Project MK23: Complex object W3(OH) at methanol line CH3OH at 12.178 GHz

Part 1

It is known that methanol is also maser transmitting at 12.178595 GHz.

This frequency lies in the standard Astra LNB range 10.7012.75 GHz. So, I decided to try to detect it. Some specs of the LNB are:


airy interface-w3-12

Fig.1 - Airy interface between horn and LNB

The expected output should be; 12.178595 GHz - 10.6 GHz= 1.578595 GHz, and this is in the range of the RTL sdr dongle.

Small problem is that the LO frequency is not exact or stable. Normally the TV receiver can adjust this because the signals are strong.

Another small problem is that the sky frequency is higher because W3() is approaching us.
I used my earlier W3(OH) measurement as a reference, because these OH clouds have thesame velocity toward us.
I used the ratio from 1.6GHz to 12.1GHz and adjusted the dongle receiver frequency to 1.581 GHz.
So, the central frequency in the receiver is 12.181 GHz

For the software I used cfrad2.exe again; giving me every 5 minutes the average of 100000 FFT spectra.
The first two X5 minutes the dish was off target and this result is used as a reference; lines green and red.
In stead of subtracting the base line as I see some amateurs do, I divide the on target result by the off target result.
This is done in python; all the software is free and can be downloaded from

Because the LNB LO is not stable, the output shifts in frequency over time/ambient temperature, so if you average a lot of data blocks, then the result will be smeared out and disappeares.
The end result here is averaged over 16X5 minutes.


Fig.2 - Overview W3(OH) @12GHz

16X5min average

Fig.3 - Sum of 16x5minutes integration @12GHz

The literature result gives a horizontal flipped over notation

literature w3-12

Fig.4 - W3(OH) @12GHz from literature; high frequency side is on the left

Part 2

Given the advice by Harke and Marcus I purchased an Avenger LNB with pll control.
The frequency stability of this LNB is much better, however there still is an amplitude variation somewhere.
Also I fully mounted the original horn; no airy interface anymore.

no airy interface-w3-12

Fig.5 - Full horn

The resulting edges are much steeper because frequency jittering is less.

steeper edges

Fig.6- Steep edges

Now, back in the nl, we have analysed the measurements some more.
In portugal, we made a run with a bandwidth of 1.024MHz, so we could expect more frequency and km/s resolution.
Funny to catch a wave, swaying around by the doppler effect, but pointed exactly to by the calculated sky frequency.
I calculated the sky freq with my own script, but there also are sites like to do that for you.

As I said earlier; still there is this strange amplitude variation and slow frequency shift over time as can be seen in fig 7.

all lines w3-av

Fig.7- All lines

We compensated that again by bin shifting each 5 minute cfrad result as function of time, and only then they were added/averaged.

all lines shift correction

Fig.8- All lines shift corrected

sum zoom

Fig.9- Sum of lines

W3(OH) is a star forming region; this means that a gas disk is rotating around the centre where a star can be born.
Over several years time it has bee noted that the gas masers move outward. This implicates that they are forced to do so, because in principle they move inward to form the star.

This gives the indication that the star has ignited, and expelling solar wind, wich moves the gas ring close by outward, and clearing the nearest parts of the disk by that wind and uv light so the disk becomes a torus.

Still, the star cannot be seen on a visual picture.see

no opt w3

Fig.10- No optical W3(OH)

It even does not have a filled WIKI page; so unpopular are non optical things.

Although the SGRT beam is 0.15 deg, the masers cannot be resolved of course; we are looking at a solar system XXL size.
But that problem the radio astronomers also had in the past.
They tried to separate the peaks by assuming gaussian sources.

Later with VLBI, better 'pictures' could be made, and separate local maser regions (dots) could be seen.

w3oh contour

Fig.11- W3(OH) contour with maser dots

As an amateur association we still lack that luxury, or coorporation to make a VLBI, so I tried to separate the peaks with calculations.
First we tried excel. Below you see how the resolver function can adjust the Gauss curves in frequency, peak hight and peak width, so summed all together gives you the original curve.
The problem is how many sources are there hiding in the result. I tried to estimate that by differentiating the curve twice. In this way all the bends in the curve show as peaks, trying to hide their existance.
In total I counted about 20 gaussians above noise level. This number was taken as a starting parameter for the solution.
Another initial seeding which has to be made is the size of the cloud, or the width of the gaussian.
I took the size of the smallest peak as a standard; reasoning that the turbulance in the disk/torus was thesame everywhere.


Fig.12- Gaussians

So,the gaussian curves in the final solver run represent the individual masers, and when they are added gives you the curve you actually received with your dish.

gaussian fit

Fig.13- Gaussian fit excel

We now can compare this to papers in which VLBI astronomers have identified about 50 individual masers.
I omitted all the weak masers from that list and just took the 20 stongest. The result is very satisfactory.
In the picture I have connected some of the gaussians with the W3(OH) locations where they come from.

gaussian fit

Fig.14- Torus projection and the gaussian sources

You can deduct that the torus is rotating counter clockwise.
Accurate VLBI measurement shows that at the left part of the torus no masing can be seen.
So you can also deduct that we must be looking at the bottom side of the counter clockwise rotating torus.
It is more clear for me now why the front side of the torus does not emit maser signals.
Masing is obscured/attenuated by the thick part of the ring towards us.
So the masing is coming mainly from the inner part of the ring illuminated by the star.

To make things more complicated; there also seems to be a 'champagne' flow of gas from the poles.
Where does that material go; does it return in a high ecliptical orbit. Can we see that also in our solar system.

Some FAQ's;

As I said in other projects; each line is the result of 5 minutes of integration (100000 spectra averaged) with an efficiency depending on sample rate.

While measuring I moved the dish up and down, left/right increments of 0.01 degree to find the maximum signal.
I do that by monitoring one of the ten sub bands of the 5 minute FFT result. In the stacked graph fig 7 and 8, you can see that the amplitude of the source changes because of my optimal pointing trials.
The dongle is set to the AGC mode.
That also means that when the monitored track goes up, the others go down.

The measurement stopped, simply, because there were a sufficient number of 5 minute lines/results, so I gave the system instruction to go to the park position and put the clamps on, but I did not stop the capturing.

In the zip files I added all the referenced papers here in a zip file.

Michiel Klaassen november 2016