How to measure RF carriers

Hello Klem,

A spectrum analyzer is the usual instrument here. Unless you are chasing non-linearities in large amplifiers.

We're designing some wideband RF power amps, and we plan to do all the testing with sampling oscilloscopes. We're concerned with time-domain stuff like phase shift versus time, rf envelope amplitude droop over long bursts, things that would be hard to measure in the frequency domain.

John

Hello John,

But AFAIR you guys have some >10GHz scopes. The stuff with a high drool factor among RF engineers and where you have to talk to the bank before signing the P.O. Although, sometimes one can luck out. When I bid on an older HP 1GHz scope I was totally surprised when they called me to pay and pick it up. Turns out there was only one bid ;-)

What is the carrier frequemcy that you are working at? RF can mean a lot of different things to different people.

Dave

....... Phil

Yes!!! Radio 2 used to be on the Long Wave at 200KHz so that makes no sense at all.

I would be inclined to ask the Engineer how he does it. Either he will tell you or look fairly stupid.

DNA

A Tek 11801 or equivalent mainframe is usually under $1K on ebay, less if you know a trick. 12 or 20 GHz dual-channel sampling heads are typically about $700 or so, although TDR runs more. They are phenomenally precise instruments.

I have 8 or 10 of the mainframes by now, and maybe 20 heads, $600k or so at last catalog price!

A guy could equip a basement lab with everything needed to go into the picosecond products business for the price of a used Harley.

John

John Larkin wrote: ...

Things like that don't seem difficult at all to measure with modern frequency domain instruments like we build here. I don't know that you'd find them on ebay quite as cheap as the scopes you mentioned in a followup posting, though.

Cheers, Tom

Hello Tom,

Phase shift versus time can be tough (meaning expensive) in frequency domain if you are looking at just a couple degrees or so. But for droop you are right, we measure that with spectrum analyzers. That stuff is a serious concern when designing pulsed Dopplers for ultrasound. Spectrum analyzers just don't like to trigger that well, mostly they provide a BNC jack in the back for this and you have to concoct some circuitry to drive it (but only once).

Your engineer knows what it said. If you have no money for a spectrum analyzer then a good oscilloscope is the next affordable tool for that. Knowing the frequency is not enough if you are his boss and you don't know to answer him why you'll not buy one. You must also ask him which would be the amplitude levels and if will be much happy if the scope will also have a good FFT and an analogic bandwidtth twice than 400MHz.

greetings, Vasile

... Well, like I said, it doesn't seem all that difficult with modern frequency domain instruments like we build here. They can trigger on all sorts of things, including things that scopes can have lots of difficulty with--like triggering on energy above a particular level in a particular frequency band. If you calibrate out cable lengths properly, you can get phase accuracy to a fraction of a degree. How, with a scope, do you detect signals at -90dBc and determine that they are or are not harmonically related to the fundamental? How do you analyze phase noise versus frequency with a scope? How do you plot a color map of amplitude versus frequency versus time with a scope, and then display a trace of amplitude versus time for any particular frequency? I'm not aware of any scopes that digitize with the required spurious free dynamic range to do that as well as modern spectral analyzers, and I don't believe they come with software that does as good a job as what you get with spectral analyzers for frequency-domain applications. On the other hand, if you need really wide instantaneous bandwidth (perhaps greater than 100MHz), you probably ARE better off with a scope--for now.

In general, I'd say that if you are analyzing a signal and want to know its time-domain behaviour, like rise time, overshoot and the like, you're better off with a scope; but if you want frequency domain information like phase, harmonic content, accurate frequency information, and spurious versus frequency, and even the time progression of any of those things, you are better off with the proper frequency domain analyzer.

Cheers, Tom