NBFM problem with rtl_fm on Raspberry Pi 2
Hi, i'm experiencing problems demodulating (NB)FM with rtl_fm, when compiling directly on Raspberry Pi 2 with Raspbian: i just get noise! WBFM does work. And i tested many things like calibration, frequency, samplerate .. directly with a LPD (low power device). rtl_tcp works nice with SDR# and i can verify the calibration with it. In the end, i got paranoid and thought of compiler errors. So i tested the same on a Raspberry Pi (1) B+ with same Raspbian image: That works good. No problems! Anyone else experiencing the same problem? What is the solution? kind regards, Hayati
Hi Hayati, On 07/08/2015 01:24 AM, Hayati Ayguen wrote:
Hi,
i'm experiencing problems demodulating (NB)FM with rtl_fm, when compiling directly on Raspberry Pi 2 with Raspbian: i just get noise! WBFM does work. And i tested many things like calibration, frequency, samplerate .. directly with a LPD (low power device). rtl_tcp works nice with SDR# and i can verify the calibration with it.
In the end, i got paranoid and thought of compiler errors. So i tested the same on a Raspberry Pi (1) B+ with same Raspbian image: That works good. No problems!
Anyone else experiencing the same problem? What is the solution?
kind regards, Hayati
Is the same code working OK on x86 machine for you? Regards, Nikolay
Yes, same code works fine on x86_64 (LinuxMint). But i want it to run on Raspberry Pi 2 for 24/7. The Raspi B+ is just borrowed from a friend for testing .. and has less CPU power to do other things in parallel :-( kind regards, Hayati Am 08.07.2015 um 01:23 schrieb Nikolay Dimitrov:
Hi Hayati,
Is the same code working OK on x86 machine for you?
Regards, Nikolay
Hi Hayati, I'm totally not used to development on Raspbian, but: It's really strange that running the same software on the same distro with the same RF device would give different results. So, debugging is on :/ I'd say you should use rtl_sdr to get raw IQ samples, and then have a look at them on your PC. If they are ok, one would have to look a bit deeper into rtl_fm, if not, something goes wrong on the librtlsdr side of things. Best regards, Marcus On 07/08/2015 01:29 AM, Hayati Ayguen wrote:
Yes, same code works fine on x86_64 (LinuxMint). But i want it to run on Raspberry Pi 2 for 24/7. The Raspi B+ is just borrowed from a friend for testing .. and has less CPU power to do other things in parallel :-(
kind regards, Hayati
Am 08.07.2015 um 01:23 schrieb Nikolay Dimitrov:
Hi Hayati,
Is the same code working OK on x86 machine for you?
Regards, Nikolay
Hi Hayati, Could you please mention which commit are you using from here: http://cgit.osmocom.org/rtl-sdr/log/ I am doing experiments with rtlsdr codebase on both RPi and X86 Ubuntu 14.04; I can check the NBFM issue you’ve mentioned. Kind regards, Murat, TA1DB From: osmocom-sdr [mailto:osmocom-sdr-bounces@lists.osmocom.org] On Behalf Of "Hayati Aygün" Sent: Wednesday, July 08, 2015 10:39 AM To: osmocom-sdr@lists.osmocom.org Subject: Aw: Re: NBFM problem with rtl_fm on Raspberry Pi 2 Hi Marcus, librtlsdr is OK, because rtl_tcp works fine, which transfers full I/Q spectrum of ~ 1.4 MHz to SDR# on Windows/PC. The problem must lie in rtl_fm. By the way, the I/Q demodulation "raw" mode is erroneous, too, when having compiled on Raspberry Pi 2. Something else i could try: use the binary compiled on Raspi 1 B+ on the Raspi 2. I'll check this, this evening. Before going for painful debugging, it would be fine if someone else could reproduce the problem .. or say that it works fine. kind regards, Hayati Gesendet: Mittwoch, 08. Juli 2015 um 08:30 Uhr Von: "Marcus Müller" <marcus.mueller@ettus.com> An: osmocom-sdr@lists.osmocom.org Betreff: Re: NBFM problem with rtl_fm on Raspberry Pi 2 Hi Hayati, I'm totally not used to development on Raspbian, but: It's really strange that running the same software on the same distro with the same RF device would give different results. So, debugging is on :/ I'd say you should use rtl_sdr to get raw IQ samples, and then have a look at them on your PC. If they are ok, one would have to look a bit deeper into rtl_fm, if not, something goes wrong on the librtlsdr side of things. Best regards, Marcus
Hello, now tested the binary compiled on RPi 1 on the RPi 2 - after copying the full source including build directory of cmake and executing "sudo make install" on the RPi 2. The result: this works fine! Do i have to conclude, that Raspbian's g++/gcc compiler does produce wrong code when executed on RPi 2 ?! gcc/g++ (Debian 4.6.3-14+rpi1) 4.6.3 OK, now verify: make clean, rebuild and install new binary (from the build directory copied from RPi 1): this works! but, i don't understand why! Now reboot, so that kernel module reloads: still works! In the end, i'm happy that it works "somehow". But, i've no idea where the problem lies! kind regards, Hayati Am 08.07.2015 um 09:38 schrieb "Hayati Aygün":
Hi Marcus,
librtlsdr is OK, because rtl_tcp works fine, which transfers full I/Q spectrum of ~ 1.4 MHz to SDR# on Windows/PC. The problem must lie in rtl_fm. By the way, the I/Q demodulation "raw" mode is erroneous, too, when having compiled on Raspberry Pi 2. Something else i could try: use the binary compiled on Raspi 1 B+ on the Raspi 2. I'll check this, this evening.
Before going for painful debugging, it would be fine if someone else could reproduce the problem .. or say that it works fine.
kind regards, Hayati
*Gesendet:* Mittwoch, 08. Juli 2015 um 08:30 Uhr *Von:* "Marcus Müller" <marcus.mueller@ettus.com> *An:* osmocom-sdr@lists.osmocom.org *Betreff:* Re: NBFM problem with rtl_fm on Raspberry Pi 2 Hi Hayati,
I'm totally not used to development on Raspbian, but: It's really strange that running the same software on the same distro with the same RF device would give different results.
So, debugging is on :/ I'd say you should use rtl_sdr to get raw IQ samples, and then have a look at them on your PC. If they are ok, one would have to look a bit deeper into rtl_fm, if not, something goes wrong on the librtlsdr side of things.
Best regards, Marcus
Hello, there's no NBFM demodulation problem when using Debian Jessie on Raspi2 http://sjoerd.luon.net/posts/2015/02/debian-jessie-on-rpi2/ and compiling everything necessary on thr Raspi2 itself. This distri comes with gcc 4.9.2. Thanks to Murat for his tip to try Jessie. Now i get useful audio - with some noise in the background, what i have to check with raw mode. But it's now understandable audio, what wasn't the case with Raspbian (based on Debian Wheezy, coming with gcc 4.6.3). kind regards, Hayati Am 08.07.2015 um 21:17 schrieb Hayati Ayguen:
Hello,
now tested the binary compiled on RPi 1 on the RPi 2 - after copying the full source including build directory of cmake and executing "sudo make install" on the RPi 2.
The result: this works fine! Do i have to conclude, that Raspbian's g++/gcc compiler does produce wrong code when executed on RPi 2 ?! gcc/g++ (Debian 4.6.3-14+rpi1) 4.6.3
OK, now verify: make clean, rebuild and install new binary (from the build directory copied from RPi 1): this works! but, i don't understand why! Now reboot, so that kernel module reloads: still works!
In the end, i'm happy that it works "somehow". But, i've no idea where the problem lies!
kind regards, Hayati
Am 08.07.2015 um 09:38 schrieb "Hayati Aygün":
Hi Marcus,
librtlsdr is OK, because rtl_tcp works fine, which transfers full I/Q spectrum of ~ 1.4 MHz to SDR# on Windows/PC. The problem must lie in rtl_fm. By the way, the I/Q demodulation "raw" mode is erroneous, too, when having compiled on Raspberry Pi 2. Something else i could try: use the binary compiled on Raspi 1 B+ on the Raspi 2. I'll check this, this evening.
Before going for painful debugging, it would be fine if someone else could reproduce the problem .. or say that it works fine.
kind regards, Hayati
*Gesendet:* Mittwoch, 08. Juli 2015 um 08:30 Uhr *Von:* "Marcus Müller" <marcus.mueller@ettus.com> *An:* osmocom-sdr@lists.osmocom.org *Betreff:* Re: NBFM problem with rtl_fm on Raspberry Pi 2 Hi Hayati,
I'm totally not used to development on Raspbian, but: It's really strange that running the same software on the same distro with the same RF device would give different results.
So, debugging is on :/ I'd say you should use rtl_sdr to get raw IQ samples, and then have a look at them on your PC. If they are ok, one would have to look a bit deeper into rtl_fm, if not, something goes wrong on the librtlsdr side of things.
Best regards, Marcus
participants (5)
-
"Hayati Aygün" -
Hayati Ayguen
-
Marcus Müller -
Murat Tologlu -
Nikolay Dimitrov