Re: [casper] heliospectroscope
Hi Tavi, It would be best to use an SDR or a board like Redpitaya or SNAP that can sample the entire band, but if you have got an RTLSDR, I can suggest a small trick that might work (no harm in trying). GQRX might be an overkill as you don't need the demodulation part. I have written a small python code that uses pyrtlsdr to read some samples and FFT them. The centre frequency is changed and the process is repeated. The spectra are then stitched together. The code snippet is not the best and the logic has got some bugs. Nonetheless, you can start from that and develop. It is available at [ https://github.com/VU3VWB/rtlsdr_spec_analyser | https://github.com/VU3VWB/rtlsdr_spec_analyser ] ; try rtl_spec_analyser_fft_scan_3.py NB: This is my first time replying to a mailing list, please let me know if there is some silly mistake (reply, reply to all etc.) Regards, Jishnu From: "Tavi B" To: casper@lists.berkeley.edu Sent: Thursday, 14 November, 2019 08:17:31 Subject: Re: [casper] heliospectroscope Hello, Dan, Thank you very much for your detailed answer. I will read the details of suggested boards. So far I've try RTL-SDR, Airspy2+Spywerter and bladeRF x40+XB200 with gqrx. But the gqrx cannot sweep and record full bandwidth at needed sample rate. The main concern is the time of stabilization of the frequency tuned, that's why a large simultaneous bandwidth is desired. The bladeRF 2.0 is faster and have a large bandwidth (56MHz) and have affordable price. It is an integrated radio with TX and filters but we would not pay for features that we don't need. RedPitaya sounds better. Best regards, Octavian marți, 12 noiembrie 2019, 16:13:51 UTC+2, danseti a scris: hi octavian, i don't know much about solar radio spectroscopy, but i think in some solar applications, the SNR is very high, and the time variability is slow enough that you could use a spectrometer with a small instantaneous bandwidth, perhaps 10 or 20 MHz bandwidth, and then sweep this spectrometer across your 800 MHz band. if that's the case, the spectrometer could be very inexpensive because you could use a GNUradio system. if you need more instantaneous bandwidth, the red pitaya fpga board has two 125 Msps ADC's, and casper has a spectrometer tutorial for this board. you might be able to use the spectrometer tutorial design directly, or modify the design for your application. the red pitaya boards cost a few hundred dollars. [ https://www.redpitaya.com/f130/STEMlab-board | https://www.redpitaya.com/f130/STEMlab-board ] the next step up would be to use a snap fpga board, which costs about $3000, and has four 950 Msps ADC's. casper has a spectrometer tutorial for the snap board as well. for even higher instantaneous bandwidth, there are several casper ADC boards you could plug into the snap board. eg: single 2 Gsps ADC board, a dual 2.5 Gsps ADC board information on the snap board, adc boards and tutorials are on the casper wiki pages. best wishes, dan Dan Werthimer Marilyn and Watson Alberts Chair Astronomy Dept and Space Sciences Lab University of California, Berkeley On Tue, Nov 12, 2019 at 1:27 AM Tavi B < [ javascript-blocked: | tav...@gmail.com ] > wrote: BQ_BEGIN Hello, I'm new in this research, in Romania there is no scientific radio telescope yet. I'm working at the Solar Group of Astronomical Institute of Romanian Academy and this is the reason I would start with a solar radio spectroscope, CALLISTO like station. I searched an inexpensive SDR to emulate the callisto analog receiver but with fairly large bandwidth and speed. The high end ones are too expensive and have many features that we don't need like DAC, TX, filters and so on. I've learned that a 12bit ADC with high Mbsp and a FPGA can do the job, we don't need to decode or demodulate a signal, just record the noise level coming from the Sun (or other radio astronomy objects). In this search I've found your group and I wonder if you can help me to find a really cheap solution because we don't have a budget for this project right now. What modules I can buy or build myself (I have basic skills on electronics, microcontrollers and radio) to get a spectrogram of 400 MHz wide with a better than 0.1s time resolution? The Sun radio burst can be received between 10-1500MHz but with different antennas, so I would start with a log periodic dipole wide band antenna, from 120 to 800MHz. Thank you and best regards, Octavian Blagoi researcher AIRA [ http://www.astro.ro/ | www.astro.ro ] -- You received this message because you are subscribed to the Google Groups " [ javascript-blocked: | cas...@lists.berkeley.edu ] " group. To unsubscribe from this group and stop receiving emails from it, send an email to [ javascript-blocked: | cas...@lists.berkeley.edu ] . To view this discussion on the web visit [ https://groups.google.com/a/lists.berkeley.edu/d/msgid/casper/68ef8ced-
Re: [casper] Solar Spectrometer Channeliser
Hi Mugundhan, With red pitaya, can you sample 20-80 MHz without downconversion ? And if analog down conversion is required, wouldn't that restrict the usable dynamic range (unless a high level mixer is used) ? Jishnu From: "Mugundhan vijayaraghavan" To: "casper" Sent: Thursday, February 6, 2020 1:06:09 AM Subject: Re: [casper] Solar Spectrometer Channeliser Hi Dan, We have done integration times from 20ms upto a second. There are bursts that last for hours, minutes and a few ms to seconds as well. We have not tried AGC on these bursts since we were aiming to study them only. In the 8 bit design, we had 3 bits for background and remaining 5 to accommodate bursts. The red pitaya can still be used if Colm can restrict the band to 20-80 mhz, because the ionosphere starts cutting of anything below 20mhz mostly(depending on the location of course). Then this can be down converted to fit into 0-62.5 mhz base band of the red pitaya. Thanks, Mugundhan On Wed, 5 Feb 2020, 22:23 Dan Werthimer, < [ mailto:d...@ssl.berkeley.edu | d...@ssl.berkeley.edu ] > wrote: hi mugundhan, what's the time scale for these bursts? rise and fall times? can you use a AGC circuit (automatic gain control), eg: computer controlled attenuator to turn down the power going into the ADC during the bursts, so you could keep the levels going into the ADC relatively constant? if the rise and fall times are longer than 1ms (the integration time of the spectrometer), then you could adjust the power level for each spectrum, and write down where you set the attenuator for that spectrum, so you could still know the absolute power. if not, there are some 14 bit 200 Msps ADC boards, and i think the new RFSOC boards/chips have 14 bit ADC's, but you'll have to write a casper yellow interface block for this ADC, as we don't have a 14 bit 200 Msps ADC in the casper library. another possiblity is to multiplex your 80 MHz band, 40 MHz at a time into a red pitaya board, ping ponging back and forth between bands: 0 to 40 MHz for 1 ms, then 40 to 80 MHz for the next ms. best wishes, dan On Wed, Feb 5, 2020 at 8:33 AM Mugundhan vijayaraghavan < [ mailto:v.vaishnav151...@gmail.com | v.vaishnav151...@gmail.com ] > wrote: BQ_BEGIN Hi Dan, Usually quiet sun doesn't show such abrupt changes, but bursts do (easily 40-50dB or more) for bright bursts. We have built 8 bit spectrometers in 40-80Mhz, but have found then when the burst is pretty strong, saturation effects starts kicking in. Thanks, Mugundhan On Wed, 5 Feb 2020, 21:52 Dan Werthimer, < [ mailto:d...@ssl.berkeley.edu | d...@ssl.berkeley.edu ] > wrote: BQ_BEGIN hi colm, regarding dynamic range - for your solar spectrometer, do you need 14 bits of ADC dynamic range? it's very unusual in radio astronomy to need that much instantaneous dynamic range on the input. does the sun vary on short time scales in the radio band by factor of 1000 in voltage (1,000,000 in power) ? or do you have very strong bursting RFI that is 1000 times the average noise voltage (1M in power) in the whole band? as you probably know, you'll have lots more dynamic range in the output power spectrum than the dynamic range of the ADC: if you are building a 1024 channel spectrometer with 1 ms integration, you'll get about 8 bits more bits of dynamic range above your ADC dynamic range in frequency domain voltage, which is 16 bits more of dynamic range above your ADC dynamic range in power spectra. so you'll have about 20 bits of spectral dynamic range if you use an 8 bit ADC, (power spectrum dynamic range of about 1 million in 1 ms with an 8 bit ADC, setting noise at 3 bit RMS). and 24 bits of spectrral dynamic range for a 10 bit ADC, 28 bits for 12 bit ADC, and 32 bits for for 14 bit ADC). regarding boards for your spectrometer --- 1) as adam pointed out, the red pitaya is cheap, but sample rate and bandwidth don't quite get the specs you need. 2) another possibility is to use a snap board, which costs more, but can sample 3 inputs at 950 Msps, or 6 inputs at 500 Msps, or 12 inputs at 250 Msps with 8 bit ADC's. most people populate the snap board with 8 bit ADCs, but a few people have populated it with 12 bit ADC's, although the sample rate goes down by 8/12. 3) another possibility is to use a xilinx RFSOC board. the first gen has a bank of 12 bit ADC's (8 inputs at 4 Gsps, or 16 inputs at 2 Gsps), but i think the new generation has 14 bit ADC's ? the RFSOC boards cost more than snaps, but RFSOC was designed in dublin, so you can probably get one from xilinx dublin the ZCU111 board has not been fully casperized yet though. best wishes, dan Dan Werthimer Marilyn and Watson Alberts Chair Astronomy Dept and Space Sciences Lab University of California, Berkeley On Wed, Feb 5, 2020 at 5:06 AM Colm Bracken < [ mailto:co
Re: [casper] Solar Spectrometer Channeliser
Hi Hari, Good point regarding RFI below 25 MHz, though in coming years more and more AM stations may go off air ! It may be hard to find high quality, inexpensive mixers for this application having a very high dynamic range (>60dB). Perhaps Dan's suggestion of using quadrature sampling is a more cost-effective way to implement this, provided red pitaya supports that mode. Jishnu From: "Hariharan Krishnan" To: "casper" Sent: Monday, February 10, 2020 4:33:32 PM Subject: Re: [casper] Solar Spectrometer Channeliser Hi Jishnu, Down-conversion is inevitable with red pitaya given the max. sampling frequency is 125 MSPS for the on-board ADC (for the 20 - 80 MHz RF band here). The dynamic range of the mixer used in the down-conversion becomes rather important here primarily because of the RFI channels (AM transmission below 25 MHz) that will introduce IMD products. Otherwise I wouldn't expect the dynamic range for a relatively clean band to go down at single-stage down-conversion. Regards, Hari On Mon, Feb 10, 2020 at 1:00 AM Jishnu Nambissan T < [ mailto:jis...@rri.res.in | jis...@rri.res.in ] > wrote: Hi Mugundhan, With red pitaya, can you sample 20-80 MHz without downconversion ? And if analog down conversion is required, wouldn't that restrict the usable dynamic range (unless a high level mixer is used) ? Jishnu From: "Mugundhan vijayaraghavan" < [ mailto:v.vaishnav151...@gmail.com | v.vaishnav151...@gmail.com ] > To: "casper" < [ mailto:casper@lists.berkeley.edu | casper@lists.berkeley.edu ] > Sent: Thursday, February 6, 2020 1:06:09 AM Subject: Re: [casper] Solar Spectrometer Channeliser Hi Dan, We have done integration times from 20ms upto a second. There are bursts that last for hours, minutes and a few ms to seconds as well. We have not tried AGC on these bursts since we were aiming to study them only. In the 8 bit design, we had 3 bits for background and remaining 5 to accommodate bursts. The red pitaya can still be used if Colm can restrict the band to 20-80 mhz, because the ionosphere starts cutting of anything below 20mhz mostly(depending on the location of course). Then this can be down converted to fit into 0-62.5 mhz base band of the red pitaya. Thanks, Mugundhan On Wed, 5 Feb 2020, 22:23 Dan Werthimer, < [ mailto:d...@ssl.berkeley.edu | d...@ssl.berkeley.edu ] > wrote: BQ_BEGIN hi mugundhan, what's the time scale for these bursts? rise and fall times? can you use a AGC circuit (automatic gain control), eg: computer controlled attenuator to turn down the power going into the ADC during the bursts, so you could keep the levels going into the ADC relatively constant? if the rise and fall times are longer than 1ms (the integration time of the spectrometer), then you could adjust the power level for each spectrum, and write down where you set the attenuator for that spectrum, so you could still know the absolute power. if not, there are some 14 bit 200 Msps ADC boards, and i think the new RFSOC boards/chips have 14 bit ADC's, but you'll have to write a casper yellow interface block for this ADC, as we don't have a 14 bit 200 Msps ADC in the casper library. another possiblity is to multiplex your 80 MHz band, 40 MHz at a time into a red pitaya board, ping ponging back and forth between bands: 0 to 40 MHz for 1 ms, then 40 to 80 MHz for the next ms. best wishes, dan On Wed, Feb 5, 2020 at 8:33 AM Mugundhan vijayaraghavan < [ mailto:v.vaishnav151...@gmail.com | v.vaishnav151...@gmail.com ] > wrote: BQ_BEGIN Hi Dan, Usually quiet sun doesn't show such abrupt changes, but bursts do (easily 40-50dB or more) for bright bursts. We have built 8 bit spectrometers in 40-80Mhz, but have found then when the burst is pretty strong, saturation effects starts kicking in. Thanks, Mugundhan On Wed, 5 Feb 2020, 21:52 Dan Werthimer, < [ mailto:d...@ssl.berkeley.edu | d...@ssl.berkeley.edu ] > wrote: BQ_BEGIN hi colm, regarding dynamic range - for your solar spectrometer, do you need 14 bits of ADC dynamic range? it's very unusual in radio astronomy to need that much instantaneous dynamic range on the input. does the sun vary on short time scales in the radio band by factor of 1000 in voltage (1,000,000 in power) ? or do you have very strong bursting RFI that is 1000 times the average noise voltage (1M in power) in the whole band? as you probably know, you'll have lots more dynamic range in the output power spectrum than the dynamic range of the ADC: if you are building a 1024 channel spectrometer with 1 ms integration, you'll get about 8 bits more bits of dynamic range above your ADC dynamic range in frequency domain voltage, which is 16 bits more of dynamic range above your ADC dynamic range in power spectra. so
