AW: Re: AW: Re: DVB-T2

Müller, Marcus (CEL) mueller at kit.edu
Mon Aug 26 13:05:28 UTC 2019


Hi Ingo,
On Mon, 2019-08-26 at 14:35 +0200, Ingo.Wolf at gmx.de wrote:
> But if you bypass demodulator, don't you have to demodulate in software then?

Indeed. That's why you don't bypass the DVB-T demodulator if you want
to use your TV dongle as TV dongle.

> How can you just say "NO" isn't that nonsense?

Because it's not nonsense.

> Can the RTL2832U capture the 10MHz Bandwidth?

No.

> Wasn't it something like 3MB/s max on the RTL, does that mean 3MHz with 8 Bit Samples?

That's what it can transport in raw IQ sample across USB.
That has nothing to do with what the integrated DVB receiver /
demodulator can deal with. The contained MPEG data stream is much less
in bits per second than the raw samples.

>  
> Wether CPU can handle is another topic, but GHz Multicore + GPU don't look like completely impossible.

It's not completely impossible, just very hard. I'm not aware of any
DVB-T2 software decoder that would work on laptop-grade hardware. You
seem to be fond of stating a few unfounded assumptions here:

I'd recommend looking into gr-dtv, which is software decoders for DVB-
** standards, and look how far in rate you can push them with purely
recorded or simulated signals. I think you don't realize how complex
the problem of concatenated LDPC and BCH decoders is. The LDPC block
for DVB-T2 is 64800 bits long, and that means that channel decoding
means you have to find solutions x to

xH=y

where H is a 64800×(64800/2, /3, /4, /8) matrix, and y is the received
softbits vector of length 64800, multiple hundred times a second. Good
luck doing that in software without spending a lot of time optimizing
the architecture of your solver (that solver is called a LDPC decoder).
Suddenly, your fast CPU isn't fast enough for real-time decoding – by
orders of magnitude, not by a factor of 2 or 5 or so.

Best regards,
Marcus

>  
> -----Ursprüngliche Nachricht-----
> Von: "Müller, Marcus (CEL)" <mueller at kit.edu>
> Gesendet: Montag, 26. August 2019 14:11
> An: "Ingo.Wolf at gmx.de" <ingo.wolf at gmx.de>
> Cc: "osmocom-sdr at lists.osmocom.org" <osmocom-sdr at lists.osmocom.org>
> Betreff: Re: AW: Re: DVB-T2
> </osmocom-sdr at lists.osmocom.org></ingo.wolf at gmx.de>
> No.
> 
> The fact that you can use your RTL dongle as SDR device is because one
> can *bypass* most of the digital logic (demodulator, decoder, stream
> extractor…) and just get a raw IQ stream.
> Decoding DVB-T in software is a really CPU-intense problem, due to
> high-rate (in both senses of the word) channel coding, complex
> synchronization, and high-order constellations used.
> 
> Best regards,
> Marcus
> 
> On Sun, 2019-08-25 at 05:02 +0200, Ingo.Wolf at gmx.de wrote:
> > But isn't SDR demodulation in software?
> > 
> > -----Ursprüngliche Nachricht-----
> > Von: "Steve Markgraf" <steve at steve-m.de>
> > Gesendet: Samstag, 24. August 2019 18:19
> > An: osmocom-sdr at lists.osmocom.org; Ingo.Wolf at gmx.de
> > Betreff: Re: DVB-T2
> > Hi,
> > 
> > On 24.08.19 12:14, Ingo.Wolf at gmx.de wrote:
> > > can somebody confirm RTL-chip is not DVB-T2 capable not having 10MHz
> > > bandwidth needed?
> > 
> > The RTL2832U does not support DVB-T2 as it lacks the required
> > demodulater hardware. On the RTL2832P, an external demodulator can be
> > attached via a parallel MPEG TS interface. This is used by dongles
> > manufactured by Astrometa, which attach a Panasonic MN88472 DVB-T2
> > demodulator, see [1].
> > 
> > Regards,
> > Steve
> > 
> > [1] https://steve-m.de/pictures/rtl-sdr/rtl2832p_dvbt2/
> >
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 6582 bytes
Desc: not available
URL: <http://lists.osmocom.org/pipermail/osmocom-sdr/attachments/20190826/2e0624e7/attachment.bin>


More information about the osmocom-sdr mailing list