A day of troubleshooting ...

... a very unusual problem that I've never heard of before in my ham shack, and I'm not sure I have anything to share. If you've been down the road of troubleshooting intermittent problems you're familiar with the idea that you may spend a lot of time deciding if you've fixed the intermittent problem, or just made it more intermittent. 

For one thing, I don't have single picture of the failure when it happens. I've been too distracted by trying to "make it stop!" 

Let me start from the beginning. Back in about 2018, I started running the various modes offered by physicist Joe Taylor, the vast majority of the time on what's called FT8. Yes I've used others, mostly FT4 which is like FT8 but less sensitive (less able to copy very weak signals) by using shorter transmit and receive intervals. The trade-off being you give up some sensitivity for shorter contacts - for those days (like today) when maybe you hear another state for a minute. I've used its meteor scatter mode, moonbounce mode, and more. 

Over the years, I've worked FT8 on all HF bands and 6m. My station has antennas for 80 to 6m except for 60m, a small chunk at around 5.5 MHz. 

A couple of nights ago, with 6 being as dead as it has been, I dropped down to HF, but instead of going to 80/40/30 as I typically go, on a spur of the moment decision, I went to 17m (18.068–18.168 MHz). It's a small band that hams were granted in the '79 World Administrative Radio Conference, or WARC, and the three bands that came out of that are known as the WARC bands. (If you've never heard of 17m, this seems to be a good overview). 

One the things I'm used to doing as last check on my station before I start to operate is hit the tune button, which just transmits a carrier at whatever your rig is set to. The (always there inline) power and VSWR meters looked perfectly normal, but nothing else did. Let me show you a screen capture of FT8 software (tuned to 30m - that's the only thing that's different) to make some points.

There are two windows here, the top one is the audio channel from 100 to 3100 Hz, each of the colored areas at the top are individual stations. In general the orange strands that are the most orange (like around 1700 Hz on the top scale) are the strongest, fading down to the more yellow ones (around 2000 Hz and that green rectangle). The spectrum at the bottom is more like a regular spectrum analyzer and is only showing the very topmost thin line of colors. The taller the spectrum the stronger the signal. 

When the "bad" happened, that topmost line was replaced by what looked like a continuous spectrum of lots of narrow "spikes," all red. Except spikes tend to have points and bases. This was just alternating big signal and nothing. Maybe like a comb, but all the red "teeth" were the same height and width. 

It gets weirder. The needle in the inline power meter that had previously measured the normal roughly 100 W transmit signal was now not showing any power at all. The receiver sounded muted, as it does during transmit. There's a transmit/receive indicator LED on the radio that stayed in transmit (red) instead of receive. 

It took a while to get it to stop; I had to reset everything. Turn the radio off and back on, reboot the computer, and several things. 

Yesterday, I started looking at it in more detail. I found that it did this on 20m but not higher: 15 to 10. This is helpful because 20 and 17 are on the same antenna, my Tennadyne T6 log periodic (LPDA). The other bands I've been tending to operate, 80, 40, 30 and 6m, haven't shown the issue. Then playing around on 17m, I found I could operate CW or SSB with no problems showing up. The frequency sensitivity made me think something in the LPDA's circuitry might be the problem. Looking up from the ground shows nothing, but it's at 25' above ground, so the view isn't that great. An easy next step seemed to be to check the grounds around the antenna. 

So that was today's start. All I could say was that the grounds looked just like the last time I did my routine inspection last spring. Nothing broken, nothing missing, nothing looking like the ground wire was replaced by weeds. They just looked like they've been outdoors and corroded a bit. The next thing to check was connections in the shack and while checking things on 17m again, I noticed the rig control software (Ham Radio Deluxe - HRD) would not always see the radio. So I did the old "unplug it then plug it back in" trick. That's a USB connection from the radio to a little box that sets up the modulation for these WSJT modes. That made a difference. It was easier to recover from the "red comb". 

Late in the day (it was nearly 5PM by this time) I decided to retest the setup on 17 and the problem was gone. I tested a couple of times and it worked. I tested 20m - no problems, and worked my way back up to 10 without seeing the problem again. Then I went back to 17 to "soak in the success" of just unplugging and re-plugging one cable. That's when it failed again. 

What was that thing about making the intermittent problem more intermittent? 

Around the time I sat down to write this I realized, I could use far more pictures than I've got. So I went back to the shack, tuned to 17 and hit the tune button to transmit. No problem. Tried it again - surely it must fail by now? - Nope. I think I transmitted 10 times and it never had the failure. Then I tuned it to a band I've never seen it fail on, 30m, for the picture above. 

What's next? Can't say I really know. More experiments and measurements, but I sure don't know what to go after. It looks like it's in the computer control of the rig. I've tried two different versions of the WSJT software, the latest and a somewhat earlier release. I'm on the latest update of HRD.