[USRP-users] Again performance issues B210, while BladeRF performs fine

Ralph A. Schmid, dk5ras ralph at schmid.xxx
Wed Apr 1 07:57:42 EDT 2015


I will give it a try later this evening, however from the first test in the
lunch break I found that DVB-T still produces lots of U, while DVB-T2 is
stable. I did not yet test DVB-S2, but this will be one of the next
projects...

Thanks a lot for all the assistance! It is amazing what you created with all
those TV mode projects!!

Ralph.

> -----Original Message-----
> From: Ron Economos [mailto:w6rz at comcast.net]
> Sent: Tuesday, March 31, 2015 12:50 PM
> To: Ralph A. Schmid, dk5ras
> Cc: 'usrp-users'
> Subject: Re: [USRP-users] Again performance issues B210, while BladeRF
> performs fine
> 
> For the DVB-T transmit flow, delete the Rational Resampler block and set
the
> sample rate to (8000000.0 * 8) / 7 for an
> 8 MHz wide channel. For other channel widths, the sample rate formula is:
> 
> (8000000.0 * channel width in MHz) / 7
> 
> On my setup, I'm getting better performance with the B210.
> I was just doing some testing with the DVB-S2 flow, and I was able to do
36
> Msps. With bladeRF, I can't go over 24 Msps.
> 
> It all seems to be related to the USB3.0 controller. On the B210, I'm
using a
> VL80X based controller, which seems pretty fast.
> 
> On bladeRF, the VL80X controller doesn't work well at all and causes hard
> crashes (where you have to power down to recover). I have to use the
built-
> in NEC uPD720200 controller instead. It's slower, but solid as a rock.
> 
> Ron
> 
> On 03/31/2015 03:09 AM, Ralph A. Schmid, dk5ras wrote:
> > OK, with your tips the signal is less broken, but still not stable; I
> > already was using the vv009-4kfft.grc anyway. CPU is running at around
> > 80%, not critical, but who knows?! I will try tweaking a bit if I can
> > get it more stable. This is a very interesting case for fine tuning,
> > as it is right at the edge small changes get visible at once.
> >
> > Edit before sending: Now I have a solid signal, after assigning 4
> > cores to the VM. Just the question remains, why does the B210 use the
> > given resources so bad, compared to the BladeRF? And still I am not
> > sure if I'd better use four virtual cores, or two virtual cores with
> > two "virtual-virtual" ones each :) At first sight both variants behave
> identical.
> >
> > Reception of the signal needs to wait until the DVB-T2 receiver is
> > here, I expect it by next week. The DVB-T package I did not get to
> > run, it transmits, but only some garble. The gnuradio fft view looks
> > perfect, but the real spectrum is completely different, both with
> > BladeRF and B210 it is the same crap coming out of the antenna. Maybe I
> missed something...
> >
> > Ralph.
> >
> >> -----Original Message-----
> >> From: Ron Economos [mailto:w6rz at comcast.net]
> >> Sent: Tuesday, March 31, 2015 8:55 AM
> >> To: Ralph A. Schmid, dk5ras
> >> Cc: 'usrp-users'
> >> Subject: Re: [USRP-users] Again performance issues B210, while
> >> BladeRF performs fine
> >>
> >> I'm using the UHD sink.
> >>
> >> BTW, the AD9361 has built-in sin(x)/x correction, so you should set
> >> that option to "Off" in the Pilot Generator and IFFT block.
> >> It will still work fine either way, but the correct setting is "Off"
> >> for B210 and "On" for bladeRF.
> >>
> >> Also, if you still end up with performance issues, the
> >> vv009-4kfft.grc
> > test flow
> >> graph requires quite a bit less CPU performance than the
> >> vv003-cr23.grc
> > test
> >> flow graph.
> >>
> >> Ron
> >>
> >> On 03/30/2015 11:13 PM, Ralph A. Schmid, dk5ras wrote:
> >>> Hey, thanks a lot, I will give this a try later this day! What do
> >>> you recommend, the osmocom sink, or the UHD sink?
> >>>
> >>> Ralph.
> >>>
> >>>
> >>>> -----Original Message-----
> >>>> From: USRP-users [mailto:usrp-users-bounces at lists.ettus.com] On
> >>>> Behalf Of Ron Economos via USRP-users
> >>>> Sent: Tuesday, March 31, 2015 12:52 AM
> >>>> To: usrp-users at lists.ettus.com
> >>>> Subject: [USRP-users] Again performance issues B210, while BladeRF
> >>>> performs fine
> >>>>
> >>>> Just joined the mailing list to respond to this question. I'm the
> >>> developer of
> >>>> gr-dvbt2 and coincidently, I've just bought a B210.
> >>>>
> >>>> The fix for the underruns is to add some buffering. In the USRP
> >>>> Sink block Device Address field, add the following:
> >>>>
> >>>> "send_frame_size=65536,num_send_frames=128"
> >>>>
> >>>> This works well here with a VL80X based USB3.0 controller.
> >>>>
> >>>> Ron
> >>>>
> >






More information about the USRP-users mailing list