difference might be. Only 6db max = almost NOTHING between daytime/nighttime compared to kph

    Simple conclusion: My qth is much more noisy than kph - too noisy for serious DX ?
    But another finding: The ability to decode very weak stations seem to be not much better in Europe than at my NOISY location?

    Question, which comes up: Is the noise behaviour different, if the antenna has most sensitivity at high angles or if the antenna is especially designed for low angle sensitivity ??

    At kph the antenna with the biggest daytime/nighttime noiselevel difference is in fact designed for local communication, so I guess has most sensivity at very high angle radiation - we can see already the lower noiselevel diff with the Marconi-T, which might be more sensitive for low angle rad. especially noticable during the day.

    So back to my Beverages: They are almost dead during the day on lower frequencies. However during the night the noise doesn't come up greatly, while signal strength becomes generally very loud! It's great to listen... noise is only up by 6db..

    So what would happen with such antenna at a really quit QTH ? Also no big difference of day/night noiselevel on 160m either ? kph shows about 20dB with the TCI ant...

    Until now I am not really succesful, finding pronounced noise sources on 160m at my qth. I am confused, why the noise difference on these antennas is so small while they sound dead during the day. Is it still broadband noise I am missing to identify ? My feeling still is, there is a lot somewhere hidden below the noise.

    What should be the day/night noise diff be, with a real sharp low angle beam rx-ant. I tested also longer Beverages up to 400m on 160m in various directions - no change. The difference of daylight-nighttime noise in winter was never more than 6db...
    Is this typical or only an indication of noisy qth ?

    Any experience with comparisons. I never could compare my beverages with a dipole at 30m/100feet for 160m for example...
    (to find ouy if radiation angle is the big difference)

    Ulli, ON5KQ
  • The KPH site was selected by Dr. Beverage in 1920, so it should be exceptionally quiet. However we have found that our nearby suburban stations using short well balanced active dipoles can receive as well as KPH at LF/MF. Eliminating RFI introduced by common mode
    antenna +feedline imbalances has reduced noise levels by 20-30 dB at some rx sites.
  • Hello
    does anyone experience using RTL_SDR in VHF (144.489) and know the values ??to set in the noise_ca_vals.csv file.
    the antenna is a 6 elements quad
    Rene F6BIR
  • In V2.8 I am working on much improved support for RTL-SDRs and all other SoapyAPI-supported devices. You have stimulated me to add study of noise calibration to that effort.
  • Hi Rob
    I am happy to read this, do you know other one using RTL_SDR on 2 meters, I am runing V2.5a on PI-B+ (old cb) I was not able to install V2.7 on it (mkdir wspr-captures problem !)
    TU
  • Rob, please excuse my stupid question (really no computer geek!):

    "In V2.8 I am working on much improved support for RTL-SDRs and all other SoapyAPI-supported devices."

    Does this mean, I can use my Elad FDM-S2 or my old QS1R receiver with WSPRdaemon in the future ?

    I have never saw a documentation HOW to configure RTL-SDR or for example the Funcube dongle pro+ stick or even a simple audio input as a source for WD and implement it in the main .Conf file
    What do I need to install on the linux machine?
    Is there any step by step information for dummies. What do I need to do, to learn, what needs to be done ? I just don' t know where to search for it...

    Regards,
    Ulli
  • In 2.8 my goal is to support all SDRs which are compatible with the CubicSDR application program which utilizes the SoapyAPI library to communicate with a wide range of SDRs. While I am close to finishing debugging that feature, higher priority projects have kept me from understanding why the rx_fm command line utility which uses the same SoapyAPI library, outputs only silence when CubicSDR works fine.

    The current RTL-SDR support uses the rtl_fm command which I have found is not able to tune accurately enough for WSPR use above 54 Mhz, even when the RTL-SDR is driven by a GPSDO. Several WD users are using WD on 6M, but success is a matter of luck.

    In contrast, WD does have reliable support for analog analog input. For example, attach a USB audio dongle to a Pi and execute './wsprdaemon -i' to run the 'setup audio input level' command. The dongle loops the analog input to the output, so during '-i' you can listen to the audio being fed to the dongle. Then add that device to your list of rx devices in the conf file and add that rx device to your schedule.

    But there are so few users of this feature that I have invested my time in debugging the SoapyAPI support and deploying our noise and spot databases. All of these features will be implemented and enhanced, but I'm sorry I can't invest more time in WD for the next few weeks.
  • Rob, many thanks you take the time for some information...
    Ulli
  • Access to the noise plots at the Grafana website seem to require a password, now.
    How to get a password for the Grafana Noise plots ?
    https://grafana.int8.com/login

    Many thanks.
Sign In or Register to comment.