discuss-gnuradio
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Discuss-gnuradio] FFT Playback


From: Marcus Müller
Subject: Re: [Discuss-gnuradio] FFT Playback
Date: Wed, 04 Dec 2013 13:50:36 +0100
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:24.0) Gecko/20100101 Thunderbird/24.1.0

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

How do you downsample 50 to 30 Msps? With a fractional resampler?
That's highly demanding of your CPU! Do it with the recorded data, not
before, if you can afford the memory.

On 04.12.2013 13:47, Paul B. Huter wrote:
> Without the low pass filter, I was getting a sequence of D's. The
> low pass filter uses a Volk machine, and downsamples the 50M to 30M
> with a 5M transition size. I am using the GUI companion on Linux,
> but I vaguely recall that I may have had better luck recording when
> I just created a Python script on a PC. I will try the script (on
> Linux) this afternoon, and I will also try running at 25M
> downsampled to 10M.
> 
> In the meantime, any additional comments/suggestions would be
> appreciated.
> 
> Thanks.
> 
> Paul B. Huter On Dec 4, 2013 6:29 AM, "Marcus Müller"
> <address@hidden> wrote:
> 
> With most hardware interfaces, you get notifications when your 
> driver/interface starts dropping samples -- with UHD you'd see
> loads of 'O' indicating overflows. If you don't see that, there is
> something really wrong. How does your data acquisition take place?
> 50Msps is really a lot for an average PC if you do filtering; I
> don't know how your low pass filter looks, but I think it's
> probably causing to much computational load for your computer to
> process the incoming samples in real time.
> 
> Greetings, Marcus
> 
> On 04.12.2013 12:39, Paul B. Huter wrote:
>>>> There must be something with my record, because my 20 seconds
>>>> of running only produces about a 180MB file. I am not seeing
>>>> any indications of dropped data, though. I am recording to a
>>>> RAMDisk, but I get the same results recording to my HDD.
>>>> 
>>>> Does anyone have recommendations for how to ensure I get my
>>>> data recorded? I really think that if I cam record the data,
>>>> playback will be more what I am looking for.
>>>> 
>>>> Paul B. Huter On Dec 4, 2013 1:20 AM, "Marcus Müller" 
>>>> <address@hidden> wrote:
>>>> 
>>>> Hi Paul!
>>>> 
>>>> I'm gonna go ahead and rearrange parts of your message for
>>>> the ease of reply:
>>>>>>> My datafile was recorded for about 20 seconds at a
>>>>>>> sample rate of 50Msps through a low-pass filter to only
>>>>>>> capture 0-30MHz.
>>>> 50Msps * 20s = 1Gs (you're sure your file is 8GB in size?)
>>>> should be plenty enough, so my guess was totally astray.
>>>> 
>>>>>>> Looking at the configuration for my FFT, and changing 
>>>>>>> 'Refresh Rate' from the default '15' to '1M' gives me
>>>>>>> some extra playback.
>>>> what gui toolkit are you using? wx or qt? However, a refresh
>>>> rate of one million frames per second... doesn't even
>>>> remotely make sense. Since there is no way that your file
>>>> source can supply 1024Msps for the GUI sink, even without
>>>> throttling, I don't know what the (wx or qt or whatever) FFT
>>>> sink does in that case. I'd have to read the source code...
>>>> 
>>>>>>> I like the way the waterfall looks, but I'm having the
>>>>>>> same issue as the FFT.
>>>> 
>>>> Ok, this is confusing. The waterfall sink should basically be
>>>> able to have one pixel of display per sample going in. And
>>>> with 1 billion samples going in, there shouldn't really be
>>>> repitions visible, unless your data is repetitive.
>>>> 
>>>>>>> Is there some correlation between data rate on the
>>>>>>> record and FFT refresh rate?
>>>> No. Refresh rate is just the display refresh rate, i.e. the
>>>> speed at which the GUI updates its visualization.
>>>> 
>>>>>>> How is 'FFT Size' used (right now I have that at
>>>>>>> '1024')?
>>>> Um this is kinda awkward - it's the size of the FFT; how the 
>>>> samples are internally processed prior to being transformed
>>>> depends on the actual FFT sink implementation.
>>>> 
>>>>>>> On a related note - is there a way to get the FFT to
>>>>>>> start out in 'Autoscale', or do I need to hit the
>>>>>>> button every time?
>>>> 
>>>> No. "Autoscale" is (as far as I know and can remember right
>>>> now) not a /mode/ for the qt nor the wx gui fft sink, but
>>>> scales the display to fit the data that is display /the
>>>> moment the button is clicked/. So, since data can only be
>>>> available at run time, you can't do that before running the
>>>> flow graph. However, at least for the qt implementation, you
>>>> can set y max/y min, if you know it beforehand.
>>>> 
>>>> So: Verify your data is really about 1 billion samples long
>>>> and not only a few thousand; if your waterfall looks
>>>> periodic, your recorded signal is periodic.
>>>> 
>>>> Greetings, Marcus
>>>> 
>>>> On 03.12.2013 22:45, Paul B. Huter wrote:
>>>>>>> 
>>>>>>> Is there some correlation between data rate on the
>>>>>>> record and FFT refresh rate? How is 'FFT Size' used
>>>>>>> (right now I have that at '1024')? On a related note -
>>>>>>> is there a way to get the FFT to start out in
>>>>>>> 'Autoscale', or do I need to hit the button every
>>>>>>> time?
>>>>>>> 
>>>>>>> Thanks!
>>>>>>> 
>>>>>>> 
>>>>>>> On Tue, Dec 3, 2013 at 3:01 PM, Marcus Müller 
>>>>>>> <address@hidden> wrote:
>>>>>>> 
>>>>>>> Most probably your data is simply to short in relation
>>>>>>> to the fft length and the amount of samples your
>>>>>>> specific graphical FFT amplitude sink drops for
>>>>>>> display. Please review you fft length, update rate, and
>>>>>>> try out different fft GUIs (qt/wx).
>>>>>>> 
>>>>>>> You could zeropad your data generously to notice the 
>>>>>>> repitition. However, I think with what you're
>>>>>>> describing you might be better off with a waterfall
>>>>>>> graph; try them, they're fun :)
>>>>>>> 
>>>>>>> Greetings, Marcus
>>>>>>> 
>>>>>>> On 03.12.2013 21:45, Paul B. Huter wrote:
>>>>>>>>>> When I play my data file back through a throttle
>>>>>>>>>> and frequency translating FIR filter to an FFT
>>>>>>>>>> sink with repeat OFF, it seems to just show a
>>>>>>>>>> static plot. However, with repeat ON, I get
>>>>>>>>>> playback, but I can't tell when data ends and
>>>>>>>>>> starts back over. Is there a way for me to know
>>>>>>>>>> when it repeats? Or, is there a way to get
>>>>>>>>>> playback with repeat turned off?
>>>>>>>>>> 
>>>>>>>>>> Thanks.
>>>>>>>>>> 
>>>>>>>>>> Paul B. Huter
>>>>>>>>>> 
>>>>>>>>>> 
>>>>>>>>>> 
>>>>>>>>>> _______________________________________________ 
>>>>>>>>>> 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
>>>>>>>>
>>>>>>>
>>>>>
>>>>
>>
>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.15 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQEcBAEBAgAGBQJSnyUcAAoJEAFxB7BbsDrLghsH/2M1hLURhotBQpCP8GXVECsG
11X/XYyieteYTdfd8bT/BEdNBWOtefHicimrkXBwXW6oC+50jRbvCwGDQdKrVFCE
JltDYvV0YsIM5h4+BqaAE05T1ol1scKKCgJrqKgkorYsjgHMxlSo9XOZF09dRzLT
Ib1j52FyH5IZ4RgU7q5hyil5lEGwPEqCyKj4TQFDO8nFLU+l9OboAj3H5Y+G3ew0
BbHyzIe4sZWHmPKGdur+xtJ+84TcNvsVVxgn/kFBRcqQrywk68U5IX2w8Dx+7sMR
Dr/hQKXD1yO75YYlO8vXd74hH802tur5/BW/Q0SRGGjcr2sS2vnFr1akcIFq7QA=
=Irgo
-----END PGP SIGNATURE-----



reply via email to

[Prev in Thread] Current Thread [Next in Thread]