Another month has passed, in which I continued to use Starlink as my primary internet service, except when I needed to make Facetime or Zoom calls. From a subjective standpoint, I can tell you this: March was a far more frustrating experience than February.
The month started much like February ended. The graphs above are two views of the history data from Dishy. Figure 1a is a histogram of how often a disconnection (red is obstruction, blue is beta downtime, green is no satellites) or connection (yellow) of a given length happened. Fifteen thousand one-second beta downtime disconnects of one second or less in that week. About three periods of connection lasting 30 minutes or longer. Figure 1b is the "timeseries" chart: the color of each square is the status of the connection at that second: red/blue/green are disconnections as before, black is when I don't have data (either my collection script missed a run, or the dish was rebooting). White in this figure is just any random second that the connection was live. Yellow is only used if that second was part of a span of 30 minutes or longer where there was no disconnection lasting longer than two seconds. There are 1200 seconds (=20 minutes) per line; a day is 72 lines tall; the chart covers seven days, roughly midnight to midnight.
Things got suddenly much worse in the second week of the month. In Figure 2b, that darker blue/red band above the thin black line is March 10. This is the first day we had rain since installing our Starlink service. Far from our first weather event, but before this, it had been well below freezing for two months, so all precipitation was snow. We marveled at how little snow affected Dishy. Unless it was the wet, heavy stuff that clinged to the bare tree limbs, the connection hardly noticed. Rain, however, seems to be Dishy's nemesis.
On March 10, I reached out to support, because while our obstructions increased somewhat during the rain, our beta downtime increased far more. Their response was puzzling. I asked specifically about beta downtime, and their response was, "we have detected obstructions … [in] basically your entire field of view." No mention of beta downtime at all.
The only way I've been able to explain the support team's response is what I shared with the starlink subreddit: because Starlink is currently allowing the dish to use a lower horizon than they expect to use when the service leaves beta, they are marking obstructions that occur below the future horizon as beta downtime.
I already know I need to take care of some obstructions. It's just now starting to get warm enough to plan that. Having a reason to believe that the super noisy beta downtime I've experienced might also go away with fewer obstructions, and/or a higher post-beta horizon, gives me reason to believe the effort will be worth it.
For the last two weeks, service has suddenly been much more frustrating. In February and the first half of March, browsing and streaming would very occasionally hiccup for a second. In the past two weeks, it has been a somewhat frequent occurrence that browsing and streaming just stop for several seconds. I haven't looked at this view of the data until now, but it's nice to see that it backs up my subjective experience. Note the change in Figure 3b from lots of little red and blue dots to lots of red and blue bars.
I have changed nothing about my setup. Dishy is in exactly the same place I put it when I first installed. I keep Dishy on 24/7, with its own router plugged in. And, this isn't a change in the scenery around Dishy either. One of the trees to the east side of Dishy finally put out buds yesterday. The rest are still bare.
What did change was Dishy's firmware. On the morning of March 21, Dishy rebooted and installed firmware d61f015c. That's the lower blue band spanning the whole image except for the small green strip. The longer red/blue bars do seem to start the day before. The black band above them is roughly the start of March 20, but my notes say that Dishy was still running a8a9195a after that reboot. That firmware had been installed on March 12.
This is a beta program. It is expected that Starlink will make changes, and that not all of those changes will be obvious improvements. If anyone at Starlink is reading this, please note that that change was noticed, and it has not been an improvement.
Last week was not an especially great one for Starlink use. Figure 4b starts off with another new firmware: 5f1ea9d9. It did not improve my connection stats.
The histogram for March 23-29 (Figure 4a) shows a specific worsening trend that appeared in the previous week: fifteen second obstruction outages. Interestingly, fifteen seconds, and fifteen seconds only, saw a large jump. Obstructions lasting 13, 14, 16, 17 seconds saw no change. What's up with fifteen seconds, specifically?
The dish installed firmware b44f4294 on March 31. The stats look the same as the past two weeks to me: large spike of fifteen second obstruction outages, and general wide bands of obstruction and beta downtime. I'll save its charts for next month, when it will have a full week to fill the histogram.
I know someone on the Starlink subreddit is going to stamp their foot and complain about yet another person whining about their obstructions. I know I need to get Dishy up higher, and get some trees down lower. I still think it's interesting that without changing anything, my connection statistics changed so drastically. Take it as advice to refresh your own obstruction view if your connection quality suddenly changes.
Categories: Starlink
Post Copyright © 2021 Bryan Fink