Operating FT8 on the Elecraft K3

Last night on 30m

Last week, I put up a new HF antenna and this week, I set the station up to operate in FT8 mode. Below, I describe a minimalist implementation using an Elecraft K3, a computer and two audio patch cords. I am focusing only on items specific to the radio and not found in the WSJT-X documentation.

Why FT8?

Why FT8? First, because it is popular, which means that there are people at the other end listening (or sending). Since I have lived overseas since FT8 came out, just about everyone in the US is a potential new contact for me. This is a good chance for me to work on my digital WAS. Second, because my power output is limited, for a digital mode, I can’t go above 50W on HF. That’s something of a limitation for phone, but certainly not FT8. Third, and closely related, we’re nearing the bottom of a solar cycle. If I’m going to make contacts on the upper bands, I’ll either need some luck with sporadic openings or to run an error-correcting weak signal mode, or both. Finally, because there is often a lot of local EM noise. I can listen through it, particularly for CW, but turning the sound off and letting the computer sort through the buzzing is also attractive.

Why the K3?

Why the K3? It requires less equipment. Previously, I ran digital modes to my TS-2000 using a rigblaster nomic to provide isolation between the computer sound card output and the rig’s microphone input. However, the K3 has line in and line out ports built into the rear panel. Both of these ports are coupled via internal transformers, which means that both lines are galvanically isolated, not just audio input to the rig. It also means that I don’t have to either give up my front mike jack or wire up a complicated accessory port to get sound in and out of the radio.

Audio Connection

I am currently using a 2007 MacBook Pro to run WSJT-X version 2, so I am taking sound from its headphone jack and running it to line-in on the K3, and sound from the line-out jack on the K3 to the microphone input on the Mac. As with most computers, there is no line-level input on the Mac, so I needed to turn the Mac’s sound input sensitivity almost all the way down. In principle, I could have put a resistive attenuator inline, but I found that wasn’t necessary, so I saved myself the trouble.

To get the K3 to input sound from the back panel, it is necessary to go into the main menu and turn “MIC SEL” from its usual front panel mike jack setting to line input.

When the microphone selection is set to line input, the “mic level” knob on the radio’s front panel now controls the sensitivity of the line input. When you switch back to the front panel microphone, the previous setting of the microphone level is not affected.

The radio’s line output sound level is set by a configuration menu option, “LIN OUT”. I found that level 2 worked for me. Again, this setting is independent of the headphone/speaker settings, so you can dial the volume up or down, and it will not affect the level going to the computer (which is what you want).

VOX versus CAT

That’s the extent of my interfacing. I decided not to add a CAT control, although I could have done so through a USB-to-serial cable. I am using the K3 as my main CW contesting rig, so it is already plugged into another computer and I would rather not need to switch the serial line between computers. Instead, I am using VOX control, which works fine.

The trick with the VOX control is that the computer’s sound output has to be loud enough to trigger the radio’s audio detection circuit. The radio’s sensitivity is set by the main menu’s “VOX GN” setting, which runs from 1 to 100. To figure out where to set it, put the rig into test-transmit mode so that no signal goes out over the air (long press on the right side of the mode button). Then hit “tune” on WSJT-X to send a tone. Turn VOX GN high enough that the radio’s transmit is triggered and then dial it down to the point that the radio is no longer keyed. Then turn the vox gain back up a bit over that threshold to provide some margin.

Getting the audio levels right is a matter of playing with the computer’s sound card settings as well as the K3’s line in and out levels. The instructions for WSJT-X indicate that the program’s sound bar should show about 30% on a dead band, so I set the controls accordingly. As for output, it is affected by the computer volume level, the power slider in WSJT-X, and the line-in sensitivity knob on the radio. My goal was to set the level such that the ALC showed not more than four bars — the point at which ALC kicks in on a K3. Again, this meant putting the rig into “test” transmit mode and hitting WSJT-X’s “tune” button.

All settings that would distort audio should be off. So, on the transmit side, I turned compression to zero using the front panel knob. For receive, I turned off the AGC (first taking my earphones off), as well as the noise blanker and noise reduction. I dialed up the SSB filter to max width, which for me is 4 kHz. There’s no need for any sort of filtering or DSP — let the computer handle all of that as it sifts through the noise for FT8 signals.

Power

At this point, the rig is ready to go on the air. Set the band within WSJT-X and then set the rig to USB mode and the dial frequency just as shown in WSJT-X. Also: don’t forget to make sure your computer’s clock is precisely synched. Although my mac is set to pick up internet time, I usually force a synch before an FT8 session by typing “sudo ntpdate -u time.apple.com” on the command line.

Before transmitting, make sure that reception works. If all has gone well and you’re on a band that isn’t dead, you should see signals on the waterfall and their corresponding decoded text within WSJT-X.

Folks have different philosophies about power level, but a practical upper bound is half the rated power of your rig. The K3 will do 100W, but for a full duty-cycle, I won’t go above 50W; on times are only 15 seconds, but if you are having a good day, the rig might not cool well with frequent cycles, and heat definitely does shorten component life.

I’m not advocating operating at that much power, though. Statutorily, you are supposed to use the minimum necessary for a given purpose. If your goal is DX contacts, that purpose might require more power, but if you are interested in more local QSOs, less power.

Luckily, some feedback is available to help you pick the right power setting: PSKreporter (and to a lesser extent right now, the reverse beacon network). Last night, I settled on 20W for 30 meters because I could see that at higher power I was being heard by stations that I could not myself hear, likely due to differences in background noise levels. It makes no sense to spam the airwaves with calls if I can’t hear the reply. This method isn’t perfect since I don’t know how much power the other stations are running, but it’s a rough guide.

Here’s what I heard last night on 30m. I think the Pacman shape of my receive pattern is mostly due to lack of stations in that part of the Atlantic.

Similarly, for DX, having higher power is probably less important than finding clear bandwidth and being properly time-synched with respect to the other station. It’s not like the other station is listening for the loudest signal on a single frequency — it will probably respond to the first intelligible signal it detects anywhere in its passband (although the timing may be more random if it’s operating in DXpedition mode).

With this set up, I was able to fairly consistently pull fish out the pond, one after another for hours. I’m sure that will drop off as more people have me in their logs, but it was reassuring that the set up works as designed.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.