Re: [Discuss-gnuradio] GNU Radio Companion - ALSA

[Marcus Müller]

*high five*, got it to work!

So, why the rational resampler?! makes no sense, especially since both
interpolation and decimation need to be integers; since it doesn't make
sense mathematical, is it possible you did something slightly different
on Windows?

Resamplers are *only* necessary to convert a signal from one sampling
rate to another mathematically, without changing the signal's content.
The rate change is interpolation/decimation. If your original signal is
already at the right sampling rate, adding a resampler in between will
only *break* things.

The RuntimeError has something to do with how you configure the audio
sink. so, you need to use exactly the same config, ie. 2 inputs, hw:0,
sampling rate 48000. It doesn't come from having the wrong kind of
resampling before that. GNU Radio blocks really don't care at all about
what happens upstream. They get a series of numbers they process.

There's no reason for multiple resamplers. Please take this as a
constructive criticism: You must stop using a copy and paste approach,
and go for an understanding-based approach.

So let's work from the ends: We know we have an audio sink that we need
to run at 48000 as sampling rate, and in the end, we'll need to connect
that to a RTL source that gives us 2 MS/s. So in total, from source to
sink, we need a resampling ratio of 2e6/48e3. I'll do it the boring way
here, by using a prime factorization of both, to figure out how we can
group these:

2e6 = 2 · 10⁶ = 2 · (2·5)⁶
= 5⁶ · 2⁶

48e3 = 48 · 10³ = 3 · 16 · (2·5)³ = 3 · 2⁴ · 2³ · 5³
= 5³ · 3 · 2⁷

Let's divide these two numbers so that we know what our overall
resampling ratio is:

2e6/48e3 = (5⁶ · 2⁶)/(5³ · 3 · 2⁷) and cancelling out everything that's
possible to cancel
5³ / (3 · 2)

So what we need is a decimation of 5³ = 125 and an interpolation of 6.
Awesome! Let's make that happen.

We know that we have at least one block that we need to use between RTL
source and Audio sink: the WBFM receive. Let's plug that directly into
the Audio Sink. That means we know its output rate, because it *must* be
the 48k that said audio sink consumes. Ok, let's see, we need to
decimate by 5³ in total, so let's put the "Audio Decimation" to 5 here.
That leaves us with still 5² to decimate, and 6 to interpolate. (we also
can directly calculate that the *input* of the WBFM receive must be 5 ·
48kS/s = 240 kS/s). And since we have two channels on our audio sink,
connect the both Audio sink inputs to the one autput of WBFM receive.

Well, then, let's do that: add a rational resampler that has decimation
= 25 and interpolation = 6. Connect its output directly to WBFM receive,
and its input to the RTL source. Done!


Best regards,

Marcus


On 19.02.2017 19:42, Robin A. Jensen wrote:
> Hello again.
>
> Now i've a little test.
> Signal Source: sample rate: 48k, Waveform: Cosine, freq: 1k amplitude:
> 1 ->
>
> Audio Sink: Samp_rate: 48 k, Devicename: hw:0 (with 2 inputs)
>
> And that is a succes! There a sound.
> It will only run with 48k as sample rate.
> So now we know there are an output for real. :-)
>
> But if i put in Rationel Resampler: Interpolation: 1 / 48k,
> Decimation: 1 and Source is 48k.
> There is no sound..??
> No mather how i fidle around with the values.
> I've tried with different settings for the Sampler and also Rationel
> Resampler Base, with no change regard to the output.
> I find it really strange as the Rationel Resampler is working on a
> Windows 10.
>
> I then whent back to my WBFM setup and remove the resampler and run it
> direcly from WBFM to Auduo Sink,
> and then the RuntimeError came back :-(
>
> Best regard
> Robin.
>
>
> Den 19-02-2017 kl. 18:06 skrev Marcus Müller:
>> Hi Robin,
>>
>> Progress! (and you really don't have to try anything in lower- AND
>> uppercase. the names are simply case-sensitive, so use them **exactly**
>> like aplay -L lists them. Everything else can't work. There's really no
>> magic involved here! Stop googling stuff that you copy and paste. You've
>> got a discrete problem on *your* machine, and you can solve it, so copy
>> and pasting device strings that work on other people's computer's is
>> really not that helpful.)
>>
>>> It returns: check topology failed on on audio_alsa_sink(8) using
>>> ninputs=1, noutputs=0
>> Now that means you're using a device it can find. so: yeah, we're doing
>> something right. I'd really recommend you stick with "pulse", as that
>> will make sure of mixing all the audio streams other programs produce
>> together with yours and you'll not compete for exclusive access to the
>> audio device. It *should* give you at least the same check_topology
>> error. Can you please confirm that?
>>
>> Best regards,
>>
>> Marcus
>>
>>
>>
>> On 19.02.2017 16:42, Robin A. Jensen wrote:
>>> Hey Marcus.
>>>
>>> Have tried that.
>>> In documentation for the block:
>>> pulse , hw:0,0 , plughw:0,0 , surround51 , /dev/dsp
>>>
>>> I've tried every single one of them with lowercase and upper case.
>>>
>>> If i use: hw:0,0
>>> It returns: check topology failed on on audio_alsa_sink(8) using
>>> ninputs=1, noutputs=0
>>> If I use: HW:0,0
>>> It returns: RuntimeError: audio_alsa_sink.
>>>
>>> I've also tried with: HW:0,1 , HW:1,0 , HW:0.1 , HW:1.0 also in lower
>>> case.
>>> I've been tried to google with different search terms with no luck.
>>> That's why I ended up on this mailing list.
>>>
>>> GNU Radio is downloaded from rpi Jessie repository with apt-get
>>> install gnuradio.
>>>
>>> Best regards
>>> Robin.
>>>
>>> Den 19-02-2017 kl. 16:25 skrev Marcus Müller:
>>>> Hm, that looks like there's no HW in the list, so that would be an
>>>> initial debugging success!
>>>>
>>>> So, the easiest is probably if you just use "pulse" in the Device Name
>>>> field.
>>>>
>>>> Best regards,
>>>>
>>>> Marcus
>>>>
>>>>
>>>> On 19.02.2017 16:15, Robin A. Jensen wrote:
>>>>> Yes of course.
>>>>> Here we go:
>>>>>
>>>>> aplay -L
>>>>>
>>>>> null
>>>>>       Discard all samples (playback) or generate zero samples
>>>>> (capture)
>>>>>
>>>>> pulse
>>>>>       PulseAudio Sound Server
>>>>> sysdefault:CARD=ALSA
>>>>>       bcm2835 ALSA, bcm2835 ALSA
>>>>>       Default Audio Device
>>>>>
>>>>> dmix:CARD=ALSA,DEV=0
>>>>>       bcm2835 ALSA, bcm2835 ALSA
>>>>>       Direct sample mixing device
>>>>>
>>>>> dmix:CARD=ALSA,DEV=1
>>>>>       bcm2835 ALSA, bcm2835 IEC958/HDMI
>>>>>       Direct sample mixing device
>>>>>
>>>>> dsnoop:CARD=ALSA,DEV=0
>>>>>       bcm2835 ALSA, bcm2835 ALSA
>>>>>       Direct sample snooping device
>>>>>
>>>>> dsnoop:CARD=ALSA,DEV=1
>>>>>       bcm2835 ALSA, bcm2835 IEC958/HDMI
>>>>>       Direct sample snooping device
>>>>>
>>>>> hw:CARD=ALSA,DEV=0
>>>>>       bcm2835 ALSA, bcm2835 ALSA
>>>>>       Direct hardware device without any conversions
>>>>>
>>>>> hw:CARD=ALSA,DEV=1
>>>>>       bcm2835 ALSA, bcm2835 IEC958/HDMI
>>>>>       Direct hardware device without any conversions
>>>>>
>>>>> plughw:CARD=ALSA,DEV=0
>>>>>       bcm2835 ALSA, bcm2835 ALSA
>>>>>       Hardware device with all software conversions
>>>>>
>>>>> plughw:CARD=ALSA,DEV=1
>>>>>       bcm2835 ALSA, bcm2835 IEC958/HDMI
>>>>>       Hardware device with all software conversions
>>>>>
>>>>> If i use @: speaker-test -f 800 -t sinus -r 48000 -c 1-s 1
>>>>> I'll get a fine sinus tone.
>>>>> So there are sound through the system.
>>>>>
>>>>> Best regards
>>>>> Robin.
>>>>>
>>>>>
>>>>> Den 19-02-2017 kl. 15:48 skrev Marcus Müller:
>>>>>> You're right, we should tackle this more systematically.
>>>>>>
>>>>>> My problem is that I don't have a RPi3 to play around with at hand,
>>>>>> so I
>>>>>> have to trust you on the "HW:0,0"; it's not a very typical string,
>>>>>> through, as most alsa device names are lowercase. Could you share
>>>>>> the
>>>>>> output of "aplay -L" with us?
>>>>>>
>>>>>> Best regards,
>>>>>> Marcus
>>>>>>
>>>>>> On 19.02.2017 15:16, Robin A. Jensen wrote:
>>>>>>> Hello Marcus.
>>>>>>>
>>>>>>> Thanks for your reply and your welcomming to the community.
>>>>>>>
>>>>>>> I've made the changes as suggested.
>>>>>>> Also i've made a setup on a Windows 10 machine, to ensure the
>>>>>>> script
>>>>>>> runs.
>>>>>>> With the changes the sound is much better! :-)
>>>>>>>
>>>>>>> But when i run the same setup on RPi 3 / Jessie-distro i've end up
>>>>>>> with the same result, no matter what I do with the Audio sink.
>>>>>>> I'm using Gnu Radio Companion 3.7.5
>>>>>>> The error code is still:
>>>>>>>
>>>>>>> ALSA libpcm.c2239:)snd_pcm_open_noupdate) Unknown PCM HW:0,0 <-- 10
>>>>>>> times this line
>>>>>>> gr::log :ERROR: audio_alsa_sink0 - [HW:0,0]: No such file or
>>>>>>> directory.
>>>>>>> File "/home/pi/radio/top_block.py", line133, in <module>
>>>>>>>     tb= top_block()
>>>>>>> File "/home/pi/radio/top_block.py", line82, in __init__
>>>>>>> self.audio_sink_0 = audio.sink(48000, "HW:0,0", True)
>>>>>>> File
>>>>>>> "/usr/lib/python2.7/dist-packages/gnuradio/audio/audio_swig.py",
>>>>>>> line 195, in make
>>>>>>> return _ausio_swig.sink_make(*args, **kwargs)
>>>>>>> RuntimeError: audio_alsa_sink
>>>>>>>
>>>>>>> I know is alwayes is eassy to blame the software, but could it be a
>>>>>>> bug?
>>>>>>> As i said. i've tried with all the suggested lines from
>>>>>>> documentation
>>>>>>> of the block.
>>>>>>>
>>>>>>> Best regards
>>>>>>> Robin.
>>>>>>>
>>>>>>> Den 18-02-2017 kl. 18:17 skrev Marcus Müller:
>>>>>>>> Hi Robin,
>>>>>>>>
>>>>>>>> first of all: Welcome to the GNU Radio community!
>>>>>>>>
>>>>>>>>
>>>>>>>> On 02/18/2017 05:29 PM, Robin A. Jensen wrote:
>>>>>>>>> Hello all of you.
>>>>>>>>>
>>>>>>>>> I've just recieved my RTL-SDR dongle and is all new to this
>>>>>>>>> sdr-stuff,
>>>>>>>>> so please bear over with me, if i'm at the wrong place.
>>>>>>>>> I'm using GNU Radio Companion on a RPi 3 and no mather what i'll
>>>>>>>>> do, i
>>>>>>>>> can't get the sound to work.
>>>>>>>>> If a'im using rtl_fm and aplay, i'll get sound but it won't
>>>>>>>>> set on
>>>>>>>>> the
>>>>>>>>> radiostation.
>>>>>>>> aha, so that's good, the sound system as it does work.
>>>>>>>> You'll probably want to use "aplay -L" to find the possible ALSA
>>>>>>>> device
>>>>>>>> names that you can use in the GNU Radio Audio sink.
>>>>>>>>> I'll take on that later.
>>>>>>>>> I've createt a small FM Reciever in GNU Radio companion and
>>>>>>>>> everytime
>>>>>>>>> i'll execute the script i'll get an error:
>>>>>>>>>
>>>>>>>>> RuntimeError.audio.alsa.sink
>>>>>>>> Hm, I've never seen a GNU Radio error being printed like this; but
>>>>>>>> that
>>>>>>>> might just be me. However, I can't reproduce this error printing
>>>>>>>> shape
>>>>>>>> as hard as I try.
>>>>>>>>> I've been all over the internet to find a solution but with no
>>>>>>>>> luck.
>>>>>>>>> So now i'm have a hope that this mailling list can help me?
>>>>>>>>>
>>>>>>>> My suspicion is that your audio device doesn't like the sampling
>>>>>>>> rate
>>>>>>>> your trying to use, or you need to specify a device name (or
>>>>>>>> both).
>>>>>>>> Can
>>>>>>>> you make things work on the PC you use to design these flow
>>>>>>>> graphs?
>>>>>>>>
>>>>>>>> I'd start with a signal source (sampling rate == the sampling rate
>>>>>>>> that
>>>>>>>> you set in your Audio sink), configured to produce a "float"
>>>>>>>> output
>>>>>>>> sine
>>>>>>>> of 1 kHz, directly connected to an Audio sink. If that works,
>>>>>>>> move on.
>>>>>>>>
>>>>>>>> --------------------------------------------------------------------------------
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> What I say about the flow graph in the following has, as far as I
>>>>>>>> can
>>>>>>>> tell, nothing to do with the error you're getting. Still, there's
>>>>>>>> mistakes in the flow graph that would make it impossible to
>>>>>>>> successfully
>>>>>>>> run it, and thus I'd like to avoid frustration later on by
>>>>>>>> pointing
>>>>>>>> them
>>>>>>>> out know:
>>>>>>>>
>>>>>>>>
>>>>>>>> So, the main issue with your flow graph is that the sampling rate
>>>>>>>> at the
>>>>>>>> audio sink must be what you configured your audio sink for (48
>>>>>>>> kHz).
>>>>>>>> But: that rate is the result of your SDR's sampling rate (2MS/s),
>>>>>>>> multiplied with all interpolations, divided by all decimations in
>>>>>>>> the
>>>>>>>> path between.
>>>>>>>>
>>>>>>>>> My setup is:
>>>>>>>>> RTL-SDR Source: samplerate:  2M, frequency 96.5 MHz ->
>>>>>>>>> Rationel Sampler: Interpolation 4, Decimation: 1 ->
>>>>>>>> Uh, that means that you have now 8MS/s. That seems unnecessary,
>>>>>>>> since to
>>>>>>>> capture a <100 kHz wide FM channel, you wouldn't even need the
>>>>>>>> 2MS/s
>>>>>>>> you get
>>>>>>>>> Low Pass Filter: Sample rate: 2M, Cutoff freq: 100K, Transition
>>>>>>>>> Width:
>>>>>>>>> 100k ->
>>>>>>>> Which contradicts the 2MS/s used here, so you're actually getting
>>>>>>>> 400kHz
>>>>>>>> passband width, 400kHz transition width. Also, this feels like a
>>>>>>>> prime
>>>>>>>> candidate for including decimation in the filter (because the
>>>>>>>> resulting
>>>>>>>> bandwidth is 200 kHz (if you overlap the two transition widths),
>>>>>>>> and for
>>>>>>>> that you'd only need 200 kS/s of complex digital signal).
>>>>>>>>> WBFM Recive: Quadrature: 500K, Audio Decimation: 1 ->
>>>>>>>> This is now off by a factor of 16; are you sure you should be
>>>>>>>> using
>>>>>>>> "interpolation=4,decimation=1" instead of the inverse?
>>>>>>>>> Rationel Sampler: Interpolation: 500
>>>>>>>> Certainly not :) 8 MS/s · 500 = 4 GS/s
>>>>>>>>> Decimation: 48
>>>>>>>> Aside from that not even having greatest common denominator of 1
>>>>>>>> with
>>>>>>>> 500 (you can't know that this is important, not blaming you), this
>>>>>>>> would
>>>>>>>> give you an output sampling rate of 4GS/s/48 = 1 GS/s / 12 =
>>>>>>>> 83.333
>>>>>>>> MS/s. Which isn't even a multiple of 48 kHz, which you use in the
>>>>>>>> audio
>>>>>>>> sink:
>>>>>>>>> ->
>>>>>>>>> Audio Sink: Sample Rate: 48 KHz, Device Name: HW:0,0
>>>>>>>>>
>>>>>>>>> I've found that recipie on a Hack5 video and there it's working
>>>>>>>> I think there's some error in the way you configured these
>>>>>>>> resamplers. I
>>>>>>>> don't know the Video you're referring to, but the amount of
>>>>>>>> rational
>>>>>>>> resamplers used here alone, paired with the fact that you don't
>>>>>>>> decimate
>>>>>>>> in the WBFM receiver makes me kind of suspicious this might not
>>>>>>>> have
>>>>>>>> been the optimal video to take reference designs from!
>>>>>>>>
>>>>>>>> --------------------------------------------------------------------------------
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> Best regards,
>>>>>>>> Marcus
>>>>>>>>> With best regards
>>>>>>>>> Robin.
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> _______________________________________________
>>>>>>>>> Discuss-gnuradio mailing list
>>>>>>>>> address@hidden
>>>>>>>>> https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
>>>>>>>> _______________________________________________
>>>>>>>> Discuss-gnuradio mailing list
>>>>>>>> address@hidden
>>>>>>>> https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
>>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> Discuss-gnuradio mailing list
>>>>>>> address@hidden
>>>>>>> https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
>>>>>> _______________________________________________
>>>>>> Discuss-gnuradio mailing list
>>>>>> address@hidden
>>>>>> https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
>>>>>>
>>>>> _______________________________________________
>>>>> Discuss-gnuradio mailing list
>>>>> address@hidden
>>>>> https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
>>>> _______________________________________________
>>>> Discuss-gnuradio mailing list
>>>> address@hidden
>>>> https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
>>>>
>>>
>>> _______________________________________________
>>> Discuss-gnuradio mailing list
>>> address@hidden
>>> https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
>>
>> _______________________________________________
>> Discuss-gnuradio mailing list
>> address@hidden
>> https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
>>
>
>
> _______________________________________________
> Discuss-gnuradio mailing list
> address@hidden
> https://lists.gnu.org/mailman/listinfo/discuss-gnuradio