Rigol RF sig. gen.

dBc is dB down from the carrier level, whatever it happens to be just now.

It's amazing how much harmonic distortion even expensive RF generators can have. Like -20 dBc. They should be required to include the harmonics in the phase noise spec!

The Rigol stuff seems to be very good.

dBc is decibels below carrier level.

Op 09/30/2015 om 06:38 PM schreef John Larkin:

I think these are more aimed at tests where there will be a filter somewhere in the signal path later on. My HP E4421B also specifies harmonics at -30dB.

Harmonics have nothing to do with noise and won't even cause jitter, so lumping them together would usually be annoying to the (discerning) customer.

You're right that level of harmonics can be quite high even for expensive signal generators, yet it doesn't matter much in many cases as it is very easy to filter them out externally if your test setup works over a small fractional frequency range. Filtering them within the signal generator would be costly and difficult, due to the wider range of frequencies and also because you would want the filter after the output amplifier, so that the output amplifier could be run near its maximum output capability (where it generates harmonics), yet if the switched bank of filters is after the amplifier then it will mess up the impedance seen looking into the output port. So then you could add isolators, but you'd likely need one per octave, and this extra stuff could cost more than the rest of the sig gen. So it would be expensive and annoying to do over a wide frequency range, if you want to use the full power capability of the output amplifier. Anyway if you use the signal generator for driving the LO port of a mixer (which is non-linear anyway) then there is no point worring about odd harmonics, and in many cases not much concern about even ones either.

What is much more important is having no (or at least very low) close-in spurs, and other non-harmonic spurs. E.g. for a frac-N synth, integer boundary spurs etc. - One can't filter these out easily, as they are too close to the wanted signal, and they do cause jitter.

There is a great variation in the phase noise of expensive signal generators at low frequencies. e.g. the Agilent ESG series are quite expensive and can do fancy modulation, but the ones I have used are very noisy if you generate something like a 20MHz signal, as they do so by mixing together two much higher frequencies that differ by 20MHz so you get all of the noise from both of these higher frequency oscillators. The Marconi signal generators (202x, 203x, 204x) on the other hand would divide down a high frequency by some large ratio, and then filter it to a sine wave. By dividing the frequency, the noise improves.

I would very much like to know what is in the "low noise tray" assembly of the Marconi (then IFR, Aeroflex, whatever) 2042. I think there is something pretty special in there, and the manual doesn't give many clues.

A nice way to generate fairly clean low frequencies without close-in spurs would be to use a DDS chip clocked by a PLL synth, but constrain the DDS frequency word to be always a power of 2, so that all of its spurs are at least harmonics of the desired frequency. It would have the advantage over a plain digital divider, that at least the low-order harmonics would be pretty small before you even filter them. It would provide roughly the same noise improvement as a digital divider, 20log10(N).

Chris

Thanks Chris, (and all others.) So -30dBc is not unusual.

I'd mostly want to use this to modulate a diode laser. Some stuff in the 100's of MHz range. And there are hyperfine transitions in Rb. Rb87 at ~6.8 GHz is used in Rb clocks. but Rb85's is at 3.03GHz. Yeah I need a bit more, but I think I can put sidebands at

1.5x GHz and let those drive the transition. (I'm not sure all the phases work out, it's this complicated CPT thing.)

(I should write Irina and ask.)

George H.

Why would a deterministic signal be included in a noise spec? Why would a spec that is mainly for "close" noise include something that is "distant?"

Maybe that's the birdy in the generator ?

Jamie