[Discuss-gnuradio] USRP-1 and 4ch receiver?

[Mamoru Yamamoto]

Mr. Marcus Mueller,

Thank you very much for suggestions.  I have some more questions.

1) I searched for the GNU Radio mailing-list archive and found a discussion

Subject:  [Discuss-gnuradio] USRP1 4 Channel Subdev Issue
Date:  Tue, 19 Feb 2013 11:07:55 -0800 (PST)

Then it would be possible to organize a 4-ch RX with USRP-1.  (Some
limitation is reported here, too.)  But as you suggest, we can use only
real (one) input for each frequency.  Also I need to prepare the
downward frequency conversion by our selves.  Careful additional
hardware design is necessary.  I know this is not easy.

2) Your suggestion of "2 x B210" is very interesting.  A good thing is
that the frequency conversion is included in the board.  But I want to
know how to run 2 USRP boards from one application program.  Is there an
example?  If possible I want to use GNU Radio Python code, but is this
possible?  If not, do we need to control from C++?  In my application
"phase coherency" across all inputs are necessary.  "timing
synchronization" is not necessary.

3) One more question.  If 2) is possible, is it possible to run "2 x
USRP-1" from a single application on a single PC?

Regards,

-- 
Mamoru Yamamoto / RISH, Kyoto University
address@hidden
Phone +81-774-38-3814, Cell +81-90-5653-7555


On 2014/08/08 1:00, address@hidden wrote:> Message:
8 Date: Thu, 07 Aug 2014 09:41:38 +0200 From: Marcus M?ller
> <address@hidden> To: address@hidden Subject: Re:
> [USRP-users] USRP-1 and 4ch receiver? Message-ID:
> <address@hidden> Content-Type: text/plain;
> charset=ISO-8859-1 Hi Mamoru Yamamoto, What daughterboards are you using
> now? BasicRX?
>> >I know that before it was possible to organize 4-ch receiver by setting
> 2 Basic-RX on USRP-1, and use (in-total) 4 SMA connectors for inputs.
>
> The dual-BasicRX setup does give you four channels, but these 4 channels
> are*real*, not complex. All other daughterboards mix the target
> frequency down to complex baseband, which implies that you need 2 ADCs
> per input (I+Q).
> Now, the BasicRX just brings the raw inputs of your dual-ADC to SMA
> connectors. That means you can still only receive one complex baseband
> signal with a BasicRX, OR you can receive two independent real channels.
> The problem at hand is that mixing down your RF signals to baseband will
> usually result in complex signals, and thus the BasicRX really doesn't
> give you two input channels for baseband signals.
>
>> >Sure coherency between signals are important.
> Stock USRP1s cannot be made coherent across multiple USRP1s. There were
> some tricks floating around, detailing how to use a shared clock between
> USRP1s, but I haven't heard from someone doing this in a long time,
> primarily because most people have switched to devices that support
> external clocks from stock.
>
> (1) As explained before, if you really need 4 RF channels, you will have
> to synchronize 4 complex baseband receivers, which needs at least two
> USRP1s. The BasicRX approach only works for*real*  baseband signals, and
> it doesn't sound like this applies to you:(
>
> (2) There is no example code for synchronized USRP1s, because that
> requires warranty-voiding hardware modification by the user. To get the
> four real input signals of two BasicRX, just receive at 0Hz center
> frequency, and set subdevice specs accordingly.
>
> (3) Various options exist:
> a) 2 x B210 with an external 10MHz clock is the cheapest solution to
> your problem, I guess. B210 have two coherent receiver chains each and
> come with the RF hardware integrated, so you don't need daughterboards.
> Instead of an external clock distributor, you can also use GPSDOs to get
> coherent clocks; Ettus sells both.
> b) If you need higher ADC resolution, using 4 x N2x0 + 4 Daughterboards
> might be a good choice, too. You can re-use your existing
> daughterboards. You will need an external clock distributor or a GPSDO
> for each of your devices.
> c) 2 x X3x0 + 4 daughterboards. This would be the most versatile
> solution, but most likely also not the cheapest. Use an external clock
> source, GPSDOs, or try the (still experimental) clock daisy-chaining.
> X3x0 offer higher bandwidths than the other options, and a lot of FPGA
> if you want to develop your own hardware signal processing.
>
> Greetings,
> Marcus