[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: [Fwd: [linrad] Homebrew Receivers]
> The correct URL is: http://www.qsl.net/pa3ckr/hlmixer/
I made some experiments with the Fairchild switches but I was
not very sucessful. The reason is probably that these switches
do not work so well when the clock frequency is in the range
50 to 70MHz.
As it turns out I have difficulties to make a mixer that is
any better than +30dBm for input IP3. I have seen up to +45dBm
but not in circuits easily reproduced. They require delicate
tuning for the LO waveform and for balance.
A simple mixer with two J310 can give +35dBm if the gates are
driven from a saturated ferrite transformer (to make LO a
square wave). With "normal" drive to the gates the input IP3
is about 10 dB lower.
I have tried ATF33143, they are a good deal better but they
are awful to work with because they are too small.
I think I could replace 4xJ310 by a single MGF1302 for similar
performance but that does not seem attractive to me...
-- o --
It now looks like I will make the 144 to 70 MHz converter like
1) RF amplifier: G=6dB NF=2dB IIP3=27dBm
2) Mixer (16xJ310): G=-8dB IIP3=30dBm
3) IF amp: G=7dB IIP3=26dBm
The entire unit has G=5dB, IIP3=24dBm, NF=5dB, 1dB compression
There are several changes as compared to the schematic diagrams
on the Linrad site but the semiconductors are tha same. One
RX144 will contain 67 j-fet transistors J310 :-)
When the RX144 is used in front of RX70+RX10700+RX2500+Delta44,
the system NF is 12 dB with the Delta44 set for minimum gain.
The noise floor is lifted by 1.8dB as compared to the Delta44
with nothing connected so the dynamic range within the visible
passband is about 145dB/Hz. This corresponds to 118dB with the
standard bandwidth of 500Hz used by ARRL lab. As a comparison,
the FT1000D (borrowed from SM5FRH) measures 120dB/Hz at 20kHz
separation which corresponds to a dynamic range of 93dB in 500Hz
bandwidth. This is very far (!!!) from the BDR values published
by ARRL lab because the ARRL lab measurement is a measurement
for blocking, not the capability for receiving a weak signal
while a strong one is present.
-- o --
I have seen +40dBm for mixer IIP3 at several places. I do not know
what it takes to get such performance in a reproducible fashion.
Some day something like HJK-3H http://www.minicircuits.com/dg03-80.pdf
might become available for the frequencies I need. The article
by Ulrich Rodhe in Jan/Feb QEX suggests they do not have to
be narrow in bandwidth.
The RX144 mixer with IIP3=+30dBm and 1dB compression at +19dBm
is 5dB better than a TUF1-H (~USD 10)and similar to a SAY-1
mixer (~USD 60) if I trust the data book. This will be good enough
for my personal needs, I am not particularly worried about
third order intermodulation.
With IIP3 = +24dBm, NF=12dB at the RX144 input, a preamplifier with
NF=0.2dB and G=30dB will give a systen NF of 0.26dB. Then IIP3
will go down to -6dBm with the noise floor at -174dBm/Hz or
-147dBm in 500 Hz. Two signals that are 100dB above the noise
(in 500Hz) are 41 dB below IIP3 so they produce IM3 at a level
of 18 dB above the noise (in 500Hz). I do not feel this is a
problem, the FT1000D for example can not even handle a single
signal 100dB above the noise at 20 kHz.
It should be obvious that -6dBm at the feedpoint is not a reasonable
IIP3 for a 144 MHz system. The 30 dB gain of the preamplifier
should be arranged as 20dB gain followed by a good filter that
excludes anything strong that is outside the ham band. Then one
more amplifier is needed to compensate for filter and cable losses
and to add 10 more dB.
I have sent away RX10700 and RX70 for production. When I get the first
unit of each type to check that everything is ok, I will be able to
sum up the total cost. I will put it at the antennspecialisten site.
The RX144 unit will probably need two more iterations.
Leif / SM5BSZ