Miscellaneous Summer Updates

The past few weeks have been a mad dash towards an as yet unspecified (and as yet completely unscheduled) summer break, so although a lot has happened most of it will pop up in retrospective posts somewhere down the line. There are, however, some odds and ends worth pinning to dates, so I’ve bundled them together here.

After a few weeks spent with my revamped printer, I can categorically state that eSUN PLA+ is likely the most challenging material I have ever tried to print with.

It might be the lack of a heated bed, but the matte white is just infuriating to deal with and I just can’t get decent prints out of it. Cooling makes no difference. Swearing at it makes no difference. Slowing down speeds seemed to help, but it was just a fluke.

3D printing is a fun hobby, but it can be an extremely time consuming one. I’ve been sampling various brands, and all have drawbacks. For instance, SUNLU silk grey PLA curls up for no reason whatsoever and has utterly bad layer adhesion.

I tried using it for a bearing holder hoping its smoothness would decrease friction, but it just desintegrated the very first day of use, even when printed at 0.1mm layer height.

So I replaced it with Amazon Basics PETG (which is still looking like the best filament I can get for general usage), and am thinking of returning the unopened spool of silk copper I bought on a whim given I am unlikely to be printing decorative stuff.

My office light gave out in the past few weeks, and rather than replacing the dinky little fluorescent tubes in it (which are, weirdly, the same size used in aquarium lighting), I decided to splurge on a [Zigbee]zb enabled, 60x30cm Paulmann LED panel.

More conservative folk would have picked the white-only variant, but since I have been quite happy with the RGBW strips I use for indirect lighting (which have 5 LEDs per element), I decided to go with RGBW rather than a plain tunable white one on a whim.

Although I don’t see myself going all out on lurid disco-like ambiances, the tunable white mode works great, and via zigbee2mqtt you can configure the panel to reset to a specific mode when it’s powered on via the wall switch (I have the hue set to 320, which seems to maximize light output from both sets of “white” LEDs).

I had zero trouble binding it to my zigbee2mqtt setup, and although the drama (and earsplitting din) of drilling into concrete ceilings is not one of my fondest recollections, I am tickled pink (pun intended) with the results.

I have been sticking to my decision to write more Go, and since I’ve found myself again doing late night meetings and missed my ambilight-like setup under Windows but didn’t want to actually install a Windows compiler I decided to do a quick cross-platform hack (which I added to the original page for easy reference).

If Go does one thing right, it’s cross-platform–I haven’t found anything simpler than setting GOOS and rebuilding to get a working executable…

Solar Smokeping

I had to help troubleshoot a relative’s solar panel installation out in the middle of nowhere, and ended up doing it by packing a Z83ii loaded with Smokeping and Tailscale off into the wilderness, into which I remoted to set up monitoring targets for the (somewhat flaky, but mostly working) NOS LTE connection and all the LAN devices I could find.

Initial suspicions were focused on the Huawei router, but I ended up taking a close look at the SMA equipment controlling the solar panels atop a shed and the TP-Link powerline adapters the installers decided to use instead of running dedicated Ethernet wiring between the cottage and the shed.

In the process (which was a throwback to the days I was doing Ethernet MAC spoofing to debug spanning tree) I ended up going all the way from running iptraf via SSH to Wireshark via RDP and digging into the SMA protocol (of which there are a few implementations out there), but after three days of monitoring and some remotely directed plugging and unplugging of things it was pretty obvious the powerline connection was the problem – not because the adapters didn’t work (they were at their limit, but still within acceptable signal levels), but largely because the janky cheapo power supply from the NOS satellite box fuzzed out everything else.

A little judicious re-plugging of things into different extension cords straightened that out, but now that I have the Z83ii back I fully expect the LTE connection to drop out occasionally since it doesn’t have Smokeping keeping it open.

Ah well.

A bunch of family shenanigans (trips, birthdays, visiting relatives) have also been a priority, and with summer school just ended, I did what every technically-minded parent would do and started farming out some of my hobby electronics projects to my kids.

I have nothing but praise for MicroPython, but having it on someting like the WaveShare RP2040 Zero takes things to another level and makes for some pretty impressive turnaround times for prototyping. My eldest is already looking at using Arduino C++ instead for performance reasons, and I mist go on the record to state I had nothing to do with that.

Well, at least not outside of genetics.