WizardOfOZ
Posts: 76
Joined: Thu Oct 13, 2011 8:08 am

Re: Software Defined Radio

Wed Dec 14, 2011 6:58 pm

Quote from ai5eedave on December 12, 2011, 23:21
(Lots of good stuff.)
OK, sounds very promising. The Silicon labs synth chips have a good reputation, so should be a great starting point. You will still need some (a lot, really) of gain and selectivity between your quadrature mixer, and the A/D converter. One possible starting point for an inexpensive yet high performance direct conversion receiver would be the design by Rick Cambell, KK7B, published in the late nineties in QST. It is even an I/Q design, though your H-mode mixer with logic quadrature generation will be a superior solution, compared to his passive phase shift network and diode mixers. However, apart form that, you can do worse than copy his low noise audio pre-amp and ditto selectivity circuitry. He also published an earlier article, named "A High Performance Direct Conversion Receiver" (I think...), which has an even better audio diplexer design. I believe ARRL has published copies of both articles in their web library.

Another reason for running a high sampling frequency, even though you may be satisfied with a 4 KHz audio passband, is that you can use commercial A/D converters with built-in oversampling and ditto anti-aliasing filters/noise shaping. Without those you'd need some pretty ridiculous active audio filtering to ensure you *actually* have 16 bit worth of noise free operation from the A/D. I'll dig up some part numbers if you like.

Quote from Danbert on December 13, 2011, 11:46
WizardOfOz: Does it have to be that way with SDR, though?
Yes, it absolutely does have to be that way with SRD.

You'd need at the very least 120dB as a low ballpark value for the required dynamic range of the whole receiver front end *and* the A/D converter. This is assuming you want a general purpose receiver, which can dig into the local noise level, and that you don't live too close to any strong, local transmitter. So your chosen chip is about 55 dB short, and needs to be 24 bit resolution.

I'd also like to see your proposed solution for the front end amplifier. At least 120dB of dynamic range across 1-250MHz, with a maximum output voltage equal to the full scale input of your chosen A/D converter. Looks like you will only need about 30dB of gain if the FS input voltage to the A/D is 3.3V, and assuming its input impedance is 50 ohm. This assumes an input resolution of 0.1 uV, which will be way below your local noise level across the chosen band.

Finally, FFTs are boring. We want real time audio playback, please. So you need to come up with a FIR filter, which can carve out a 250 Hz bandwidth among the 250MHz input, a 1:1'000'000 reduction in bandwidth. (250 Hz resolution is rather easy to achieve with a modestly priced conventional receiver.) This needs to happen to a sample set of 24 bit resolution, entering your system at 500 MHz. Good Luck With That(TM).

Bottom line: This idea is not exactly new, it gets flogged regularly by people, who have apparently not bothered to check on the most basic requirements. Plugging The World(TM) into a high speed A/D is just a silly idea, which would require some impressive hardware. Unless, that is, you actually intend to build a professional radio telescope, and have the computational resources to match. (And they are not even trying for real time playback...)

User avatar
ai5eedave
Posts: 8
Joined: Tue Nov 22, 2011 11:48 pm

Re: Software Defined Radio

Wed Dec 14, 2011 9:45 pm

I built one of Rick Campbell’s R2s back in the late 90’s when I was in college. With the lab equipment I was able to apply the quadrature local oscillator (LO) signals and inject a test radio frequency (RF) signal, and I heard the single side band (SSB) output just fine. I never built a stand-alone quadrature LO for it, so the R2 board now lives in a box somewhere in my garage.

According to the SoftRock web page, “The approximate tuning bandwidth you can get, given a specified center frequency (CF) from the hardware's local oscillator and the soundcard's sampling rate (SR) is a range: [CF - SR/2] to [CF + SR/2]. “ So, given a 48 kHz sample rate, I should get about 48 kHz of bandwidth to utilize in the SDR software. My understanding is that you do the coarse tuning using the hardware LO (you choose the chunk of the spectrum you want to listen to), and then do the fine tuning within that chunk using the SDR software. This would indicate that you wouldn’t want a 4kHz low pass filter, but rather a 24 kHz filter. And that’s if you need filter at all since the audio amplifiers are going to do some filtering as a result of their gain-bandwidth product.

As for the gain, YU1LM’s boards have gain select of either +26 dB or +46 dB. I am not sure if his design is expecting to have some additional gain in the sound card interface, or if the SDR software is capable of dealing with a low amplitude digital input. One would assume that the best thing would be to amplify the audio signal to the point that the maximum signal amplitude is equal to the maximum voltage range of the ADC. But to do that, you would either have to be really fast on the volume control (human gain control), or have an automatic gain control circuit prior to the ADC. Since I don’t see AGC in most of the SDR radios, I think I can safely assume that it is not really necessary and the digital processor handles this.

Again, back to philosophy, I envision a board about the same size as the RPi with all (or nearly all) surface mount components, very few if any inductors or transformers, running off power supplied to by the RPi, with an SMA connector at one end and a pin header at the other connecting to the RPi board. For an RX only, decently performing system, this *should* be possible. In the future, I may design a small low power SSB modulator and power amplifier to use with this, but that’s probably a ways down the road. Perhaps a good intermediate step would be a 5 watt class E amplifier to make a small CW transmitter. I wonder if any of the NorCal guys would be interested in building a kit for any of this? I know in the recent past there was a contest to build a low cost (<$50) transceiver. A $25 SDR front end might just make that possible. (By the way, the radio that won the <$50 contest actually cost less than $50 for the parts. A kit company agreed to sell the radio as a kit, but they decided to charge $100 for it, which really defeated the whole $50 concept.)

jaha93
Posts: 1
Joined: Wed Dec 14, 2011 10:43 pm

Re: Software Defined Radio

Wed Dec 14, 2011 11:20 pm

I'm affraid I have to agree with those who say the R-pi isn't the right tool for the job. I have done some research into this myself and to my mind there are two routes to go down depending on the interface you expect to be presented with:

If you want something that behaves like a traditional radio with relatively simple controls then I would recommend taking a look at the SDX (Software Defined transponder) design that is part of the arissat1 satellite (all of the design files are available as an open source project). This is based on a Microchip PIC with a QSD and QSE. Due to the nature of the original design it would take a bit of adapting to become a useful HF transceiver, but that work would be mostly on the RF side which I am sure you are all capable of (last time I checked the whole point of having an amateur license was to learn about RF). For those interested in pursuing this option, Microchip have since released a low pin count PIC32 microcontroller which would greatly simplify PCB layouts for a homebrew solution.

If on the other hand you want something with a decent sized screen and the guts to show live waterfall type displays etc then you are realistically looking at a point where your best bet is something like a softrock kit (for the receive side, not sure what to do for TX) and a Laptop. By the time you have a big enough screen for a sensible user interface the laptop looks attractive.

Just my thoughts,

James M0GUR


User avatar
ai5eedave
Posts: 8
Joined: Tue Nov 22, 2011 11:48 pm

Re: Software Defined Radio

Wed Dec 21, 2011 10:37 pm

Again, I am really shooting for the $50 price point. A $25 RPi SDR front end goes a long way towards that goal, so I would very much like to design and build the RF and sampling board for around $25. Most of the alternatives suggested so far are closer to the $200 price point.

So, here are some questions for the group:
1. Is a 16 bit ADC sufficient to produce acceptable performance? I know that a 16 bit ADC will provide a signal to noise ratio of ~97 dB (as opposed to a 24 bit ADC of ~146 dB SNR). When I listen to my radio at home, I can pick out a station one S unit (6 dB) above my noise floor, but I am not sure how ADC SNR relates to listening performance or strong signal rejection. Is a 10 or 12 bit ADC sufficient?
2. What tradeoffs are involved in the sample rate selection? Is it more than just the bandwidth of the received audio? I know that a higher sample rate will require more processor time from the RPi, and that there is a limit to how much data can be stuffed down the SPI link.

Any thoughts or suggestions are appreciated, including the ones that say I am crazy for even trying to do this :)

error404
Posts: 351
Joined: Wed Dec 21, 2011 11:49 pm

Re: Software Defined Radio

Thu Dec 22, 2011 12:00 am

Since audio ADCs are probably the only viable way of getting high enough quality I/Q data into the RPi, and it seems like I2S will be available to us - which is usually and seems to be in this case multiplexed with SPI - why not just hook an audio ADC like the WM8786 (192KHz, -111dB SNR @ ~$4 in singles) directly to the RPi and forget the micro? I understand the hardware capabilities of the final device are not really known yet, and the lack of documentation is a potential show-stopper, so perhaps we're shut out before we start, but it seems like the approach to take. Some external clock generation might be required, but that's pretty easy (have done it, can provide schematics). Much cheaper and higher quality than a USB ADC and easier than converting I2S -> SPI, which something like an 8-pit PIC probably doesn't have the bandwidth to keep up with. CPLD might be an option here if the I2S isn't actually an option, rather than a PIC.

I love the idea of doing SDR with RPi, and I don't think it's at all out of the realm of possibility, even at high bandwidths. Access to the DSP would obviously be a huge boon for this kind of usage, but it should be fast enough for enough software DSP to pull off a basic SSB/AM/FM receiver. Certainly worth pursuing, I'd say start with the USB audio route with an existing radio and prove out that it's fast enough for the DSP, optimizing if necessary.

bradburts
Posts: 341
Joined: Sun Oct 02, 2011 7:07 am

Re: Software Defined Radio

Thu Dec 22, 2011 1:02 am

SPI will give you more bandwidth than I2C.
The Pi can do SPI.

error404
Posts: 351
Joined: Wed Dec 21, 2011 11:49 pm

Re: Software Defined Radio

Thu Dec 22, 2011 1:06 am

I2S, not I2C. I2S is a serial audio protocol that pretty much any audio DAC/ADC will implement. It is similar to SPI, but is clocked somewhat differently and uses an extra left/right clock for stereo audio, and generally no chip select.

It is designed specifically for this task, and would obviate the need for a shim between the audio ADC (the only hardware good enough at a reasonable price *) and RPi. If the necessary access/driver is provided, anyway.

* I'm not well-versed enough on the DSP and SDR to really comment on the bit depth requirement for this, but it seems from other designs and the discussion upthread that the viable 10-12bits isn't good enough. 14-20 bit ADCs not specifically designed for audio are horrifically expensive.

bradburts
Posts: 341
Joined: Sun Oct 02, 2011 7:07 am

Re: Software Defined Radio

Thu Dec 22, 2011 1:34 am

My bad. I learn something new :)

n9vv
Posts: 1
Joined: Tue Dec 27, 2011 11:16 pm
Contact: Website

Re: Software Defined Radio

Tue Dec 27, 2011 11:35 pm

I am very interested in the concept of using a microprocessor (beagleboard, Raspberry? Panda? ....) to break the tether to the desktop PC. I have been told

that it will take a PIC32 and DSP32 to deliver enough horsepower

to meet the needs of a modern DDC receiver (like the Pegasus, RFSpace SDR-IQ, NetSDR, or even the new Hermes from the OpenHPSR.org group). But perhaps

the RaspberryPi can be used with an LCD panel and encoder to power the front

panel of a standalone DDC SDR receiver?

Here is a list I compiled of 47 INexpensive SDR Receivers and kits. Perhaps

one will meet the needs of the Raspberry pilots?

http://www.n9vv.com/collected-.....-kits.html

thank you for allowing me to participate in this extremely interesting discussion.

73 de Ken N9VV

wf2v
Posts: 1
Joined: Sat Jan 07, 2012 2:02 am

Re: Software Defined Radio

Sat Jan 07, 2012 2:45 am

I will have fun with the funcube dongle sdr with the raspberry.

Same here! I have the FCD and this would make a supurb low power portable sdr receiver. Lots of potential experiments! Lets put one in a microsat!!!

kj6dzb
Posts: 11
Joined: Thu Jan 05, 2012 12:01 am
Contact: Website

Re: Software Defined Radio

Sun Jan 08, 2012 11:13 pm

Vita and  VITA-49 provide transport layer for the CPU to crunch provide a DDC/DUC feed a reasonable section of RF band with. I and others have come to a point were 192khz doesn't cut it for were we see SDR's developing to.

http://wdv.com/Electronics/Ref.....B-VITA.pdf

Alot of people find this thing interesting for SDR use. Starting a Group for all this chatter should shape up mid week ive got to work Monday & Tuesday.

73 Mathison KJ6DZB

W9RAN
Posts: 33
Joined: Wed Feb 08, 2012 10:39 pm

Re: Software Defined Radio

Wed Feb 08, 2012 11:07 pm

Just joined the forum, and my first search was to see what folks were thinking about with respect to using the R-Pi for SDR.    Looking thru this thread I was surprised that Tony Parks SDR stuff hasn't been mentioned, he's kind of the guru of cheap and powerful SDR toys: http://kb9yig.com/

Like R-Pi, kits are sold out quickly but there's a very active Softrock group on Yahoo:

http://groups.yahoo.com/group/.....oftrock40/

The newest designs use the si570 for frequency agility and some include transmit capabiilty as well as receive.

I'm no SDR guru but I've read that guys are using Softrocks under Linux running apps such as Quisk and GHPSDR3.   I don't know if the R-Pi has the horsepower needed but I guess we'll find out!

Bob W9RAN

error404
Posts: 351
Joined: Wed Dec 21, 2011 11:49 pm

Re: Software Defined Radio

Wed Feb 08, 2012 11:46 pm

Well, most ham SDR gear is designed around the same concept – mixing the RF down to baseband and then sampling it with audio hardware. So the exact SDR rig you choose to use doesn't matter much, they all operate in essentially the same fashion. There might be some complications with regard to how you tune them and so on, but other than that it doesn't matter if you're using a standard HF rig, commercial SDR or something like the Softrock.

The key issues here IMO are getting the data in (since Pi doesn't include an audio ADC) at decent quality and bandwidth, and whether it will be fast enough to provide a good experience. If I manage to get one this is one of the first things I'll be trying out, though I only have gear for 48KHz operation at the moment.

USB2SDR and SDR-Widget are some nice ways to get high quality data in, but they're also pretty expensive.

rjtucke
Posts: 3
Joined: Tue Feb 21, 2012 6:24 pm

Re: Software Defined Radio

Tue Feb 21, 2012 6:40 pm

Sign me up. I don't have the experience to lead a project like this, but I am _very_ interested in using RPi in ham radio applications. I plan to purchase an Elecraft KX3 in a few months and would love to be able to have them talk. Anyway, if anybody is leading a project that needs a volunteer for ham radio/RPi crossover projects, let me know.

73,

Ross, NS7F

ve3gkn
Posts: 1
Joined: Thu Feb 23, 2012 8:43 pm

Re: Software Defined Radio

Thu Feb 23, 2012 8:53 pm

Great discussion on SDR, although I'm not a programmer, my interest would be to use the R-PI along with my Flex-1500 rig.

Any thoughts along this line?

73, Gord

kj6dzb
Posts: 11
Joined: Thu Jan 05, 2012 12:01 am
Contact: Website

Re: Software Defined Radio

Fri Feb 24, 2012 5:14 am

Im swaying tords using the rasbarry pi as a platform to gauge the prepossessing power of the arm CPUs to do DSP sampling with DAC like the MAX5879 and LM97593 ,  ADC  with the ADC12D1800, or DUC/DDC with the GC6016.  Is there any one in the United states interested in designing test circuits to test this sort of chip, for RF sampling. on the Raspberry Pi or other TI DSPs cpus?

73 kj6dzb Mathison Ott

RAThomas
Posts: 7
Joined: Mon Feb 06, 2012 7:03 pm
Contact: Website

Re: Software Defined Radio

Fri Feb 24, 2012 10:26 pm

Not SDR related, but still ham related: I'm kicking around using the R-Pi in an FRS/GPRS radio hack where I hijack the existing tuner-IC-to-microcontroller serial link (among other signals) to wind up with a programmable radio for at least the 70 cm band.  Granted, applying a microcontroller to the task would be cheaper, but...

bobaruni
Posts: 2
Joined: Sat Feb 18, 2012 1:24 pm

Re: Software Defined Radio

Sun Mar 04, 2012 11:37 am

Well, I for one am very interested in something like this coming to life.

I have built a couple of simple SDR front end circuits that interface to the PC's audio in and am very impressed at the ability hear very very distant radio stations surpassing anything I have ever seen that is super-het or super-regen. It would be nice to not have to cart an entire PC around

I have seen a simple SDR design somewhere using a PIC 33F DSC (40 MIPS fixed point DSP) connected to a codec chip.

Surely even if we didn't get access to the RPi's DSP core, we still have much more than 40 MIPS to play with to decode the I/Q but there's still the problem of getting the ADC interfaced.

Laid
Posts: 2
Joined: Mon Mar 12, 2012 4:51 pm

Re: Software Defined Radio

Mon Mar 12, 2012 5:06 pm

Hi!

As a radio amateur myself and also a SDR designer, I have long waited for something like Raspberry to become available, so I can design a radio "shield" for it. As Raspberry seems to do the same to high-end development range as Arduino did for basic automation for artists and students, it seems the right one to go with.

So, decided to give a little try for crowdfunding and created a project for basic SDR board what would be an inexpensive entry-level radio module for Raspberry.

http://www.rockethub.com/proje.....spberry-pi

The more high-end project I have, the SDR MK1.5 http://uvb-76.net/p/sdr-mk15-andrus.html could benefit from processor upgrade as well, but the professional receiver is likely too complicated and expensive concept for successful crowdfunding, so the one I am creating is more a entry-level.

The architecture will consist of preamp, analog downconverter with programmable clock synthesizer, a good quality 24-bit ADC and some filters and glue logic, and perhaps couple of buttons and rotary encoder to make it truly standalone unit.

The feature requests are more than welcome, but pleas emid the price!

Best Regards,

Andrus.

ai5eedave said:


As a ham radio operator, electrical engineer, and part time amateur radio designer hack, I have been kicking around the idea of a small low frequency (30 kHz to 100ish MHz) sampling device for the front end of a software defined radio (SDR). I envision the front end being similar to the system designed by Sinisa Tasic (YU1LM out of Serbia), except built exclusively with surface mount components to minimize size, and laid out on a board the size of the Raspi. See http://yu1lm.qrpradio.com/SDR&.....-YU1LM.pdf for a detailed description of what the RF front end would likely resemble.

I imagine that a PIC could be used to perform the analog sampling of the down-converted RF, and to convert the samples to I2C or SPI (since there is no analog capability available directly on the board. The PIC would also talk to a programmable crystal oscillator to be used for the local oscillator. A small amount of ham radio SDR software for Linux already exists, and it may be executable directly on the Raspi board. If not, some coding would have to be done, possibly making use of the onboard DSP once the DSP folks let us see behind the curtain.

This seams like a natural thing to do with a board built around a cellphone processor anyway…


jonititan
Posts: 30
Joined: Sat Aug 06, 2011 2:05 pm
Contact: Website

Re: Software Defined Radio

Wed Apr 18, 2012 1:43 pm

Perhaps a USB TV tuner could be used.

A very cheap SDR receiver has been demonstrated using certain of the ezcap range

http://sdr.osmocom.org/trac/wiki/rtl-sdr

zemned
Posts: 2
Joined: Mon Apr 23, 2012 9:12 am

Re: Software Defined Radio

Mon Apr 23, 2012 10:19 am

I have successfully got rtl-sdr running acquisitions on my RPi at 2MSPS and used offline processing to verify I was receiving valid signals by tuning it into radio 1.  I have also been able to port across the gnuradio toolkit to RPi ( a few makefile hacks required).  I'm looking at the best way of performing demanding DSP operations on the RPi at present with an aim to creating a DAB receiver running on the RPi in the longer term.

kj6dzb
Posts: 11
Joined: Thu Jan 05, 2012 12:01 am
Contact: Website

Re: Software Defined Radio

Tue Apr 24, 2012 3:29 am

WOW!!! Looks like you Have your PI....

As for your boment about DSP progressing the GHPSDR3 group at the sdr-widget google group is porting the FTT too openCL and openGL, in order to stream line the CPU load and primary to off load the waterfall to the GPU. This will help when the 3d waterfall is in the QT Radio. http://napan.ca/ghpsdr3/index....../Main_Page

If you would please join the group Alex has the openCL and GL code ready to go, but as that I dont know when I will get the Pi. Is worth the time to be the first on in the group to give the new code a run on the Pi.

73

Mathison kj6dzb

User avatar
ai5eedave
Posts: 8
Joined: Tue Nov 22, 2011 11:48 pm

Re: Software Defined Radio

Tue Apr 24, 2012 10:20 pm

Yep, I am insanely jealous that you already have your Pi... I was up a 1:00am local time on launch day trying to "register my interest", but I didn't get an official confirmation page from Element 14 until about 3:00am after many, many tries. That probably puts me at about 100,000th in line. The longer I have to wait for my Pi to show up, the more anticlimatic it will probably be

I just might have to buy an EzTV666 tuner - It doesn't cover the HF ham bands, but at EUR19.99 (USD$26.ish) you can't beat the price. Maybe I should design a simple upconverter rather than an HF Quadrature Sampling Detector (QSD).

gsh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 1337
Joined: Sat Sep 10, 2011 11:43 am

Re: Software Defined Radio

Thu Apr 26, 2012 7:06 pm

It might be worth waiting for one of my guys to finish the project they're working on right now, which is a USB front end with all the important stuff running on the vector processor.  (i.e. the GPU)...

That way it all happens without having to kill the ARM

Gordon
--
Gordon Hollingworth PhD
Raspberry Pi - Director of Software Engineering

Return to “Other projects”