Magnus Danielson wrote: > > Hmm, do you get a feeling that I am actually object very much to just > toss it into the processor. I think you are right. :) I suppose you are familiar with the old American adage that says that to a man with a hammer every problem looks like a nail.... :-) Each of us is more familiar with one or another technology (broadly speaking), and tend to see it as the best way to solve problems. I am not immune from this... > Nothing wrong with software, but use it wisely. Build the test-benches > as if you where doing a ASIC or full-custom design and thus also think > about each compile costing you milions of dollars and a pipe-line depth > of many months (6-9). Given the rate of compiles that sometimes I do especially when near to find the solution of a problem that bugged me for a long time, I would be bankrupted since long, should each compile cycle cost me thousands or millions of dollars, even if bogus dollars... :-) :-) > I guess I am becomming more conservative by the day. From my own and > others mistakes and succsesses. This is a privilege of becoming older and wiser :-) Cheers, Alberto I2PHD

Hi all Thank you all for you advice and suggestions re my request. At this stage it does not look like there is a simple solution of a readily available USB sound card that can be locked to a 10 MHz GPSDO reference. The constraints of portable operation with a Laptop rule out a number of solutions. I have tried the software solution using Spectrum Lab but ran into problems and perhaps this just needs more work. One of the main problems is that in working at milli-Hz binwidths the FTT word length needs to be very long to cover even a few tens of Hz range and we run into memory problems. So there is little room to have a reference frequency spaced well away from the frequency range being used. The SP DIF solution seems promising if I can generate the required input. This could perhaps be done with a product that already provides the SP DIF word output and locking that. But that could be just as hard as locking the sound-card in the first place. So at this time I think I will put some more effort into locking the sound card and let you know how I go, hi. Injection locking as suggested by some of you may be the answer. Some people asked for more details of the cloud scatter experiments which area at http://reast.asn.au/optical.php There is also a series of articles on our work in the last 3 and next issue of DUBUS. On the question of Doppler shift from clouds - this is much less than a mHz and not an issue due to the fact that we are using base band and the Doppler only applies to the audio frequency. In addition the very narrow beamwidths (around 2 degrees) mean that the possible paths are all very similar in length. Thanks again to everyone for their input. 73 Rex VK7MO

> -----Original Message----- > From: time-nuts-bounces@febo.com > [mailto:time-nuts-bounces@febo.com] On Behalf Of Rex Moncur > Sent: Tuesday, June 02, 2009 3:06 PM > To: 'Discussion of precise time and frequency measurement' > Subject: Re: [time-nuts] Sound Cards for locking to GPSDO 10 > MHz references > > Hi all > > Thank you all for you advice and suggestions re my request. > At this stage it does not look like there is a simple > solution of a readily available USB sound card that can be > locked to a 10 MHz GPSDO reference. > > The constraints of portable operation with a Laptop rule out > a number of solutions. I have tried the software solution > using Spectrum Lab but ran into problems and perhaps this > just needs more work. One of the main problems is that in > working at milli-Hz binwidths the FTT word length needs to be > very long to cover even a few tens of Hz range and we run > into memory problems. So there is little room to have a > reference frequency spaced well away from the frequency range > being used. The reference frequency can be right on top of your signal, as long as it's within the dynamic range. You can subtract it out before doing the FFT, after having determined where it is and how big it is. > Some people asked for more details of the cloud scatter > experiments which area at http://reast.asn.au/optical.php > There is also a series of articles on our work in the last 3 > and next issue of DUBUS. > >

Hi Rex, Here is a plot of my sound card. Maybe I'm just lucky with this particular sound card/computer, but the drift was only about 250 micro Hertz over a four hour period. Also for critical measurements I try to run at 200 Hz center frequency rather than 1000 Hz. Cuts the error by five. Maybe that's not practical for you modulation schema. I did not bother to calibrate the sound card before I started the test so there is about 550 micro Hertz of static error when the test starts. www.k5cm.com/soundcard.htm 73, Connie K5CM

Alberto di Bene skrev: > Magnus Danielson wrote: >> >> Hmm, do you get a feeling that I am actually object very much to just >> toss it into the processor. I think you are right. :) > > I suppose you are familiar with the old American adage that says that to > a man with > a hammer every problem looks like a nail.... :-) Yes. :o) > Each of us is more familiar with one or another technology (broadly > speaking), and tend to see it as the best way to solve problems. I am not immune > from this... This is why I try to find more tools and more approaches. >> Nothing wrong with software, but use it wisely. Build the test-benches >> as if you where doing a ASIC or full-custom design and thus also think >> about each compile costing you milions of dollars and a pipe-line >> depth of many months (6-9). > > Given the rate of compiles that sometimes I do especially when near to > find the solution of a problem that bugged me for a long time, I would be bankrupted > since long, should each compile cycle cost me thousands or millions of > dollars, even if bogus dollars... :-) :-) I think you would learn important lessons in test-benching and overall design before the "compile". >> I guess I am becomming more conservative by the day. From my own and >> others mistakes and succsesses. > > This is a privilege of becoming older and wiser :-) Whiee... I got so much to learn then! :) Cheers, Magnus

Lux, James P skrev: >> -----Original Message----- >> From: time-nuts-bounces@febo.com >> [mailto:time-nuts-bounces@febo.com] On Behalf Of Rex Moncur >> Sent: Tuesday, June 02, 2009 3:06 PM >> To: 'Discussion of precise time and frequency measurement' >> Subject: Re: [time-nuts] Sound Cards for locking to GPSDO 10 >> MHz references >> >> Hi all >> >> Thank you all for you advice and suggestions re my request. >> At this stage it does not look like there is a simple >> solution of a readily available USB sound card that can be >> locked to a 10 MHz GPSDO reference. >> >> The constraints of portable operation with a Laptop rule out >> a number of solutions. I have tried the software solution >> using Spectrum Lab but ran into problems and perhaps this >> just needs more work. One of the main problems is that in >> working at milli-Hz binwidths the FTT word length needs to be >> very long to cover even a few tens of Hz range and we run >> into memory problems. So there is little room to have a >> reference frequency spaced well away from the frequency range >> being used. > > The reference frequency can be right on top of your signal, as long as it's within the dynamic range. You can subtract it out before doing the FFT, after having determined where it is and how big it is. Considering the length of these traces, the local oscillator vs. the reference will shift around. What I would do is to ensure that the reference signal and input signal is either on very different frequencies or different channels. Then, I would in the sampling phase frequency convert the receive signal and reference signal using digital fixed NCO/quadrature oscillators (cos, sin) and do integrate and dump (synchronous dump for both receive and reference signals) processing for low pass filtering and reducing sample rate. Since the signal was fairly narrow banded, this digital receiver approach would significantly reduce the amounts of data while requiring a very reasonable amount of real-time processing. The remaining sample stream still contains the crutial information if sufficient bandwidth is maintained after the integrate and dump processing. Cheers, Magnus