
John's homebrew pages
A 13cm (2.3GHz) transverter with 432MHz IF
I decided to try to move up the microwave bands one by one - so 13cm
comes after 23cm. Whether or not I'll be able to resist jumping
straight to 3cm at some stage we'll have to see.
After using multiplier designs for the local oscillators of my 23cm
transverters, I decided to try a PLL-synthesiser approach for 13cm. I
had decided to use a 432MHz IF some time ago so had obtained a suitable
VCO - I now have ideas for doing some home brew VCO experiments as
well. I had also obtained a couple of the Analog Devices ADF4113
PLL-synthesiser chips.
I have done the design for the local oscillator, with the loop
filter design from the Alanlog Devices design tool. It uses a 20MHz
TCXO (temperature compensated crystal oscillator - a bit like a micro
crystal oven) as the frequency reference, and the synthesiser will be
programmed by a PIC on start-up, which then goes to sleep. I have added
an MMIC buffer amplifier to give enough drive for diode mixers, though
at the moment I plan to try some mixer ICs.
Here's the component side of the LO board - it measures 60mm by
37mm. The board layout was drawn out using PCB again, and the board
produced optically
on pre-sensitised 0.8mm FR4.
The PIC will be inserted in a socket on the groundplane side; to
avoid stress on the PCB tracks, the PIC socket is epoxied to the board
surface (as are the 12V and "PLL lock" pins - the holes drilled are a
touch too large!).
It's taken a while to get the circuit built because we've been busy
in the house - but here it is ready for testing. The 20MHz reference
oscillator has been tested, but nothing else yet. The VCO is mounted at
a slight angle because of a small error - failure to read the data
sheet carfully enough! The connection locations on opposite sides of
the VCO package are slightly offset, rather than lined up as I had
assumed. Anyway it still fitted. The 470k wire ended resistor is there
because I didn't have a surface mount type, and couldn't even find one
on my collection of old boards! The output buffer amplifier is fitted,
and should give an output between about 6dBm and 9dBm.
The groundplane side has leads to connect the +5V and +3V to parts
of the circuit.
The small amount of code needed for the PIC is now written, so
that the PLL-synth is programmed on power-up. The oscillator now works,
but at this stage with a test external 20MHz reference oscillator.
De-bugging took a little while - the main problem was extremely simple,
discovered by examining the ADF4113 on the board very carefully. The
pins for the serial programming interface (clock, data and latch)
showed connectivity with the board tracks using the test meter, but
looked like possible dry joints. I re-soldered them and found that the
charge pump reference voltage is, after power-up, now at the correct
voltage (it was zero), and after the programming words are loaded the
oscillator is set to the required 1890MHz.
It was a while before I had time to return to this and complete the
LO. Finally I got there; first the clock
slowdown in the programming code (put there to check the serial line
signals) was removed, and the progamming stopped by making the PIC
sleep; these both worked fine. Next I tried the on-board TCXO 20MHz
oscillator instead of the
external reference, and it also worked. This was something of a relief
as the signal level from the TCXO is only marginally in spec (it's a
bit low in amplitude) for the reference requency input for the PLL!
It's now being boxed up; here's a photo of the board partly boxed, with
an SMA output connector and the power input (just visible over the top
of the box) in through a feedthrough capacitor in the side of the box.
Now it's working, here's the circuit diagram as built. The loop
filter doesn't quite have the right values in (as given by ADIsimPLL)
but it locks fine. Note that I have removed the pull-up resistor on the
lock detect line, and used a logic output rather than the open drain
output. Click on the diagram to get the full size version.
The next step is to build the main board.
For the design there were a few new ideas I wanted to try. First,
commercial passive double balanced mixers for the microwave region are
expensive, whereas I had found some IC mixers costing just a pound or
two each. That was clearly an idea worth trying. Secondly, I needed
some selectivity at 2.3GHz, and commercial helical filters are also
rather expensive; so I thought I'd try some stripline tuned circuits
(and pipe cap filters are another definite possibility).
I also had quite a lot of A06 and A03 MMICs on surplus boards from
commercial equipment, so thought I'd make use of what I have there as
well. With these design ideas in mind, I drew up a circuit using the
gEDA Schematic Editor.
Here's the PCB after etching; my most successful optically produced
PCB so far!
With the receive and transmit sections soldered up, I have been
doing a few tests. From home this was not very successful. Here's the
test setup outdoors; the transverter board still isn't boxed so it's
rather a lash-up.
I'd set up for a Tuesday evening SHF (2.3GHz and up) activity
contest, but heard no-one. I'm not surprised, my home location is far
from ideal for microwaves, with a clearish direction only to the north.
However, with my home brew 12 element DL6WU Yagi (still to be written
up) I was able to detect RF "mush" from a communications mast about 2km
away, and also my little mini beacon (just a 25.1750 MHz block
oscillator with a 30mm bit of wire attached to the output pin; I detect
the 92nd harmonic in the 13cm band!) was detectable from the garden.
Here's
the antenna:
The antenna certainly seems to behave very well indeed, and it will
be great for /P operation.
Having failed to detect any amateur signals, I decided that I should
go somewhere I would have an excellent chance of doing that, providing
the receiver was working as I hoped. So I set out for Cairnpapple Hill
in West Lothian, which is line of sight to the GB3CSB beacon cluster.
Once the test setup was assembled, I could hear the 13cm beacon easily
- even before the antenna was pointing at it. In fact, I could hear it
without an antenna! (Since the transverter still isn't boxed.) With the
antenna pointing at the beacon, there was a very strong signal indeed -
S9+.
The beacon tone was absolutely clear (very musical in fact with the
JT4G signal alternating with the CW), so there are not going to be
noise problems resulting from my use of a PLL/synthesiser local
oscillator. In fact, the 20MHz TCXO reference oscillator I have used
seems to be very stable and also very accurate; the beacon frequency
indicated on the FT-817 was only about 60Hz out, not bad at 2320MHz!
The next step was to finish the transmit side, then put the whole
thing in a box. The transmit side seems to be a little more problematic
than the receive side; it seems that the filters have rather more loss
than I'd expected, since the output level is not at all as high as I'd
hoped. However, given large filter losses (6dB in each?) the observed
output (which might be as much as 0dBm!) is not too surprising, given
the MMIC combination I'm using. So I decided to go ahead, box it up,
and add the PA (a salvaged commercial board which should have 40-43dB
gain!) to see if it works.
Here's the underside of the main board, with the ends added to take
SMA connectors, and screws in the regulators to bolt them to the
sub-chassis for cooling.
Here are the circuit diagrams. First, the power switching which is
on-board:
Next, the receive side of the transverter. I forgot to change the
grey background on this one!
Finally the transmit side:
The bottom of the box has the sequencer board (controlled by a PIC -
details will follow eventually), the coaxial relay (obtained at a
rally) and space for the PA board. There will eventually also be a
bandpass filter between the antenna relay and the antenna N connector.
A heat sink bolts to the other side of this chassis under the PA board.
The main transverter board and LO are mounted on a sub-chassis; this
makes for a more compact overall system than my mark 2 23cm
transverter. I didn't know at this stage whether or not I needed a lid
on the main
board box.
Under test, with the PA board added, I found that the system took
off (oscillations) under transmit, even on SSB, with the main board
open. A bit of anti-static foam over the box stopped that for low
levels of drive but not higher levels; and with the main box lid on it
was not stable at all on transmit.
Here's a photo of the system under test. You can see the anti-static
foam just placed over the transverter board box. The bench could be
tidier!
I made a proper tinplate lid for the main board box, and that
cured the oscillations completely with the overall box lid off, so I
felt we were ready for some on-air tests.
The monthly RSGB UKAC for SHF was due, so I went out to a local hill
(the Braid Hills) which is easy to access and gives me a much better
location than at home. This turned out to be really successful; I could
even hear the GB3CSB beacon, and had four good contacts over 40 minutes
(I tend to chat!) including 93km to Jon GM4JTJ, with some good signal
reports. I thought this was not
bad for probably 4 Watts (as estimated from the DC power input to the
final stage and the device efficiency) and a 12 element Yagi.
The next steps were firstly, to put in a proper screened output N
connector (which unfortunately made no difference!), then secondly to
box in the PA board, to try to
stabilise the whole system under transmit. I made up a screening box
from tinplate which fits between the box base and within a few
millimetres of the sub-chassis; it fits tightly round the PA board.
It's a bit rough but I was in a hurry!
Here's the screening box fitted around the (surplus commercial) PA
board. I was quite surprised and absolutely delighted when I
reassembled the system with the sub-chassis to find that the unwanted
oscillations had all gone away; I can now operate the transverter with
the overall lid in place. I thought that I might need to add some braid
around the bottom of the screening box, to make a better contact with
the box floor, and to add a proper lid. I was even prepared to add
absorbing foam inside the screening box! However these steps proved not
to be needed.
So here's the whole transverter with the lid on. It still needs some
rubber feet! The two power connectors allow two external 12V batteries
to provide both 12V and 24V easily.
I still want to add a bandpass filter of some sort, and the transverter would probably benefit from being tuned up again now all the lids are done, but it works and gives me a few Watts on 2.3GHz. I'm ready to take it out on the next RSGB SHF UKAC now!
That's pretty well the end of the project - I'll add anything
further if and when it happens. Down the line I would like to build my
own PA board but that's really a separate project.