Posts by G0MJW

    Trolling the internet? I hope you meant Trawling. The nominal transponder edges are defined in the config files where you set up each satellite. The actual centre frequency is not a parameter but you can work it out from upper and lower. To find it during a pass you will need to compensate for Doppler. The offset is displayed in real time.


    Unfortunately, not all amateur satellites carry frequency standards and drift quite a bit with temperature. Consequently it can be really difficult to know where the centre is.


    ERP - None at all until you receive the beacons, then the minimum to hear yourself come back, but never ever more than the beacon. In practice a watt or two to a small yagi is sufficient.


    Mike

    This is not quite true but in this case close enough. Firstly, none of your power will be lost but only half what could be received will be received. This means you are wasting power, as you are also be using a smaller dish and not focusing the power as tightly. If you had a pencil beam and could aim it well enough you would only need microwatts of RF power.


    The propagation loss for horizontal or vertical is not exactly the same all the time. The largest effect is due to the shape of raindrops. At 2.4 GHz this difference is almost negligible but not totally so. Vertical might have ever so slightly less loss, fractions of a dB in heavy rain. I could look up the ITU-R model but it's really not worth it. and would unhelpfully boost my technobore score. When we get to up linking at 10 GHz and 24 GHz this starts to matter.


    Mike

    I am using this system - I have a new PCB design for the PTT too. It is in KiCad will be on my Github.



    I built a prototype. It's working well. I re-boxed my Pluto and it fits inside the case/




    I am looking at another design with a low profile reed relay did sits on the Pluto GPIO header. This might fit inside the standard Pluto case, it's a tight fit though.




    Back to the point ..


    I use UDP from OBS rather than RTMP. It's much more flexible. As I have an Nvidia hardware encoder it allows me to do H265, so full HD in 333ks. Audio at 32k is the minimum but the Pluto can re-encode the audio with later firmware. If you don't have an H265 encoder you can always use H264. It's pretty good too.


    There are several other ways to encode the TS to send to the Pluto, including the Jatson Nano and the H264/H265 encoder boxes. The easiest way for me is to use the Nvidia card, they are not so expensive. Instructions are appearing on the BATC Wiki as we get around to it.

    Several possibilities - the MER is not an exact analogue to the SNR


    Either the transmitted signals are distorted:


    No modulator is perfect - DC offsets, IQ imbalance

    Low number of samples compared to the symbol rate

    Unwanted phase modulation in amplifiers,

    Issues with the roll/off and filtering - inter-symbol interference,

    Constellation compression for APSK

    Phase jitter


    Or the received signals are distorted: See above +

    local oscillator phase noise

    other signal interference / intermodulation


    Most probably both transmitted and received signal distortions are happening and for some signals worse than others.


    Quite a few narrow band signals are 8PSK or 16APSK and the distortions become more important. If you can't decode low SR but you can decode higher SR then it's probably the LNB not being stable enough. DRO LNBs won't generally be good enough for low SR. Some PLLs are not all that good either.


    You seem to get lower MER on wider signals in the examples so that might be your phase noise. If you are reference locked a different reference frequency, even only slightly different may make a difference.


    The other thing to look for are spurious signals - I have an LO spur that's quite large on one channel and receiving there is harder than in adjacent channels. You will not see this on the Goonhilly display if it is a function of your own RX and it's well worth looking at a wideband sweep of the whole transponder - e.g. with a LimeSDR.


    Mike

    This is what I mean by doing some simple metalwork to adapt the mount so that the reflector does not hit the clamp. Recently I had the same issue all I needed to do was make a small adaptor plate and adjust the feed arm to be slightly lower. It's the sort of metalwork that can be done with a drill and files.


    In your case, it looks like you just need to raise the height of the POTY and then bend down the feed arm to get it back in the correct place vertically.




    Above is what I did last week for a lash up system using a free 1m dish. A small aluminium plate rests on the original LNB feed holder (bottom half) and is held down with the original screws. The new LNB holder (plumbing supplies) is mounted further back so now the lens is the correct position relative to the focus and the feed is higher so it clears the feedarm. To compensate this the arm is angled slightly lower - it bolts to the back bracket with 4 bolts and this allows easy adjustment by slotting the bolt holes.


    In practice I don't think the small vertical error if you do not adjust the feed arm angle will be all that significant for a 0.7 f/D dish. It's much more critical with short focal length dishes but with an offset you can simply adjust the elevation to compensate for this., within reason. The use of several LNBs mounted horizontally to receive multiple satellite slots is an example and seems to work fairly well over a few degrees.


    Mike

    Does anyone know, where the focus point of the Venton lens ist?

    As far as i remember, the PE1CKK lens has it about 11mm towards the dish...


    Ich denke, es ist ungefähr in der Mitte, aber wenn man es flach auf das POTY-Pflaster legt und das Objektiv in der gleichen Position, in der es sich mit dem Ventron-LNB befunden hätte, sollte es ungefähr richtig sein. Dies erfordert normalerweise einige geringfügige Änderungen am Vorschubarm - das ist nichts, wovor Sie sich fürchten müssen, sehr einfache Metallarbeiten. In einigen früheren Beiträgen finden Sie Anregungen zur Unterstützung


    I think it is about in the middle but if you have it flat to the POTY patch and the lens in the same position it would have been with the Ventron LNB is should be about right. This will usually require some minor modifications to the feed arm - that's nothing to be afraid of doing, very simple metalwork. See some earlier posts for ideas for support - plumbing.



    Then it's not working for you. If you are not happy with the POTY then try something else. With an ex-rocket LNB lens it works fine for me and many others but some people appear to have difficulty, at 10 GHz it's usually from an inappropriate lens. I only recommend the ex-rocket lens or the HB9PZK rexolite lens.

    Even though you may not see any interference, it's worth trying a filter in front of an SDR as you never know if there is degradation until you try it. While this interference is producing lines, other interference may appear as a rise in the noise floor. As the NB transponder is very strong and with a 1m dish you see it's noise floor there might be interference below the noise so it won't matter much but on wideband it might be impacting the signal to noise ratio a bit more. For €6 it's a worthwhile investment.

    Interesting. DATV is using a modem and 33ks and 66ks are getting towards narrow band. Maybe we can see some convergence. DVB-S2 is a very efficient coding method close to the Shannon limit. No reason it can't be done in 2.7 kHz with the advantage of efficient FEC, LDPC. It might be possible to use the DVB-S2 modules in gnuradio rather than the standard modulators and demodulators in Daniel's flowgraph.

    I just had a thought - what sort of dish is it and which lens are you using? You might also find one LNB polarisation works better than the other. That's because they are not really designed for 10.5GHz and the probes are not ideally placed with reference to the back-short.

    You need to have the right sort of lens in just the right position for it to work well. Too far in or out and it will no longer be at the focus. You should not be losing as much as 3 dB with a rocket LNB lens, perhaps a dB. Phase noise is a problem too. Was your bare LNB unmodified and the POTY one modified for an external reference?

    G0MJW Yes, the same goniometry / math is also valid for offset dishes. An offset dish is nothing else than a prime focus dish but then another part of the parabola. The higher you get into the parabola, the larger the f/D. Although the illumination may look optically 'squint', the effective surface of an offset dish is equal to a 'prime focus dish' with the diameter of the width of the offset dish, see below.


    Yes, I know this for a long time since I first came across the concept in the 1980s. What I don't know is if the curvature calculation still holds when working out the angle - I could work it out but... in my dish the curvature it rather shallow so it matters not.

    I would suggest going down to 1 Ms rather than 1.5Ms. The quality that can be sent in 1 Ms H264 is really good. Even at 2/3 FEC. As the beacon is a fixed sequence, the coding could be done in advance and just a pre-prepared TS with all the optimisation thrown at it.

    The idea is to have the feed illuminate the dish so it is -10dB at the edge of the dish. This is sort of compromise between efficiency and sidelobes. Ideally you want the whole dish illuminated evenly and nothing beyond. That's not possible as it violates EM theory, so -10dB is a realisable goal that gives good results doesn't upset physicists. A rule of thumb, not too much power blasting past the dish and doing nothing useful (over-illumination) and not too little of the dish used (under-illumination)


    The good thing about rules of thumb is that like real thumbs you can bend them, but only so far. For the uplink you might want a little less than -10 dB at the edge. It's a trade off which is how the POTY works OK with most dishes 0.4-0.7 f/d - not brilliant but good enough, especially when RF power is plentiful.


    The angle of the edge of the dish from the point of view of the feed depends on the f/d. Higher f/d is a narrower angle. 90 degrees is about right for common satellite dishes but you can work it out with a little geometry.


    Well OK, you need complex geometry to get it properly but approximately is fine. E.g. for 0.6 f/d the make a triangle from dish centre to edge as the opposite and 0.6 x d as the adjacent. So that's 0.5 x d (as the centre to the edge is 1/2 the total dia) and 0.6d. Recall Tan(theta) = opposite/adjacent = 0.5d/0.6d = 0.833. Inverse tangent tells us theta = 40 degrees. That's a half angle so the feed beamwidth needs to be 80 degrees to fully illuminate the dish. I am not sure where 90 came from, except it's the value for a 0.5f/d dish and perhaps also account's for some of the liberties I just took in the calculation.


    Think too much about this and you will tie yourself in knots. The bottom of the dish vs the top, but it's closer, near field, diffraction..oh no, hard sums, panic... At this point invest in some simulation tools or go and have a beer.


    A helix with the same beamwidth should be easy to design but you must pay attention to the phase centre as it needs to be close to the focus. The 10 GHz focus alignment is most critical and with the LNB at the focus there is a good chance 2.4 GHz feed won't be optimum. I don't know where the phase centre of your Helix is but it's worth checking the design you plan to use with one of the reputable online calculators.


    Mike


    EDIT - following the link I see where the 90 came from. Remco did a more accurate geometry which I think is fine for prime focus but I am not sure it holds for an offset. Anyway, the idea is the same.


    [Blocked Image: https://pbs.twimg.com/media/Dyd5IijXgAILiaP?format=jpg&name=medium]

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

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


    https://twitter.com/TheRealMike/status/1074326098752745474


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


    https://twitter.com/g4bao/status/1075456339005059074


    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.


    https://twitter.com/TheRealMike/status/1077166330195951616


    Here is the original RemcoCADTM

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