Hi Jack, You're right that when testing a wide-band device like an amplifier you have to filter out the generator harmonics in order to get a valid third-order IMD test. But when testing a narrow-band device like a receiver the harmonics are normally not an issue.
Those Coilcraft low-pass filters look very nice, especially for he price. Where can you buy them? Alan N1AL On Thu, 2011-04-14 at 16:54 -0400, Jack Smith wrote: > Alan: > > One are where a square wave RF source can get you into trouble is 2nd > and 3rd order intermodulation measurements. > > Mini-Circuits Ap Note 00008 discusses the reason why. > http://www.minicircuits.com/pages/pdfs/AN00008.pdf > > I usually use an outboard low pass filter on my HP 8657A signal > generators when making IMD measurements of high performance amplifiers, > and it's not a bad idea even with HP 8640B generators which are 60 dB+ > down on the harmonics. > > As far as a low pass filter goes, CoilCraft has some very inexpensive > ($3 range) 7th order elliptical low pass filter modules, the P7LP > series, with cutoff frequencies from 300 KHz to 500 MHz. > http://www.coilcraft.com/pdfs/lcfilt.pdf These are perfectly fine for > output filtering of a single generator, but should not be used with > multiple signals as they can themselves be intermod generators. I use > them here for various tasks including signal generator clean up. > > If anyone is interested, I have a few extra printed circuit boards with > provisions for PCB-mount BNC connectors and a P7LP filter and can make > them available at a reasonable price (PCB only, you supply the > connectors and filter modules). Usual double sided, silk screen, solder > masked boards. I have installed a socket strip in one as an experiment > so that one might just plug in the appropriate filter, but that's not a > great idea as the filter modules are not all that mechanically robust > and the extra lead inductance from the socket degrades stop band > performance at higher frequencies. I can send a photo of the board to > anyone interested. > > Jack K8ZOA > > > > > > > On 4/14/2011 2:26 PM, Alan Bloom wrote: > > Question 1: The frequency is based on the period of the square wave, in > > other words, the frequency of the fundamental. If you program a > > frequency greater than the maximum 200 MHz, software automatically > > selects the proper fundamental so that the harmonic comes out at the > > desired frequency. > > > > Question 2: I wasn't one of the XG3 designers, but my understanding is > > that the amplitude is that of the fundamental (the sine-wave portion) > > and does not include the harmonics. When you are operating on a > > harmonic, the amplitude is uncalibrated. Although theoretically the > > third harmonic is 1/3 the amplitude (-9.5 dB) and the fifth is 1/5 the > > amplitude (-14 dB) it is not very accurate because of the finite > > rise/fall times of the square wave. > > > > For most things you use a signal generator for, the harmonics don't hurt > > anything - they are filtered out by the receiver or other device under > > test. Even expensive professional signal generators typically only > > specify 30 or 40 dB harmonic suppression. > > > > Alan N1AL > > > > > > On Thu, 2011-04-14 at 04:14 -0400, John Ragle wrote: > >> The more I have glanced through the specs for this device, the more > >> puzzled I have become. I grew up on signal generators that put out a > >> (good approximation to a) sine wave. This device is described as an RF > >> "square wave" generator, which I take to mean that the output wave shape > >> is a trapezoid with very fast rise and fall times. As everyone knows, > >> the harmonic content of an ideal square wave falls off as 1/N. In other > >> words, this device is a harmonic generator, par excellence, very rich in > >> harmonic content. To get a sine wave, one has to run it through a > >> bandpass filter. In this case, the bandpass filter is the receiver to > >> which you connect it. > >> > >> So there are two questions. > >> > >> 1. What is the "frequency?" This term usually refers to a sinusoid, in > >> which case the answer is trivial. Presumably the "frequency" meant here > >> is the fundamental component of the "square" wave, but the true answer > >> depends on the shape of the on/off switching, which presumably depends > >> on the "frequency" one has dialed into the device. Is one getting more > >> or less a 1/N dependence on amplitude at all frequencies? > >> > >> 2. When one refers to the calibrated "levels" of output, is one > >> referring to the level of the square wave or of its fundamental > >> sinusoidal component? Clearly, these are related...but which is meant? > >> What about the high frequency end of the operating range, where the > >> spectrum may well depart substantially from 1/N (because the trapezoid's > >> edges are less "sharp"). > >> > >> Neither of these questions has anything to do with the "phase noise" > >> specification. > >> > >> John Ragle -- W1ZI > >> ______________________________________________________________ > >> Elecraft mailing list > >> Home: http://mailman.qth.net/mailman/listinfo/elecraft > >> Help: http://mailman.qth.net/mmfaq.htm > >> Post: mailto:[email protected] > >> > >> This list hosted by: http://www.qsl.net > >> Please help support this email list: http://www.qsl.net/donate.html > >> > > > > ______________________________________________________________ > > Elecraft mailing list > > Home: http://mailman.qth.net/mailman/listinfo/elecraft > > Help: http://mailman.qth.net/mmfaq.htm > > Post: mailto:[email protected] > > > > This list hosted by: http://www.qsl.net > > Please help support this email list: http://www.qsl.net/donate.html > > > ______________________________________________________________ > Elecraft mailing list > Home: http://mailman.qth.net/mailman/listinfo/elecraft > Help: http://mailman.qth.net/mmfaq.htm > Post: mailto:[email protected] > > This list hosted by: http://www.qsl.net > Please help support this email list: http://www.qsl.net/donate.html > ______________________________________________________________ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:[email protected] This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html
