Introduction

[string]

Hello I’m Steven (string) from the Deep South, US. I am retrofitting two 3.7m dishes positioned 37m apart as part of a planned remodel my parents old house next door. Luckily, I had one of these running 300ft to my house in the C-Band days so I used oversized dishes.

 Now I'm looking for a new purpose for this unusual asset. Many trees have grown in since these were in use. I can easily cut back enough to do zenith drift scans. One day I could do a simple AZ/EL to the south but that's about all the sky these will ever see in this location.

I became interested in Astrophotography during the lockdown. My hobby expanded quickly to include an Observatory with Tandem RASA8s. However, I'm a RA newb so I can neither afford or even comprehend building/running a state of the art radio telescope. 

Recently, the emergence of inexpensive SDRs and parallel GPU technology has given me the Amateur RA bug. I am currently building a Jetson Nano design being tested by Evan Mayer but I'm very interested in what you're doing as well, David. I am very open to input from anyone else that has experience with building an Amateur interferometer. I'm super happy to find to find this group.

[David Lonard]

Hello Steven,

The two 3.7m dishes should really do a nice job on C band and the 37 meter baseline sounds very nice also. If your dishes are already permanently positioned, the long baseline will present some challenges (and capabilities) in that fringes will come fairly fast and for other technical reasons you will have to more carefully control the length of coax running to each dish. A cheap time delay reflectometer such as a Mini1300 antenna analyzer or a nano-VNA will come in very handy for this. My more recent code that does polynomial fitting should help with long baseline delay issues.

For a lower cost path to getting some results, I would use a LimeSDR and use the code that I've posted as the interferometer's back end. For the front end, you could use a pair of Titanium C-1PLL LNBs modified to accept an external 25 MHz oscillator. 25 MHz TCXOs can be purchased on Ebay for around $25. New RG-6/U cable should suffice for the cable runs.

I am using a i9-9700k CPU/RTX3090 GPU for my interferometer (using a XTRX SDR with 50 MHz bandwidth), but I've also been able to achieve a 10 MHz bandwidth with an i5 laptop that has a GTX1050 GPU. I'm not sure about the capabilities of the Jetson Nano.  

If you already have old Chaparral feeds, I would splurge and get upper C band Norsat LNBs to fit to them. They will have a cleaner spectrum with far less RFI. I would add that the standard Titanium LNBs can be useful for engineering purposes since C band signals can be used for dish alignment and tests of system coherency. Aiming a 3.7m dish is going to be tricky. Once getting away from the Clark Belt, you won't have any clear references for pointing. I would recommend getting a Witmotion 3 axis inclinometer for aiming. 

Trees will really hurt system performance, so near-zenith targets are a good idea. I would recommend Taurus A as a good 'starter' object at 22 degrees DEC. Another good target is Orion A. From the southern US, Orion A should follow very close to the Clark Belt and if you point both your dishes at a bird with a satellite test receiver, that should get both dishes synched up for this bright target. 

David

[string]

David,

I am excited to read your response. The project that you describe is in a different league than the Nano project I was tinkering with. I did not have a CUDA capable GPU to experiment with so I picked up the Nano. It's impressive Compute Capability Score is 5.1 out of 8.6... if you can picture a $60, Ubuntu, CUDA capable Rasberry pi on steroids. However, It's not a substitute for a CUDA Ready PC. I am already planning to build just such a box. That is another rabbit hole in and of itself.

Although, the SDR RA equipment that your great code supports is in a different league, it's still not too extravagant. I can budget enough for this if I feel I can do something worthwhile with my limited view. The targets you mention are tantalizing indeed but I'm not sure if my location justifies the expense. I would appreciate your opinion on that?

The dishes are mounted better than most TV installations of the day. The dishes themselves have a heavy support sub-structure. Both will need some work getting back in proper shape but I think it can be done? They are mounted very close to the house as you can see. There is really is no other place for them so I'm stuck with that. I have manually moved one to about the zenith but the mount is not really designed for that. Do you have knowledge of a slip over pole (EL only) mount that would handle these big dishes even manually but with some control?

I have gutted the entire house down to studs so it will be extremely easy to bring the cables right into a desk at the center point. (See red rectangle in pic) The cables should be the same length give or take a few feet extra on one. Will that help?

Regardless I would like to run 4 high quality cables to the desk while it's so easy to do. For this wire installation, I will splurge... Is there something MUCH better than RG-6/U that I should consider?

I read what you said about 4 channel SDRs. I'm very interested in that. I already have two directTV dishes on the house (see red circles) I don't suppose those are of any real value here? I do have space for a couple of horns or small dishes in addition to the big ones. I would love to hear more about that just for the sake of the remodel wiring job at hand.

My studio office is behind the right dish in the pic and not part of the remodel. I'd like to get started playing with that dish. So far I've moved it to the Zenith, with my Canntena. I've run A USB Repeater out there and have a RTL-SDR/Sawbird + H1 in the mount box. I'm running GQRX/gnuradio on a modest i3 Ubuntu desktop with a GTX460. I have not seen any evidence of data yet as one would expect. Is there anything I can do with my starter setup?

I have no problem splurging on Norsats one day if I can get results with the dish placement. For now, I would like to start with one Titanium. Is the the oscillator mod a simple task? I would like purchase one immediately to replace my DIY feed.  There are a few models. This is the only one that has the correct letters: Titanium Satellite C1-PLL Phase Lock Loop C-Band LNBF. Is that the model I need?

Do you think the new LimeSDR Mini would be enough for two dishes?

One of the dishes still has the old chaparral feed with two LNBs perpendicular to one another. I'm not sure if that has any merit.

Witmotion 3 axis inclinometer is just brilliant. I'll get one of those now.

You have packed more info on the SDR based interferometer in one ReadMe and this post than I have seen thus far all over the net. Sorry for long response and so many questions. I can't wait to hear what you say next.

Thank you,

Steven

[David Lonard]

Steven,

I think your biggest asset is the big dishes that you have that I expect would exceed the capabilities that I have available to me. The little dishes you mentioned could work, but I would focus on your big dishes that will be far more capable. The LimeSDR mini is only a 1x1 transceiver, so it can't do interferometry. I would stick with the 'full sized' LimeSDR itself. Ordinary RG-6/U cable with good compression fit F connectors will work just fine since the LNBs have enough gain to overcome cable losses.

I've attached a link for the Titanium LNB mod.

https://osmocom.org/projects/miscellaneous-projects/wiki/C_Band_LNB_Ext_Ref_Mod/12

I would focus on the Titanium C1-PLL. They do make a wide band version (C1W-PLL), but it will likely get desensed by 4G LTE signals, unless you live in a very remote, RFI-free location. With Titanium C1-PLL LNBs, you can tune to frequencies between 4200 and 4300 MHz to get some sorta clear spectrum, especially if you are out of the way of aircraft flight patterns.

With the Jetson Nano, you likely could modify my code to reduce the size of the 2D FFT blocks down to the point where you no longer see memory errors. There will be some point where it will do the job, but I'm not sure what its final bandwidth capability will be. 

I would save the Chaparral orthomode feed that you have for later if you really decide to go down the radio astronomy rabbit hole. The pros typically use two polarizations and if you had a matching orthomode feed on the other dish (4 Norsat LNBs total), it could be a nice setup. Since each polarization is separate, 2 LimeSDRs could be put to work on each polarization without the need for phase synchronization. Synching two LimeSDRs for 4x4 work isn't easy and I haven't tried myself. From a single LimeSDR, 2x2 synchronization works as a turn key solution. For more baselines, something like a XTRX Octopack will be needed, but it has some support issues itself and the cost is pretty high.

The big unknown and something you might want to think about before you commit to interferometry, is that while bright radio sources like SgrA, CasA, CygA and TauA are within reach, other sources may not be practically achievable. This would leave you with only a single or maybe two radio sources that could be detectable at any given declination during a drift scan. With my 2.1 meter dishes, this is all that I can achieve for a single day drift scan. Integration over multiple days might do a better job, but this is still a work-in-progress for me. System stability (LNB temperature, coax temperature-dependent phase delay changes, etc.) are the next hurdle I face. With 3.7 meter dishes, you might be able to improve on this to the point where fainter objects come into play. At some point in the future, better SDRs with higher bandwidths, say 300 to 500 MHz, should open the door for smaller dishes.

David

[Leif Johansson]

Skickat från min iPhone

9 juni 2021 kl. 15:59 skrev David Lonard <dmlo...@gmail.com>:

Steven,

I think your biggest asset is the big dishes that you have that I expect would exceed the capabilities that I have available to me. The little dishes you mentioned could work, but I would focus on your big dishes that will be far more capable. The LimeSDR mini is only a 1x1 transceiver, so it can't do interferometry. I would stick with the 'full sized' LimeSDR itself. Ordinary RG-6/U cable with good compression fit F connectors will work just fine since the LNBs have enough gain to overcome cable losses.

You can provide an external clock to the mini tho so I guess

in principle it would be possible to correlate externally in sw

after fft in the fpgas?

--
You received this message because you are subscribed to the Google Groups "Amateur radio interferometry" group.
To unsubscribe from this group and stop receiving emails from it, send an email to amateur-radio-interf...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/amateur-radio-interferometry/da456aa8-d30a-4300-9fdd-79476dbc1324n%40googlegroups.com.

[David Lonard]

Yes, it should be possible to use LimeSDR minis if they share a common clock, although there will be some issues related to synching the streams due to latency in the USB interface.

I previously did something similar using BladeRFs that you can look over in some of my earlier posts and if you search for old messages in the SARA Google Groups archive (search for N5OIQ). I synched three radios but this was done as a post processing approach using data streams saved to SSDs using Gnuradio. 

Stream alignment can be a challenge, but I posted code that can be used to accomplish this task using FFT/IFFT based cross correlation (see some of my earlier posts). This does open up the possibility for doing VLBI with a collaborator and I've talked with a few people about doing this, but we never actually got around to doing it.

David

[string]

I knew that the small dishes would not be in the same ballpark as the large ones. Just making sure they had no added value. I'll probably only run two wires through the walls.

I'm trying to understand how much bandwidth to expect from the Lime vs the XTRX and how powerful a computer must be to provide the maximum bandwidth with each choice. I was planning to experiment with Effex with the Nano but I must also build a decent unix desktop. I'm looking for the best bang for the buck CUDA Capable machine. I need to understand more about the bandwidth to know what I might build.

It's somewhat difficult to shop for the SDRs. Could you please provide the link that you would use to purchase the XTRX or LimeSDR?

I have ordered the TCXO's on Ebay. Shipped from China. I had already ordered a QO-100 Bullseye TCXO LNB before I talked to you but it's also on a slow boat from China. I'd like to know what you think about the usefulness of this LNB. I did not know what to order and this was on the RTL-SDR site.

https://www.rtl-sdr.com/product/qo-100-bullseye-tcxo-lnb/

I have received the Titanium C1-PLL. I see that out of the box it requires 12-20v. I'll have to read about the mod.  I'm running everything with the RTL's 4.5v bias tee right now. I'm not sure if I can do anything with the C1-PLL right away. I'd be happy to do something like this with one dish: 

https://www.rtl-sdr.com/cheap-and-easy-hydrogen-line-radio-astronomy-with-a-rtl-sdr-wifi-parabolic-grid-dish-lna-and-sdrsharp/

I won't really have permanent interferometer installation until I do the remodel. However I can run cables over the ground. I want to learn as much as I can, build and experiment with all the bits as I shop and build my ultimate system. I need the required research to better understand the front end of this as well as the back end.

[string]

David,

UPDATE: I took a leap of faith and started my build around an AMD Ryzen 5 5600X 6 core. I hope that will be adequate for the bandwidth that I can achieve with the LimeSDR. What do you think?

Steve

[David Lonard]

Steve,

For me, I can sustain 10 MHz bandwidth with an i5 laptop that has a Nvidia GTX1050 GPU. I haven't tried using the CPU myself, but I am guessing that the Ryzen could do 10 MHz. Graphics cards are almost impossible to get right now, but you could always upgrade to one in the future after seeing how the Ryzen core does. 

I got my LimeSDR from Crowd Supply. I see that the LimeSDR is in 'pre-order' status right now, as is the Mini:

https://www.crowdsupply.com/lime-micro/limesdr

The QO-100 Bullseye sounds interesting, but would still need to be modified for interferometry. With the Bullseye, you could use it directly with an RTLSDR to measure sun and moon noise. Some of the other cheap Ku band lnbs can be modified and I have used them successfully for interferometry (see some of my earlier links). At Ku band, your large dishes might not focus well enough for such a high frequency and the mesh may be too coarse as well. 

You have likely seen posts from others that have detected neutral hydrogen and I would consider this a good thing to consider as well and will work well with a RTLSDR backend. You will need a pair of L band feeds for this and you should be able to get nice fringes at 1420 MHz with a LimeSDR. Since your dishes are large enough, a choke ring at 1420 MHz will really improve system performance as well. One thing to keep in mind though, is that neutral hydrogen won't produces any fringes itself, but compact, continuum sources will.  

The low bandwidth of the RTLSDR is well matched for neutral hydrogen observations, so you could confirm that all of the front end parts of your setup are working well.

David

[string]

David,

For the rear end, I've pre-ordered the Lime SDR. I'll have the computer built long before anything arrives. The front end is just a one dish experiment at this point. The C1 w/mod, bullseye w/mod, a L-band I just ordered. and I'll take down the other dish and have a look at the Chaparral feed and give the Norsat LNBs some thought as well... Then I have to decide on a matching pair.

GPUs prices are crazy right now so I am building around an old GTX460 with a CUDA score of 2. Maybe prices will get back to MSRP by the time my front end is completed. FYI The Ryzen 5600x is 6 core and in the range of a Rocket Lake i7. I'm hoping it it will fall between your two test systems with a new GPU and lots of RAM. I'm future proofing the build for a next gen Ryzen just in case.

I have a lot of research to do starting with what you have written. I am open to any required reading that you might suggest to take me from zero. I'm really going to give this a go, David.

Thanks so much!

Steven