Sliding window fft imaging

22 views
Skip to first unread message

Mike Otte

unread,
Nov 20, 2025, 11:05:50 AM (23 hours ago) Nov 20
to Society of Amateur Radio Astronomers
Jim Abshier wrote:
"The attached image was produced from the fringe data shown in the plot by taking a running FFT over a width that is about the same as the extent of the fringes. Starting at some point before the fringes, define a window that is about the same extent as the fringes and take an FFT of the data in the window. Save the FFT data as a column vector for the image. Then shift the window to the next point (which could be several samples ahead) and take an FFT of fringe data in that window. This then becomes the next column of the image. Repeat this procedure to a point beyond the fringes and you will have produced an image with the data compressed in the vertical direction by the FFT and imaged in the horizontal direction at roughly the same resolution of your antenna beam. I think of it as a hybrid aperture synthesis imaging technique."

I would like to try this.
But i have a couple questions.
1)  "define a window that is about the same extent as the fringes"  So is this one period of the phase  OR the whole detection period?  See pictures attached.

2)  the sampling here is 2sec , so moving the window could be 60seconds?  Bigger? smaller?

There is stuff online of course for "Sliding Window Discrete Fourier Transform

Thank You
Mike w9ys
OneCycleWindow3C145_M42_3X_AD8302_Figure 2025-11
OneBeamWidthWindow3C145_M42_3X_AD8302_Figure 2025-11

Eduard Mol

unread,
Nov 20, 2025, 11:48:15 AM (23 hours ago) Nov 20
to sara...@googlegroups.com
I think Jim means the extent of the whole fringe “wiggle” so essentially the main beam width. Here is an example of a sliding window FFT on simulated fringes from a 2- element interferometer at 10 metre baseline. 
I used the Acycle programme for the sliding window FFT analysis, time units are in minutes. 


Op do 20 nov 2025 om 17:05 schreef Mike Otte <mike....@gmail.com>
--
--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
---
You received this message because you are subscribed to the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sara-list+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/sara-list/6de32f2a-d5d6-478d-a15a-1de17f4d9aben%40googlegroups.com.

Marcus D. Leech

unread,
Nov 20, 2025, 12:14:48 PM (22 hours ago) Nov 20
to sara...@googlegroups.com
On 2025-11-20 11:47, Eduard Mol wrote:
I think Jim means the extent of the whole fringe “wiggle” so essentially the main beam width. Here is an example of a sliding window FFT on simulated fringes from a 2- element interferometer at 10 metre baseline. 
I used the Acycle programme for the sliding window FFT analysis, time units are in minutes.
Nice simulation!

I think you should be able to use this for multiple spacings as well.   You can convolve the fringes prior to performing the FFT, or convolve the FFT results prior to binning.

That would be a useful simulation as well--perhaps for 10m and 50m baselines.


James Abshier

unread,
Nov 20, 2025, 12:39:22 PM (22 hours ago) Nov 20
to sara...@googlegroups.com

Mike,

The window is sized to cover all the fringes, or essentially the null-to-null width of the antenna beam. You could also make the window somewhat smaller, for example the half power beam width, but this will broaden the frequency response size. The reason that you might want to reduce the window size is that it will reduce the width of the response function in RA. I recently needed to do this to provide better separation between two close sources. But as noted above, decreasing the RA response size (By decreasing the window size) increases the Dec. (frequency) size. (Thus misery is conserved)

Jim

Eduard Mol

unread,
Nov 20, 2025, 12:40:59 PM (22 hours ago) Nov 20
to sara...@googlegroups.com
Hi Marcus, 
I wanted to see what results the FFT would give with different baselines, source declinations and FFT window size. Since I don’t have an interferometer (yet) I wrote a python script for these simulations today. expect a SARA journal article at some point in the near future :-) 
By the way, what do you exactly mean with “convolving the fringes”? Do you mean adding the fringes from different baselines together, like the 10m and 50m you mentioned? 


Op do 20 nov 2025 om 18:14 schreef Marcus D. Leech <patchv...@gmail.com>

Marcus D. Leech

unread,
Nov 20, 2025, 12:45:05 PM (22 hours ago) Nov 20
to sara...@googlegroups.com
On 2025-11-20 12:40, Eduard Mol wrote:
Hi Marcus, 
I wanted to see what results the FFT would give with different baselines, source declinations and FFT window size. Since I don’t have an interferometer (yet) I wrote a python script for these simulations today. expect a SARA journal article at some point in the near future :-) 
By the way, what do you exactly mean with “convolving the fringes”? Do you mean adding the fringes from different baselines together, like the 10m and 50m you mentioned? 


Multiplying, rather than adding, but basically, yes.

If you multiply a whole bunch of fringe sets together, with lots of spacings, what you approach is a single response as if you were using a single dish the same size as the
  largest baseline.


Eduard Mol

unread,
Nov 20, 2025, 12:46:50 PM (22 hours ago) Nov 20
to sara...@googlegroups.com
Okay, so then you’re basically doing aperture synthesis but only in the RA direction if I understand it correctly?

Op do 20 nov 2025 om 18:45 schreef Marcus D. Leech <patchv...@gmail.com>

Marcus D. Leech

unread,
Nov 20, 2025, 12:48:54 PM (22 hours ago) Nov 20
to sara...@googlegroups.com
On 2025-11-20 12:46, Eduard Mol wrote:
Okay, so then you’re basically doing aperture synthesis but only in the RA direction if I understand it correctly?
Yes, basically.



James Abshier

unread,
Nov 20, 2025, 2:10:25 PM (20 hours ago) Nov 20
to sara...@googlegroups.com

I think you need to add (or average) the interferograms to do the 1D aperture synthesis. See example in attached images.

Taurus_A_1D_Synthesis_1.jpg
Taurus_A_1D_Synthesis_2.jpg

Marcus D. Leech

unread,
Nov 20, 2025, 2:45:22 PM (20 hours ago) Nov 20
to sara...@googlegroups.com
On 2025-11-20 13:59, James Abshier wrote:

I think you need to add (or average) the interferograms to do the 1D aperture synthesis. See example in attached images.

Hmmm, interesting.  I clearly remember multiplication from the NRAO introduction to this stuff, but having just run a quick simulation in Gnuplot, adding produces
  better results, particularly if the baselines are not harmonically related.

I shall have to re-read it at some point.

Here is the result of adding/averaging 3 non-harmonically-related sin functions all multiplied by a SINC funciton to simulate the primary beam envelope.


Mike Otte

unread,
Nov 20, 2025, 2:56:24 PM (20 hours ago) Nov 20
to Society of Amateur Radio Astronomers
Thank You  Eduard, James and Marcus,

I have added the 3 in a previous observations and Jame's observation with Tau A and IC 443 is why i choose  Tau A last nite.  To see if they could be differentitated.
I am excited!   Back to exploring.
Have to run up to the observatory and get the TauA data.   

thanks
Mike w9ys

Marcus D. Leech

unread,
Nov 20, 2025, 3:36:18 PM (19 hours ago) Nov 20
to sara...@googlegroups.com
On 2025-11-20 11:47, Eduard Mol wrote:
I think Jim means the extent of the whole fringe “wiggle” so essentially the main beam width. Here is an example of a sliding window FFT on simulated fringes from a 2- element interferometer at 10 metre baseline. 
I used the Acycle programme for the sliding window FFT analysis, time units are in minutes.
The image this produces is, I *think* just a kind of fancy way of gridding 1D fringe data.   It kind of assumes that the object in question is a point source (it really will be for
  an amateur-scale interferometer) AND that the beam response is smoothly symmetric in both RA and DEC.   Basically, taking RA data and rotating it into the DEC axis.
  Assuming, of course, a strictly east-west baseline, and run times that are short enough that earth-rotation doesn't muddy the waters in the baseline orientation too much.

Not an awful approximation for an amateur-scale instrument.

I haven't seen Vanessa's article on this, and I wonder if anyone would be willing to share?


Reply all
Reply to author
Forward
0 new messages