[Discuss-gnuradio] Short pulses and ringing
Many of you know that I'm experimenting with pulsar detection and monitoring tools for Gnu Radio, as part of a larger radio-astronomy toolkit for Gnu Radio. Pulsars produce narrow pulses with quite fast rise times, with pulse repetition rates of less than 1Hz up to tens or hundreds of Hz. The pulses tend to have fairly-fast rise times. My pulsar profile monitor uses the audio system to monitor the detector output of a USRP-based RF front end. That front-end does broad spectral analysis, and also produces an 8Khz sampled audio signal for use by the pulsar profile monitor, which uses epoch folding. In order to test the pulsar profile monitor, I have a small program that produces short pulses, buried in random noise (just like you'd find with a real pulsar). What I'm finding is that the displayed pulse profiles bear very little resemblance to the generated profile, with ringing and undershoot. What I need to understand is whether it's the audio subsystem that's ringing, or the FIR filter that I use in the pulsar profile monitor. How to FIR filters react to fast-rise-time pulses? Do they ring like analog filters? ___ Discuss-gnuradio mailing list Discuss-gnuradio@gnu.org http://lists.gnu.org/mailman/listinfo/discuss-gnuradio
Re: [Discuss-gnuradio] SCHED_FIFO and NPTL
On Wednesday 08 March 2006 14:33, Eric Blossom wrote: > I was burned by ext3 with regard to streaming disk throughput. > I ended up remounting the relevant filesystem as ext2 to avoid the > problem. I have no info regarding CPU and/or preemption issues during > journal posting. If we never go to the disk, it might not matter at > all. Our box here is a Dell PowerEdge 2850 with the PCIe PERC 4e/Di U320 hardware RAID controller. The data drive is a RAID0 of 4 146GB 10K RPM U320 drives, and sustains, with ext3, over 140MB/s. We have successfully recorded 2 hours at 32MB/s without any known issues. Need to test further, though. > I installed the Reiser FS on my new laptop. I'll let you know if I > see anything hinky. Haven't tried XFS. I suspect it's OK. > SGI built truely *monstrous* media servers using it. Be careful which XFS version you use. The 2.6 kernel-supplied XFS code has some problems, and you really need the latest SGI code instead. The CentOS 4 group is working on an XFS-enabled kernel with the newer code. Not sure where SuSE and Ubuntu/Debian stand. -- Lamar Owen Director of Information Technology Pisgah Astronomical Research Institute 1 PARI Drive Rosman, NC 28772 (828)862-5554 www.pari.edu ___ Discuss-gnuradio mailing list Discuss-gnuradio@gnu.org http://lists.gnu.org/mailman/listinfo/discuss-gnuradio
Re: [Discuss-gnuradio] frequency tuning word quantization
On Wednesday 08 March 2006 11:35, Eric Blossom wrote: > Here are a couple of specific suggestions: I'll second the suggestions, and add a third for those who want to do FPGA development: 'Digital Signal Processing with Field Programmable Gate Arrays' 2nd Ed, Uwe Meye-Baese. Published by Springer, 2004. ISBN 3-540-21119-5. -- Lamar Owen Director of Information Technology Pisgah Astronomical Research Institute 1 PARI Drive Rosman, NC 28772 (828)862-5554 www.pari.edu ___ Discuss-gnuradio mailing list Discuss-gnuradio@gnu.org http://lists.gnu.org/mailman/listinfo/discuss-gnuradio
Re: [Discuss-gnuradio] Short pulses and ringing
On Thursday 09 March 2006 08:45, Marcus Leech wrote: > Pulsars produce narrow pulses with quite fast rise times, with pulse > repetition rates of less than 1Hz > up to tens or hundreds of Hz. The pulses tend to have fairly-fast > rise times. > How to FIR filters react to fast-rise-time pulses? Do they ring like > analog filters? The only filter type I know off that has the step response you want is the Bessel, which is doable with an FIR filter with the right coefficients. All the others (Butterworth, etc) ring to one extent or another. The tradeoff is the Bessel's incredibly slow rolloff and absence of flatness in the passband. That said, sound cards with inputs that go below 20Hz are rare as hen's teeth, and the rolloff that low is likely to be severe, with significant group delay and ringing of very low frequency pulses. This is why I previously mentioned not trying to use the audio output; sound cards' audio output and input linearity in that frequency range stinks. A USRP with an LFRX would be a better solution, but even then if there is a series cap anywhere the lows are going to be horrible. But Matt says DC-30MHz, so maybe there are no series caps. -- Lamar Owen Director of Information Technology Pisgah Astronomical Research Institute 1 PARI Drive Rosman, NC 28772 (828)862-5554 www.pari.edu ___ Discuss-gnuradio mailing list Discuss-gnuradio@gnu.org http://lists.gnu.org/mailman/listinfo/discuss-gnuradio
Re: [Discuss-gnuradio] SCHED_FIFO and NPTL
On Thu, Mar 09, 2006 at 10:55:55AM -0500, Lamar Owen wrote: > On Wednesday 08 March 2006 14:33, Eric Blossom wrote: > > I was burned by ext3 with regard to streaming disk throughput. > > I ended up remounting the relevant filesystem as ext2 to avoid the > > problem. I have no info regarding CPU and/or preemption issues during > > journal posting. If we never go to the disk, it might not matter at > > all. > > Our box here is a Dell PowerEdge 2850 with the PCIe PERC 4e/Di U320 hardware > RAID controller. The data drive is a RAID0 of 4 146GB 10K RPM U320 drives, > and sustains, with ext3, over 140MB/s. We have successfully recorded 2 hours > at 32MB/s without any known issues. Need to test further, though. That's good throughput. My tests were about three years ago, so perhaps the problem's been addressed in the meanwhile. > > I installed the Reiser FS on my new laptop. I'll let you know if I > > see anything hinky. Haven't tried XFS. I suspect it's OK. > > SGI built truely *monstrous* media servers using it. > > Be careful which XFS version you use. The 2.6 kernel-supplied XFS code has > some problems, and you really need the latest SGI code instead. The CentOS 4 > group is working on an XFS-enabled kernel with the newer code. Not sure > where SuSE and Ubuntu/Debian stand. Thanks for the advise ;) Eric ___ Discuss-gnuradio mailing list Discuss-gnuradio@gnu.org http://lists.gnu.org/mailman/listinfo/discuss-gnuradio
[Discuss-gnuradio] flex board updates
If you have or will have a flex400 or flex2400 board, you should upgrade your gr-usrp to the latest CVS (as of a few minutes ago). It will now work with the new Flex2400, and I made some performance improvements for the Flex400. I don't know if sourceforge still has a delay for anonymous CVS, but if so, you might want to wait a few hours. Matt ___ Discuss-gnuradio mailing list Discuss-gnuradio@gnu.org http://lists.gnu.org/mailman/listinfo/discuss-gnuradio
Re: [Discuss-gnuradio] Short pulses and ringing
Lamar Owen wrote: > On Thursday 09 March 2006 08:45, Marcus Leech wrote: > >>Pulsars produce narrow pulses with quite fast rise times, with pulse >>repetition rates of less than 1Hz >> up to tens or hundreds of Hz. The pulses tend to have fairly-fast >>rise times. > > >>How to FIR filters react to fast-rise-time pulses? Do they ring like >>analog filters? > > > The only filter type I know off that has the step response you want is the > Bessel, which is doable with an FIR filter with the right coefficients. All > the others (Butterworth, etc) ring to one extent or another. The tradeoff is > the Bessel's incredibly slow rolloff and absence of flatness in the passband. > > That said, sound cards with inputs that go below 20Hz are rare as hen's > teeth, > and the rolloff that low is likely to be severe, with significant group delay > and ringing of very low frequency pulses. This is why I previously mentioned > not trying to use the audio output; sound cards' audio output and input > linearity in that frequency range stinks. A USRP with an LFRX would be a > better solution, but even then if there is a series cap anywhere the lows are > going to be horrible. But Matt says DC-30MHz, so maybe there are no series > caps. There are no series caps in the LFRX. You do need to turn off the automatic DC removal in the usrp fpga. (can be done from python) You also have to take care that DC is really DC. (I added a big series cap (2200 uF) for my application (25 Hz - 40 MHz) because of the DC-offset from my signal source) Greetings, Martin ___ Discuss-gnuradio mailing list Discuss-gnuradio@gnu.org http://lists.gnu.org/mailman/listinfo/discuss-gnuradio
[Discuss-gnuradio] USRP Update -- March 9th, 2006
Design Files The schematics and layouts for all the new daughterboards, including the LFTX, LFRX, Flex400, and Flex2400 are now available on the download page of ettus.com. If you find any missing files, please let us know. -- Daughterboard News -- The Flex2400 board, which was released this week covers 2300-2700 MHz. There is a filter on board which covers the ISM band (2400-2483 MHz), but it can be bypassed easily, giving coverage to the full 2300-2700 MHz band. The additional bands covered include satellite, military, ham, and other fixed wireless bands. Due to a lack of demand at this time, there will not be a Flex1200 board released for the 1.2 GHz Ham band. Instead, those interested in that band can purchase the Flex2400 board and modify it. Only a few passive components need to be changed to cover 1150 MHz to 1350 MHz. Please contact us for more details on how to do this if you are interested. The next daughterboard to be released will be the Flex900. It will cover 800-1000 MHz, which includes an ISM band and cell phones. With easy modifications, it will cover the 1.6 to 2.0 GHz range, giving coverage of more cell phones (PCS in the US), satellites, and other fixed wireless bands. Expect to see the Flex900 in about 1 month. The LFTX is now in stock, as are all other released daughterboards. There are no specific plans at this time for daughterboards covering other bands. If there is a band you are interested in which is not covered (like 5 GHz), please let us know. Antenna News Up until now, the YAGI900, YAGI1200 and PATCH2400 PCB antennas have not come with coax or connectors, and so were not usable "out of the box". As of now, these antennas will come with 1 foot of hardline coax (.141") with an SMA connector on it. Nothing additional will be necessary except some solder. The prices have been adjusted to account for this. If you have already purchased these antennas and would like to purchase the coax alone, it is now available from the standard ordering page. Mailing List As always, please let us know if you want: To be taken off this list Your email address to be changed A colleague to be subscribed Thanks, Matt Ettus Ettus Research LLC [EMAIL PROTECTED] 650-814-8943 ___ Discuss-gnuradio mailing list Discuss-gnuradio@gnu.org http://lists.gnu.org/mailman/listinfo/discuss-gnuradio
