[Fwd: [Fwd: Re: [Microwave] Building a noise source/head]]

Fri, 19 Sep 2003 12:08:04 -0500

Dr. Gerald N. Johnson, electrical engineer wrote:
> 
> Subject: Re: [Microwave] Building a noise source/head
> Date: Thu, 18 Sep 2003 10:34:45 -0500
> From: "Dr. Gerald N. Johnson, electrical engineer" <[email protected]>
> Organization: Gerald N. Johnson
> To: [email protected]
> CC: Scott Townley <[email protected]>, [email protected],
>      [email protected], [email protected],
>      [email protected]
> References: <OF97DF692E.C939B1D1-ONC1256DA5.002891C3-C1256DA5.00298501@diamond.philips.com>
> 
> In general, the output of a noise source is too low to read with a
> spectrum analyzer or volt meter.
> 
> Its changes in impedance of the noise generator that change the gain of
> the stage under test, much worse with FET preamps than with bipolar and
> tube or bipolar with lots of inverse feedback. It would be a better
> initial test to compare a preamp whose gain wasn't affected
> significantly (such as the 2 db NF at VHF of the transformer fed back
> bipolar transistor) by the source impedance. Then with that initial
> assessment of the noise level of the noise source, try a very low noise
> but source impedance sensitive preamp like that of a GaAsFET. That's the
> whole reason for the 20 dB precision attenuator. It drastically reduces
> the impedance change of the noise source. And the diode noise generator
> usually easily works at 25 dB excess noise to make a decent test.
> 
> Even various versions and models of HP noise heads are not perfect with
> respect to the changes in impedance and those who bring them to VHF
> conferences, take care to keep the same noise head with proven
> performance for those tests in order to minimize the effects of
> different noise heads. Not just the model, but the proven NOISE HEAD.
> 
> Because FET preamps give their best NF when their gate source impedance
> is significantly mismatched, their consistent measurement is very
> difficult. Calibration with hot/cold is more difficult because of the
> low noise difference between practical temperatures and the difficult of
> matching the hot and cold load's impedance to better than 40 dB return
> loss. Even using the same load for both temperatures (then demands gain
> stability in the preamp over time) can't be truly relied on because of
> the likely change in impedance with temperature.
> 
> Often the unknown errors of NF from impedance change are greater than
> the NF measured. And a little regeneration in the preamp under test can
> make the effects far greater.
> 
> The most practical way to calibrate the home made NF is to compare a
> bunch of preamps at different frequencies using the BEST of the
> professional sources and to adjust the ENR settings of the home made NF
> equipment at each frequency to agree with the professional source. This
> is one of the functions at US VHF conferences, like the Central States
> VHF conference. Not only is there contest, there is comparison of
> portable preamps with NF equipment for calibration of the home NF.
> 
> Comparing multiple professional sources may lead to more confusion
> though if there are enough, some statistical pattern may emerge, if ALL
> the professional sources have good impedance stability. Otherwise the
> data may end like the man with two watches, never sure of the time.
> 
> 73, Jerry, K0CQ
> 
> --
> Entire content copyright Dr. Gerald N. Johnson, electrical engineer.
> Reproduction by permission only.