The VLA is doing different kind of processing, since they're recovering the spatial properties of the illuminating field, rather than demodulating a time-varying signal. They have very precisely surveyed antenna positions and propagation delays, and very careful time synchronization, and it still takes them a lot of post-processing to calibrate out the error terms. The VLA would struggle to image most satellites, because they're not stationary and their signal isn't constant over time. (Which is not to say they don't contribute noise and interference to radio astronomy observations.)

He also has a friend called chamdler

The LNB feeds aren't exactly RF as there is down conversion to L-band inside the LNB. Doing 4 different down conversions with 4 different reference oscillators is where this is pretty much guaranteed to fail. Now if you could use a single reference oscillator and use lengths of coax as delay lines to get the phasing correct he might be able to pull it off. I would shop for a 6-8 foot dish myself.

Would be cool if all of that could somehow be DIY'ed

@@kyanhluong FPGAs aren't the most expensive thing in the world, there's even an Arduino FPGA (MKR Vidor 4000, 75EUR) so it absolutely could; just need somebody who knows FPGAs and DSP to make the chip do the thing

You could probably do this with four independent SDR receivers. I'm sure you could whip up something in GNU-radio to attempt to introduce delay and get them synchronized. (By training on the received signal.)

​@@Ziraya0 however programming FPGAs are 20x harder than programming an Arduino.

@@Ziraya0 Don't see why you couldn't do it with SDR, while FPGA would be more power efficient and definitely do it in real time, for this kind of playing around SDR would be much easier proof of concept.

There's multiple issues, phasing of the cabling being one, but another huge one is each dish has an LNB with a local oscillator in it and none of them are phase, or even frequency aligned - they'll all be slightly different. Usually for consumer sat tv reception, the LO's in the LNBs are +/- 1MHz (which is a LOT of frequency error) and have pretty poor stability too - aka they drift all over the place. The satellite receiver just follows the LNB as it drifts, which is fine if you have one dish / LNB, but is going to be a nightmare with anything more than 1.

@@gorak9000 Yes you would need to have a single oscillator provide a feed to each dish, to lock their oscillator in place. Probably easiest to take one LNB and make a amplifier for the LO resonator, and feed to the other dishes, coupling to the LO resonator there so it acts to amplify and filter it.

@@SeanBZA Typical external reference LNBs use a 10MHz reference oscillator and a PLL and or FRAC-N to multiply that up to the frequency needed, as 10MHz is much easier to transmit / distribute than the actual LO oscillator frequency used in the LNB which is 11.250 GHz. That's why it's really not practical to just amplify and distribute the oscillator of one of them to the others - it's rather high frequency! Is it doable? Yes. Is it doable on a backyard screwing around with old junk budget where the net benefit is still worse than getting 1 easily obtained larger dish? No

@@gorak9000 I have no clue about satellites but I have a question. What about using a raspberry pi to modulate the on board LMBs with more granularity?

@@gorak9000 Wouldn't it be possible to share the 10MHz signal to other LNBs?

You have to phase lock the downconverters together first.

The next issue is that like you said, the dishes are very small, and thus only useful for receiving DSS satellites which are crazy strong to begin with. You're much better off getting a single 1M, or 1.2M, or 1.8M dish and using that for any FSS satellite stuff you want to do. The next logical idea is to use them to track a drone or something to increase the wifi range, but with the dishes being "small" at 11GHz, they're even smaller (electrically - in terms of dish diameter vs wavelength) at 2.4 or 5Ghz for wifi. You could use them to point cameras, or replace the LNB with a microphone and have remotely pointed microphones you could use to listen into what people are saying walking around your neighborhood. Or you could paint the dish with metal paint, or coat it with a mylar "emergency blanket" (aka flexible mirror) to use either as a steerable spotlight, or a steerable solar collector.

Yep, the array isn't practical, I just wanted to try it since I borrowed that 4-way signal combiner and had 4 of them! I have another use for these coming up soon. I also have some larger dishes that I'll be posting about soon!

This, you need a common source for the LO base freq and have the ability to phase shift and make corrections dynamical to account for temperature and distance shifts over time. Analog Devices make a whole line of these type of shifters just for this purpose. Go look at "the signal path" on his tear down of the Starlink station that uses beamforming to see the kind of effort takes.

It's a tall order but nothing that can't be worked out in ADS or QUCS with some hard work and dedication. Keep the reflectors and home brew the block converter system. Synchronization can be achieved by a separate dedicated GPS dish receiving the global 10MHz synchronization signal. Only thing is it might be a pricey PCBWay order.

@@4DRC_ This is the way.

I've had someone suggest running them through 4 SDRs and processing the signal that way, it might be something for a future experiment!

This is correct. Most arrays are combined at RF, before any down conversion or heterodyning. The local oscillators at the separate block converters are not phase locked, and the beating between the slightly different LOs is giving you the fluctuating signal level you see. It's also making it hard for a demodulator to track the carrier (which is also being modulated by the different LOs.) You also don't show any calculation of the correct relative positioning of the antennas, so I suspect you're getting essentially random phase contributions due to propagation delays. This will account for a weaker signal, but that's a time-invariant effect; the variability and "interference" you're seeing is the asynchronous local oscillators in the individual LNBs. It is very hard to synchronise widely separated oscillators, as you need to equalize phase delay in the distribution of the reference signal, which is at least as hard as controlling the phase delays in your beamforming/combining network. The signals can be combined after down-conversion if you demodulate before combining, and use the recovered carrier (or a similar signal feature) to re-synchronise the data signals. This takes some sophisticated signal processing, and requires that there be sufficient SNR at the individual dishes to do carrier recovery. A more usual approach is to combine the signals at RF (before any conversion), then there's only a single LO in the chain, and no beating between different LOs. Note that for this to work, you need to control not just the cable lengths but the differential free space delay in propagation time to the satellite. That is, you need to know the individual antenna positions very precisely, then cut the cable lengths appropriately. The required precision varies with frequency, but, in my experience, 5% of a wavelength is approaching the point of diminishing returns in practice (~3.75 mm at C band, ~1.5 mm at X band). Note that an error of 50% of a wavelength (i.,e. 37.5 mm at C band, 15 mm at X band) will give destructive interference, that is, total cancellation of the signal. Typically, some variable delay element (e.g. phase shifters) go in each path to do fine tuning in the field (even if it's not a fully electronically steered array.) Also note that it's not sufficient to point the individual dishes at the satellite, you also need to point the array. The relative propagation delay to each dish is a function of satellite position, so you need a different set of cable lengths for each satellite.

Lots of interesting radio theory in the comments of this one! I understood a couple things you said, but I'll have to do more research before I try this again. In the meantime there should be another fun video with one of these next week :-)

Interesting ideas! I think the ones without USB are an older model with an earlier circuit board, at least one of them has an RJ11 phone jack but I haven't figure out if that's some kind of programming interface or what. I have a couple other projects to work on before I get back to this one, but it would be cool if I could make it work!

I started taking one of the dishes apart to fix one that had bad motors, so now I have two in pieces... We'll see if I get things back together again!

I don't have an oscilloscope yet, that's something on my list to find and learn how to use.

The splitter isn't intended for these, it just happens to be in a similar frequency range. I'm definitely using them for unintended purposes, I think I've done just about everything with these little dishes *except* watch TV!

Thanks! I've been working on my narrations, my early videos are a bit painful and awkward to watch now :-P

Nice! I keep hearing from local people involved with this and it's kind of fun to hear everyone's take on it.

You'll never get even 2 of them aligned because their independent downconverters are free running at slightly different frequencies.

Close, it's a UAID. A Useless Array of Inexpensive Dishes.

A couple people have suggested that, I'll have to try it if I can afford one!

These use little brushed motors, not steppers unfortunately. I might have a video on that coming up, I need to replace the motors on the one with water damage.

All of my cables were identical, the problem seems to be that the wiring inside the dishes is not (they're all slightly different models), and the local oscillators are all tuned slightly differently.

Cool! I noticed that these Tailgaters have an automatic elevation hold, so the dish tilts up and down to compensate for the entire unit tipping. That *should* help with minor rocking of a boat at a dock. They don't seem to have an automatic azimuth / turn function, so they probably wouldn't work under way or anchored when the boat is potentially changing direction.

yea that was my concern trying to find a marine one for cheap.

Or move your antenas until obtain a geometric configuration that compensate the dephasing of each one by 2pi.

You could aligne the antenas in the normally to the satellite direction and use the same cable lenght for all aligned antenas.

Thanks! I have some other projects with these, one is working so well that I might have a video for it next week!

they all have active LNBs in them which amplify and shift the incoming satellite RF engergy in the 11/12GHz band down to the 1GHz band - looking into the output side of them with a VNA isn't going to show you anything useful in the slightest

@@gorak9000 I was thinking "passive" like the antenna design more-so and maybe thinking like sections of the board if funky since can be tuned like an antenna is from my experience. Right, the active side of the system components is 100% like you note, at least I think offhand. Maybe a spectrum analyzer like the TinySA Ultra might show something useful in regards to outputs of the active side. Wondering however, if all are tuned properly and only out of phase maybe, what would be able to test?

I haven't tried the VNA, I don't think it will go that high. I'm also not really sure how to use all its feature.

@@saveitforparts If you ever get into making your own antennas or testing others, including thinking everything like you know is like an antenna as you mentioned with electronic characteristics, the IMSAI Guy and joe smith youtube channels details well and for Smith Charts w2aew has about the best presentations regarding interpreting. I think I have a stupid video or two where I was amazed the aluminum duct tape covered dowels performed so well. Though the driving element is old copper tube. Sucks theres not really a good english language comparison table I just realized. I might have to make one like I did with the upgraded SDR spreadsheet I linked somewhere in a hackaday comment and still have if your interested in though links don't post usually. Basically, all the NanoVNA's with a dash something go up to 3GHz and above like to 4GHz with some models. Theirs a 6GHz one as well I forget offhand. TinySA Ultra reads like thats up to 5GHz where at least can downconvert to view like a 1GHz range. Yeah, I just realized I can look around to see if any crowdfunded site made a table like the SDR projects had and maybe copy and paste that into cloud spreadsheet thats opensource and add the other makes and models. Or translate the French one I found that looks best.

You're a little late. Earth has been sending out many space probes like Voyager and such. 😆

I'm actually using a security camera pan/tilt as an antenna aimer on other projects. It's not really ideal since the range of motion isn't very large and it won't hold heavy antennas, but it's a starting point.

The Russian one stopped working last year, so no good there. The NOAA satellites do have a higher resolution channel, I just need to set up a motorized antenna to track them for that one to work.

You know, I'm not sure. I've heard of people taking multiple photos through a telescope and then combining them in software to get the best image. However, I think they're mostly trying to fix distortion from the atmosphere and they want photos at different times, not from different angles.

In theory yes, but not in practice

I think it's more about the internal signal paths and wiring being slightly different lengths. At these frequencies, any variation in wiring is going to affect the phasing quite a bit.

@@saveitforparts but thats my point, at 1 ghz the wavelength is less than 30cm thats less than a foot, seems to me if there is a variation in the equipment you might offset it just by moving an antenna less than a foot. in fact as a good starting point take the wavelength and multiply by 2 or 3 and set the distance from horn to horn to that.

I'd need a variety of cable lengths, right now I just have a bunch that are all the same.

I was looking into this some more. There would be some pretty substantial challenges. 1. 10 ghz satellite signals means that we would need to capture data at 80ghz. 2. we would need an atomic clock to get accurate time-stamps at 80ghz 3. say we spaced the dishes 10meters apart and got millimeter accuracy then we could simulate a 10 meter dish. -Resolution = 1.22 * (wavelength / dish diameter), so about 0.0036 radians -the moon is about 0.0105 radians, so thats like 9 square pixels to cover the whole moon.

@@forTodaysAdventure #2 is not necessary. Just train/synchronize on the preamble. (Er, as long as your delay isn't more than one packet. Unless packets have an ID?)

​@@fitybux4664 I was thinking more like for applying this to astronomical observation. Imaging quasars, etc...