Posts by s-band

    I use general purpose analysers, including a Wiltek 9102 portable. Although they won't demodulate they are more useful for beacons and interesting non-DVB things. Setting skew accurately can also be easier by nulling signals without demodulation. To set up dishes on anything, including DVB, I use a phone as a viewer via a VNC connection. That allows easy connection of any equipment indoors and viewing at the dish without having to put something in-line.

    All examples on 26E QO-100

    Lab spectrum analyser (runs Win XP so has a VNC server installed)

    Crazyscan on PC with TBS cards

    The same PC with TBS and Minitioune on 2 different signals

    It would be handy to have a fully loaded Promax with an Ethernet connection but they are too expensive compared to surplus lab test gear.

    Look for a generic log periodic. There are many military ones covering 30-88MHz but they are expensive. This is reasonably priced but in short supply in UK.

    CLP-5130-1N Log-Per Antenna
    CLP-5130-1N, CREATE, broadband log.-per antenna
    stecker-shop.net

    There is also this one: https://stecker-shop.net/LP028150S/en

    Suchen Sie nach einem generischen logarithmisch-periodischen Antennenstecker. Es gibt viele militärische Antennenstecker für 30-88 MHz, aber sie sind teuer. Dieser ist preiswert, aber die Antennenstecker scheinen knapp zu sein

    femi The rectangular to circular transitions used for C120 to WR75 are usually 1/4 wave transformers. These are not a wide band match although generally good enough for our purposes. The match and losses are important when measuring noise figure and are not defined. One can guess at the loss (probably <0.1dB) but the effect of a mis-match is hard to quantify. Anything put between the calibrated noise source and the test unit is a potential source for error.

    Regarding the square to circular transition that is part of the Invacom, it does not affect our measurements as it is always present. We are interested in the noise figure of the whole thing as seen at the C120 flange. A square to circular transition is actually quite a good wide band match when compared to a rectangular to circular transition.

    Die für C120 bis WR75 verwendeten rechteckigen zu kreisförmigen Übergänge sind normalerweise Viertelwellentransformatoren. Dies ist keine Breitbandübereinstimmung, obwohl sie für unsere Zwecke im Allgemeinen gut genug ist. Die Übereinstimmung und die Verluste sind bei der Messung der Rauschzahl wichtig und nicht definiert. Man kann den Verlust schätzen (wahrscheinlich <0,1 dB), aber die Auswirkung einer Fehlanpassung ist schwer zu quantifizieren. Alles, was zwischen der kalibrierten Rauschquelle und dem Testgerät liegt, ist eine potenzielle Fehlerquelle.

    Der quadratische zu kreisförmige Übergang, der Teil des Invacom ist, hat keinen Einfluss auf unsere Messungen, da er immer vorhanden ist. Uns interessiert die Geräuschzahl des Ganzen, wie sie am C120-Flansch zu sehen ist. Ein Übergang von quadratisch zu kreisförmig ist im Vergleich zu einem Übergang von rechteckig zu kreisförmig tatsächlich eine recht gute Breitbandübereinstimmung. <note: I don't think 'match', meaning return loss, translates correctly>

    I find the noise figures quoted by non-professional suppliers most annoying. The Invacom LNBs are some of the better ones but perhaps someone can explain to me how a 0.3dB noise figure device and give 0.2dB noise figure (without using cryogenic techniques) . The noise figures quoted for devices are generally the de-embedded figures. i.e. they have all circuit losses removed by calculation. The best noise figures achieved in real-world designs are in the region of 0.7dB at room temperature and that is often for a relatively narrow band design at 10GHz.

    The only LNBs that are reasonably easy to measure are standard WG types. I have found that those from the pro manufacturers generally meet their specifications. For example SMW, Norsat etc who usually quote around 0.8dB on Ku band. It's hard to measure a C120 LNB in the lab since the transition to rectangular guide is an unknown. The simplest way to measure LNB noise without test gear is to use natural sources as described here. I usually use Sun noise to test the whole system. However, Sun noise itself is variable so is of most use for comparisons made within an hour or 2 of each other. This is a good paper on natural noise.

    If comparing LNBs, it's not just the NF you need to worry about, especially for modes such as 32APSK and very narrow low SR signals. For those the LO must have a low phase noise. The Invacom is actually quite good for a free running DRO. When testing using IQ monitor/Crazyscan etc, check a selection of SR and modulation types.

    It may be worth trying different devices in LNBs. However, in my experience, it is hard to even keep the original performance. I repaired an SMW LNB which had a dead 1st FET and it only just managed to tweak it to 0.8dB NF using patches on the tracks. I have got down to 0.7dB, narrow band, on an 8.45GHz LNA (MGF4919G FET).

    <edit> Here's a useful article from CEL. Note the final comment:

    Quote

    In actual product development, the design process is much more complex due to various imperfection in components/devices and constraints on circuit implementation. While the simulation shows a noise figure of about 0.5dB, a product specification of 1dB or slightly lower for noise figure is generally considered excellent when everything is taken into account. Nevertheless the considerations outlined in this note are relevant in real world applications.

    qwer  satesco

    I think that there are big problems with the BWEI LNBs on Ext KU. They appear to have very poor image rejection which gives the effect you see. It is especially bad on 42E due to the strong signals below 12.75GHz. I don't think they have much RF filtering. There is also the problem of unwanted sampling modes in TBS and other cards.

    One way to check if you are 'seeing' an image is to look at the Spectrum box on CS which usually shows 'Inverted' on most signals. If it shows as non-inverted, it's likely to be an image response.

    My modified SMW is a 12.25-12.75GHz PLL which had an LO of 11.3GHz. I have changed the LO to 11.75GHz and tried to improve the performance on ext Ku. However although the NF is 2dB, the gain is 20dB down compared to its response at 12GHz. I use a 1.5GHz high pass filter followed by an 18dB gain amplifier in the IF. I still see an increase in noise when pointing at the strongest satellites and this is probably caused by front end overload.

    1.8PF 2dB NF LNB

    I don't remember seeing the approx 800kHz wide signals at 13625.5, 13627.5 and 13605.5H which are all much too low to get a lock. The peak at 13526H is from 42E and the signals at the low end of the band were described here. As before, the transponder noise can be seen.

     

    strannik I wonder if you are getting a false positive from the pair of closely spaced carriers on 13584 H & V. If the polarisation is not set exactly, both are present at the same time.

    13497H still there but in and out of lock as clouds pass by.

    The mid-transponder beacon has the 2, closely spaced frequencies similar to 5A but the vertical one is at a much lower level in this case.

    The transponder bandwidth can be clearly seen on V & H. However, I can find no evidence of any other signals, from near London, and the method used here is quite sensitive.

     

    1.8m PF 2dB NF LNB

    femi What is your ext Ku LNB model number and LO frequency?

    SRmin = 30 kS/s :

    As you can see, the constellation has become very clear, and SNR = 12.3 dB! SR = 92 kS/s, and at the corresponding frequency of the first harmonic

    11452 MHz such a powerful signal is not observed :73:

    I'm not sure what that can be. Is it a terrestrial link?

    Today jaes (SatsUK) found colour bars on 13497H 13330ks/s. I can only just lock this on a 1.8m dish. There is also a carrier on 13526.00073MHz horizontal.

     

    The confusion over the polarity of 13584 is caused by there being 2 carriers spaced by 900Hz. The lower one is vertical and the upper is horizontal. If you use a wide bandwidth, you'll see a small dip as the skew is swept through 45 deg offset. In an audio BW the 900Hz tone should be audible.

    For the plots below I used a modified SMW 12.25-12.75GHz LNB with an LO of 11.75GHz (the system is locked to a Rb source). The NF is about 2dB and the dish a 1.8m PF. The actual frequencies are on the plots.

    I carried out some tests to compare a Bullseye LNBF with a SMW Q-PLL Type C on a 1.8m PF dish.

    SMW Q-PLL Type C on IRTE C120 adapter
    Bullseye LNBF with feed removed fitted to C120 flange from Invacon LNB

     

    The plots here used Crazyscan's blind scan 1 with 2MHz steps. This gives more repeatable results than BS2 but doesn't catch all signals. There is more data than presented here but it needs processing. I would normally compare systems using Sun noise but the tracking dish is out of use at the moment. Notes:

    • Cloud cover changes resulted in 0.5dB variations, maybe more.
    • 7E chosen as it was of interest at the time
    • 10804 and 10845H give random values sometimes
    • Skew set by nulling beacons. SMW had >32dB X polar rejection and Bullseye <27dB
    • 10720H varied by the most with varying cloud (to be expected)
    • Number of feeds varied through day
    • Might be better to do 7W as that has most low level signals
    • High SNR signals give poor indication of differences as 1dB RF change results in <<1dB SNR change when > about 15dB SNR.
    • BS1 used with 2 MHz steps, BS1 & BS2 scans saved as .ini