[USRP-users] Phase Noise for USRP N200 with LFTX/LFRX

khalid.el-darymli khalid.el-darymli at mun.ca
Thu Dec 11 10:23:09 EST 2014


Thanks very much Marcus for the detailed explanation.

>> When you're using an external REF clock, your phase noise is utterly
dominated by the phase-noise of your external reference.  So, if your
  external reference has worse phase noise, then your signals will have
worse phase-noise.  But, I want to ask, why use an external reference
  with worse phase noise?

-- I don't want to use a reference signal with worse phase noise, but I
have access to a self-made device for generating the external reference
clock. I didn't make that device myself, and I don't have access to neither
its phase noise specs nor a high-quality spectrum analyzer to measure it.
I'll have to find a way to characterize its phase noise.

>> Images should be well-suppressed, as far as I know--the upconversion is
done digitally, so it's like a "perfect" image-reject mixer.

On another note, since the reference clock for the whole system will be the
external 10 MHz signal, the digital images would be situated at multiples
of this 10 MHz frequency (i.e., m*10 MHz - [e.g., carrier frequency] 15
MHz; m=-/+1, -+2, ...). Hence, for this case, the first few images will
fall within the bandwidth of the built-in LFTX/LFRX LPF filter. However,
based on your earlier explanation,these are totally taken care of
'digitally' and I don't have to worry about any of that, am I right?

Best,
Khalid




On Thu, Dec 11, 2014 at 11:25 AM, Marcus D. Leech <mleech at ripnet.com> wrote:

>  On 12/11/2014 08:04 AM, khalid.el-darymli wrote:
>
>  I would like to thank you all for this information. You are always a
> great help!
>
>  - Now, given that my FMCW chirp is digitally synthesized (i.e., using
> the digital DDS architecture of N200) before it is up-converted to an
> analog form, is it correct to say that the temperature-based drifts I am
> seeing (i.e., in the Tx FMCW chirp directly loop-backed to the Rx) are not
> due to the Tx chain but rather they are solely from the Rx chain (with the
> exception of the DAC from the Tx side)? Or they are (roughly *EQUAL*)
> mixture of temperature drifts in both Tx (i.e., DAC) and Rx (i.e., ADC)
> chains? In other words, what are the major contributors of these
> temperature drifts? and are they 'roughly' equal in both the Tx and the Rx
> sides?
>
> It's hard to say, piece by piece, what contributions to very-small analog
> gain drifts can be ascribed to which piece.  The LFRX/LFTX are very similar,
>   electronically, so I would any temperature coefficients to apply roughly
> equally to both sides.  Keep in mind that in a production system, your
>   coax cables will also be contributing to both gain and phase drift, as
> they change temperature.
>
>
>  - On another note, given that the system is driven by a 100 MHz
> reference clock, images will be produced at some multiples of the clock
> frequency. For example, if the requested center frequency was 20 MHz, then:
> 1st +image @: 100 MHz-20=80 MHz
> 2nd +image @ 100 MHz+20=120 MHz
> 3rd +image @ 200 MHz-20=200MHz-20=180 MHz
> and so on.
>
> Images should be well-suppressed, as far as I know--the upconversion is
> done digitally, so it's like a "perfect" image-reject mixer.
>
>
>  Given that the LFTX daughter-board has an in-built 30 MHz low-pass
> filter, is it correct to say that all such images will be filtered-out, and
> it is 'guaranteed' that the generated FMCW chirp (e.g., with a BW of 200
> KHz) up-converted to some specific center frequency (e.g., 20 MHz), is
> guaranteed to 'ONLY' be within that BW around the center frequency? In
> other words, do I need not to worry about generating an out-of-band
> noise/interference (for whatever reason such as images, phase truncation
> spurs, harmonic spurs due to DAC nonlinearity, clock feed-through, etc.)
> that entail installing an additional external analog band-pass filter?
>
> At the end of the day, you'll have to measure in the laboratory what your
> undesired output signals are.  It is nearly always the case that
> app-specific
>   filtering needs to be applied by the end-user in an SDR implementation,
> since the platform is designed not for any specific application.  Assuming
> that
>   in end use, you use an RF amplifier, you'll have to revisit your
> filtering *after* said amplifier, because they nearly always introduce some
> small
>   undesired products as well.
>
>
>
>  - Finally, what effect does the external REF clock (used for syncing
> multiple N200 devices) has on the internal clock (CVHD-950) of the N200? In
> other words, is the common external clock meant to OVERRIDE CVHD-950? If
> so, if the phase noise of my external clock is worse than that of the
> CVHD-950, will that affect the phase noise of the signal generated/received
> in USRP N200?
>
>
>   When you're using an external REF clock, your phase noise is utterly
> dominated by the phase-noise of your external reference.  So, if your
>   external reference has worse phase noise, then your signals will have
> worse phase-noise.  But, I want to ask, why use an external reference
>   with worse phase noise?
>
>
>
>  Thanks.
>
>  Best regards,
> Khalid
>
>
>
> On Tue, Dec 9, 2014 at 11:59 AM, Marcus D. Leech <mleech at ripnet.com>
> wrote:
>
>>  On 12/09/2014 07:53 AM, khalid.el-darymli wrote:
>>
>>  Thanks very much for this information. According to the datasheet you
>> mentioned in the link below [1], the supported oscillating frequency by
>> CVHD-950 is 50 MHz to 130 MHz. The PN specs are provided for some center
>> frequencies in this range.
>>
>> However, I am interested in the HF band for which the LFTX/LFRX support
>> center frequencies up to 30 MHz. For requested center frequencies in (DC,
>> 30 MHz], do you digitally down-sample the center frequency of CVHD-950? Is
>> there any description on how exactly you do that? and what effect does this
>> may have on the phase noise of CVHD-950?
>>
>>  No, that is the reference clock for the entire platform--driving all
>> timing in the platform, including the 100MHz sample clock for the ADCs. The
>> schematics
>>   are here:
>>
>> http://files.ettus.com/schematics/n200
>>
>>
>>
>>
>>  I am working on an ocean-based HF radar. Basically, the system measures
>> the Doppler frequency of ocean waves for different sea states. It is
>> important that I characterize my system to know its limits and capabilities.
>>
>> Thanks for your help.
>>
>> Best,
>> Khalid
>>
>> [1] http://www.crystek.com/home/oscillator/vcxodetail.aspx?pn=CVHD-950
>>
>> On Mon, Dec 8, 2014 at 7:35 PM, Matt Ettus <matt at ettus.com> wrote:
>>
>>>
>>> With the LFRX and LFTX, there is no mixer, so all phase noise is from
>>> the ADC clock.  The oscillator used is a CVHD-950 from Crystek and you can
>>> find the phase noise spec for it on their web site.
>>>
>>>  Matt
>>>
>>>  On Mon, Dec 8, 2014 at 2:11 PM, khalid.el-darymli via USRP-users <
>>> usrp-users at lists.ettus.com> wrote:
>>>
>>>>    Hi,
>>>>
>>>>  According to the specs provided in [1], the RF performance of USRP
>>>> N200/N201 is characterized for the WBX daughter-board. Particularly, phase
>>>> noise measurements are given for three offsets from a center frequency of
>>>> 1.8 GHz.
>>>>
>>>>  My question is, are there any such measurements for the LFTX/LFRX
>>>> daughter-board?
>>>>
>>>> If not, is there a recommended procedure for doing these measurements?
>>>>  Among others, I am particularly concerned with the phase noise and its
>>>> short-term effect on my Doppler measurements.
>>>>
>>>>
>>>> [1]
>>>> https://www.ettus.com/content/files/07495_Ettus_N200-210_DS_Flyer_HR_1.pdf
>>>>
>>>>  Thanks,
>>>>  Khalid
>>>>
>>>>
>>>>  _______________________________________________
>>>> USRP-users mailing list
>>>> USRP-users at lists.ettus.com
>>>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
>>>>
>>>>
>>>
>>
>>
>>
>>
>>
>>  --
>> Marcus Leech
>> Principal Investigator
>> Shirleys Bay Radio Astronomy Consortiumhttp://www.sbrac.org
>>
>>
>
>
>
>
> --
> Marcus Leech
> Principal Investigator
> Shirleys Bay Radio Astronomy Consortiumhttp://www.sbrac.org
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/attachments/20141211/57234ab4/attachment-0002.html>


More information about the USRP-users mailing list