Wednesday, May 29, 2013

The Reason I Hate Windows

It was going to be a typical casual contest weekend. I had the new 160/80m Inverted L ready to go. I had hoped to work a few more countries on 80m for DXCC, as I inch ever closer to 5BDXCC. My employer even let us out a little early, so I should have plenty of time to get everything set up. Right?


It seemed simple enough. Just crank up the Acer and set the N1MM software for the WPX CW contest.   The Acer started ok, but for some reason, the N1MM software would not run. Strange, it ran just fine last time I tried it. Well, that's ok, the copy I have is a few months out of date anyway, I ought to update the software to the latest before the contest.

So, uninstall N1MM, go out to the site, grab the base (2011) installer, install it, then install the latest build. Try running. Oh, it says it has to reboot before you can run. Go to reboot - the Acer waits a long time at Logging Out....

And it was about time to go off to marital arts class, so I just left the computer Logging Out.... I figured it would be done by the time I got back, and maybe I would finish the setup then. But, after three hours of class, I was pretty beat and figured I would tackle that job in the morning.

So, around 1300z, I hop out to the shack and crank up the Acer -- and it is STILL Logging Out.... Well, it spent so long trying to log out that the computer went to sleep. I'm out of patience at this point. So, I pull the battery, and unplug the power, and poof, it is ready to reboot.

Turns out, the reason it was taking forever to log out was because it was installing updates. You see, somewhere in the infinite wisdom of the Windows designers, they decided that the best time to install updates was when you were trying to log out of the computer. It obviously never occurred to them that you might want to log out only to log back in on some other account, or perhaps just reboot the machine so some software you just installed would work. No. You say you are done using that account, and it's time for Windows to take over and install the almighty updates.

Oh, and if you happen to bypass that by yanking the power and rebooting the machine, Windows takes care of that by installing the updates when you boot up. But, at least during the boot-up phase, it gives you progress information as to what it is doing, rather than simply saying it is Logging Out....

Half an hour later, it finally finished the update process, and I could successfully run N1MM. Set it up for the WPX CW contest and....

It can't talk to the K3.

OK. It was working just fine back in March -- what changed? Well, nothing, really -- except I had just installed an upgraded copy of N1MM. The Elecraft K3 Utility has no trouble talking to the K3 at all through that same port. Hmm. Maybe I should go back to the old version that was working, except when it didn't work for some weird reason.

So, uninstall N1MM, go out to the site, grab the base (2011) installer, install it, then install the latest build. Go to reboot - the Acer waits a long time at Logging Out.... Hey, I just installed all those updates! Pop battery, pull plug, restart. OK, there's just a handful of updates, should only take a few minutes.

And, once Windows had finished updating AGAIN, run N1MM and.... It can't talk to the K3. Looking more closely, it is repeatedly reporting 8020 errors. A bit of searching the internet, and this appears to be a problem related to the serial port drivers. Seems some USB serial port drivers work fine with other applications (and the Elecraft K3 Utility had no trouble talking tot he K3)

So -- what changed? Looking at the Plugable site, the latest driver is v1.8, and that's exactly what's installed. However, from my work back in January, I remember a different version -- v1.7. Maybe that's what happened -- one of those many updates must have updated the serial port driver in a way that's no longer compatible with Visual Basic.

Easily fixed, right? Just install the old driver. Well, first you have to FIND the old driver. Unfortunately, I did not make a copy of it when I installed it in January. Fortunately, after about 15 minutes of searches, I did find a copy of v1.7. Uninstall, reinstall, and pray that plug-and-play doesn't just go update it all anyway.

Behold, start up N1MM and -- it can talk to the K3! It's now almost 1500z, I've wasted two hours of contesting time fighting with the stupid computer over what was, in reality, a dependency of the N1MM software on a serial communications driver that hasn't been supported by Microsoft for five years now. (All Visual Basic support ended in March 2008)

I need to wean myself off Windows software. If only there were decent contesting software for the Mac. Hmm.

Tuesday, May 28, 2013

160m / 80m Inverted L

Just left and below center you can see the trap hung up on
a branch at around 42 feet.
From the minute I put up the 160m Inverted L, I had planned to add a trap for 80m. First, however, came more radials. With the original four 125 foot radials, the antenna seemed rather quiet, and that should have been a sure sign it was too lossy.

Adding four more 125 foot radials made a big difference. Noise level went up, along with the performance. Eight more 62.5 foot radials followed, for a total of sixteen - eight long and eight short ones. For any antenna with ground-mounted radials, sixteen should be considered the minimal number of radials. At least, for any antenna not mounted near salt-water.

I used this antenna to work and confirm two new countries on 80m phone in the WPX Phone contest. 

OK, so radials are easy. Not cheap, since a 500 foot spool goes for $45 these days. The next step was to add an 80m trap. The easy way to do this is to simply create a resonant trap on the operating frequency and stick it into the antenna by trial and error.

80m trap wound and set up for
trimming, Note the temporary solder
connections with the capacitors.
However, that's not the most efficient. W8JI wrote an excellent article about making efficient trap antennas. Two important lessons from this article: traps work best when they are made from very high-Q components, traps should never be resonant at the operating frequency. 

With my previous 80/40m trap antenna, I had followed half of this advice -- but still used coaxial traps. W8JI found that these traps are much more lossy than those made with discrete components.

And since wire is so expensive, trial and error isn't the best way either. The tricky part about using traps that are not resonant at the operating frequency is that some antenna current flows in all parts of the antenna at all frequencies. This means adjusting one part of the antenna affects the resonance at all frequencies. And while a trap at resonance offers an effectively infinite impedance regardless of the actual values of capacitance or inductance -- off-resonance impedance is definitely affected by the choice of capacitance and inductance.

So, how does one figure out all these variables? Antenna modeling! I used CocoaNEC, developed by Kok Chen W7AY. While it is pretty easy to use the spreadsheet model for very simple wire antennas, I ran into some bugs trying to model trap antennas. After e-mailing Chen, I discovered that Chen recommends the NC interface (a C-like programming language) for modeling, rather than the spreadsheet.

Once I figured out the NC interface, I started to get better results. Then the virtual trial and error part began. Lots of programming, running and bug-fixing later, I had a pretty good idea what was needed to build this antenna.

I found some 100 pF 15 kV ceramic disc caps from Mouser, and used two of them for 200 pF. I used a piece of 3 inch schedule 20 PVC for the coil form. I computed that it would take about eleven and a half turns on this form of close-wound 14 gauge THHN wire. THHN is really designed for house wiring inside a conduit, but it is relatively cheap and easily obtained at your local home improvement store. The wire is secured to the form by drilling 1/8 inch holes through the PVC.

Proper technique for measuring trap
resonant frequency. Note wooden work
surface and nothing metallic nearby.
Unless you have a vector impedance analyzer (and who does?), the easiest way to see if you trap is even close to the right frequency is to use a Grid/Gate/Emitter dip oscillator. Mine is a Heathkit HD-1250. I found it at a hamfest years ago for about $30 including all the coils and carry case. It was modified to add a switch to test the battery condition on the meter, and whoever did the mod did a great job. Only two things wrong with this unit: the lettering around the meter has completely rubbed away, and the foam inside the case to hold the coils down completely disintegrated. (I have never seen a HD-1250 carry case where the foam has held up)

Once you have your trap built, careful technique is necessary to measure the frequency. The coil end of the dip oscillator couples to the trap. Couple too closely, and the trap will pull the oscillator, and it will be difficult to find the exactly frequency. Couple too loosely, and you won't find the dip at all.

I set my trap on a wooden workbench and cleared away everything metal for at least 12 inches around. First, couple the coils very closely to be sure you can find a dip in the meter indications somewhere close to what is expected. Then slowly move the meter away for less and less coupling. The best spot is where you get a good meter indication, but it doesn't pull the oscillator too much -- which you can see as you tune across the dip. From the photo, you can see that the best spot was found with the dip oscillator coil just outside the trap form.

Great -- so you've got a good dip on the meter with the right amount of coupling. What frequency are you on? Good question. Unless your dip oscillator has an output for a frequency counter (another useful mod), the easiest is just to spot the oscillator frequency in a nearby receiver. In reality, the exact frequency of the trap doesn't matter so much, so the dip oscillator dial is probably good enough.

Completed trap ready to install.
I modelled my traps at 3450 kHz, but the trap I built was resonant at around 3350 kHz, which I deemed close enough. I drilled a four more 1/8 inch holes for securing the antenna wire to the trap.

With the trap ready, it's time to install the trap into the antenna. My model told me that the trap should go around 47 and 1/3 feet above the feedpoint. I started at 48 feet and used an MFJ-259 antenna analyzer to spot lowest SWR. About three trims later, I have 44 feet to the trap bringing the low SWR just below 3800 kHz.

Completed trap temporary installed for trimming. No solder
used on the connections yet, just twisted together. Note
how the antenna wire (black) is looped up and back down
the trap. This will hold it securely in place.
The modeling work told me that the 80m frequency wasn't affected terribly by changes to length of the upper portion of the antenna. I did a couple of quick calculations on my phone and cut the upper portion to 67 feet. Low SWR came right in at 1830 kHz. Perfect. 

Using the analyzer, the actual antenna measurements don't exactly match the model. For one thing, the NEC 2 model software assumes perfect grounds, and the actual ground is a bit more lossy. The model shows very sharp resonances, but the antenna measures more broadly -- indicating expected ground losses. 

Using NEC 4 would allow more realistic ground models, but I didn't feel like it was worth the $300 to get a license to it. In any case, the modeling did exactly what I expected -- it guided me to produce a workable antenna design in the field.

How does it play? Works pretty well on 160m still, and I used it in the WPX CW to work another new country on 80m with 100 watts. Of course, it's not the right season for low band work right now, but I think this antenna has promise for next fall.

Of course, I could slip in another trap for 40m. Hmm. Let's fire up that antenna modeling software....