21cm band. Refurbishing the dish with a new Cantenna.

[SteveBz]

Hi People,

I've decided to start work on refurbishing my profile photo.  This is quite a long and arduous voyage as obviously I have to start with the rust.  I've decided to drop the EQ5, as it's not really strong enough and revert to the original motorised mount with it's RA motor rotated by 90 degrees.  Ie the Earth is the RA motor and the RA motor is the Dec motor (RA is the new DEC, I could say). I spent most of this week with the wire brush attachement to my drill cleaning up, coating with Hammerite rust treatment, undercoating and finish coating the mount while waiting for parts from Amazon US and UK, AliExpress, Ebay to arrive.  In particular:

• 150 mm x 500 mm aluminised pipe + separate 150 mm end cap
• RTL.SDR software radio dongle
• SAWbird + H1 LNA
• SMA Connector and various others.

The dish is 130 cm diameter with a 47 cm focal length, making it f/2.8 if that means anything here.  😀

Here are some photos:

Adjusted, painted, motorised mount.

Hammerite rust treamement and some undercoat.

Fully undercoated support plate (it's VERY rusty)

Topcoat, also Hammerite.

Cantenna, cut down to 28 cm with angle grinder and then sharp edge ground down with emery disk.

Cantenna with end cap.

Cantenna undercoat.

Cantenna topcoat.

There are a number of measurements which it is hard to get right.  The length of the 'can' is variously given as anything from 14cm to 38 cm (including the choke).  The correct answer seems to be 28 cm. It's calculated as 'at least' 3/4 of the wavelength in the waveguide.  Apparently the wavelength in the waveguide is not 21.1 cm, but more like 37 cm after other interactions are included.  28 cm seems to be the right answer.

I'm waiting for the topcoat to dry before assembling the dish and mount for the next exciting installment.

Any warning, advice, observations or encouragement happily received.

Kind regards

Steve.

Edited February 6, 2022 by SteveBz
Add photos

[ZiHao]

Nice work.
There's this excel sheet provided here http://www.setileague.org/hardware/feedchok.htm that calculates the feedhorn dimensions. I have built a 1.4m geodesic dish recently, using a feedhorn with diameter 145mm and length 180mm without choke. I didn't follow the exact length as given in the excel spreadsheet to minimize the weight as much as possible. The length of the feedhorn should be twice or three times as long as the distance from the probe to the end cap as a rule of thumb according "The Radio Sky and How to Observe it" by Jeff Lashley. 

feedhorn.xls

[SteveBz]

1 hour ago, ZiHao said:

Nice work.
There's this excel sheet provided here http://www.setileague.org/hardware/feedchok.htm that calculates the feedhorn dimensions. I have built a 1.4m geodesic dish recently, using a feedhorn with diameter 145mm and length 180mm without choke. I didn't follow the exact length as given in the excel spreadsheet to minimize the weight as much as possible. The length of the feedhorn should be twice or three times as long as the distance from the probe to the end cap as a rule of thumb according "The Radio Sky and How to Observe it" by Jeff Lashley. 

feedhorn.xls 22.5 kB · 0 downloads

Hi Zi Hao,

I hope you're well.  Thanks for your input.

Yes, that's about what I calculated/found.  28 cm overall length, 9.3 cm distance from monopole to back, 15cm diameter, 4.9 cm length of monopole.  Nealy finished.

I'll post a photo of the antenna when it's done.

Kind regards,

Steve.

Edited February 7, 2022 by SteveBz

[SteveBz]

And here is the finished feedhorn/cantenna:

Monopole connected to port.

Monopole/port mounted 9.3 cm from base.

Final product with NooElec SAWbird H1 Premier LNA and RTL.SDR.

Now comes the challenge of mounting it at the focal point.

S.

[SteveBz]

I can't remember where I found this paper,  but it covers all the calculations very nicely and compares antennae.

Scroll to section 7.5.

Steve.

 

Hydrogen_1.pdf

[SteveBz]

And here's the finished product. I guess I'll need to dismantle it to get it through the door 🙂.  Then it needs a bit of a tidy-up.

The next question is software.  I've managed to get @Coto's Virgo software running on my PC, but not on the RPi3b+.  I don't want to damage my current setup, so I have a 128GB SD card on order from China to see if I can get it working with a fresh install.  The other option is @Victor Boesen H-line-software, which I got running on the RPi3b+.  In both cases I'm not quite sure how to turn the sampled output into an integrated signal (ie the total flux in 1420 +/- 50 MHz), for each sample.  I'd like to get a skymap of the 21cm line intensity in the way both the above authors have on their respective pages.

Edited March 6, 2022 by SteveBz

[Victor Boesen]

I'm surprised I haven't seen this thread before now! Your build looks excellent

On 06/03/2022 at 18:51, SteveBz said:

In both cases I'm not quite sure how to turn the sampled output into an integrated signal (ie the total flux in 1420 +/- 50 MHz), for each sample.

I'm glad you got my software working! I'm currently in the proces of rewriting a lot of the code structure to make future improvements easier. Unfortunately an unexpected faulty PC has set back my work by quite a bit, and I haven't had the opportunity to work on the code for some time. I will hopefully have the ability to do so soon! With regards to your question, the software is only able to compute the signal to noise ratio of the hydrogen line peak in dB. If you want to create a map from this signal intensity, you could take each signal to noise ratio and plot it together with the right ascension and declination of the observation. Unfortunately, there is currently no simple way to do this yet, but from the beginning of the project, it has been my intention to add a feature in the software that will do this automatically. However, due to life's circumstances I have not found the time to implement this yet, but I hope to do so at some point

I think you would find the following article quite interesting if you haven't seen it already. It's basically the same procedure as I explained above:
https://www.rtl-sdr.com/jobs-radio-telescope-hydrogen-line-northern-sky-survey-with-rtl-sdr/

I hope I understood your question correctly? If not, feel free to ask again!!

Victor

[SteveBz]

11 hours ago, Victor Boesen said:

the software is only able to compute the signal to noise ratio of the hydrogen line peak in dB.

That's perfect.

If I run

python3 H-line.py - t

on my RPi, then I can run a query from my pc every 5 minutes or so and plot the result.

I don't really think it needs a rewrite, it's quite clear, but I'll happily chat about improvements if you like.

Kind regards,

Steve.

[Victor Boesen]

55 minutes ago, SteveBz said:

That's perfect.

If I run

python3 H-line.py - t

on my RPi, then I can run a query from my pc every 5 minutes or so and plot the result.

I don't really think it needs a rewrite, it's quite clear, but I'll happily chat about improvements if you like.

Kind regards,

Steve.

Thanks Steve, appreciate it! The re-write is mostly for my own sake I made some sneaky solutions back then that I would like to fix/tidy up, before I would go ahead and advance to new features. I would also like to increase performance slightly. I know the duration of observation is mostly due to the actual sampling by the SDR itself, but when performing digital signal processing on a Raspberry pi I am quite confident I can reduce the execution time by a significant amount.
Please get in touch if you experience any major issues or have a great idea I should try to implement I may not be able to fix/add it immediately, but I will add it to my TODO-list.

Victor

[SteveBz]

1 hour ago, Victor Boesen said:

Thanks Steve, appreciate it! The re-write is mostly for my own sake I made some sneaky solutions back then that I would like to fix/tidy up, before I would go ahead and advance to new features. I would also like to increase performance slightly. I know the duration of observation is mostly due to the actual sampling by the SDR itself, but when performing digital signal processing on a Raspberry pi I am quite confident I can reduce the execution time by a significant amount.
Please get in touch if you experience any major issues or have a great idea I should try to implement I may not be able to fix/add it immediately, but I will add it to my TODO-list.

Victor

For me the performance is slowed down by uploading the whole sample file from the Pi to my pc. If it just returned the answer requested (Eg the max S/N ratio) that would be much quicker over the network. Or indeed the integral.

I might play with it and let you know.

Tx

Steve.

[SteveBz]

So here's the dish all plumbed in and ready to go:

And here's the first connection. 

It seems to be the same as with no aerial, so I think maybe there is a broken connection.  I'll try a continuity test. But as it's dark, I'll just cover up the LNA and do it at the weekend.  I might even buy a shorter cable as this is old and 25 m long.  Maybe a new 2 m cable would be better.

It goes:

Dipole=>LNA(NooELec H1)===========25 m cable===============>RTL.SDR=>Powered USB hub=>RPi with H-line-software.

Regards,

Steve

[SteveBz]

On 07/02/2022 at 15:36, ZiHao said:

Nice work.
There's this excel sheet provided here http://www.setileague.org/hardware/feedchok.htm that calculates the feedhorn dimensions. I have built a 1.4m geodesic dish recently, using a feedhorn with diameter 145mm and length 180mm without choke. I didn't follow the exact length as given in the excel spreadsheet to minimize the weight as much as possible. The length of the feedhorn should be twice or three times as long as the distance from the probe to the end cap as a rule of thumb according "The Radio Sky and How to Observe it" by Jeff Lashley. 

feedhorn.xls 22.5 kB · 1 download

Hi Zi Hao,

I forgot to ask, rereading your post,  did you manage to get any results from this or have you not deployed it yet? I'd love to see your build and setup including software.

Kind regards

Steve.

[ZiHao]

On 11/03/2022 at 16:43, SteveBz said:

Hi Zi Hao,

I forgot to ask, rereading your post,  did you manage to get any results from this or have you not deployed it yet? I'd love to see your build and setup including software. 

Kind regards

Steve.

Hello Steve,

The genuine rtlsdr just arrived few days ago, so you asked at the right time! I was using a clone rtlsdr which has a lot of noise before this, and the genuine one shows a lot of improvement. I am using a very simple python script with pyrtlsdr for data acquisition, this is similar to Victor's method. My setup is LNA-->5m LMR400-->Rtlsdr-->Rpi. Still working on a cleaner baseline for easier data processing and hopefully able to mount the dish on my equatorial mount soon.

 

Edited March 13, 2022 by ZiHao

[SteveBz]

8 hours ago, ZiHao said:

Hello Steve,

The genuine rtlsdr just arrived few days ago, so you asked at the right time! I was using a clone rtlsdr which has a lot of noise before this, and the genuine one shows a lot of improvement. I am using a very simple python script with pyrtlsdr for data acquisition, this is similar to Victor's method. My setup is LNA-->5m LMR400-->Rtlsdr-->Rpi. Still working on a cleaner baseline for easier data processing and hopefully able to mount the dish on my equatorial mount soon.

 

Hi Zi Hao,

So impressive. I love the ultra light dish.  Did you make it yourself or buy it? The waveguide is also very nice. I like how you've mounted it with the collar.  I might try the same idea.  You have the same NooELec H1 as me, so I'm pleased to see that it works. You're definitely ahead of me. The fact that the LNA comes out at a right-angle could put extra torque on the port.  I've put an elbow on mine to avoid that.

I also seem to have short circuited my rtl-sdr. There is no dc voltage on it at all, even after I switch it on🙄, which I didn't realise you had to do.

I've ordered a second one.  I just added a baby antenna direct to the sdr without an LNA -  for testing the software. Bad idea.

Can I ask you, have I understood that right? Did you need to active the bias tee?  No one else has mentioned this at all.

Good luck.

Steve.

[ZiHao]

21 minutes ago, SteveBz said:

Hi Zi Hao,

So impressive. I love the ultra light dish.  Did you make it yourself or buy it? The waveguide is also very nice. I like how you've mounted it with the collar.  I might try the same idea.  You have the same NooELec H1 as me, so I'm pleased to see that it works. You're definitely ahead of me. The fact that the LNA comes out at a right-angle could put extra torque on the port.  I've put an elbow on mine to avoid that.

I also seem to have short circuited my rtl-sdr. There is no dc voltage on it at all, even after I switch it on🙄, which I didn't realise you had to do.

I've ordered a second one.  I just added a baby antenna direct to the sdr without an LNA -  for testing the software. Bad idea.

Can I ask you, have I understood that right? Did you need to active the bias tee?  No one else has mentioned this at all.

Good luck.

Steve.

Hello Steve,

I made the dish myself, the idea is developed by JA6XKQ. No pre-bending work required, the structure will form by itself. You are right, I can see a little bending when I connect the cable to LNA so seems like I have to order a right angle sma as well. I haven't tested the waveguide with a vna, so most of it is not really optimized for hydrogen line I believe.

The user guide for the bias tee is here https://www.rtl-sdr.com/rtl-sdr-blog-v-3-dongles-user-guide. There should always be a actual powered device connected to the sdr before turning on the biast. So always check you have the LNA connected. When it is turned on, you should see a white led light on your LNA.

Links for the geodesic dish if you are interested: https://www.reddit.com/r/RTLSDR/comments/7ksfba/diy_parabolic_dish_reflector_antenna/

Zi Hao

 

[SteveBz]

So my SDR was indeed shorted.  The new one seems to work in that the LNA light comes on.  Good news.  However, my signal is amazingly noisy.  I have no idea why.  There is no signal whatsoever, even though I'm apparently in the middle of Cygnus:

If I ramp up the Median Smoothing I still get nothing.  I can't believe the antenna doesn't work, it's so simple!  Noisy cable, maybe?

I'll just replace it and see.

Kind regards

Steve.

[Victor Boesen]

1 hour ago, SteveBz said:

There is no signal whatsoever, even though I'm apparently in the middle of Cygnus

How many samples ("n" option) are you collecting?

And also, have you tried listening to regular FM broadcast with your antenna feed? That could quickly conclude whether the SDR is even receiving any signal. Note you should try this without the H1 SAWbird in-line of course 

Victor

[SteveBz]

5 hours ago, Victor Boesen said:

How many samples ("n" option) are you collecting?

It's like this:

python3 H-line.py -s 2400000 -r 11 -n 1000 -l 51 -g 0 -z 0 -a 90 -e 192.168.1.116 -m 3 -i 0

I gradually ramped up m to 200, which smoothed the line, but still no signal.  I've got a new cable on order, should arrive tomorrow.

Otherwise I managed to get the DEC motor working today, so that should help.

I don't have a spare LNA, so I can't really check the SDR without shorting it again!  Unless you think I should do it by turning off the bias T?

Kind regards

Steve

[Victor Boesen]

44 minutes ago, SteveBz said:

 

I don't have a spare LNA, so I can't really check the SDR without shorting it again!  Unless you think I should do it by turning off the bias T?

Just turn off the bias T, remove the LNA from the receiving line and then try to have a look around the FM band in a software like SDR++ or SDR#.

44 minutes ago, SteveBz said:

It's like this:

python3 H-line.py -s 2400000 -r 11 -n 1000 -l 51 -g 0 -z 0 -a 90 -e 192.168.1.116 -m 3 -i 0

Try something like this instead:

python3 H-line.py -s 2400000 -r 11 -n 10000 -l 51 -g 0 -z 0 -a 90 -e 192.168.1.116 -m 3 -i 0

It will make the spectrum look cleaner but not fix your problem, if your SDR isn't receiving anything from the antenna, hence my suggestion with the FM band.

Victor

[SteveBz]

59 minutes ago, Victor Boesen said:

Try something like this instead:

python3 H-line.py -s 2400000 -r 11 -n 10000 -l 51 -g 0 -z 0 -a 90 -e 192.168.1.116 -m 3 -i 0

It will make the spectrum look cleaner but not fix your problem

Tried it with no success.  I'll try the new cable tomorrow and then I'll try

1 hour ago, Victor Boesen said:

then try to have a look around the FM band in a software like SDR++ or SDR#.

Tx.

[SteveBz]

Hi People,

So amazingly I found the problems. Both were manufacturing faults:

1) My RTL-SDR was not an original model and the DC Bias never worked.  The second one had all the manufacturers details and worked perfectly (apparently).

2) My elbow connection had no pins in it, so there was no current!!!!

Both have now been fixed and look at the result.  But I have these huge spikes (at about 1420.0 & 1420.8) which distort everything.  What to do?

Otherwise there seems to be a mild hump which I feel I can built on.

I'm running a 24 hour scan now with i=1.25, m = 10 and n = 1000.

Advice please.

Tx

Steve.

Edited March 17, 2022 by SteveBz

[Victor Boesen]

1 hour ago, SteveBz said:

Hi People,

So amazingly I found the problems. Both were manufacturing faults:

1) My RTL-SDR was not an original model and the DC Bias never worked.  The second one had all the manufacturers details and worked perfectly (apparently).

2) My elbow connection had no pins in it, so there was no current!!!!

Both have now been fixed and look at the result.  But I have these huge spikes (at about 1420.0 & 1420.8) which distort everything.  What to do?

Otherwise there seems to be a mild hump which I feel I can built on.

I'm running a 24 hour scan now with i=1.25, m = 10 and n = 1000.

Advice please.

Tx

Steve.

Looks great Steve!! Glad you found the problem;)

To increase the resolution of the observation, I would try increasing n to 10000 (10k). Each observation will take longer, but it should look even nicer
Look forward seeing your 24h observations!

Victor

[SteveBz]

1 hour ago, Victor Boesen said:

Looks great Steve!! Glad you found the problem;)

To increase the resolution of the observation, I would try increasing n to 10000 (10k). Each observation will take longer, but it should look even nicer
Look forward seeing your 24h observations!

Victor

Just restarted with:

python3 H-line.py -s 2400000 -r 11 -n 10000 -l 51 -g 0 -z 0 -a 90 -e 192.168.1.116 -m 10 -i 2.5 -d

Note n=10000, m = 10 and i = 2.5 (same as every ten minutes).  I've hacked the 'd' option to dump in 24 hr mode too.

Tx for the ongoing help!!

Kind regards

Steve.

[SteveBz]

So here's a quick update of the rig with it's 'hat' to keep rain off the electronics.  Is it enough?  What about condensation?

Secondly (and it's dark here now),

I've just added a superjack actuator to the DEC axis.  It seems to be a bit under-powered.  36 v and 1 amp.  Maybe 1 amp is not enough.  What do people think?

Kind regards

Steve

Edited March 17, 2022 by SteveBz

[SteveBz]

So the 24 hour mode is performing quite nicely.  You can see the bumps corresponding to Cygnus, more or less.  It's still quite noisy with those big spikes.  Does anyone have any idea what is causing them?  Maybe I should just truncate the data. 

This one here seems to be showing me three spiral arms.  One at 1.4205 GHz, and then two more a bit further away.  Have I read that right, or is it just noise? I don't know whether that is 'Perseus', 'Outer' and 'Centaurus, or Sag, Per and Outer.  Any ideas.  And even is that a fourth out at about 1.4212GHz?  In which case I could answer the question.  But really, I need to improve the data.

Steve