10 GHz

This will be a collection of notes and a timeline of sorts as I work on exploring a new band. I will include both successes and failures.

Click on any of the small images to get a big version.

January, 2022

I have been intrigued by the idea of trying microwaves for years but I thought it would be near hopeless trying to make contacts from my location. The vast majority of 10 GHz activity is portable where one can select favorable locations. Operating portable is not an option for me. It would be my home station or nothing. I am not on a hill, my horizon is imperfect and I am a long way from any other 10 GHz stations. I was aware of rain, snow and aircraft scatter but perusing the various web pages and articles I had access to did not turn up enough information to assess my chances. I found a good amount of information on how these propagtion mechanisms work but little to nothing regarding what one could realistically expect for a given station setup. When I did find web sites or videos with information about specific QSOs that were made, there was often no information given about power or antenna gain. Over the last year or two there has apparently been a surge in 10 GHz activity. It seems this has resulted in an uptick in videos being posted to platforms such as Youtube. Still, information about equipment used was often incomplete. I also read an article by KA1GT about a very low cost 10 GHz receiving setup. This, combined with the loss of 2200 meter operation pushing me to look for a new challenge, led me to seriously consider trying 10 GHz. The other challenge is getting above the trees. Fortunately the trees top out at about 95 feet and I have two towers taller than that. One is 96 feet, the other 104 feet. While rebuilding from storm damage in 2021 I reserved a spot on a mast at 105 feet for a small 10 GHz dish.

Since I was seriously considering starting out with a low cost receive setup to explore the potential of my location, I became particularly interested in beacons. Since most 10 GHz activity is from portable operations and the majority of QSOs are coordinated in some way, it didn't seem likely I would be hearing much unless I could hear beacons. I had questions about this since beacons often have much less antenna gain than a typical 10 GHz station. I went on a search for information about hearing beacons on rain, snow, or aircraft scatter but found next to nothing. Did this mean beacons are too weak to be heard at distance using these scatter modes, or was it simply that no one was posting information?

My horizon profile from planned location of 10 GHz dish generated by HeyWhatsThat

The next thing I did was join several 10 GHz and microwave forums. The people on the forums are very friendly, outgoing and generous. They seem eager to help. There I started asking questions about propagation and what to expect from my imperfect location given the size of station I felt I might be able to put together. Based on information gathered in that way, 10 GHz sounded much more promising. Over several months digging through forum archives, reading new posts and watching videos shared there, my perception of what is or might be possible continued to evolve. Even so I felt I wasn't getting a solid picture. I developed a strong perception that what is known within the microwave community regarding possibilities and expectations is largely not known outside the community. Is 10 GHz one of the best kept secrets out there? I hoped to find out.

At this point, after several months with some view "inside" the community (only by way of the forums) I still feel I have a poor understanding of what to expect, though it has greatly improved. I still perceive a large gap between knowledge inside the community and information that gets outside it, particularly in the area of possibilities and expectations. Only time and experience will show whether this perception is accurate. I may be coming to this in an unusul and particularly isolated manner. Once again I only have perceptions to guide me, but from what I can gather most newcomers to microwaves have some opportunity to connect in person with those areadly involved, including being a participant or observer in portable microwave operations. I do not have such opportunities, which makes me far more isolated and out of the loop.

I haven't asked but I might have access to a potential beacon site as well. The tower is owned by a local ham radio club. I own one of the repeaters on the tower. I know I can't afford it but I can't get the idea of a 10 GHz beacon out of my mind. In these days of very affordable 10 GHz receive systems, might beacons play an important role in attracting newcomers?

Potential beacon horizon profile generated by HeyWhatsThat

I don't want to start off on the wrong foot with the 10 GHz community I hope to join but I am one to tell it like I see it. If I am being honest, I give the 10 GHz and microwave community a failing grade on getting information out there to help prospective newcomers understand the potential. There is not enough out there to help someone who is isolated or not solidly connected to the community evaluate the prospects of the band. I find this unfortunate and disappointing. I am increasingly finding that a lot of stuff that would be very helpful in this regard, such as videos of actual reception or QSOs, is only being posted temporarily and its availability is only known within the microwave community. I will be on a mission to do whatever I can to improve the situation. As I learn and discover what can be done from my location I will post things here, on my blog, and on Youtube. I hope to have a receive-only system ready to go on the tower in the early Spring. A full transceive system is a long way off due to budget constraints.

January 13, 2022

Today I did initial testing of a 10368 MHz signal source and the LNBF I hope to use for a receive-only system. The signal source is an old Qualcomm synthesizer board (thanks N1JEZ) modified to produce 1152 MHz. It should be producing harmonics that fall in several ham bands: 2304, 3456, 5760, 10368 and possibly 24192 MHz. The 10368 signal is quite strong listening with the Bullseye 10 kHz LNBF in the next room. The LNBF converts 10368 down to 618 MHz which I run into one of the cheap RTL2832 SDR receiver sticks. I mounted a switch on the side of an equipment rack to control 12V to the LNBF. At this point nothing has been calibrated. I am still using the original 10 MHz OCXO reference on the Qualcomm board, I am certain the RTL2832 stick is off frequency a bit and though the Bullseye 10 kHz is one of the more accurate if not the most accurte LNBF in terms of frequency, it is likely not perfect. Getting some of the frequency errors and drift under control is next on my list of things to do.

February 22, 2022

Completed work on the weak signal source. Removed 10 MHz OCXO from Qualcomm board, replaced with external Bliley NV47A1282 on a VA3TO board (thanks to VA3TO). After a day or two of testing I feel that I know where I am to within 30 Hz at 10 GHz. Oscillator aging may require frequency adjustment to maintain that level of precision but I can easily do that since I have a GPS ereferenced HP counter.

March 15, 2022

Over the past few weeks I have been working on a power and SWR (return loss) meter. The first trial run used a $25 power sensor board from China using the AD8317 chip. Others have had good luck using boards like this but mine was terrible. It was sort of OK at 1 to 2 GHz but not a good fit to the curves for 900 and 1900 MHz in the AD8317 datasheet. 2 GHz is the limit of my in-house testing capability. It was unreasonably poor at 144 MHz. That plus the board being .062" thick which I understand is not good at 10 GHz led to me lose all confidence in that board. I suspect getting a good one is a matter of luck. For the first attempt I was using the Arduino Nano. Here is a picture of this failed atteampt at the project.

I decided to try the much more pricey ($144) Analog Devices AD8317 eval board. It looked much better. At 900 and 1900 MHz its performance closely followed curves for those frequencies in the AD8317 datasheet. At 144 MHz it looked far better than the cheap boards, though calibration was significantly different than at the higher frequencies. I also was not happy with the code I was using for the Nano. It was going to require a lot of work to get it where I wanted to end up. I used to be a fair PC software developer back in the 1990s but had not written a line of code in 20 years prior to taking on this project. I then learned of another approach and obtained different code from another 10 GHz operator who is working on a similar project. As a matter of convenience I switched to Arduino Uno R3 with LCD keypad shield. This greatly reduced the breadbording reauired, as well as conveniently provding several buttons for controlling the meter. Here are a couple pictures of the test setup with this implementation. The top line of the display shows the reading (here power in dBm) while the bottom line is a bar graph which can be very useful when tuning things.

April 19, 2022

Progress has been a little slow but I have several projects in the works. Today I did final assembly on the sensor head for power and return loss measurements. A connectorized DC power feed might have been preferable but was not within my capabilities in very limited space. Small size and light weight were important goals. The connection is well strain relieved and should hold up for quite a while. The DC power feedthrough filter is a #4-40 size from Downeast Microwave. Thanks to W1GHZ for generous assistance. Several parts have been sent out for characterization on a VNA, including the directional coupler, SMA short and SMA termination for this project.

I calibrated the power sensor for all bands 1.8 MHz to 903 MHz. I also calibrated at 1000 and 2000 MHz, which is as close as I can get to the 1296 and 2304 MHz bands, but should be close enough. The reading has about 2 dB of jitter at 1.8 MHz. All higher bands are OK. Calibration changes by 3 or 4 dB as I work my way up through the bands. Useful dynamic range is 60 dB on all bands checked so far (-60 to 0 dBm). It will be less on 10 GHz but should be at least 40 dB there. I will need to send it out to be calibrated for 10368. If whoever I send it to has time, I will have it calibrated for 3400 and 5760. It would also be interesting to have it checked at 11 GHz to see if it is still useful there. That might be good to know when it comes to making return loss measurements. The calibration process was boring to some...

I never seem to take the shortest or quickest path to anywhere. I decided to put a beacon on the PARC/N1BUG repeater tower. Club members and leaders have been enthusiastic and very supportive. The beacon is a major expense and that slows down progress toward getting my home station 10 GHz capable but it will be eductional and helpful to me and perhaps others. Initially the beacon will be 10mW, 2x12 slot antenna and GPS locked. There may be a future upgrade path to 2 or 3 watts. Such an upgrade would have to be discussed with the club. I will consider that if the beacon would be of use to those outside the line of sight area.

April 30, 2022

I love it when a project finally comes together! The power and return loss meter is complete. All that remains is to send it out for calibration at 10368 (and perhaps 3400, 5760, 11000). Thanks to W1GHZ, W3IP, N8ZM and others for helping with this project.

May 3, 2022

Barn find. With help from local hams I just dragged home this 6-foot dish found in a barn just five miles from my house as the crow flies. What are the odds of that in rural Maine? I have no idea what frequency this was designed for or how unkind years of storage and transport have been to it. But it was free so why not? I also got about 10 feet of WR90 waveguide with flanges I don't recognize. The rectngular flanges have eight holes but are significantly larger than the eight hole flanges on my other piece of WR90. There is also a few feet of what appears to be WR159. Thanks to NC1Y for the dish, KB1WEA, K2KJ and KB1WRZ for help with transport.

May 24, 2022

I ran into a snag on the beacon project. Having tried a VHF Design PLL module as a beacon exciter I did not like the hard, clicky keying. I decided to take a different approach using a W1GHZ Personal Beacon board. I believe I can get the keying on that to sound good. Since I had two Leo Bodnar GPS reference clocks laying around not serving any purpose, I decided to try using one at 576 MHz, into a doubler, then into the beacon board. Here is what I discovered. When progrmmed for 576000000 (10,368.000) the output of the Leo Bodnar is clean with no close in spurs. When programmed for any frequency that puts the beacon on a reasnable frequency it is not clean, having several strong spurs on either side. Here is an example programmed for 576020000 (10,368.360) showing spurs at +/- 20, 40, 60, 80 and 100 KHz. The first spurs are only down 33 dB. These would be on 10,368.000 and 10,368.720. I don't like that! The search is on for a signal source that can provide clean, GPS locked output at 576.020 or 1152.040. I cannot examine the latter with my spectrum analyzer directly but may be able to look at the final 10 GHz output using a satellite LNBF into my spectrum analyzer or SDR receiver.

Later in the day I decided to test the VHF Design board at the same two frequencies. I found no difference between the two frequencies. On both frequecies it looked reasonably good with a bit more "grass" on either side than the Bodnar. This may be phase noise or something else. I am not sure.

June 1, 2022

Just when I thought the only way I could tell anything about the spectrum of a beacon under test was to listen to it with a receiver, I was reminded one can use a satellite LNB into a 1 GHz spectrum analyzer to get a look at 10 GHz signals (thanks W1GHZ and W1FKF). I tried it today and while there may be some uncertainties it does work. There were too many reflections with the setup inside so I moved the signal source under test and LNB outside, with a cable running in to the spectrum analzer in its usual spot in the test equipment rack. I used the VHF Design PLL board that is built into my power meter as a test subject. I programmed it to generate 10,368.360 and looked at its radiated output spectrum using its internal TCXO, an external OCXO (I conveniently used the one in my microwave weak signal source) and with a Leo Bodnar GPSDO as the external reference. Ignoring slight frequency differences, the spectrum was virtually identical on all three, perhaps a couple dB lower phase noise with the GPSDO. Spurs were not a significant issue with the strongest ones being about 65 dB down. The plot shown here is with the GPSDO. I do not like the CW keying. There is a residual signal that is only about 45 dB down when "key up". Key clicks are also significant. Much work remains to be done building additional modules and testing possible beacon configurations to find the best one (that I can afford).

June 4, 2022

Christmas in June! Thanks to W1EX, W1FKF and W1GHZ for the pile of stuff W1FKF kindly brought to me yesterday. There is a HP 432A RF power meter with cable and 478A thermistor mount, a Kuhne 10 GHz transverter, a Kuhne 2 watt amplifier, DEMI preamp, T/R relay, an old Icom 290A 2 meter all mode transceiver which could serve as an IF radio if I were to get a chance to operate portable somewhere, various cables, a Radio Waves two foot dish with radome, a dish feed and many other small items. There were also several items for 1296 MHz not shown here. I am working on cleaning things and checking everything out.

June 21, 2022

I got a Winegard DS-2076 76cm offset dish and a Bullseye 10 KHz LNB which I hope to use as a 10 GHz receiving setup. I wanted to play with the LNB position to try to optimize it, so I needed to set up the dish close to the ground where I could work with it to measure noise ratios: sky to sun and sky to ground. Since I have no budget for materials now, I needed to use whatever was on hand. The only sturdy mounts I have are my towers, so I decided to mount it to the side of my short Rohn 45 that is not being used. I clamped an 8 foot mast alongside. I did not want to cut that mast down as I will probably use it on a tower at some later date, but this meant the dish is about 9 feet off the ground - too high to reach. So I needed a work platform which was easily made with a utility trailer, some left over scraps of plywood and some dimensional lubmer that was laying around. Then I needed a platform for the laptop to bring it up to where I could see it while moving the dish. A small folding workbench and a cardboard box solved that problem. Finally I needed to be able to see the laptop display in the bright outdoor conditions. A tarp draped over the laptop and my head got the job done, though it was annoying to keep it on as I moved around doing things, and even the slightest breeze blew it away. At the top of the 8 foot mast I have a back assembly unit salvaged from an old, beat up DirecTV dish. The back assembly unit has fine adjustments for azimuth and elevation, so I can set a coarse position by hand and then fine tune it with the BAU. This worked well once I got used to it.

The LNB was feeding a RTL SDR with AGC disabled. I was using HDSDR and Spectravue, connected by VB Cable (a virtual audio cable). I saw only very minor differences in sun to sky and ground to sky Y factors as I moved the LNB in one centimeter steps. The best, by a slight margin, was with the LNB at its closest position to the dish. Here I measured approximately 3.9 dB sun to sky and 3.1 dB ground to sky. I am not certain how to interpret the measurements, but playing around with VK3UM's EMECalc a preliminary guess is that the LNB noise figure is around 2.5 dB and the dish is under illuminated. This is only a guess! There is a big variable in that EMECalc only seems to provide one feed that gives reasonabe efficiency for this dish and I haven't the slightest clue what that feed is! Opt dual-mode F2L x L3.10L means nothing to me. I do not know what type of feedhorn the LNB uses. Things are further confused by two different sets of numbers obtained from EMECalc at different times using input data that looked identical. I don't have a clue what happened there. When peaked on the sun, the LNB shadow was not as far onto the bottom edge of the dish as I expected, based on what I have read. Also it had a slight horizontal offset from bottom center.

June 22, 2022

More fun and games trying to verify and ajust this LNB and dish. Today I tried the LNB vertical probe instead of horizontal. There is a difference in the probe construction and at least some LNBs have better noise figure with the vertical probe. I found the Y factors were indeed better, with sky to ground showing 4.0 dB and sky to sun 5.0 dB. This still looks fishy to me and quite by accident I found a potential problem with my measurement setup. The first thing I did today was a control set of measurements to see if yesterday's result on horizontal polarization could be repeated. The Y factors came in about half a dB lower than yesterday. At the time I didn't understand this and went on to testing vertical polarization. Later I returned to look at horizontal again and noticed the level into Spectravue was a couple dB higher than yesterday and a few dB higher than today on vertical. The RTL SDR gain setting is very touchy, acting almost like an on/off toggle, but I managed to get it down a few dB. I was then able to replicate the results from yesterday. Now I question whether all of my results to date may be into a gain compression region for the RTL SDR and therefore not accurate measurements. Next time I will add an attenuator between the LNB and the RTL SDR and attempt to adjust the RTL SDR and HDSDR gain levels lower.

June 24, 2022

OK, I'm not sure what I know at this point! I went over the RX setup with my signal generator and calibrated gain to ensure that no stage was being run outside its linear range. I am confident I got that right. The biggest sky to ground and sky to sun Y factors seem to be with the LNB at its closest position to the dish and reamin near 4.0/5.0 dB. I might be getting some noise contribution from spillover when looking at sky. The dish can only see sky (or sun) to the east and southeast with elevation above 20 degrees or so. There are tall trees behind the dish to the west and northwest. I suspect these fill most or all of the spillover area and may be contributing noise when I am trying to measure sky. Whatever the case may be, this seems to be as good as it gets with this setup. Perhaps I could gain another tenth or two on the Y factors if I modified the LNB holder arrangement to get it even closer to the dish but I doubt that is worth the effort. Note: as of July 3 I have not had a chance to get back to this. I have been very busy with other things.

July 8, 2022

The PE1RKI 2x12 slot beacon antenna arrived!

August 15, 2022

I heard my first 10 GHz signal - DL0SHF!!! I added a borsight to the dish and hastily constructed a work platform since the utility trailier is in use elsewhere. Last night I spent four hours standing on the platform trying to hear the DL0SHF EME beacon. I know for sure where the dish is pointing now, but there were a few challenges. It turns out on a night with a large, bright moon the boresight isn't needed. I can aim the dish by watching the position of the partial LNB shadow on the lower edge of the dish! Either the LNB or my old Nooelec RTL-SDR stick (the type with no TCXO) is prone to periods of drift. It will move two to three kilohertz in three or four minutes, then reverse direction and drift back to the starting point, more or less. During those times it was not possible to decode the Q65-60E transmissions from DL0SHF, but in between there were periods of relative stability where decodes were possible. It was a thrill to see the signal clearly on the waterfall and get the decodes! At first I wasn't sure in which direction to rotate the LNB to compensate for the roughly 55 degree spatial polarization offset, but that was easy to figure out -- rotate in the wrong direction and there is no signal! Another challenge was getting on frequency. I used my 10368 marker with the Qualcomm board to provide a reference, but it is not possible to calibrate out errors in the LNB LO and the RTL-SDR LO using HDSDR. I had to mentally calculate where to tune for the doppler shift (which changes very rapidly at 10 GHz!) and then subrtract whatever offset I had in the receive chain from that. Typically the latter was in the 14 to 17 kHz range. I was worried about this but it really wasn't all that difficult. Even with these challenges I was able to see the signal on the waterfall within five minutes. The first decode was about 15 minutes after that as the LNB warmed up and its drift slowed down. I did have to recalculate the frequency and retune often due to receiver drift and doppler shift.

Getting digital decodes is all well and good, but of course I had to see if I could hear the signal and just maybe, somehow, copy the CW! I don't like SDRs for weak signal CW reception, especially when there is drift involved. Despite hundreds of hours trying to master it, I cannot adjust the tuning and filters quickly enough to follow the signal with my finicky old ears. Give me good old knobs and dials any day! The only thing that saved me was that the 10 GHz EME signal is rather wide due to doppler spreading, so very narrow filtering won't work anyway (at least not for me). Nevertheless I found keeping the signal in the "sweet spot" for my old ears was the biggest challenge of the night. Being able to see the signal on the HDSDR AF spectrum helped considerably. I found that I could hear the signal nearly all of the time. I could hear the changing tones of the Q65-60E, and I could hear the CW. Copying it was another matter! I could barely get a few characters from time to time. Being able to copy even a few characters was a thrill I will long remember! I wish I could squeeze another two or three dB S/N out of this system. In this clip I can copy "0SHF" and "DL0".
Update: Listening to this file with different headphones or room acoustics I am able to copy the full call sign!

DL0SHF CW audio sample

Finally I had to see if I could measure any moon noise. I didn't take time to tweak all of the settings to get the most stable measurement, but sure enough it is there, perhaps about 0.15 dB.

The system is the Winegard DS2076 dish (76cm offset), unmodified Bullseye 10 kHz LNB, old (non-TCXO) RTL-SDR stick and HDSDR running on a laptop that is well past its life expectancy. The laptop adds a layer of challenge. It is very slow and sometimes maxes out the CPU which seems to result in visible and audible audio drop-outs. For whatever reason it cannot successfully accept one of the big Windows 10 updates, but Microsoft insists on retrying frequently. Sure enough, when it was time to shut down at 0300 local time, the update was waiting to install. The battery is long deceased, so I had no choice but to let the update install on shut down. I left the laptop outside doing that while I came in for some sleep. In the morning it was finished, so I brought the laptop inside to retrieve files from it for this post. Of course the other part of the update had to run. As always it ended in an error and had to reverse all of the changes. That process takes about three hours, during which time I was impatiently waiting for access to my files from the night's adventure.

For information about the DL0SHF beacon, try here: The DL0SHF 10GHz EME Beacon

August 23, 2022

I tried to compare my three LNBs by looking at sky to ground noise Y factor. This is dicey at best, as I have a very limited view of the sky. I did my best to make sure all three LNBs were pointed exactly the same when looking at sky and ground, but there may be differences in beamwidth of the LNBs and I cannot be certain they weren't seeing the tops of some trees. These tests should not be taken too seriously, but it appears any noise figure differences between the LNBs is relatively minor? The first is the Avenger KSC 321S-2 LNB. The second is the Bullseye 10 kHz LNB I have been using for previous tests, including EME reception. The third is another Bullseye 10 kHz LNB I recently obtained as a guinea pig for hacking.

August 30, 2022

As noted in my comments about EME reception, drift is a problem with the Bullseye 10 kHz LNB. My old non-TCXO RTL-SDR stick was likely contributing to the problem as well. Last week I attempted to modify the Avenger KSC 321S-2 LNB to use an external LO following KA1GT's notes. I was successful in destroying an Avenger when two very fine PCB traces ripped off the board during my attempt to remove the crystal! Oops. OK, moving on then. I was curious to see if it would be possible to use an external LO with a Bullseye 10 KHz LNB. These have the second type F connector for internal LO monitoring. The TCXO output passes through an attenuator to this connector. Would it be possible to feed an external LO into this connector without further modification after removing the TCXO? Would the LNB lock to 25.476923 MHz to get a 432 MHz IF?

The Bullseye was easier to disasseble than the Avenger. On the former I had broken several of the little clips holding the two halves of the plastic case together prying it apart. On the Bullseye I was able to reach inside through the oval cutout around the connectors with a small screwdriver and pry open four of the six clips while squeezing the sides of the top half of the case. It came apart with no broken clips. A bigger challenge is the white silicone sealant around the perimeter of the metal case and edges of the PCB. Removing that required careful cutting with a knife and lots of picking at it. It is necessary to be gentle about prying the PCB loose, as it may be easily damaged. The TCXO output is a sine wave. Using an oscilloscope I measured the voltage at the TCXO oputput pin at approximately 700 mV peak to peak. The heavily attenuated level at the red F connector was much less. My attempt at removal of the TCXO with the little PCB still in the metal case was not successful. Apparently there is fairly good thermal transfer from the PCB to the case. My hot air station was not able to overcome that to melt the solder. Only the two F connector center pins need to be unsoldered to remove the PCB. I was able to do that with my soldering iron and solder wick. Once the board was out, hot air easily removed the tiny TCXO.

I tried injecting 25,000,000 (for 618 MHz IF) and 25,476,923 (for 432 MHz IF) from a Leo Bodnar GPS reference clock, but the Bullseye did not lock to it. The Bodnar output is square wave but the waveform at the old TCXO output pad was a mess, neither square nor sine wave. It had multiple peaks and looked more like 2fo than fo. The peak to peak level was only about 300 mV. I didn't know if failure to lock was due to the waveform, insufficient level, or both. Next I put a Mini-Circuits BBP-27R5+ band pass filter on the output of the Bodnar. This is a 24 to 31 MHz filter, resulting in a nice looking sine wave. I now saw a clean sine wave at approximately 300 mV peak to peak at the old TCXO output pad. The Bullseye had no problem locking to this at 25,000,000 or 25,476,923. Up to this point I had been running the Bodnar at its maximum output level with drive set to 32 mA. I found that I could reduce it all the way to the lowest drive setting of 8 mA and the LNB remained locked. Progress! I connected the LNB to my Q5 Signal 432 transverter and switched on my 10,368 weak signal source. I had a nice strong signal about 180 Hz above 432.000. The small error is most likely the transverter, which is using its internal TCXO. (Note: 25,476,923 Hz times 390 comes out to 9935.999970 MHz, or 30 Hz below target. This is the best we can dp with 1 Hz steps on the Bodnar. During a two hour observing period there was less than 20 Hz drift. It appears the Bullseye LNB can be locked to an external LO via the red F connector with no modification other than removing its internal TCXO. Great! There is some phase noise evident and a spur only 20 dB down about 60 Hz below the fundamental. Further investigation is warranted but for weak signal work such as EME I don't think this will be a problem. With a signal about 35 dB above the noise floor, phase noise did not appear to present any problem for receiving signals a few kHz removed from the moderately strong one. I did look at the Bodnar output spectrum around 25.476923 on the spectrum analyzer and found it to be clean (but that spur at only 60 Hz removed is too close for me to see if it is present). I should have taken photos of the waveforms and levels noted above, but didn't. I was too interested to get on with the work and see the results. Sorry about that!

Update: I ran into some problems with this! It turns out the noise figure was significantly degraded operating at 432 MHz IF. See the next post for details.

September 14, 2022

As noted in the last post, I ran into some problems getting the LNBs to work with 432 MHz IF. I realized something was wrong when I checked the DL0SHF beacon and found SNR to be 5 dB lower than it had been with the stock LNB. Both the Bullseye and Avenger LNBs do work when moved to 432 IF, but performance suffers. They still have plenty of gain and definitely receive signals, but noise figure takes a dive. The follwing Cold sky to ground noise plots show the problem. Less difference between cold sky and ground means poorer performance. The good news is performance at 618 MHz IF remains the same after modification. So they can be stabilized using a GPSDO or other accurate external reference, and the reference level does not seem to affect performance as long as it is enough for the chip to lock to.

First the Bullseye 10 kHz LNB (BE01):

And now the Avenger KSC321S-2 LNB:

The Bullseye appears to have a lower noise figure than the Avenger, both stock and modified as long as the IF remains 618 MHz. It is also far easier to modify for external reference, making it the preferred choice in my opinion. Interestingly the Bullseye suffers a much greater performance reduction when moved to 432 IF than does the Avenger, but neither LNB is happy there.

However, remember these are results with LNBs with my 76cm dish. The results may be misleading. If the Avenger has a wider beam width, it may simply be picking up more noise from ground and trees behind the dish, making it appear to have a higher noise figure. I am not confident that any LNB can be properly tested alone in my environment due to the limited sky window. A bare LNB may always see some trees.

Is there any hope for improving performance of the Bullseye LNB at 10.368 GHz? Possibly. I did a quick check at 10.968 GHz to 1218 MHz using the nominal 25.000 reference frequency. The cold sky to ground Y factor improved 0.5 dB over the 10.368 to 618 using the same reference. Bear in mind this is with the LNB on my 76cm dish. The dish should have about half a dB more gain at this frequency. I have not tried to work out the calculations to see whether this improvement in cold sky to ground Y factor is all due to more gain or partly due to better performance / lower noise figure of the LNB. Adding a tuning screw opposite the vertical probe in the Bullseye might offer some improvement at 10.368. An experiment for another day!

September 15, 2022

I had been digging around the internet off and on for weeks looking for some way to implement WSJT-X doppler tracking with HDSDR and my RTL-SDR dongle. I had found a couple of methods that were quite complex and required installing more software. This was one reason I wanted to use 432 MHz IF, so that ultimately the receiver would be my FT-2000 which is easy to control directly from WSJT-X. The other reason had to do with not liking SDR for extreme weak signal CW copy. Eventually small hints here and there come together in my foggy head and I realized it should be possible to control HDSDR directly from WSJT-X if I had a connected pair of virtual COM ports. Having prior experience with VSPE from my 2200 meter days, I chose it to create a pair of virtual COM ports. Configuration was easy.

First, start VSPE and create a pair of virtual ports:
Click create new device.
Select Pair for device type, then click Next.
Choose any two available COM ports. Leave emulate baud rate unchecked. Click Finish.

Configure HDSDR:
Options>CAT to HDSDR>Port, select either of the two virtual ports of the VSPE pair.
Options>CAT to HDSDR>baud rate, make a selection (I chose 9600).
Options>CAT to HDSDR>activated, just click on it.

Configure WSJT-X:
File>Settings, Radio tab.
Choose Kenwood TS-2000 for radio type. This is the command set HDSDR understands.
Choose the other COM port of the virtual pair.
Select the same baud rate you chose in HDSDR. Ohter port options can be left at Default.
Select Fake It for the split option. Although not used to control HDSDR, this is required by WSJT-X.

That's all there is to it. Open the WSJT-X Astronomical data window, click the box for doppler tracking, and select the type of doppler compensation you want. It should just work at this point. Receiving EME signals is much easier with a GPS referenced LNB and automatic doppler tracking! Compare the nice straight line on the waterfall with the one from August 15.

September 18, 2022

I had clear skies for locating the moon on Friday night, which happened to coincide with one of the ARRL EME Contest microwave weekends. I was able to decode the DL0SHF beacon at -12, W3SZ -16, GB2FRA -8, IK0HWJ -16, HB9Q -15, and OZ1LPR -15. It was very gratifying to see that my simple LNB based setup is capable of receiving a number of stations on EME! GB2FRA had a really nice signal and was clearly visible on the HDSDR waterfalls.

Right at the end of my moon window, GB2FRA was working DB6NT on CW. I knew I didn't have time to switch to CW mode in HDSDR and adjust filters to get optimum copy and the dish needed to be moved so I did the best I could in SSB mode. I am very disappointed by having to use SDR as they are so awkward for me to adjust for very weak signal copy by ear. A conventional receiver would have made this so much easier. I was able to "easily" copy both call signs, RRR, TNX, and 73 but I never really copied the signal report during the QSO. Here are some sound clips:

Early in the QSO, 800 Hz pitch (not good for my ears!)

Later, 500 Hz pitch (ah, that is better!)

End of the QSO, 500 Hz pitch

September 24, 2022

I was getting a lot of questions about the EME RX setup so I made a Youtube video wherein I describe it in some detail. One thing I did forget to mention in the video is polarization. I rotate the LNB in the clamp to set the correct polarization for the station I am listening for. Europe mostly transmits vertical, and I am using the vertical probe in the LNB because it has lower noise figure than horizontal. So for Europe I simply rotate the LNB the specified number of degrees for the Dpol given by WSJT-X. A negative Dpol means rotate the LNB counter-clockwise as would be seen looking from the dish surface or behind the dish. It's slightly more confusing with North American stations who usually transmit using horizontal polarization. First I rotate the LNB 90 degrees to get horizontal plolariztion, then rotate it according to the given Dpol figure from there. Clear as mud?

Last update September 24, 2022