HI Farhan,
Very nice. I've bern working on a setup if near identical topology. I
chose different gain stage
format and the IF is 25mhz as that is a common crystal. The filter at
25mhz works well. However I didn't
give up 10M. in addition to the low pass filter I used a switchable
bandpass for 10M (I skipped 12M).
Your choice of 24mhz is a good one. One item you may find the TX gain
is likely low. I needed
about 6-8db (including filter losses) to get the balanced mod level up
to near with the mixer
can take as that gives better signal to carrier leakage with low IMD.
For that LO and those diodes
a input level for signal is around 3-5dbm before compression sets in and
that is much stouter than
a TUF3 or SBL-1 commercial part.
I arrived at the same problem with the KISS mixer its single balanced
design means at least
one port will must have too much of the wrong stuff. I also found out
early on notch filters
are a pain to tune and never selective enough 20 if was just to close to
the 15m band.
I did experiment with 12mhz and that did work as the notch was narrow
enough but the
2nd and 3rd order mixer products were daunting.
The mixer is the same DBM save for a minor change. The RF port (as your
schematic has it)
which is usually the IF port has the winding spit and a 100 ohm port
with the wiper going to the
cap and the low pass ( and mine has bandpass for 10M). Makes it easy to
achieve higher orders
of balance (TenTec uses it in many of their radios) and the diodes are
1n4148. Rather than the
single balanced modulator I went with a repeat of the mixer with the IF
port wiper being the
audio in/out carrier balance is easilly better than 50db with loosely
matched diodes.
SI570 for LO and crystal based carrier osc as its easy and switchable
for USB/LSB/CW.
The current version only has the Arduino doing the Si570, as most of the
other functions
are easily done with random logic implemented with transistors. I'm
trying to avoid
anything like menus. Main functions (USB, LSCB, CW, attenuator,
preamp(rx), and
power level) will be buttons with led annunciators. I may do a variant
of another radio
I have where there is an up and down button and a digit select button.
Up and Down
changes the selected digit and the digit select high lights (blinks the
selected digit)
the desired digit for increment or decrement. That makes large
excursions and
small ones easy.
Still working on the TX amp and its testing and running with a smaller
3w. The goal
is a switchable 5W and higher like 10 to 20W with high efficiency. The
experimental
version of that amp is single ended driving push pull bipolar, driving
push pull bipolar
like MRF406 or 2sc1945s as I have some I have some. Maybe RF16HHF. I'd
considered IRF510s but at 12V they are not optimal (at 24V they work
very well).
The goal for that stage is to fit it in a 2x3x1.5" volume including a
finned heatsink.
First break I get to assemble a schematic I will post it but for the
most part nothing
unusual.
Allison
On 05/20/2015 03:13 PM, Ashhar Farhan wrote:
comrades,
i know that it has been a very long time since i posted anything
substantial on this list. but i have been quite active measuring,
thinking and trying out various things for the minima. I have finally
arrived at some conclusions that i'd like to present to you all.
I have produced a new version of the Minima that is substantially
simplified, easy to work with. However, it comes at two major
trade-offs (that is why we are engineers, not scientists : we work to
a budget). First, the transceiver works from DC to 21 MHz. I had to
drop 10 meters - a personal favourite. Second, I have switched to a
diode mixer resulting in a drop of IIP3 performance down to around
+15dbm. This is still superb. But not in the same league as before.
Finally, I have spent the last two days using the rig. It is a really
sweet sounding transceiver. Easily the best I have used. The circuit
is simplified to the extreme. It is even simpler than the BITX.
Here is my long story about it :
Measurements
I realized that I didn't have the equipment to actually test and
measure IIP3, loss, etc. Hence, I spent a few months building
equipment. I now have a spectrum analyzer, entirely home-built that
has excellent dynamic range. It is based on the same Arduino + Si570
combo as the 1st oscillator. The rest is an evolution of the W7ZOI's
spectrum analyzer. Along the way I learnt to sweep VHF filters, and
measure IIP3. This analyzer can step in 1 Hz steps (thanks to you guys
for having developed a better Si570 library for radiono) and I have a
narrow 500 Hz and wide 300 KHz filters. But that is an entirely
separate topic for another article.
I also made a two-oscillator setup by pulling 14.318 MHz crystals
apart by 20 KHz. And combined their buffered output in a 6db hybrid
combiner and took the output through a an LPF cut for 14 MHz. Thus I
had the ability to measure loss, intercept, band-pass. I still lack
the ability to measure noise figure due to a lack of calibrated noise
source.
Equipped with this, I set out to hack the KISS mixer. I have spent a
great deal of time trying to build them with discrete devices. I tried
everything : from 2N3904s through 2N7000s to J310s. I tried five
different biasing schemes. I have documented it all in my notes. The
summary is simple : the KISS mixer lacks enough suppression of the LO
to be used in the middle of a passband. It is an excellent mixer for
high performance receivers. One could add some narrow band filters to
the Minima and a Linrad backend to beat the living daylights out of
K3S. Btw, I measured more than 30 dbm IIP3 on the KISS mixer, original
version. I say 'more than' because my -10dbm per tone signal source
was hitting the noise floor on the specan.
1. Giving KISS a miss
So, the KISS mixer has to be parked aside for the a minmal rig like
the MInima. That leaves us with the old favourite : the diode ring
mixer. I built a diode ring mixer with 1N4148 diodes that measured
15dbm IIP3. This can be as good as any of the higher performance rigs.
(http://www.elecraft.com/K2_perf.htm#Main RX Table)
However, the standard mixer circuit took the IF from the center tap of
the tranformer that was driven by the LO. This leaked the LO to the IF
(which we use as the RF port). By grounding the center tap of the LO
transformer and taking the IF from the center tap of the other
transfomer (the one connected to the RF port), the LO dropped
substantially. It went down by almost 57-60dbc (below the carrier).
The diodes will have to be matched to the last millivolt : easily done
with a two dollar DVM.
2. Dropping 10 meters
I had written earlier that i was fooling around with 24 MHz crystals.
These commonly available and inexpensive too. By moving the IF to 24
MHz, we achieve a number of things. First, an 4 section LPF cut for 21
MHz will receive everything from DC to 21 MHz. Second, it offers
reasonable attenuation to IF. It comes at the cost of dropping the 28
MHz band. (We can add an 'extra band' with relays that provides a
BPF based narrow band coverage of any one other band (the diode mixer
will mix from 144 MHz to 28 MHz).
3. Post IF amp
The diode mixer needs a robust termination to work well and the
crystals of 24 MHz were quite lossy. Both these factors lead to adding
of a post-mix amplifier ahead of the crystal filter. I know, it is
kinda 'old world'. But look : sensitivity is up, crystals need not be
expensive.
4. IF amp - not really required
Given that we have gain ahead of the crystal filter, we really dont'
need much gain before the audio detector. Hence, just an emitter
follower to buffer the signal from the crystal filter to the audio
detector.
5. Simpler audio system
I replaced the three-transistor W7EL style audio preamp with an old
fav from the BITX20 using a single transistor. The power audio
amplifier is no longer the discrete power amp. Instead, I am using a
TDA2822 power amp chip. The original Minima audio needed more gain.
This chip has 40db of voltage gain.
6. Improved tuning
Each time I used the original Minima tuning i wanted to kill the guy
who wrote it. I rewrote the tuning system. Now, it tunes like a normal
tuning knob for 100 KHz in 100 Hz steps. However when u hit the band
edges it starts to 'scan' first in 20 Khz steps, then 100 KHz and
finally in 500 KHz steps. It works well. I must add some visual alert
when it starts to scan. But that apart, the system is quite workable.
I am attaching a very rough and incomplete picture of the circuit from
my lab notes that shows the changes.
- farhan
