39.4GHz converter for AlphaSat
Having worked my way up in frequency topping off at 32GHz for the NASA Kepler mission, I needed to refine my microwave construction skills once again and try to build a converter for higher frequencies. I had looked around at 38GHz microwave links but terrestrial stuff is very strong and therefore pretty easy to receive. I needed to find some signal from space around this frequency. Having a dig around on the Internet I found a Czech site detailing their commercial 39.4GHz AlphaSat receiver here. Looking at the antennas showed that with relatively small apertures reception would be possible. AlphaSat's primary mission is maritime communications in the L and Ka bands but it hosts a number of Technological Demonstration Payloads, the most interesting one is Payload 5; there is a summary write up on the ESA here
To start on the down converter project, I had a dig around in the various boxes of junk and found some Phillips 40GHz MVDS equipment (left picture below) including production-run down converters and a prototype converter unit (right picture below) with a low-noise amplifier and lens antenna. First, I looked at how the 40GHz converters worked, they have a ~14GHz DRO oscillator; this was replaced with a SMA socket and an external LO was injected just to see how wideband the front end was, needless to say there wasn't a lot of success at 39GHz. A local 4.75GHz 'brick' oscillator, locked to a 10MHz GPSDO was used with a diode multiplier to generate spurs up to 39GHz for the initial tests.
Next on the list were the XP4 microwave link units made by Stratex. I have a few of these out-door-units including some for the 38GHz band. The transmit side is shown below and is very much similar in board layout / design to the receiver.
The receiver has a few PSU requirements +5v / -5v, 8v and was quickly run up with a test PSU. The IF output was examined on a spectrum analyser and showed plenty of bandwidth extending up to around 2.6GHz. The inbuilt local oscillators VCO was fed with an external 0v to 5v supply and runs around 1GHz. Extending the tuning voltage input up to 8 or 9 volts tunes up to 1.2GHz so for the LO we are getting into the right ballpark. For my converter project I wanted a conversation LO of 40.000GHz with a high side mix making the resulting 39.4GHz signal from AlphaSat appear at an IF frequency of about 600MHz. All doable and within the passband of the converter. (I wasn't too concerned about image noise as I didn't expect 40.6GHz signals to exist)
Initial tests were made with the LO free running and connected to a multi-turn pot to adjust the tuning voltage between 0 and 8 volts. The local test carrier could be seen on the spectrum analyser but it was basically an unstable mess. The general setup with the manual LO adjustment is shown below. The lens antenna has 37dB of gain at 42 GHz so one might expect a dB less at 39GHz given its broadband nature. These are similar horn antennas that appear on eBay from time to time but they are not cheap.
A very quick test from AlphaSat with the free running oscillator did produce a signal, as expected it wobbled about a bit. The picture below shows the setup for the very first test.
The FFT below shows the result of having the entire converter assembly wrapped in bubble-wrap and left to stabilise for 10 minutes.
The solution to the wobble would be to phase lock the VCO to a good 10MHz reference. I had previously had the luck to find German Ham DF9NP http://df9np.de/ and he makes and sells great quality, low-cost PLL boards that can generate the frequency of your choice from a few hundred MHz to 6GHz or so; this was the answer so a custom PLL board was ordered to generate a 1111.1111MHz signal when locked to a 10MHz source.
The PLL board can be seen in the above image in the lower left hand side. Its internal VCO has been disconnected and the 1st LO output running around 1GHz from the 38GHz module connected in its place. Since the PLL chip only tunes between 0v and 5v an op-amp level shifter is made to add about 3v to the tuning voltage which is sufficient to get the PLL to lock the 38GHz modules LO. The note when checked with a communications receiver sounded pretty good and no signs of massively bad 'noise' could be seen in FFT.
I decided to add the broadband LNA from the pre-production Phillips MVDS unit in front of the 38GHz module since it was operating about 1GHz away from its optimised frequency, the assembly is shown below;
A horn antenna, again from the Phillips MVDS system, was used as the antenna, the thinking here being that it would have a fairly wide beamwidth at 39GHz so antenna pointing should be less critical than with a dish. I set up the converter on a tripod and fed it with volts, 10MHz reference and returned the IF to an AirSpy software radio. The appropriate azimuth and elevation was set and the SDR tuned around 600MHz to see if the signal could be improved over the initial attempt.
As expected, with the PLL locked 40GHz LO the signal is massively improved and easy to copy with just a small lens antenna. The FFT above shows the AirSpy SDR tuned to the IF frequency of 597.985MHz which corresponds to an off-air signal downlinking on 39.402013GHz. Incidentally, I did meet Volkmar HB9DUK, a fellow millimetre wave chap, at Friedrichshafen HamRadio in 2015 and he had quite a few 40GHz amplifiers and modules present, so it was good to catch up and chat about the possibilities of receiving AlphaSat and other EHF signals.
To summarise, an interesting experiment resulting in good 39GHz DX! Next on the list is to measure sun noise and cold sky / ground to guesstimate the system noise figure.