RF signal generator recommendation

A lot of the 8640s have bad or failing output amplifier hybrids that are not available for repairs. The output starts to drop, and there isn't enough gain for the output leveling to work reliably. I think that is the model with the plastic cams that crumble, too.

--
Service to my country? Been there, Done that, and I\'ve got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida

I've got a couple of Marconi 2019A units. They must be good, the lab at Racal when I worked there a few years ago was full of them. They are rather heavy and bulky, though. They cover 10 kHz to 1.2 GHz. I paid

450 GBP for one of them, and 150 GBP for the other. It has a minor intermittent fault with the reference oscillator caused by one of the pins corroding, it's somewhat common. That's about the only weak point in them.

Leon

8657s are not quiet. Their phase noise sidebands are *very* disappointing. For quiet, I usually use the tracking generator of an HP 70000 spectrum analyzer.

Cheers,

Phil Hobbs

Marconi instruments / Aeroflex model 2042 is excellent but probably out of your budget.

The 203x and 205x models which I am aware of are many dBs noisier but still not bad. The 2024 is small and quite nice too but not as good as a 2042.

All of those use a high frequency synthesiser and divide it down by factors of 2 to get to lower frequencies, then filter the output to a sine wave. By dividing down, the phase noise in dBc/Hz at a given offset gets better and better down to about 21MHz, then it switches to a different mode so that frequencies below 21MHz aren't quite so quiet, (e.g. to get 10MHz they mix something around 100MHz with something at about 110MHz and filter the result.)

By contrast, the HP / Agilent ESG series are much much much noisier, because they would make 30MHz by mixing together one synth at about 1GHz with another synth at 1.03GHz, rather than by dividing down and filtering.

Chris

The second log-log plot on the page at

has an informal comparison between several HP signal generators, including an HP 8656A and 8657A. You can click on it for an expanded view.

In absolute terms, the traces are reasonably accurate at offsets below

20 kHz. They're within a couple dB of reality for most of the signal sources, except perhaps the 8662A, which is probably better than what the graph shows.

The traces aren't very meaningful at all beyond 20 kHz.

8657As aren't at the top of the heap, noise-wise, but I don't think it's fair to call them "very disappointing." They sold for about 1/5 the price of the 8662A, after all. And there's certainly worse gear out there... like the 8656A.

-- john, KE5FX

Phil Hobbs wrote:

(Reposted to forestall the usual GG top-posting flames; sorry 'bout that.)

The second log-log plot on the page at

has an informal comparison between several HP signal generators, including an HP 8656A and 8657A. You can click on it for an expanded view.

In absolute terms, the traces are reasonably accurate at offsets below

20 kHz. They're within a couple dB of reality for most of the signal sources, except perhaps the 8662A, which is probably better than what the graph shows.

The traces aren't very meaningful at all beyond 20 kHz.

8657As aren't at the top of the heap, noise-wise, but I don't think it's fair to call them "very disappointing." They sold for about 1/5 the price of the 8662A, after all. And there's certainly worse gear out there... like the 8656A.

-- john, KE5FX

One generator I have intimate knowledge with: the older Fluke 6071a. Once had to fix a palletload of them.

They have very good low-noise performance, and superb ergonomics,

WHEN THEY WORK.

Unfortunately the low-noise performance is due to a very complex and delicate delay-line clean-up PLL loop board. Lemme tell you more than you want to know about this board (1) Takes up about 19 by 20 inches, a good 1/4 of the generator's guts.

(1,5) Board is chock full of unobtainable HP VHF amplifiers, hot-carrier matched foursomes, septifilar-wound triple-balanced klein-bottles, and even harder to find parts.

(2) Has many very delicate adjustments, most of them interacting in inexplicable ways.

(3) Has about SIX very delicate adjustments that can only be made at the factory (the manual says).

(4) Has a few other adjustments that are burned into a EPROM (this was in the days before flash memorys).

(5) The loop tends to work fine most of the places, but loses lock at say 344.6 MHz thru 347.544, making it really hard to verify whether you've fixed the loop lock problem.

and

(6) The loop will lock or unlock if you touch any one of 74 test points with a ten meg scope probe, making those test points unhelpful for establishing lock., and oh

(6.5) the adjustment manuals are expensive, and apparently dictated by Paris Hilton's dumber sister to Stevie Wonder, by the looks of the language and formatting.

(7) Anything may change once you put on the covers, with their 37 screws per cover, or after a few days of it idling, or if the temperature changes.

So I can recommend this signal generator only IF you're very lucky or only need it to work over some narrow band where you can get the noise-reducer to lock.

snip

Hi,

So if you'll need such generator like the originator of the question asked in quite vague question, let say with at least -140dBc/10Khz in the whole frequency range, what did you choose ?

thank you, Vasile Surducan engineer, consultant Sandbridge Technology Inc