Don’t forget there is gain in the signal paths. That first AUX input stage is
a Norton configuration amp with gain controlled by the unknown transformer
ratios. The two J-310's that sum the signals also appear to have a few dB gain.
A full analysis of the circuit is necessary to derive what the insertion gain
of this device would be. Or a measurement might be in order with gain pots at
max. It could be amplified input device noise at the output.
Lee K7TJR
-----Original Message-----
From: Topband <topband-bounces+k7tjr=msn.com@contesting.com> On Behalf Of Dave
Cuthbert
Sent: Monday, March 16, 2020 2:05 PM
To: n2ic@arrl.net
Cc: Topband <topband@contesting.com>; Guy Olinger K2AV <k2av.guy@gmail.com>
Subject: Re: Topband: Slightly OT - amplifier noise
Steve,
I performed a hand calculation of the MFJ-1025 14 MHz noise and it's (only)
9 nV/Hz^0.5, or 0.2 uV in 500 Hz. This is -120 dBm, or S-1. But you measure
S-4 noise. What is going on?
Questions to determine what is going on:
Question 1) What model is the radio?
Question 2: What bandwidth is the noise measured with?
Question 3: What band?
Question 4: What does the radio measure connected to the MFJ-1025 but with
the MFJ powered OFF?
*MFJ-1025 termination *
With the two MFJ-1025 antenna terminals left open the two JFET inputs are
terminated in 0-78 ohms depending on the gain pot settings. Note that gain pots
reduce the signal but not JFET noise and cutting the gain degrades the S/N
ratio. With the gain pots at maximum the unterminated input noise from the
resistors is (only) 1.1 nV/Hz^0.5.
*MFJ-1025 noise calculations*
Let's add up all the noise sources and derive the 14 MHz noise.
*Q4 output buffer*
Q5 voltage gain, 1
1 nV/Hz^0.5 transistor (my estimate; I can calculate it but it's low enough
to just call it 1 nV/Hz^0.5)
1.9 nV/Hz^0.5 for R8 (220 ohms)
Total input noise 2.2 nV/Hz^0.5 (the transistor noise is my estimate)
Total output noise 2.2 nV/Hz^0.5
*Q5, Q6 active combiner*
Q5 voltage gain, 1.1
R9||R26, 1.1 nV/Hz^0.5
Q5 gate, 0.6 nV/Hz^0.5
Input noise, 1.3 nV/Hz^0.5
Output noise, 1.4 nV/Hz^0.5
Q6 voltage gain, 2.3
R13, 1.9 nV/Hz^0.5
Q6 gate, 0.6 nV/Hz^0.5
Input noise, 2.0 nV/Hz^0.5
Output noise, 4.6 nV/Hz^0.5
Q5 + Q6 noise
Output noise, 4.8 nV/Hz^0.5
*Q7 phase splitter*
Q7 voltage gain, 1
Q7 gate, 0.6 nV/Hz^0.5
R13, 1.9 nV/Hz^0.5
Output noise, 2.0 nV/Hz
*Q8 AUX amp *
Q8 voltage gain, 2 (that's a guess)
R20||R27, 1.1 nV/Hz^0.5
Q8 gate, 0.6 nV/Hz^0.5
Input noise, 1.3 nV/Hz^0.5
Output noise, 2.6 nV/Hz^0.5
*Adding it up*
The noise into Q6 is the AUX amp (2.6 nV/Hz^0.5) + the phase splitter (2.0
nV/Hz) = 3.3 nV/Hz^0.5
The noise out of Q6 is X2.3 the input noise = 7.5 nV/Hz The noise out of Q5 is
4.8 nV/Hz^0.5 Adding two these together the noise into Q4 is 8.9 nV/Hz Add 1
nV/Hz^0.5 for Q1 and we have 9.0 nV/Hz
The MFJ-1025 *output noise is 9 nV/Hz^0.5*
In a 500 Hz BW this is 200 nV, or 48 dB below S-9. *This is S-1*.
The AUX-referred noise is ~1.7 nV/Hz^0.5 The MAIN-referred noise is ~8 nV/Hz^0.5
*12V Power is well decoupled*
The 12V power is well decoupled with 14 MHz PS rejection ratio of 110 dB.
This is calculated for bypass cap C7 having 10 nH lead inductance. So, I don't
see that noise on the 12V source makes a difference.
*LTSpice J310 noise model issue*
I ran sims of a J310 JFET in a simple source-follower circuit and LTSpice
reports noise that is so far off it's unusable. Instead I use a claimed noise
figure of 3 dB. If I use the specified 100 Hz noise of 10 nV/Hz^0.5 the output
noise rises enough to make S-4 noise. But, I believe the 100 Hz noise is below
the J310 1/f frequency and is much higher than the HF noise.
I think I can calculate the 1/f frequency to make sure. About 10 years ago I
designed a test setup to measure MOSFET noise and ran thru the calculations for
1/f and noise.
*References*
MFJ-1025 manual with schematic:
https://www.mfjenterprises.com/support.php?productid=MFJ-1025
J310 datasheet:
http://pdf.datasheetcatalog.com/datasheet_pdf/intersil/J308_to_J310.pdf
J310 RF amp noise figure: http://receiverforjupiter.tripod.com/rfamp.htm
Dave KH6AQ
On Mon, Mar 16, 2020 at 8:33 AM Steve London <n2icarrl@gmail.com> wrote:
> I did that - terminated in 50 ohms.
>
> 73,
> Steve, N2IC
>
> On 03/16/2020 12:27 PM, Guy Olinger K2AV wrote:
> > Depending on the exact circuit and device, to measure noise from the
> device
> > itself, don’t you have to terminate the inputs to put the normal
> impedance on
> > them? The circuits are often designed with the coax Z zero in mind.
> >
> > 73, Guy K2AV
> >
> > On Mon, Mar 16, 2020 at 12:47 PM Steve London <n2icarrl@gmail.com
> > <mailto:n2icarrl@gmail.com>> wrote:
> >
> > This has been a very interesting thread - Thanks for all the input.
> >
> > Perhaps I have set my expectations too high.
> >
> > A typical application is on 15 meters, late in the opening,
> > working
> JA's from
> > here in SW New Mexico. Absent any local QRN, the band is very quiet.
> Any local
> > noise covers the bottom layer of 5 watt JA's calling me.
> >
> > I did some more testing with the MFJ-1025. With no antennas
> connected, the box
> > raises the noise floor about 10 dB, irrespective of the Aux
> > Antenna
> Gain or the
> > Main Antenna Gain. I haven't yet tried powering from a battery,
> > to
> make sure
> > the
> > power supply isn't the source. Assuming the J310's are quiet,
> > that
> leaves the
> > 2N5109 emitter follower, or the back-to-back protection diodes.
> Might also try
> > disconnecting the RF sense circuit.
> >
> > 73,
> > Steve, N2IC
> > _________________
> > Searchable Archives: http://www.contesting.com/_topband -
> > Topband
> Reflector
> >
> > --
> > Sent via Gmail Mobile on my iPhone
> _________________
> Searchable Archives: http://www.contesting.com/_topband - Topband
> Reflector
>
_________________
Searchable Archives: http://www.contesting.com/_topband - Topband Reflector
_________________
Searchable Archives: http://www.contesting.com/_topband - Topband Reflector
|