dspira-lessons/forum/dsp/ #4
Comments
|
Discuss DSP |
|
Hi dr. Langstone and good Sunday, I have a question for you. First of all I'm not a gnuradio expert. In the "spectrometer_w_cal" grc schema, the vector length is set to 4096 samples. The first gnuradio "Integrate block" has a decimation factor of 16. At output we get 4096 real samples, at input we have 4096 real samples. But if we decimate with value 16 we should get 256 output samples. Only if we have 16384 samples at the input do I get 4096 samples at the output with a decimation factor of 16. I can't understand. |
|
Sorry dr.Langstone, ...4 blocks of 16384 samples, 65536 total sample to get 4096 samples using 16 value of decimation. |
|
Dear Gianni,
Thanks for your interest in Radio Astronomy and our Milky Way Galaxy.
You’re asking a software design question. Now I did not write the spectromter_w_cal.grc program,
but have some guesses as to why they wrote it that way.
Remember that with these programs, we’re getting 10 Million I+Q samples a second,
so computer efficiency is important, to eat all that data every second.
They decided to gather up blocks of 4096 samples and process these blocks in
parallel, which works well if you have a computer with many CPUs. (I use Raspberry PI computer,
which have 4 CPUs).
The code is designed to gather 4 blocks of data and send them to the computationally expensive FFT block.
I think this was done to make computers with 4 CPUs (called cores) use these cores more efficiently.
Now I use the examples in “nsf” directory, including NsfIntegrate60, NsfDetect60 and NsfWatch60, with an Airspy mini.
See the documentation for the other versions of this code.
This code does not decimate but just has one data path for the FFT. This code can eat all 6 Million I+Q samples/second
with a Raspberry Pi, when using an Airspy Mini, but drops some samples when running at 10 Million I+Q samples/second.
So, to summarize, I’m guessing it was for computer efficiency that they wrote the program that way.
Glen
From: GiaCet ***@***.***>
Sent: Sunday, March 5, 2023 9:40 AM
To: WVURAIL/dspira-lessons ***@***.***>
Cc: Subscribed ***@***.***>
Subject: [EXTERNAL] - Re: [WVURAIL/dspira-lessons] dspira-lessons/forum/dsp/ (#4)
This email originated from outside of the National Science Foundation. Do not click links or open attachments unless you recognize the sender and know the content is safe.
Hi dr. Langstone and good Sunday, I have a question for you. First of all I'm not a gnuradio expert. In the "spectrometer_w_cal" grc schema, the vector length is set to 4096 samples. The first gnuradio "Integrate block" has a decimation factor of 16. At output we get 4096 real samples, at input we have 4096 real samples. But if we decimate with value 16 we should get 256 output samples. Only if we have 16384 samples at the input do I get 4096 samples at the output with a decimation factor of 16. I can't understand.
Thanks a lot
Best regards
Gianni
—
Reply to this email directly, view it on GitHub<#4 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AADRXGGAKJY3JB527UZ2XBDW2SQSXANCNFSM4T36MOMA>.
You are receiving this because you are subscribed to this thread.Message ID: ***@***.******@***.***>>
|
|
Hello Glen Kind regards |
|
Maybe a stupid question (and maybe not the right forum), but what are the units for the y-axis (signal) in the "Simple Spectrometer" or even the uncalibrated spectrum of the "spectrometer_w_cal"? Thank you, |
|
For the uncalibrated spectrum, the units are an arbitrary 'power' received by the telescope. After doing the calibration procedure, then those units become a temperature in Kelvin. |
|
Hello Aficionados!
We’ve just released a marvelous summary of how to build a “Pail of Milky Way”
horn radio telescope. This is described in LightWork memo 32.
https://github.com/WVURAIL/lightwork/blob/master/memos/LightWorkMemo032-PailOfMilkyWay-r5.pdf
The memo concludes with a plot of one day’s observations, showing
what you can see with a horn radio telescope. The Doppler shifted
Cygnus and Perseus Spiral arms of the Milky Way are visible.
Comments and improvements welcomed.
Data are all obtained using Gnuradio designs available from the DSPIRA web site.
https://github.com/WVURAIL/gr-radio_astro
Thanks to Kevin Bandura at WVU and the Green Bank Observatory Staff.
Cheers
Glen
|
|
Hi Isabella,
You’ve asked a good question. The answer is that all computer measurements
are actually digital “counts”, based on adding enough amplifier gain (hardware) to get
the voltages to be compatible with the software defined radio (SDR) device you’re using.
The counts are converted to physical units by comparing the counts of the sky
with the counts of an object of known temperature, usually the ground. Measure the
calibration by recording the counts when pointing your telescope straight down.
Then we multiple the Kelvins/counts factor we measured by the observations of the Sky to get the Sky brightness, in Kelvins. (This is done in spectrometer_w_cal).
Good luck with your observations.
Glen
This assumes the amplifier gains don’t change between measuring the ground and the sky.
… On May 1, 2023, at 8:53 PM, isabellad23 ***@***.***> wrote:
Maybe a stupid question (and maybe not the right forum), but what are the units for the y-axis (signal) in the "Simple Spectrometer" or even the uncalibrated spectrum of the "spectrometer_w_cal"?
Thank you,
Isabella
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>
|
|
Hi dr. Langston |
|
Hello, |
|
Dear Gianni,
Thanks for your emailed question. Sorry I did not notice this before.
Your setup should work, but you’ll need a somewhat more powerful computer to take all that
data, from two AirSpy R2 devices at 10 MHz sample rate (I+Q).
There will likely be some delays internal to the gnuradio buffers. So you might need a
timing pulse to synchronize the data samples, or expect some offset between the two data
streams.
You could check this by spitting a single input single into the two Airspys and seeing
what size delays between samples gnuradio introduces.
John Makous, who contributes to DSPIRA is a real expert on interferometry. Maybe he has
comments?
Best regards,
Glen
PS FYI, the gnuradio software system puts the samples in largish buffers before
handing the data to the next software threads. You might be able to tag the
data streams.
John
… On Sep 4, 2023, at 1:47 PM, GiaCet ***@***.***> wrote:
Hi dr. Langston
I have two airspy-R2 (10MHz) and an external clock generator (10MHz clock). Then I have two LNA prof.Kevin Bandura design, two VBF-1445+ band pass filters Minicircuits, and two inline amplifier etc etc. So, I would be to make a radio-interferometer using gnu-radio for produce two signal source. Then using a mixed schema from "simple_spectrometer_w_cal" before FFT, and "interferometer_simplespectrometer Multiplying" after FFT. Could be ok?
In this moment I cannot to buy limeSDR.
Best regards
Gianni
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you commented.Message ID: ***@***.***>
|
|
Hi Steve,
I’m finding John Makous’s report on line.
Please check this link: https://wvurail.org/lightwork/memos/LightWorkMemo031.pdf
Best regards
Glen
Unfortunately I’ve not tried the other multi-port DSPs. Kevin Bandura reported
planing to test the “ANT SDR”. I’m not sure how that’s going.
… On Oct 25, 2023, at 2:37 PM, 19smuh23 ***@***.***> wrote:
Hello,
I am new to radio astronomy and find the DSPIRA project very helpful. My questions: Clicking on the lesson modules for interferometry result in a 404 error message (can't find page). Can you repost? Also, has anyone tried using other multiple port SDRs in addition to Lime (such as 2-port RSPduo or 5-port KrakenSDR) with GnuRadio?
Thanks,
Steve
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you commented.Message ID: ***@***.***>
|
|
I had a general question for anyone familiar with the programming used in the spectrometer_w_cal.grc program. Thanks, |
|
Hi Brian,
Thanks for your experimenting with DSPRIA code. You've made
a good suggestion. I've tried this a bit. But was not happy with the
results. The sound was not great. If doing it again, it might be
better to think more musically.
The key, of course, is to down convert the 1.420.4 +/- 3 MHz signal into
the audio range 200 to 1000 Hz. This might be doable by creating a set of a few
sine waves at good intervals, say maybe 8 piano keys, and mapping the
intensities in the spectra to different frequencies. Maybe higher notes
for higher doppler shifts towards us and lower notes for doppler shifts away.
This would not be an exact map, but I think it would be more satisfying.
Good Luck!
Glen
PS there are, of course, excellent Audio out blocks in GnuRadio.
Sum up all the sine waves and push to audio.
…________________________________
From: Herbhuskr ***@***.***>
Sent: Thursday, June 20, 2024 1:33 PM
To: WVURAIL/dspira-lessons ***@***.***>
Cc: Langston, Glen ***@***.***>; Comment ***@***.***>
Subject: [EXTERNAL] - Re: [WVURAIL/dspira-lessons] dspira-lessons/forum/dsp/ (#4)
This email originated from outside of the National Science Foundation. Do not click links or open attachments unless you recognize the sender and know the content is safe.
I had a general question for anyone familiar with the programming used in the spectrometer_w_cal.grc program.
Since it utilizes a SDR module typically used to output sound modulated on a given frequency, would it be relatively easy to add a toggle button, or the like, to direct/output sound to the PC speaker for the frequency selected within the program? It would probably just be static, but if audible maybe some louder static around the hydrogen line would be interesting to hear. Maybe if really lucky you could hear another WOW signal around 1420.4556 😉
Thanks,
Brian
—
Reply to this email directly, view it on GitHub<#4 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AADRXGF232ZLADKHI2VLWGLZIMG6TAVCNFSM6AAAAABJUMZJSKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCOBRGIYDANRTHE>.
You are receiving this because you commented.
|
|
Glen,
Thanks for getting back so quickly, and your suggestions.
I will think about your suggestions and see how they might be utilized.
I am still on a steep learning curve, but this stuff is really interesting, and I am intrigued with radio astronomy applications.
Thanks again for taking the time to respond!
Best Regards,
Brian
…________________________________
From: Glen Langston ***@***.***>
Sent: Thursday, June 20, 2024 5:42 PM
To: WVURAIL/dspira-lessons ***@***.***>
Cc: Herbhuskr ***@***.***>; Comment ***@***.***>
Subject: Re: [WVURAIL/dspira-lessons] dspira-lessons/forum/dsp/ (#4)
Hi Brian,
Thanks for your experimenting with DSPRIA code. You've made
a good suggestion. I've tried this a bit. But was not happy with the
results. The sound was not great. If doing it again, it might be
better to think more musically.
The key, of course, is to down convert the 1.420.4 +/- 3 MHz signal into
the audio range 200 to 1000 Hz. This might be doable by creating a set of a few
sine waves at good intervals, say maybe 8 piano keys, and mapping the
intensities in the spectra to different frequencies. Maybe higher notes
for higher doppler shifts towards us and lower notes for doppler shifts away.
This would not be an exact map, but I think it would be more satisfying.
Good Luck!
Glen
PS there are, of course, excellent Audio out blocks in GnuRadio.
Sum up all the sine waves and push to audio.
________________________________
From: Herbhuskr ***@***.***>
Sent: Thursday, June 20, 2024 1:33 PM
To: WVURAIL/dspira-lessons ***@***.***>
Cc: Langston, Glen ***@***.***>; Comment ***@***.***>
Subject: [EXTERNAL] - Re: [WVURAIL/dspira-lessons] dspira-lessons/forum/dsp/ (#4)
This email originated from outside of the National Science Foundation. Do not click links or open attachments unless you recognize the sender and know the content is safe.
I had a general question for anyone familiar with the programming used in the spectrometer_w_cal.grc program.
Since it utilizes a SDR module typically used to output sound modulated on a given frequency, would it be relatively easy to add a toggle button, or the like, to direct/output sound to the PC speaker for the frequency selected within the program? It would probably just be static, but if audible maybe some louder static around the hydrogen line would be interesting to hear. Maybe if really lucky you could hear another WOW signal around 1420.4556 😉
Thanks,
Brian
—
Reply to this email directly, view it on GitHub<#4 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AADRXGF232ZLADKHI2VLWGLZIMG6TAVCNFSM6AAAAABJUMZJSKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCOBRGIYDANRTHE>.
You are receiving this because you commented.
—
Reply to this email directly, view it on GitHub<#4 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/BJK4JSZXAUHZJJI55NOJOL3ZINEF3AVCNFSM6AAAAABJUMZJSKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCOBRGYYDQMZWGY>.
You are receiving this because you commented.Message ID: ***@***.***>
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Digital Signal Processing Forum – Digital Signal Processing in Radio Astronomy - Lessons Portal
Lots of lessons
https://wvurail.org//dspira-lessons/forum/dsp/
The text was updated successfully, but these errors were encountered: