Hi Daniel, On 12/25/2015 09:51 PM, Daniel Pocock wrote: > > On 25/12/15 19:44, Marcus D. Leech wrote: >> On 12/25/2015 09:18 AM, Ralph A. Schmid, dk5ras wrote: >>> Hi, >>> >>> First of all, the USRP radios are kind of experimental radios, using >>> them for real ham radio operation on antennas will require filters and >>> PAs. "Out of the box" it will only be some proof of concept when you >>> create a ham radio application with it. >>> >>> All 50 ohms, no limitations other than upper and lower border, >>> regarding frequency. But connecting a roof-top antenna will most >>> likely not work, due to the lack of preselection, you will receive >>> lots of images and other garble. >>> >>> Ralph, dk5ras. >> I think it boils down to "good RF hygiene is still important". Even if >> you had massive-amounts of excess dynamic range in the digital >> domain, you don't necessarily have that headroom in all the analog goo >> that is in front of all the digital "stuff". >> >> When I'm doing RF projects with SDRs where I care about the reliability >> and consistency of the results, I pre-select with either commercial >> or DIY filters. >> >> In the TX direction (which I don't do much of), you'll almost always >> need at least harmonic filtering--there isn't a synthesized superhet >> TX chain on the planet that doesn't have *some* amount of unwanted >> harmonic and spur energy, and if you're going to be amplifying it, >> cleanliness of what you're amplifying will not only reduce the amount >> of wasted power, it will eliminate complaints from other >> users of the spectrum, and possibly eliminate unwanted visits from the >> authorities. >> >> Purpose-built radios have already taken care of this, at least in the TX >> direction, by using output filters that nicely knock-down unwanted >> mixing products and other spurs. That's nearly impossible to do in >> a "general, not-designed-for-specific-application" radio like your >> typical SDR. >> > > On its own the SDR device may not be ideal, but when used in conjunction > with some external input filter or power amp, can these issues be > managed by the average (or slightly above average) hobbyist? I do think so, yes. Imagine this situation: You want to operate on X MHz , so you either go out and buy a filter that has a cutoff frequency between X and 2*X, or you go ahead and build an LC (if the frequency is still in a range that easily allows you to do so) with four blobs of solder and two discrete components. > I fully understand no one filter or amplifier will support the full > spectrum from 70MHz to 6GHz, that is OK. > > It is not just harmonics that came to my mind when asking about this > before, there is also the risk of software bugs. There is no such thing > as perfect software, ... or perfect hardware ... > sooner or later somebody may configure a flow graph > that transmits in a band they didn't intend Well, good news is that by design, only hardware imperfections can lead to you transmitting out of the physical sampling rate-wide band around your local oscillator. Basically, these imperfections manifest in harmonics and clock intermodulations, so it's not completely surprising where they'd end up. Of course, a bug in UHD might occur (we're not producing perfect software -- we're just trying as hard as possible to do so, have code review and extensive product testing in place etc), which might misconfigure your LO. Of course, you can also just send a signal that doesn't conform to your regulations in your band. > and it would be nice if the > power amp or other components were attenuating such things or even > better, alerting the user. That sounds pretty much impossible without "listening" on your own transmission. The USRP will allow you to do anything that's power-limited and band-limited with your spectrum, and that includes the signals you're allowed to transmit by regulation, e.g. a 5kHz narrowband FM speech signal, and such you're not allowed (e.g. a 10MHz OFDM signal), all in the same band. Detecting the difference between the two would require the amp to contain a scary lot of signal intelligence.
However, you'll quickly learn what "normal" and hence, legal, signals look like when observed for example with a very simple visualization flow graph. Since the USRPs are full duplex you can listen in on your transmission (of course, you'd want to avoid frying your RX amplifier -- these things can extract GPS out of background noise -- by pushing in PA output; but that's simply a matter of turning off you when you RX. That way, you can quickly verify legal operation. Regards, Marcus _______________________________________________ Discuss-gnuradio mailing list Discuss-gnuradio@gnu.org https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
