SteveBz

I also think that my mount is off by about 18 degrees, so at low Az there is a greater RA Error than at High Az.  Rather than waiting for the next clear night, I'm trying the correction RA_New = Delta_RA/cos(lat). For me that translates to Delta_RA ~ 25 degrees.  I get a move in the right direction, but not completely.  So I think I may leave the mount as it is and just update the Az input to H-line.py and also, retrospectively amend the RA data for my previous scans.

So, just to be clear, I'm doing RV = LSR_Adjustment - Doppler_Speed.

Kind regards.

31 minutes ago, ZiHao said:

Interesting pattern, might be the pointing error, or the code itself, not too sure... I have only one set of data for now so can't really test it reliably. We need other sets of data🧐

Yes, I'm doing the Dec measurement with a digital level and I think it's always within 1 degree.  But the RA seems to vary (according to @Victor Boesen's graph) by up to 5 degrees.  Because of all the iron in the mount it's very hard to use a compass to align it.  Next clear night, I'll align it with Polaris. 

Actually, it seems to be nearly linear in Dec.  

In fact here (https://www.reddit.com/r/RTLSDR/comments/2pkue7/best_way_to_calibrate_with_a_known_signal_source/) they say:

'I purchased a Nooelec dongle and have been very pleased with it. Right out of the box, I am picking up air traffic as far as Philadelphia from Baltimore. I noticed, however, that when using gqrx in Linux that some stations appear to be off by 50khz or so.'

I think that's about 10 km/s, so about the same error as I'm getting.  I don't know how close the Nooelec is to the RTL-SDR Blog v3, however.

Steve.

I wonder @Victor Boesen, should I have calibrated my rtl-sdr?  The guy in the video suggested by @ZiHao talked about calibration.  What do you think? ZiHo, did you calibrate yours?

Steve

16 hours ago, ZiHao said:

c=299792.458
rest_freq=1420.40575e6
#freq=frequencyarray
velocity=c*(1-(freq/rest_freq))
corrected_vel=velocity*u.km/u.s

rv=c*((rest_freq/rest_freq)-1)*u.km/u.s

t = Time(, format='mjd')

loc = EarthLocation(lon=,lat=,height=)
ra=*u.deg
dec=*u.deg
sc = SkyCoord(ra,dec)

vcorr = sc.radial_velocity_correction(kind='barycentric', obstime=t, location=loc)  
rv = rv + vcorr

my_observation = ICRS(ra=*u.deg, dec=*u.deg, \
        pm_ra_cosdec=0*u.mas/u.yr, pm_dec=0*u.mas/u.yr, \
        radial_velocity=rv, distance = 1*u.pc)

new_rv = my_observation.transform_to(LSR()).radial_velocity

corrected_vel=corrected_vel+new_rv

Attached above is my attempt on LSR correction, I chose rv as zero so it is easier to add the final rv (new_rv) to the frequency array as I find it takes quite long to compute for individual frequency point. I don't think there will be significant difference in the final result using this method, I will check this again later. You can try the code above, just enter the corresponding info required. I have checked this with the LAB survey with my limited data at hand. The main peak in my data is around +9km/s, which agreed with the survey. 

OK, I just used your code.  It improved my code by about 7 km/s in the right direction, but I think I'm still about 10 km/sec too fast.  See here (the blue is corrected and the orange uncorrected).

Tx

Steve.

16 hours ago, ZiHao said:

The formulae should be correct, attached is the derivation.

I should have said, it's very nice to see this all written out by hand!  Your notes are much neater and more organised than mine! 🥴

On 16/04/2022 at 14:00, ZiHao said:

I have attached a general illustration of roughly how the arrangement of colored dots should look like, the picture is from DARA https://www.youtube.com/watch?v=RQiLMxQHxr8&t=2946

This was really useful, although I thought the guy was quite patronising to his his audience. 'Work hard, write up your results' etc.

He gets this bit wrong.

No brackets and wrong V.

3 hours ago, ZiHao said:

The formulae should be correct, attached is the derivation.

Perfect. That's what I got too.

What I don't have is the derivations for the other quadrants or outside the solar circle. Do you think they are different?

3 hours ago, ZiHao said:

c=299792.458
rest_freq=1420.40575e6
#freq=frequencyarray
velocity=c*(1-(freq/rest_freq))
corrected_vel=velocity*u.km/u.s

rv=c*((rest_freq/rest_freq)-1)*u.km/u.s

t = Time(, format='mjd')

loc = EarthLocation(lon=,lat=,height=)
ra=*u.deg
dec=*u.deg
sc = SkyCoord(ra,dec)

vcorr = sc.radial_velocity_correction(kind='barycentric', obstime=t, location=loc)  
rv = rv + vcorr

my_observation = ICRS(ra=*u.deg, dec=*u.deg, \
        pm_ra_cosdec=0*u.mas/u.yr, pm_dec=0*u.mas/u.yr, \
        radial_velocity=rv, distance = 1*u.pc)

new_rv = my_observation.transform_to(LSR()).radial_velocity

corrected_vel=corrected_vel+new_rv

Attached above is my attempt on LSR correction, I chose rv as zero so it is easier to add the final rv (new_rv) to the frequency array as I find it takes quite long to compute for individual frequency point. I don't think there will be significant difference in the final result using this method, I will check this again later. You can try the code above, just enter the corresponding info required. I have checked this with the LAB survey with my limited data at hand. The main peak in my data is around +9km/s, which agreed with the survey. 

OK, very nice. My code only has the last three statements of yours. I'll double check it.

Tx

Steve.

22 minutes ago, SteveBz said:

Yes I agree about the PDF.  The data and the formulae are very suspicious.

I also agree about validating the data from scratch.  I'm working through it manually and indeed I'm finding a lot of errors.

Here's the baseline.  Before I took all the samples within 5 degrees of the galactic equator and that's why I had so many apparently nice points.  Now I've gone through the whole set of data again manually and I've just taken the single point from each scan with the highest signal intensity.  I've only examined samples from Quadrant I, so I have about six data points in total.

Here is the base scan for the first point:

Then I adjusted it for Local System of Rest and flipped the axis to get radial velocity and Gaussian fitted it.  Here it is. The blue line is the original line. The green and pink lines are the result of subtracting the standard Gaussian curves. 

Then I looked at the University of Bonn result:

If you look at the Bonn result there are 2 clear peaks at about -55 km/s and about 10 km/s which correspond to the values on my chart of -38 km/s and 27 km/s.  Ie my results are about 20 km/s too large.  Maybe I didn't get the LSR adjustment right.

I feel I have to get this bit right first.

Any thoughts welcome.

Kind regards

Steve.

My LSR correction looks to be about 40-50 km/s and that seems quite large.  I would have thought a value nearer 30 or less would be better.  In fact, if I ignore the inflection at + 30 or 40 km/s, which is hard to evaluate precisely, the difference between the baseline result and the Bonn result seems to be about 30+/-5 km/s.

On 16/04/2022 at 14:00, ZiHao said:

Very interesting idea, especially the color that shows the amount of redshift. I think the equations used to plot the cartesian view of the milky way applies for all cases, you will get two positive r due to the two intersection points along the line of sight for quadrant I and IV. For quadrant II and III, I believe you will get one negative r and one positive r, I might be wrong. For the two positive values of r,  seems like we have to reject the data as shown in your attached pdf, I wonder how can we further validate the values. 

Yes I agree about the PDF.  The data and the formulae are very suspicious.

I also agree about validating the data from scratch.  I'm working through it manually and indeed I'm finding a lot of errors.

Here's the baseline.  Before I took all the samples within 5 degrees of the galactic equator and that's why I had so many apparently nice points.  Now I've gone through the whole set of data again manually and I've just taken the single point from each scan with the highest signal intensity.  I've only examined samples from Quadrant I, so I have about six data points in total.

Here is the base scan for the first point:

Then I adjusted it for Local System of Rest and flipped the axis to get radial velocity and Gaussian fitted it.  Here it is. The blue line is the original line. The green and pink lines are the result of subtracting the standard Gaussian curves. 

Then I looked at the University of Bonn result:

If you look at the Bonn result there are 2 clear peaks at about -55 km/s and about 10 km/s which correspond to the values on my chart of -38 km/s and 27 km/s.  Ie my results are about 20 km/s too large.  Maybe I didn't get the LSR adjustment right.

I feel I have to get this bit right first.

Any thoughts welcome.

Kind regards

Steve.

1 hour ago, SteveBz said:

Hi Victor,

I'll try to plot the errors afterwards.  I haven't even begun to think of that 😅

In the meantime I have 4 nice curves that are all in the wrong place!!! They should be centred on the galactic bulge, by they really aren't.  They don't even have a shared centre of curvature and I don't understand, either, why my coordinate does not seem to be centred on the Sun, when that is how it is defined.  There is even another level of points, that I haven't been able to extract yet because of the effect of electrical noise in my vicinity.

  

 

So it seems to me that the yellow line is about right.  The red and orange line is incorrect because R>R0 and I can't see how to fix that.  It's going faster, so it should be outside the yellow line.  Yes? No? The blue line and the short yellow lines are incorrect because they are in the wrong quadrants, also I don't see how to fix that.  Maybe I can just change the sign of the sin(l) for R>R0.

13 minutes ago, Victor Boesen said:

I'm not so sure actually. Shouldn't be too much different for the fourth quadrant (270-360deg longitude). However, the outer quadrants are a little tricky, hence the huge error bars you see on this image for areas further away than the sun.

Hi Victor,

I'll try to plot the errors afterwards.  I haven't even begun to think of that 😅

In the meantime I have 4 nice curves that are all in the wrong place!!! They should be centred on the galactic bulge, by they really aren't.  They don't even have a shared centre of curvature and I don't understand, either, why my coordinate does not seem to be centred on the Sun, when that is how it is defined.  There is even another level of points, that I haven't been able to extract yet because of the effect of electrical noise in my vicinity.

7 minutes ago, Victor Boesen said:

Yes, it is. You can only use the radial method, I believe it's in one of the papers you shared, for galactic longitude between 0-90 degrees (in reality more like 20-75 because of the galactic bulge and etc).

Radial method above.

Victor

And what can we use for the other quadrants?  In fact the paper I shared already has at least one fatal error in it's equations.  Maybe I need to go through them again.

17 hours ago, SteveBz said:

In the meantime, because I can pick out the main peak, but not the secondary or other peaks, I can only plot one Galactic Arm.  Here it is (I I don't know what the little blue one is and I don't quite know where the Sun is - any suggestions?):

It seems to me that the orange arm is the Perseus arm.  It looks about right.  Galactic Centre seems to be at (0,0) and we are at (0, 8.5).  

I got these just finding the peak dB value and doing the velocity calculation to calculate 'R' and then plotting - while converting (R, l) to Cartesian Coords.

At first I thought the inner, blue arm was just noise, but now I think it's a real arm, I just got the coordinate conversion wrong.  Do you guys know, is the maths different in each Quadrant?

So to push the limit then I subtracted the main Gaussian peak from each 'Data' profile and looked for the 'Max' again (as above).  Now I have a whole series of beautiful arms (like 4 different arms), but all in the wrong place 🥴

Steve.

In the meantime, because I can pick out the main peak, but not the secondary or other peaks, I can only plot one Galactic Arm.  Here it is (I I don't know what the little blue one is and I don't quite know where the Sun is - any suggestions?):

2 hours ago, SteveBz said:

Ok, I'll look later and let you know. 😄

So I understand what you said now.  Yes, you've answered my question but I may have some suggestions.  I think I probably need to look at the code properly.

Steve.

11 minutes ago, Victor Boesen said:

If you only using the debug file and processing that yourself then you'd need to perform a slight modification to your own stuff.

The data which the software plots is the "SNR Spectrum" as a function of the "Frequency list". The new version even includes the doppler and SNR in the debug file as well under "Observation results".

Hope that answered you question! Please correct me if I misunderstood your question

Victor

Ok, I'll look later and let you know. 😄

28 minutes ago, Victor Boesen said:

Very interesting plot! It looks correct to me. It's a little difficult to comment on the questions you have comparing the paper to your plot though😅 Even equatorial coordinates can be hard to interpret/understand sometimes, so galactic coordinates are even more difficult!

I have some good news! I've just been testing the new rewrite of the software and compared it to the current software. I have managed to get a pretty consistent performance increase, which I can imagine will only be even better on slower machines (raspberry pi ant etc).

I'm still testing the rewrite until I merge the rewrite branch with the main branch. If you want to, however, you can try the rewrite if you pull/clone from the rewrite branch. I would advise you to do so in a separate folder, so you don't overwrite your current files.

To clone the rewrite branch:

git clone https://github.com/byggemandboesen/H-line-software -b rewrite

There will most likely be bugs, hence why it's not on main branch yet, but feel free to test it and take note of any bugs that may occur. Then I will fix them

Victor

Hi Victor,

Very good.  Are the two versions compatible?  I would certainly like the performance improvement and I haven't modified the host version on the RPi, but I have modified the version on the client - mainly to to save the .json file.  I'll have a look at my mods.  If you've included them all, there's no reason I shouldn't upgrade the whole thing.  In fact it would be much better. I always run in debug mode and save the .json file.  As far as I'm concerned, it's the main output.  The .png will be more interesting when I start plotting the arms of the Milkey Way.

Kind regards

Steve.

10 minutes ago, SteveBz said:

And here is my attempt by Galactic Coords.  Hmmm.... The colours look good, but the rest looks a bit scary 😅

The Gaia plot on the ESA website looks a little nicer 😄

Steve.

Questions:

According to the paper attached.

1) At l = 125 degrees the radial velocity seems to be a minimum.  That looks right.

2) At l = 0  degrees, the Galactic centre, the radial velocity should be zero.  Not shown.

3) At l = 200  degrees, the radial velocity should be large and away from us.  It's large and towards us.

4) At l = 50  degrees, the radial velocity should be large and towards us.  It's large and away from us.

Again, I think maybe I have the sign the wrong way round.

I think I might also play with the plot a bit to make it easier to read.

What do you both think?

Kind regards

Steve.

Mapping_the_Spiral_Structure_of_the_Milk.pdf

And here is my attempt by Galactic Coords.  Hmmm.... The colours look good, but the rest looks a bit scary 😅

The Gaia plot on the ESA website looks a little nicer 😄

Steve.

56 minutes ago, Victor Boesen said:

Thanks Steve, your testing is very valuable to me!

You're welcome.  I enjoy it and your software is very useful.