A while ago I did build my first 30m WSPR-RX to participate in the WSPR-challenge. I must admit it is very addictive! It did not took long before I was building my second WSPR-RX. This time a little bigger since the first one was too much miniature. Also I wanted to add some tweaks.
WSPR is a very small band signal. Actually it only covers 200 Hz on (30m) 10,138.70 MHz. When you build an x-tal oscillator for this receiver, it has to be quite stable to be able to hold the WSPR-RX between its lines. And the original design drifts. Too much for me. At some point I could read the rooms temperature from the drift of the decoded WSPR-signals.
In the original PA2OHH-design of this receiver there is an Negative temperature coefficient thermistor (NTC) with a varicap to compensate for frequency drift. I think Remco PA3FYM has left these parts out for simplicity but I wanted to build it back in.
Voltage regulated fine tuning
Furthermore to put the oscillator on frequency is a little challenging by itself. Since the original design almost never hit the 1500Hz band middle of the WSPR-band right away, you have to tune a little to get it right. Great, but how do you tune an oscillator that has no tuning options?
So I did build (with tips and hints from Remco PA3FYM) a voltage regulated add-on. As it turns out, you can re-engineer an 7805-voltage regulator to a variable regulator pretty easy. I did use this concept to design a voltage regulator for the whole design. I’m able to set the voltage between 11 to 14 volts which is enough to set the oscillator somewhere between 1400 and 1600Hz.
In the meantime I am able to reach in the top 20 of the WSPR-challenge. Consequently to get into the top 10 seems very hard! I need all the (WSPR)-ears I can get. As a result I took everything I could find (including Roeland PA3MET‘s Sony ICF-SW7600G short wave radio) to create as many WSPR-receivers as I could in addition to the home made WSPR-RX’s. But it is still not enough to reach the top. I need to come up with another plan.
My latest acquisition to the shack is a RaspberryPi-3. A great birthday gift to get! I already owned two RaspberryPi-1’s and one RaspberryPi-2. One of the RaspberryPi-1’s is doing a great job to put out the telemetry of my weather station with Xastir. I wanted the RaspberryPi-3 put to work as a WSPR-decoder as extension on my latest project: the 30m WSPR-RX. Still working on tweaks for the RX though.
My first hunch putting an OS on the RaspberryPi-3 was to use a special light version of Ubuntu (Ubuntu Mate). Unfortunately WSJT-X (the software to decode WSPR-signals) as a Ubuntu package is pretty old (version 1.1 or something while the latest is 1.6). So I decided to compile from source. That turned out to be a little laborious since you need a lot of extra source code packages that all named different from the documentation. After an hour or so searching the source was compiling. When WSJT-X was ready it did run. But after a few starts it gave errors which I couldn’t fix. It drove the RaspberryPi-3 to a load of 100% and couldn’t recover. Back to square one. I changed Ubuntu Mate for Raspbian. And I was already compiling again from source when I stumbled upon this link. A simple PPA with the latest and the greatest of WSJT-X: version 1.7!
I did a few changes to the standard setup of Raspbian; since I run the RaspberryPi’s kind of headless, I installed tightvncserver. I just switched off the default lightdm and enabled the vncserver from boot.
Something else I did was adding WSJT-X as startup application. This way after a power failure or a spontaneous reboot, WSJT-X would start automatically again to pick up it’s task.
I still need to add a cronjob to check every 5 minutes or so if the program stil runs and if not: start it again.
After a few hours playing around with the new RaspberryPi-3 I very much like the better speed! The RaspberryPi-1 is just plain slow, the RaspberryPi-2 is a little better, but this latest edition is very good workable as all rounder in my shack. I put in a cheap Chinese USB-sound stick to decode the WSPR-signals from the WSPR-RX. It works problem-less. Booting is actually pretty fast and Raspbian is very usable. The RaspberryPi-3 isn’t getting warm or anything (you read about these errors lately) I run it on a good but simple Apple 5v/1A adapter which runs fine. Tonight I’ll do another adjusting session at the club’s lab of the WSPR-RX and then I’ll let this combination run for a few weeks.
There is buzzing a new phenomenon at PI4RCG (our local HAM club): the WSPR-challenge. One prominent club member, Remco PA3FYM, founded this WSPR-challenge with Richard PE1ITR. Essentially they download the whole WSPR-spot-database once every night and parse it in a top 50. The result changes per day. This challenge only consist of reported spots to wsprnet.org. So you can sneak peak at wsprnet.org site to see your uniques. Since wsprnet.org only logs your reported spots, this challenge is actually a RX-only contest.
Of course you can leave your main rig running 24×7 to participate in this WSPR-challenge. I did just that for a few days to see where I would end up in the WSPR-challenge. I know a few amateurs who took Remco’s design of a really simple WSPR-receiver and build it their selves. This receiver performs really really well. All builders are in the top 20 of the WSPR-challenge. According to Remco’s explanation big part of it’s success is to find and eliminate all distortion source in your area. Also use of beverages, loop-antenna’s etc. could help you to drive up your unique spots in the WSPR-challenge.
I did a little reintroduction of my horizontal-loop-wire to push up my spot-rate. Unfortunately (or maybe luckily) there wasn’t much performance gain in comparison to my ZS6BKW. Which conclude into two possibilities: or the noise level on 30m is so high it is the same on the loop and ZS6BKW, or the noise-level is already low and isn’t improved by a closed loop.
For the last part of your WSPR-challenge setup you need a computer to decode the WSPR-signals and upload them to wsprnet.org. The absolute best performing decode software for WSPR is WSJT-X. To really run it reliable you need to run it on linux. If you don’t want to leave your shack pc on continuously, you can use a Raspberry Pi 3 for this task.
I’ve learned a lot about WSPR-signals with this experiment so far. I’ve also noted that a closed loop antenna doesn’t perform better then a symmetric dipole necessarily (which I suspected). Since I wanted back my rig and shack-PC I temporarily stopped the participation in the WSPR-challenge. But I will back! On my to-do-list: buy a Rasberry Pi and build on of those PA3FYM-WSPR-receivers. That combo on an dedicated RX-antenna should be a winner.