Topband: Hi Gain Preamps, Noise Figure Comparison (KD9SV 994 x5, W1FB, W7IUV preamps)

[Don Kirk]

Hi JC,

I did not publish any absolute Noise Figure values, my measurements that I
published are all just differences in Noise Figure.

It actually is much easier to measure the difference in Noise Figure
between preamps that have such high gain (approximately 40dB in this case)
since the noise floor generated by a 50 ohm resistor at room temperature is
easily measurable (at least indirectly) after 40 dB of gain (It's -106.89
dBm for a 40 dB preamp using a bandwidth of 500 hertz if the preamp adds no
additional noise).  Even if you can't accurately measure -106.89 dBm level
signals, you just need to be able to be repeatable in your measurements
between preamps (accuracy and repeatability are not the same thing), and
your system must be able to detect differences in output signal from the
preamp in units of dB (easy to do with an SDR receiver at these levels
using attenuators to verify changes in dB on the SDR display).  In my case
all of my measurements were made with my Kenwood TS-180s operating on 160
meters, and then making measurements via the wideband IF output port of the
Kenwood TS-180s that's connected to an SDR receiver (the HDSDR software
also allowed me to do long term averaging, and it even allows you to
display power spectrum density).

My second method of measurement used the S meter on my TS-180s, but just as
an indicator (actually digitized the voltage supplied to the S meter).  I
recorded the signal strength on the S meter that the preamp generated when
the preamp input was terminated with the 50 ohm resistor, then I went back
and adjusted my signal generator connected directly to my TS-180s via
attenuators until it provided the same S meter reading.  Using the signal
generator and precision attenuators I was able to calculate the voltage
(and therefore power) into the TS-180s that the S meter recorded when
driven by the preamp.  I used this technique to compare Noise Figure
between preamps, and I made numerous measurements over a 6 day period and
then averaged the data to reduce my variation using this method.

I did in fact test the use of #31 mix cores (10 to 14 turns) on various
lines (power supply lines, 3 foot short feedline between preamp and
receiver input, etc. and saw no difference in signal levels.  I even
eliminated the Kenwood power supply, and fed the TS-180s with a lab grade
linear power supply to eliminate slight changes in supply voltage over time.

P.S. the Kenwood TS-180s is one of the better radios made in regards to
noise floor (measured -139 dBm by Sherwood).

It would certainly be much easier to make these measurements using a
spectrum analyzer, and I probably will acquire one of the lower cost Rigol
spectrum analyzers in the near future.  In the meantime I would love to
have someone independently make measurements comparing these amplifiers
that either supports or disproves my measured values.

73,
Don (wd8dsb)


On Tue, Nov 8, 2016 at 7:24 PM, JC <n4is at comcast.net> wrote:

> Hi Don
>
> Measuring noise figure below 1 db is a very complicated work.
>
> First you need to remove all common mode noise, your noise floor should be
> better than  -135 dBm at least with a 40 db preamp connected to the radio
> input and a 50 ohms load.
>
> You need chokes everywhere and most important shield on 12V, RF in and RF
> out. Check the AM band you hear carrier coming in , noise is also coming
> in.
>
>  Second the NF is very different on 1.8 MHz , measurement near 10 Mhz, does
> not tell you the right NF on low bands.
>
> In order to have Noise Figure Uncertainty near .2db it is necessary a good
> quality lab equipment, calibrated and two hours per measurement for each NF
> reading. This assuming the temperature does no change during the test.
>
>
> Your Noise Figure Uncertainty based on your video could be 5db or more.
>
>
> 73's
> JC
>
> N4IS
>
>