Low noise IF3601/02 don't meet specs

Sure, but who said the FFT was scope based :-)

The analyzer I use is a nice HP3563A. BTW all the measurements done, either with the conventional HP3585B spectrum analyzer, with my true RMS HP3456A (integral measurement done through a calibrated (measured) HP-LP filter to keep only a spur free region of the spectrum), or even with the DSA601 (either FFT or true RMS), agree within better than 5%.

What could we ask more? Oh yes, I also have the nasty habit to do sanity checks of my measurement chain and I can measure Johnson noise of my 50R and 75R BNC terminations within better than a couple of percents WC of the theoretical value. Also the following x1000 preamplifier measured input noise agrees very well with the opamp datasheet values. I guess this should clear the remaining doubts, don't you think? :-)

I've not done it but I'm pretty sure that if I put the termination at the end of a cable and put it outside I can read the outside temperature. Hey, that's a great setup for an ambiant thermometer. Now I've a meeting 100km from here but I'll do that when back this evening :-)

Those IF3602 measurements look wrong and the ratios don't work out. Maybe something to do with what the unused half is doing?

The IF3601 ratios look promising though. The gm should increase in proportion to the square root of the Id current ratio, and the einv should reduce with the reciprocal of the square root of gm. For a current ratio of 2.1:1 then that would be x1.5 and x0.83 respectively.

Is there any chance that the JFET's dVds/dVgs during the noise measurement is higher than your measured Gain, (presumably measured at the input of the capacitive attenuator, and calculated to get dVds/dVgs)?

If not.... The ratios working out reasonably well suggests that the source of the excess noise could be at the input, either in your external circuitry or (say) an actual apparent 1R5 in series with the Gate, inside the JFET.

The 'spec' gives you little in the way of production spread information to calculate with..... the Idss is only given as "30mA Min", which means nothing because the source chip is 50mA to 1000mA anyway, and the 300pv/rtHz noise is only given as a Typical figure, so those apparently noisy devices you have still meet the 'spec'.... and so on.

You can get anything through a nice wide barn door. :)

"Tony Williams" a écrit dans le message de news: snipped-for-privacy@ledelec.demon.co.uk...

Don't think so. I had no answer from Interfet on this point but it seems from the process datasheet that the 3602 is a dual die and it's left unconnected on the setup.

No, the final setup has the FET cascoded.

Well I'd hate to think so because that would indicate sloppy design from Interfet. And for the IF3602 that'd be nearly 6R. Arghhh...

or Toshiba 2SK369 0.71 nV/sqrtHz @ 10 mA, 10V, 1KHz, TO-92 They are cheap.

BTW the 2sk369 data sheet is not clear if the pinout is top or bottom view. At least DC-wise, there seems to be no difference.

I think, I'll get a few of those IF3601 for my sideband noise measurement rig..

regards, Gerhard