[USRP-users] SBX daughtercard bandwidth problem..!!

salman dinani salmandinani at gmail.com
Wed Feb 15 00:16:25 EST 2012


Hi,


Thanks for your answers, what I have understood from all that discussion is
that :


1. I cannot set the decimation or interpolation factor below "4" in
Simulink's  SDRu block. It's a     limitation. Am I right in this
conclusion??


For elaborating and explaining my problem in a much better and clear way,
I have attached Snap shots of a very simple SDRu Transmit project in which
a sine wave block is connected to SDRu Transmit block (see *Sine_prj.png*).
The frequency of the sine wave is set to be 10MHZ , the configuration of
sine block is shown in pic *"sine_block.png"*. The centre frequency is set
to be "1 GHz" and the interpolation is set to "4", SDRu block setting is
shown in pic "*SDRu_block.png"*.  The output is taken on a Spectrum
analyzer and the snap shot of that is shown in *"Spectrum_anylzr.png"*. It
clearly shows two peaks ,10MHZ apart from centre frequency.


Sir, the problem starts from here that this all process works well up to 12
MHZ. But when I increased the sine wave frequency above  12MHZ (that's why
I was emphasizing on word 12 MHZ in my last email), let say 15MHZ or 20MHZ,
the output doesn't show peaks at 15 or 20MHZ apart from centre frequency,
In fact the peaks starts moving back as I increase the frequency of sine
wave and at 25MHZ, both peaks meet each other on zero on spectrum analyzer.


I hope sir I have explained my problem better than previous. The DAC in
USRP N200 is of 400MSPS and despite of interpolation is set to "4", the
sampling rate becomes 400e6/4= 100MSPS, and there is also complex sampling
involve. Obviously I won't able to get 100MSPS because of the reasons as
You have mentioned in your earlier email, but I am not able to get even
25MHZ of output. I don't know why it's happening like that. I have checked
and varied all the parameters but the response remains the same.


 Please help me :


1. what is going wrong? or where I am wrong that I am not able to get even
25 MHZ of output?


2  Is this because of any kind of limitation in software or hardware?


Thanks


Regards


Salman Dinani



On Tue, Feb 14, 2012 at 6:10 PM, Ettus Research Support
<support at ettus.com>wrote:

> **
> On 14/02/12 02:38 AM, salman dinani wrote:
>
> Hi,
>
>  I have attached a picture which shows my problem that even I set the
> decimation factor to 1, in order to get the sample rate of 100 MSPS, but
> the device value remains at 4, which means :
>
>  RF_Font_end_Sample_rate= 100e6/4=25e6,
>
>  which means that only waves up to Max 12 MHZ can be sampled..
>
>  we have to buy more daughter boards for our application which require
> 100MSPS and BW of 40MHz and we bought this product because of its these
> features but unfortunately we are not able able to use it that way.
>
>  1. Am I write in this conclusion? Am I configuring the card wrong or it
> is the inherit problem?
>  2. So than what is the use of ADC of 100 MSPS, if we can use it only up
> to 25MHz??
>
> If you need 100MSPS, you'll need to do processing in the FPGA.  Even if
> there were a 4-bits-per-sample
>   mode over the 1GiGe, there's no way you're going to be able to usefully
> *do* anything with a sample
>   stream at 100Msps into an ordinary PC computer--even and expensive
> multi-chip server.
>
> DSP systems *very often* oversample the analog bandwidth, in order to make
> filtering easier, which is
>   the main reason that signals are sampled at 100Msps.  The vast majority
> of applications out there
>   use much lower delivered bandwidth than 100Msps, which means that the
> digital filtering and
>   decimation done in the FPGA produces very good out-of-band attenuation.
>
>
>   3. It is written on application notes that Bandwidth is 40 MHZ, but why
> its is only 12MHz??
>
>  Two points here:  with the "sc8" mode, you can send 50MHz of bandwidth
> towards the host, but apparently
>   SDRu doesn't yet support the "sc8" wire mode.
>
> Further, you keep mentioning "12MHz", which leads me to believe that you
> don't understand complex sampling.
>   In a complex-sampled system, the actual bandwidth is equal to the
> sampling rate. So a sample rate of
>   25Msps delivers 25MHz of bandwidth--unlike a real-mode-only system,
> where 25Msps would only deliver
>   12.5MHz of bandwidth, due to Nyquist.  So, with the "sc8" mode, you'd be
> able to make use of the entire
>   roughly 40MHz of analog bandwidth coming out of the daughtercards.
>
> Again, though, I'll caution that actually *doing* anything with a sample
> stream arriving at 50Msps or more
>   is going to be a significant challenge if you propose to do that on a
> regular PC host computer.
>
>
>  Regards
>
>  Salman Dinani
>
> On Mon, Feb 13, 2012 at 6:10 PM, Mike McLernon <
> Mike.McLernon at mathworks.com> wrote:
>
>>  Hi Salman,
>>
>>
>>
>> The Simulink SDRu driver is built on top of the UHD driver, so any
>> limitations imposed by UHD will be encountered in Simulink.  Also, note
>> that the SDRu Receiver block supports int16 complex data outputs.  That
>> math implies a max rate of 25 Msps ( (16 bits/sample real + 16 bits/sample
>> imaginary) * 25 Msps = 800 Mbps.
>>
>>
>>
>> Mike
>>
>>
>>
>>
>>
>> *From:* usrp-users-bounces at lists.ettus.com [mailto:
>> usrp-users-bounces at lists.ettus.com] *On Behalf Of *salman dinani
>> *Sent:* Sunday, February 12, 2012 11:01 PM
>> *To:* Ian Buckley
>> *Cc:* usrp-users at lists.ettus.com
>> *Subject:* Re: [USRP-users] SBX daughtercard bandwidth problem..!!
>>
>>
>>
>> Thanks Josh and Ian Buckley....:) So it means:
>>
>> 1. I cannot set the decimation or interpolation to "1" in simulink SDRu
>> block? or in other words i        cannot get the sampling rate of 100
>> MSPS?? :(
>>
>>
>>
>> 2. Is there any way using Simulink SDRu block??
>>
>>
>>
>> Regards
>>
>>
>>
>> Salman Dinani
>>
>> On Mon, Feb 13, 2012 at 2:06 AM, Ian Buckley <ianb at ionconcepts.com>
>> wrote:
>>
>> Clearly I needed more coffee!
>> You confused me there referring to it as a 16bit mode rather than 8bit
>> complex.
>>
>>
>> On Feb 12, 2012, at 9:47 AM, Josh Blum wrote:
>>
>> >
>> >
>> > On 02/12/2012 09:36 AM, Ian Buckley wrote:
>> >> Josh,
>> >> It's possible you didn't have your Sunday morning coffee yet, or
>>  possibly that the laws of Physics have been shattered...
>> >> but did you really intend to state that you can now receive 16bits @
>> 50Msps in UHD...because Cisco would like to offer you a job if so :-)
>> >> I'm assuming you intended to say 8bits at 50Msps.......or is it not a
>> complex sample stream in this new mode?
>> >
>> > sorry, no coffee yet, but the math seems solid
>> >
>> > 1 byte real, 1 byte imaginary = 16 bits per sample (complex baseband)
>> >
>> > 16 bits per sample * 50 Msps = 800 Mbits per second, which is well
>> > within gigE rate of 1000 Mbits per second.
>> >
>> > -Josh
>> >
>> >>
>> >> -Ian
>> >>
>> >> On Feb 11, 2012, at 10:58 PM, Josh Blum wrote:
>> >>
>> >>>
>> >>>> 1. Am i write in this conclusion??
>> >>>
>> >>> Correct.
>> >>>
>> >>> However, it is possible to increase sample rate at the expense of
>> >>> dynamic range. On the UHD master, you can receive 50Msps with 16 bit
>> >>> samples. This feature is not yet a part of the SDRu driver.
>> >>>
>> >>>> 2. If yes, than why it is written in application notes tat it has BW
>> of 40
>> >>>> MHz??
>> >>>
>> >>> This is an analog bandwidth filter. You cannot receive signals outside
>> >>> LO +/- 20Mhz due to the filter. Consider that some may tune the LO
>> away
>> >>> from the center frequency for various reasons.
>> >>>
>> >>>> 4. Why i am not able to set decimation or interpolation to 1.??
>> >>>
>> >>> The bandwidth bottleneck is the gigabit ethernet port. Given the
>> recent
>> >>> advent of 16 bit samples, a decimation of 2 is really the minimum.
>> >>>
>> >>> This is a user mod to burst 100Msps packets, which you may find
>> helpful:
>> >>>
>> https://lists.gnu.org/archive/html/discuss-gnuradio/2011-03/msg00303.html
>> >>>
>> >>> -Josh
>> >>>
>> >>> _______________________________________________
>> >>> USRP-users mailing list
>> >>> USRP-users at lists.ettus.com
>> >>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
>> >>
>>
>>
>> _______________________________________________
>> USRP-users mailing list
>> USRP-users at lists.ettus.com
>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
>>
>>
>>
>
>
>
> --
> Marcus Leech, Technical Support, Ettus Researchsupport at ettus.comhttp://www.ettus.com
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/attachments/20120215/e6edb188/attachment-0002.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Sine_prj.png
Type: image/png
Size: 37518 bytes
Desc: not available
URL: <http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/attachments/20120215/e6edb188/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: sine_block.png
Type: image/png
Size: 43566 bytes
Desc: not available
URL: <http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/attachments/20120215/e6edb188/attachment-0001.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: SDRu_block.png
Type: image/png
Size: 36965 bytes
Desc: not available
URL: <http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/attachments/20120215/e6edb188/attachment-0002.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Spectrum_anlyzr.jpg
Type: image/jpeg
Size: 416231 bytes
Desc: not available
URL: <http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/attachments/20120215/e6edb188/attachment.jpg>


More information about the USRP-users mailing list