Re: [Discuss-gnuradio] Two USRP B210s with Octoclock-G (Samith Abeywickrama)

[Samith Abeywickrama]

Thank you Marcus. If I have a point source somewhere in the room, the channel gains (both magnitudes and phases) from the point source  to 4 antennas should be approximately equal. Otherwise, we are not correcting actual phase differences of two USRP B210s. Isn't it? 

On Wed, Sep 7, 2016 at 6:21 PM, Marcus Müller <address@hidden> wrote:

Both, wired and separate transmitter, are viable solutions. If you use wired, you'd have to calibrate your cabling first, though, then calibrate the cabling used when attaching antennas against that. So, in the general case, you're probably better off if you just have a "point source" somewhere in the observed room, because the calibration done with that is directly related to the signals you actually want to observe.

Best regards,

Marcus

On 09/07/2016 11:39 AM, Samith Abeywickrama wrote:

Hi Luca,

Thank you very much for your in detail explanation and it was really helpful for me. The question now I have is, how can I feed calibration signal? Is it a separate transmitter or some kind of wired link to usrp rx inputs? Thank you.

Samith  

On Sep 7, 2016 5:11 PM, "Luca Pascale" <address@hidden> wrote:

Hi Samith,

to calibrate phase offset you can use a "calibration signal".
Using a (phase) balanced power divider you can feed the 4 rf input with
this calibration signal and use it to estimate differential phasors
between the 4 channels taking one of them as reference (i.e. the first).
You will have 3 significant differential phasors.

You need to do this after each retune of the boards on a given frequency.
Then you can apply the correction factors to the channels 2,3,4 once you
acquired the signals you want to use to estimate DOAs.

Calibration Signal:
- CW tone if you are interested on a single specific frequency (not real
world condition)
- wideband noise
Also depends on your implementation...(are you using a FFT Filter-bank
and estimate DOAs for each FFT bins or ... are you working "narrowband").
To estimate correction factors I suggest to do some averaging.

Moreover if your DF algorithm is (in some measure) sensitive to signals
amplitude I suggest to perform also amplitude calibration not just phase
calibration.
You can estimate it using the differential phasors.

Regards,
Luca




>Well, that's a bit up to your implementation. Probably, you want to have
>something that you know the direction of and use it to find the
>individual phase offsets, and find a complex de-rotating factor.


>On 06.09.2016 10:52, Samith Abeywickrama wrote:
>> Hi Marcus,
>>
>> Thank you very much. Btw, how can we calibrate the relative phase
offsets?
>>
>> Samith
>>
>> On Tue, Sep 6, 2016 at 4:34 PM, Marcus M?ller
>> <address@hidden <mailto:address@hiddenom>> wrote:
>>
>>     Yes, you can  Offering synchronization using external 10MHz and
>>     PPS signals is the purpose of the Octoclock-G !
>>
>>     However, the B210's LOs will have a random phase after every
>>     tuning, so you must calibrate the relative phase offsets every
>>     time you tune
>>
>>     Best regards,
>>
>>     Marcus
>>
>>
>>     On 06.09.2016 10:21, Samith Abeywickrama wrote:
>>>     Hi all,
>>>
>>>     I'm going to implement MUSIC(Multiple Signal Classification)
>>>     algorithm to find Angle of Arrival of signal using two USRP
>>>     B210s. Therefore I have 4 antennas. But I need to synchronise my
>>>     two USRP B210s. May I know that can I use Octoclock-G with my
>>>     USRP B210s without any problem? Thank you in advance.
>>>
>>>     --
>>>     Best Regards!
>>>     Samith
>>
>>
>>
>>
>> --
>> Best Regards!
>> Samith

--

Best Regards!

Samith