[USRP-users] Data corruption with X3x0/UBX-160 receive at 200 Msps

Michael West michael.west at ettus.com
Wed Mar 9 19:00:06 EST 2016


Hi Mike,

It does seem like the ADC issue since it seems the corruption only happens
on the 4 MSBs.  If it is the ADC issue, perhaps the best solution is to run
the data through a filter on the host side.

Regards,
Michael

On Wed, Mar 9, 2016 at 3:01 PM, Mike Williams <jmw711 at verizon.net> wrote:

> I’m seeing a sample error rate of > 5e-7 (about 5 samples in 10 million).
> The test signal only toggles the lower 8 bits and the glitches show up when
> one of bits 11:8 is set.
>
> What is the best way to send a data capture file?
>
> Mike
>
> *From:* Michael West via USRP-users <usrp-users at lists.ettus.com>
> *Sent:* Tuesday, March 08, 2016 6:30 PM
> *To:* Marcus Müller <marcus.mueller at ettus.com>
> *Cc:* USRP-users at lists.ettus.com <usrp-users at lists.ettus.com>
> *Subject:* Re: [USRP-users] Data corruption with X3x0/UBX-160 receive at
> 200 Msps
>
> Hi Mike,
>
> I spoke with some of my colleagues and was told that the X300/X310 ADC has
> a known issue on the 4 MSBs that results in a bit error rate of 1e-9.  I'm
> not sure whether or not it has anything to do with what you are
> experiencing, especially because your error rate is much higher, but I
> thought you should know.  If the majority of the bit errors you see are in
> the 4 MSBs, it may be a cause.
>
> Regards,
> Michael
>
> On Tue, Mar 8, 2016 at 12:57 PM, Marcus Müller <marcus.mueller at ettus.com>
> wrote:
>
>> Hi Mike,
>>
>> can you get us an upload of the samples somewhere?
>>
>> Short, rough calculation regarding probability of the gaussian
>> noise-induced bitflip:
>>
>> At a thermal noise floor of -174dBm/Hz, you'd end up with
>> -174dBm/Hz*(80MHz)+NF[dB] noise power per I and Q.
>> According to our measurements[1, p 44], for 20 dB gain, depending on
>> frequency, NF is ca 10dB.
>>
>> So, noise power becomes (-174+79+10) dBm = -85dBm, which yields in total
>> an SNR of 25 dB, i.e. the noise variance is -25dB of the tone power:
>> [image: $\sigma^2 =10^{\frac{-25}{20}}\quad
>> \frac1{2\pi}\int\limits_{-\pi}^{\pi}\,A^2\sin^2(t)\,dt$]
>> [image: $= 10^{-1.25} \quad \frac12A^2 $]
>> [image: $\approx 0.028\,A^2 $], i.e.
>> [image: $\sigma \approx 0.168\, A$].
>> The Q-function at [image: $\frac1{0.168}\approx 5.96$] is about [image:
>> $1.3407\cdot10^{−9}$], i.e. the probability that noise is higher than
>> signal amplitude is about 1 in 745 million, if we only have thermal noise
>> in the receiver, and not the transmitter. 745 Million is not quite the
>> probability you see, but it's in the same ball park, at least! Notice that
>> change in SNR can have quite a significant effect on the "outlier"
>> probability here.
>>
>> Thus, looking at the samples would not only be interesting from a
>> debugging, but also from a SNR estimation point of view :)
>>
>> Cheers,
>> Marcus
>>
>> [1]
>> http://files.ettus.com/performance_data/ubx/UBX-without-UHD-corrections.pdf
>>
>>
>> On 03/08/2016 08:57 PM, Mike Williams via USRP-users wrote:
>>
>> Hi Michael,
>> It's not clipping.  I'm testing with a -60 dBm tone from a signal
>> generator and a gain of 20.  The peak I/Q values out of the X3x0 are < 256
>> except for when this data corruption occurs about every 1 to 10 million
>> samples.   It looks like a bit flip in either the I or Q samples.
>> For example, the I samples in one instance are:  0x0054, 0x00AB, 0x028C
>> (bit 9 is a 1 but it should be 0) , 0x007B.
>> Thanks,
>> Mike
>>
>> On 03/07/16, Michael Westmailto:michael.west at ettus.com
>> <michael.west at ettus.com> wrote:
>>
>> Hi Mike,
>>
>> The errors could also be due to clipping.  Your gain settings could be
>> too high.
>>
>> Regards,
>> Michael
>>
>> On Mon, Mar 7, 2016 at 7:42 AM, Mike Williams via USRP-users <
>> usrp-users at lists.ettus.com> wrote:
>>
>>> We are using an X300/UBX-160 programmed UHD-3.9.2 to record data at 200
>>> Msps using the rx_samples_to_file example program.  We are seeing data
>>> corruption that appears to be due to bit errors in some I or Q samples.
>>> The self cal of the adc capture and transfer delays works fine so maybe
>>> this is a timing error elsewhere in the FPGA?
>>>
>>> We've tried an X310 as well and have the same problem.
>>>
>>> Mike
>>>
>>> _______________________________________________
>>> USRP-users mailing list
>>> USRP-users at lists.ettus.com
>>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
>>>
>>>
>>
>>
>> _______________________________________________
>> USRP-users mailing listUSRP-users at lists.ettus.comhttp://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
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/attachments/20160309/7f895128/attachment-0002.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: tblatex-12.png
Type: image/png
Size: 1240 bytes
Desc: not available
URL: <http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/attachments/20160309/7f895128/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: tblatex-27.png
Type: image/png
Size: 2299 bytes
Desc: not available
URL: <http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/attachments/20160309/7f895128/attachment-0001.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: tblatex-21.png
Type: image/png
Size: 1101 bytes
Desc: not available
URL: <http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/attachments/20160309/7f895128/attachment-0002.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: tblatex-25.png
Type: image/png
Size: 1121 bytes
Desc: not available
URL: <http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/attachments/20160309/7f895128/attachment-0003.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: tblatex-19.png
Type: image/png
Size: 1143 bytes
Desc: not available
URL: <http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/attachments/20160309/7f895128/attachment-0004.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: tblatex-18.png
Type: image/png
Size: 1064 bytes
Desc: not available
URL: <http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/attachments/20160309/7f895128/attachment-0005.png>


More information about the USRP-users mailing list