VE2ZAZ - My 3.2m EME dish project

Owning a large parabolic dish antenna has always been the ultimate goal for me. EME (moonbounce) has given me the excuse to proceed! Besides, two years of EME on 432 MHz has made me realize that my QRP setup would not allow me to work tons of stations. I work more or less the same stations in rotation. And I am not willing to grow my 4 x 4.3 wavelength yagi array for several practical reasons. The other consideration is that there are many more stations on 1.2GHz EME than on 432MHz. And I need a new project to work on!

A 10.5-foot (3.2 m) dish will provide a better performance on 1296 MHz (10+dB sun noise vs. 6dB with my 432 array) and an even better performance on 2304 MHz.

I will spare you the details about negotiation with my wife, the swimming pool, neighbors and so on. The bottom line is that thing are progressing faster than expected!

Click here for my 1296MHz EME logbook.

CHRONOLOGICAL UPDATES

Fall 2008

My friend Robert - VE2ASL calls me up to inform me that he spotted a 3.2 m all-aluminum TVRO mesh dish. It lies on the ground in good shape and is available for 50$. It is the dish only, no hardware, no mount, no feed; just the dish. I say yes! This marks the beginning of my adventure...

So my friend brought it back to his place on a boat trailer during the winter season. My first contact with the dish was through this picture to the right. It would have to do for now...

This project was supposed to be a long term one. I had agreed with my wife to wait until our old above-ground swimming pool wears out beyond repair before installing the dish. This could be a few years...

Summer 2009

I take possession of the dish and bring it home on our camping pop-up trailer. The dish mesh is in pretty good shape. The screws that hold the mesh in place are rusted though. I contemplate replacing them with stainless steel hardware.

On another front, I have the opportunity to purchase two more TVRO dishes! They are a 2.7 m dish and a 3 m dish, both complete with all the hardware, the actuators, the feeds and mounts. Even the receivers and the cabling come with it... 75$ take them home. I don't really need the dishes, but I can certainly use the mounts and the actuators! Now I have what I need to get going, at least I have a polar mount I can adapt for EME use and a linear actuator to steer the dish.

Upon inspection, I notice that one of the linear actuators took water. I completely disassemble it dry it out and lubricate the gears and motor. In the end, it was a good learning exercise to understand how it is built and controlled.

September 2009

The multi year wait is turning into an active project. It occurs that there is a housing development just getting going right behind our backyard fence. So with my wife's approval, I decide to quickly erect one of the dishes (non-functional for now) as a way to "mark my territory". The chosen location for the dish is not ideal for west-pointing EME, but this is a temporary location (for a few years, until the swimming pool disappears).

One critical decision before I start is to select the type of moon tracking to implement. I elected to use polar tracking for my EME operation. Implementing a Azimuth/Elevation mount requires heavy duty sprockets, chains, motors, bearings and lots of welding. Since I am an illiterate in welding, this choice will save me a lot of pain and money in the future. Besides, the ice and snow here in Canada complicate everything even more if 12-month operation is desired. Polar tracking will be much easier to implement, but will have limitations and will require manual interventions, like changing the actuator position to cover more of the moonrise or the moonset. I am willing to deal with this for now.

The first step is to get a pipe to mount the dish on. A quick visit to a recycling yard yields a 10 foot (3 m), 2 3/8 inch (6 cm) OD, thick wall steel pipe. This is not as big of a pipe as I would have liked, but it will work in my environment (houses that break the wind, not living in a windy area). So I dig a 42 inch (1.1 m) deep, 15 x 15 inch (38 x 38 cm) square hole, a long and painful job since we live in on rocky land. I then pour 8 cubic feet of concrete and also fill the pipe with the mix.

After a week to let the concrete set, I install one of the 10 foot (3 m) dishes with the un-modified mount and without feedpoint or actuator. This is not the dish that I will be using for EME; it is merely a placeholder. I will work on the real dish this upcoming winter.

After installation, I noticed that the pipe bends a little bit from vertical due to the dish weight. It is not much for now, but once a feed is installed, it will become more significant. I may decide to add a counter-weight or guy wires. I will adapt when I get there.

November 2009

My decision to install a temporary dish to "mark my territory" was a good one. Two houses are currently under construction behind my backyard fence! Close call...

I have completed the replacement of the 500 rusted screws that hold the dish mesh in place. I have stainless steel screws in place now. Oh what a job! Regular steel and aluminum do not get along well... It took me about 5 hours of work divided down into ten half-hour sessions to complete the job.

I have also started to put together the VE4MA feed. For the cylinder cavity, I used a Tx-Rx Systems VHF duplexer cavity. Cheap and rugged (0.100 inch wall!). I cut off the welded top end cap; that yielded a nice 6.4-inch diameter, 24-inch long cylinder. I kept the removable bottom end cap, but added more screws to fasten it. I drilled the N connectors and polarization screw holes as per VE4MA's drawings. Note that, as recommended by VE4MA himself, I did not implement the nulling post. It apparently provides little improvement in TX-RX port isolation. The result is what you see on the right hand side pictures.

The initial testing of the feed without scalar ring and pointing to the zenith gives promising results, with better than -20dB of return loss on both port at 1336MHz. Port isolation is better than 28 dB. The peak performance should normally show up at 1296MHz though, so there will be a need for some tweaking. Axial ratio (circularity) will be measured later.

The list of things still to do is long:

Build a PHEMT LNA
Build a 200-300W Power Amplifier
Design changes to the polar mount
Design and construct the feedpoint arms
Design the Actuator controller
...

August 2010

A lot of things have happened in the last 10 months! Here is a summary:

Once again, what a good idea to put up a placeholder dish ASAP last fall. Now all the lots behind my property are occupied by fully-constructed houses! I have heard of no complaints as of today from any of my neighbors. One of them actually offered to help out on the dish installation!
I first built an ATF54143 PHEMT LNA (Preamp). The results are 18dB of gain and a Noise Figure of less than 0.3 dB. This will be a good preamp for EME!
I have performed more tests on the dish feed. The results are the same: I am having problems bringing the resonance down from 1336MHz to 1296MHZ. I have lengthened the probes and I have built a scalar ring made of cardboard and aluminum but without success. Extending the cylinder length forward with aluminum foil seem to improve the S11. This behavior is unexpected. I will be spending more time on it once the hardware/mechanical on the dish is completed.
I have found the type of tracking scheme I will be using to follow the moon. It is called Azimuth-over-Elevation (as opposed to the classical Elevation-over-Azimuth). Because my mount can easily tilt an almost-full 90 degrees in elevation, this becomes the easiest way to do tracking without any modification. The conversion of the Az-El data from a tracking program to Az-over-El involves a few trigonometric formulas. This is now understood and documented here. The toughest math part is to figure out the current dish position as a function of the actuator lengths. This required some serious kitchen table trigonometry work... I think I have what I need and will be putting it to the test soon...
The Az-El moon data will be provided by Nova for Windows - a wonderful tracking program - using the Windows Dynamic Data Exchange channel, which is supported by the software. Thanks to the author, Michael Owen, for documenting the DDE feature for third-party usage. It is very easy for any Windows program writer to use this feature.
The control of the antenna actuators will be performed as shown in the figure below. This involves creating a Windows software that interprets the Nova data, the actuators pulse counts and that steers the antenna accordingly. The remote controller is located outdoors. It controls the power relays to the actuator motors and it reads the pulse counts as the actuators move. Both pieces are linked by a serial port link at 2400 bps. That link can be several hundred feet and will still work fine, as tested on the At Last T/R Sequencer.

I completed a first version of the Custom Windows software (see picture). The calibration process is also supported in the software. I used Delphi (Object-oriented Visual Pascal) to accomplish this. I am currently struggling with divide-by-zero and square root of negative value errors. My calculations have to be accurate and robust!
I built and tested the remote controller (see pictures). I wrote the firmware code in assembly language. It runs on a PIC18F1220 micro-controller. A lot of code re-use from my other projects was done. As physical support for this controller, I used a "Rechargeable Battery Cycler" PCB. This provided the voltage regulation, RS-232-to-TTL conversion and PIC micro support, all "pre-wired" on a small PCB. When wiring the power relays, I used a scheme that flips the actuator motor wires for direction selection and that interlocks CW-CCW and Up-Down directions respectively, i.e. opposite directions cannot be invoked simultaneously, thanks to DPDT relays.
I replaced the black 3.0m placeholder dish with the final 3.2m mesh dish. The feedpoint arm brackets are installed and the arms are ready to be installed and are ready to receive the feed. I even power-washed the dish so that it shines in the sun... My wife is convinced that the natural aluminum color shows less than the black. If she's happy with this, then I'm happy!
I also installed and tested the two actuators (see picture). Cool to see the dish move! The Inclination actuator (18 inches long) provides a 0 to 85 degree range. The Rotation actuator (24 inches long) provides a 100 to 190 degree azimuth range (with the mount oriented due-south). This is not as much as I would have liked. When looking at the rotation actuator in fully deployed position, there is still a lot of westward movement that could be achieved with a longer actuator. This may or may not be important since Pacific Ocean coverage is useless, and I would not reach Asia or Australia anyway without transferring the actuator to the other side of the dish. I also have the ability to turn the mount on its post to adjust the range for better coverage of a point of interest. All I have to do is take this into account in the Windows software. This is likely a better avenue. I can see two positions of the mount that would give me all the coverage that I need. Hmmm....well, well.

The list of things still left to do:

Make the Windows software bullet-proof,
Prove the math by checking for correct orientation using the sun,
Build a 200-300W Power Amplifier (a winter project),
Make that feed work!,
Modify the TS-2000 for split RX-TX ports,
Run all the required cables,
Build an At Last T/R Sequencer,
...

September 2010

I have been bugged by actuator motor EMI/RFI noise problems. Essentially, the pulse counting circuit gets triggered by DC motor brushes noise. It flows all over the place on the controller board! I am wasting precious time on this issue. I sent an email on Moon-Net crying for help and got lots of good suggestions to reduce that noise. Obviously, the noise should be reduced as close as possible to the source, so at the motor end. RF choke and capacitive decoupling is recommended. I tried a common-mode choke connected at the motor terminal screws and it reduces the noise significantly. Hopefully, this will be sufficient, along with running a separate shielded cable for the reed pulse switch signals.

Despite the noise problem, I have been able to see that all the math formulas to derive the rotation and inclination angles from the pulse counts do work. A few evenings under the bright moon were enough to prove this using the shadow generated by the feed arms. Some more work is required to null out any offset or non-linearity. Objective is to be within one degree of target.

I still don't have the cabling run between the shack and the antenna, so every time I want to experiment, I need to bring computer, power supply, controller board, necessary cabling, etc outdoors and connect to the actuators. It represents a lot of handling, but to be honest, I am not comfortable yet working remotely. I want to see, hear and feel what is happening with the dish. One thing I will probably do soon is wire up everything permanently, but bring out a netbook PC and have a VNC remote session running to control the shack desktop PC.

On another front, it looks like I have covered enough exceptions in the math formulas to make the Windows software work reliably. This is not my main concern at this point though; the outdoors work is prioritized.

Finally, I have built and installed the super-scalar ring on the feed. I am currently testing up the feed for VSWR, isolation and axial ratio. More on this next time!

First-Half of October 2010

We have liftoff! I receive sun noise. During the last month I did a lot of work:

I definitely solved the actuator DC motor EMI noise by installing a common-mode choke on the motor leads inside the actuator covers. The "Pulse P0502" toroid choke is available from Digikey for around $5 US. A well-worth investment! I also ran the pulse signal wires in a separate cable. Now I don't mis-count pulses anymore. Problem solved!
I managed to tune the feed for good reflection loss (23dB) and I also adjusted the axial ratio to around 1dB. One important note when trying to adjust for axial ratio: Rotate the feed, not the reference antenna. Otherwise reflections will fool everything. I suspended the reference antenna 3.5m above ground and used an office chair to rotate the feed (pointing upward) under the reference antenna. Worked like a charm.
I ran the control and RF cables to the shack. I installed the controller in a weather-tight box on the dish post.
I installed the feed on the dish and positioned it so that the focal point ends up inside the feed by 1-2 cm.
I hooked up everything on the Rx port (SPDT relay and preamp) and ran inside to make sun noise measurements. The first results were 8.3dB of Sun-to-Cold_Sky noise. This is around 2dB less than what was expected using VK3UM's EME calculator program. I varied the feed distance from the dish but could not improve. After discussing with Barry-VE4MA, it appears that the 17.5dB of gain on the LNA (preamp) is insufficient when considering the feedline losses and the receiver noise figure. A quick check with VK3UM's EME calculator program (Receiver performance tab) confirmed this. I need at least 30dB of gain in order to "stabilize" receiver performance and get the full dynamic range. So I will build a second ATF54143 LNA to put in cascade with the first one. At least I know the recipe...

Second-Half of October 2010

I managed to run a 70 feet-long Heliax 7/8" from the dish to the shack. This is the Tx line. The Rx line is heliax 1/2". Weatherizing of the feedpoint will be temporary but that will suffice for this winter. Let it snow!

I found out that using the TS-2000X audio fed into the PC sound card is not the good way of measuring the sun noise. The radio (likely the I.F. DSP stage) saturates and does not provide a linear audio output vs. RF noise input, that is despite turning off the AGC! No wonder I was measuring low in sun noise! I instead used a 1296-to-144MHz K7RR downconverter, added two 144MHz preamps and a low-pass filter on the transveter output and I fed this signal into a HP 70000-series spectrum analyzer. There you go! Linear response.

So I still built a second 1296 MHz preamp and installed it at the feed. This made an improvement. The actual measured sun noise is 11.3dB, pretty much in line with the VK3UM Calculator program. I plan to extract the I.F. signal (first or second I.F., not decided yet) and send it outside the TS-2000X for spectrum analyzer measurement, a much simpler setup. This shall work.

So the real reward came on October 30th-31st during the ARRL Contest. My face flashed a big smile when I first heard CW stations while tuning around. During a 3.5-hour period, I copied 22 stations and lots more partial calls. I even saw a pileup of 5 stations. I didn't know where to start. All of this was done with the bottom half of the dish covered with thick wet snow. Yeah! My system definitely works. Pure pleasure, the ham way :-)

I must now prepate to transmit. My first QSOs will be performed with a 40W P.A. brick in the shack. We'll see how things go...

November 2010

I had made several receptions of various JT-65 and CW signals since the contest and felt everything was pretty much optimal for a real bi-directional QSO. So I modified my TS-2000x for Split-Tx-Rx ports, I built the T/R sequencer and I hooked up the PA brick to the Tx feed line. A first attempt one Saturday evening was a partial failure. I could be copied at the other end, but I had no copy on my side, not even the "birdies" I usually get at known frequencies. A bit of search and I found what was wrong. My transmit power amplifier brick oscillates and produces wideband noise that gets picked up by the receive chain. My receive capability was totally wiped out by this noise. A simple solution was to kill the PA DC supply with a power relay when in receiving. It worked like a charm...

So I had MY FIRST QSO, and it was in CW, not in JT-65C! Joe - K1RQG is the man. Easy copy on my side even on the SSB filter, and he reported a good copy on his side despite the tiny 25W I produced at the feed. Whoohoo!

Here is a video I made. It is a tour of my dish antenna, feed, mount, coaxial lines and control software.

I am now in chase for more QSOs. And I will concentrate on the last (but not the least) piece of indoors work, which is build a 200-300W Power Amplifier (a winter project). Stay tuned!