pa0r

The PERL ARQ module I am writing for the PSKMail system is doing well. Today I finished the TX side, and I have included a module to test it. PERL is predestined for a task like this; its reqexp features cannot be found in any other language. I plan to release it as a library as soon as it is has been tested in our experimental setup at the TU Eindhoven.

I have spent most of the rest of the day refreshing the CVS for tlf, on popular demand. This had been dormant since the crash of the servers of Savannah in October 2003, I have cleaned everything and added the complete source tree of version 0.9.23.

Tlf's CVS repository can be checked out through anonymous CVS over SSH with the following instruction set. . When prompted for a password for anoncvs, simply press the Enter key.

The SSHv2 public key fingerprints for the machine hosting the cvs trees are:

RSA: 1024 80:5a:b0:0c:ec:93:66:29:49:7e:04:2b:fd:ba:2c:d5
DSA: 1024 4d:c8:dc:9a:99:96:ae:cc:ce:d3:2b:b0:a3:a4:95:a5

Software repository :

export CVS_RSH="ssh"

cvs -z3 -d:ext:anoncvs@savannah.nongnu.org:/cvsroot/tlf co tlf

You can also use your web browser to browse through the source at:
http://savannah.nongnu.org/cgi-bin/viewcvs/tlf/

I hope someone will need it :)

Thanks to the help of Joerg Ostertag from the gpsdrive team, I have been able to locate the problem in gpsdrive. The hickup is probably caused by the gpilotd deamon trying to access the tty at the same time as gpsdrive, which hasn't been 'hardened' against a case like this. The gpsd (GPS deamon) is immune against it, so when I use the daemon instead of the direct serial connection I can now use the program again. This is a big relief, as we tend to use it a lot. I thought I only upgraded UBUNTU since our last trip, but after a lot of logical thinking (which programs use the tty?) and checking (what's running?) the villain was demasked. I use the Palm for reading ebooks, and I needed the pilot link to get the books on it... I can now switch gpsd on and off without killing gpsdrive. And even better, gpsdrive is now running at 20% CPU without moaning... which is good for the batteries.

On our trip to the harbour I tried to use gpsdrive, which had served us well during the spring trip to spain (Jan-Mar). It bombed. It will log on to the gps unit, which is directly connected to the ttyS0 port, produce some NMEA errors and SEGFAULT within a few minutes. This behaviour was reproduceable. The setup is exactly the same as during the spring trip, except that I have upgraded from UBUNTU 4.10 to UBUNTU 5.04...
I have spent most of the afternoon testing and trying, without finding anything. Then I remembered I had given gpsdrive only 25% of the CPU power... I then increased that to 60% and the program ran stable for at least 1 hour. Maybe UBUNTU 5.04 is stealing more CPU power? According to 'top' there is plenty of power left (72% idle). Anyway, there seems to be a timing issue in the serial gps driver, otherwise it would not segfault. To be further investigated.

We have done our yearly duty to prepare the boat for the summer. A steel boat needs lots of paint every new season. And ours is already 33 years old, so the rust under the many layers of paint takes its toll. We are a lot more flexible now we can use the camper as our own hotel. I remember the many years we had to do all the painting in rainy, stone-cold weekends when your hands were freezing in the cold wind. This time we were lucky to have 5 glorious, sunny, blue-sky-days. Still an assault on muscles that are normally used in a different way... We used 1 litre of primer and 3.5 litres of anti-fouling on the under water ship, and I got to ride the bike a lot when it turned out we would run out of paint just before the job was done... the nearest paint shop is 5 km away. Here is the before-after evidence... "Ritzelaer" is now back in te water waiting for us to go out to sea.

I am still getting mail about my interface circuit of the OMNI VI to the QSK5 + ALH 811 which I published on the web some 5 years ago... Now the OMNI VI's are changing hands as a second hand item (I have one for sale if you need one), the problem resurfaces. It turns out that the ALH 811 is not the only linear which shows wrong timing when interfaced with the TX OUT/ TX ENABLE connectors on the OMNI. A quote from the article:

The OMNI (+ the AMP) still produce output when the TX OUT output is already high, switching the QSK5 to the receive position. Luckily, this effectively destroys the small fuse lamps. To delay the switch point from TX to RX a capacitor of about 10 microfarad and some diodes (1N4148) are used. (The TX enable must not be delayed).

It turns out that the same circuit is necessary for interfacing the ORION to the QSK5/Ameritron combination
so I am republishing it here. This circuit gives me full QSK at > 40 wpm with the ALH811 + QSK5.

Hmmm?? QSK with the ALH 811??? Read all about it here...

The cw sprint seems to rouse less participants... a lot of regulars were either missing, or I could not hear them. I did not make any qso on 80, and only a few on 40. So instead of the adrenaline-generating event it used to be, it was more of a relaxing experience. Maybe the rx was dead? I did have S5 noise on the band. I worked a few USA stations without problem on 20. Anyway, I made 60 qso's, and the largest number received was below 100, where in previous sprints the top was 190.... Condx??

I missed 1 feature in tlf. The most efficient way to work the sprint is to:
call someone in S&P mode,
then give QRZ in CQ mode,
work the caller and
find the next CQ'er.

It would be nice if after the first QSO tlf would switch to CQ mode automatically. Won't take much effort I think.

Today I have ugraded the software on the shack computer to tlf-0.9.23 and hamlib-1.2.4, in preparation for the eusprint contest tomorrow. Tlf did not compile out-of-the-box, it wanted the ncurses-dev library. After I installed that, everything went without a problem. I only forgot to compile tlf with the --enable-hamlib directive, so I had to do it twice :).

Hamlib-1.2.4 also built without errors. This latest release includes support for my ORION, and I was eager to know if it worked as well as the driver I had written for tlf. I was not disappointed. Everything runs like clockwork!

I also wrote a small script to start to start cwdaemon. On UBUNTU-0.5 it reads:

#! /bin/sh
echo "starting cwdaemon"
rmmod lp
cd /dev
mknod -m 0660 parport0 c 99 0
cwdaemon -x b

#### end #######

We are back up... Just connected the 40/80 mtrs dipole to the new trx without a tuner.... SWR is 1:9 (we have a long cable on the anenna) and the signal at my QTH (20 km distance) is still good enough for testing! The output of the trx is 25 Watts.

99.9% copy! And the frequency is stable enough (3Hz after a 30 minute transmission). The frequency is 10.149, LSB, waterfall on 1034.

Wonder what happens when we connect an antenna next monday...

There is progression on the PSK front. I got an email from Martin, PA3DSC this morning (01:36) that the transceiver is ready, and can be built into its rack. So we will be at the Eindhoven Technical University this afternoon.