I’ve wanted one of the curvy bronze-keyboard G3 Powerbooks since they were new – I’ve always been kind of taken by the design language (half business, half iBook). I got to play with a university-owned Wallstreet for a while as a kid, so I remember them beyond looking in catalogs, and uh… I really like a bunch of classic Mac games and want a convenient late-classic machine to run them natively, because a few are glitchy in emulation.
So, I occasionally lowball bid promising auctions when one comes up somewhere. A few months ago (in late February) there were a succession of them on ShopGoodwill, and I tossed a $50 max bid on a Pismo in the 500MHz/128MB/12GB/DVD configuration in unknown electrical and OK but not perfect cosmetic condition. And, surprisingly, won. It ended up being about $67 with shipping/handling/tax/etc. Since working condition examples tend to be around $200, this seems like a decent deal. It survived the typically awful shipping, and, in long form below, it turned out to be a good buy.
I’ve had a few projects recently that needed 3D printed parts, went to use a roll of black eSun PLA+ that I accidentally left out in my basement for some time, and it produced the telltale dull, rough, stringy, gappy print quality of hydrated filament. Since I have some energy for projects right now, I decided it was finally time to get a filament dehydrator setup.
Pretty sure PLA printed to a glass bed shouldn’t look like this…
I went reading up about filament dehydration. The purpose-made filament dryers mostly seem to have a problem with actually removing moisture, most of them are dry boxes with no exhaust path, and the liberated water vapor has to go somewhere. Many seem to rely on instructions for the user to periodically open the lid during drying (which is …not automated), and a few contain a large amount of desiccant that will be more-hydrophilic than most filaments at drying temperatures, then itself require dehydration. Bambu seems to be doing basically the right thing with the AMS2 Pro and it’s automatic exhaust valve, but I’m not in the Bambu ecosystem and not looking to spend that kind of money.
So, instead, I looked at the DIY options. There are a variety of clever schemes with PID controllers and heating elements… but lot of folks seem to just repurpose inexpensive food dehydrators, and I went with a variation on that plan.
I picked up a $40 Elite Gourmet EFD770WD dehydrator; I paid a few extra bucks for a model with a digital thermostat/timer. It gets me a heater, a blower, and thermostatic control in a package that is at least theoretically safe for consumer use, which is really all that is called for. There are many directions on the internet for cutting the grilles out of several trays on this style of dehydrator to make a spool-sized cavity, and many designs for large, elaborate, and almost inevitably multi-part printable extension tubes, which must be printed in a filament still rigid at the highest drying temperature you expect to need. The former seems wasteful and the latter seems like a tedious hassle.
A little foamcore and tape to make a suitable cavity.
I did something much lazier and made a tube out of foamcore. Just bent it using the little tabs for aligning the trays and taped. Mine is two pieces because none of my scrap pieces were quite large enough to do it in one. The foamcore is slightly insulating which seems like a minor feature, and this method isn’t many hours of printing or destructive, which is a major feature. I also saw several folks modifying a 3.5gal or 5gal paint bucket with a diameter around 12.5″, and I may try that in the future as a more polished solution, but didn’t have a suitable empty on hand.
This particular dehydrator only lets you pick specific temperatures, but 115 and 125F are options right to either end of the suggested range for drying PLA variants. As for effectiveness, the main subject roll of black eSun PLA+ that sat out for a couple months, whose behavior is pictured above, went in for 8 hours at 115F, and initial results were really promising.
The first experiment looked really promising; Same printer, same gcode, top print before drying, bottom print after 8H@115F. Almost all defects were gone from the first layer.
There were still some hydration-looking defects in areas (and the thermocouple I had shoved in through one of the lid slots of the dehydrator as a safety was reading a little low), so I gave it another 8 hours at 125F. And the results didn’t really change, but there was no obvious degradation. It has a dead spool producing parts that are usable if a little textured, which is worthwhile.
Similar parts printed wet, dried for 8 hours at 115F, and dried an additional 8 hours at 125F. It’s unclear if the subsequent/hotter drying offers any significant benefit.
PLA+ is always a bit of a mystery material, I’ve generally supposed that much of it is doped with a couple percent PBT, but the relevant eSun PLA+ MSDS just shows 2-4% calcium carbonate (which apparently just provides nucleation sites to improve the crystallization) and 2-5% “other,” (likely pigment). CaCO3 isn’t very soluble in water, but who knows how it moved around or altered crystallization or whatnot during the wet-dry cycle. I ran some Inland “Egyptian Blue” regular PLA that had sat out in the basement for a while through a similar 8h@125F cycle and it did seem to reduce the surface irregularities (pips, especially on corners) between the previous and next part I printed in it, but not in such a dramatic way.
From a somewhat-amateur reading of the relevant literature, it seems like not all the hydration induced changes should be reversible. The most relevant thing I could find was Beyond Biodegradability of Poly(lactic acid): Physical and Chemical Stability in Humid Environments (2017) which looks at degradation due to liquid and vapor phase water infiltration, and found pretty substantial chemical changes especially from vapor at higher temperature. The literature in general is a little spotty, there are more liquid phase studies (eg. ref), but studies like the earlier one comparing liquid and vapor phase water infiltration indicate they aren’t entirely comparable. The literature on drying is “thin,” and the relevant Internet content is thoroughly astroturfed by vendors trying to sell you gadgets (which is becoming a real problem in the 3D printing market in general; good luck finding un-sponsored information about anything). I’m sure some of the commercial (bio)plastic manufacturers/processors have detailed internal documentation, but they aren’t sharing.
In the same order as they dryer, I picked up some indicating desiccant packs to improve my ability to monitor and dry filament in bags. I’m so distrustful of Amazon junk now, I stuck one in the bathroom to absorb shower steam to see what the indicator hydration process looked like, and it is slowly turning pink after being repeatedly exposed to shower steam. Hopefully storing filament with known-dry desiccant will help keep it from going bad – at least as long as I continue avoiding any of the truly hydrophilic materials like Nylon that require special handling.
All in all: Hydration is absolutely a problem for PLA and adjacent materials, drying is imperfect but effective, slightly modified food dehydrators that exhaust the vapor do a fine job, and keeping material dry is better than trying to dry it.
I played The Legend of Zelda: Echoes of Wisdom (as a distraction to avoid losing my mind reading the last giant batch of lab reports for the semester) since I finally found a copy for a vaguely reasonable price a few weeks ago, and will play anything from the franchise. It’s… OK. Thoughts:
The main echo gimmick is novel and mostly positive;
it’s narratively a neat way to have the player character (who is Zelda) not get their hands dirty
it’s pretty clear when you can learn something, and which things on screen are your echoes
it’s fun to try applying different echoes to situations
…but it involves a ton of menuing, which their attempts to make affordances for don’t quite cover up
and there are quite a few points where it’s aggravating to get an echo to do what you clearly need it to
The game is pretty short. A bit unsatisfyingly so.
The vast majority of puzzles are highly telegraphed and fairly trivial (most are just interesting enough to be satisfying) and a few are “go look up what obnoxious detail makes the obvious solution frustrating to execute”
The minor characters and side quests are not very interesting, which is something Zelda games are usually better at
I went in and played a bunch of side content after I beat the game, just looking for interest.
Several of the mechanics just…aren’t very useful:
If you can find a safe spot, you can always heal with a bed, so smoothies matter once for cold resistance and otherwise only if you need to heal during a boss because you did something dumb.
Once you have clouds, the other platforming-related echoes are largely irrelevant.
I… didn’t even bother with Dampe and the automatons until I went back to mop up stuff I missed after I finished. It feels more like an alternate pitch to the echo system that they didn’t quite let go of.
It’s not a bad game, but don’t pay full price. Given that the 3D Zeldas for the Switch (BoTW and ToTK) were excellent, do those first if you haven’t. And if you specifically want a 2D Zelda-like and you haven’t played Tunic (which is a 2D Zelda influenced game whose magenta otherworldly corruption feels like it might have fed back into the franchise with this one), do that first, it’s more interesting.
I got the upgrade to GIMP 3.0 yesterday, went to do a few basic image editing tasks, and it’s great they’ve fixed many of their long-standing issues and added things like NDE for filters, wider color depth, etc. … but I was so annoyed by the dialog-buttons-in-title-bar situation caused by the move to GTK3 that I stopped what I was doing and looked for a workaround.
Dialog-buttons-on-top is backwards from reading order, backwards from every other piece of software I use, and backwards from almost every other piece of software made in the last 40 years – I’m not interested in arguing if the GTK folks repeatedly doubling down on their weird bullshit is legitimate, I just want to be rid of it.
It turns out there isn’t really a runtime workaround on Wayland, your choices are force it to use the X back-end via `GDK_BACKEND=x11 gimp` so it will respect the gtk-dialogs-use-header=FALSE setting, or patch and rebuild GTK3. Because that setting gets overridden with the Wayland back-end. Unless you delete like 4 lines of code and rebuild the whole library, since the “technical reason” it had to be hard-coded for wayland is apparently just some bad behavior with CSD bars (which are also undesired behavior – I want consistent window controls and labels plumbed by my environment) that come up when running under gnome.
…And since you’re replacing your system gtk3 with a rebuilt library anyway, there’s gtk3-classic. Which isn’t packaged, so you have to build locally. From some random github repo and an AUR package. But it is pre-patched, and gets rid of a bunch of the other misfeatures like the search-instead-of-typeahead and some weird CSDisms at the same time.
I moved from daily driving xfce to kde in like 2017 because GTK3 was awful (and the KDE folks got their resource usage in check), and the stragglers (gimp, inkscape) moving to gtk3 is making me think about it again, because it is now even more awful. At least GTK3 is basically frozen at this point, so I won’t have to build the replacement too often.
I regularly have this problem where I have hobbies for which the equipment should be locked up when unattended, and I pick locks as a hobby and work as a computer engineer, so I know how bogus most of the options for locking stuff up are.
I bought one of those little portable lockboxes (Specifically a “Amazon Basics Portable Security Case Lock Box Safe, Combination Lock, Large, Black” ASIN B077K3FJHC), 50% to have a more responsible storage option for some small items, and 50% to see how crap it is.
I knew what I was buying for $18, and it’s still a hilarious piece of junk, because the gates are visible from the top so you can decode it by sight:
Combination is set to “0,1,2” and the dials are turned to “0,9,2.” Note that you can see a slot next to the wheel on 0 and 2 and not on 9? That’s because you can always see the gate when the correct number is dialed, so it’s trivial to decode.
A few points because it can’t be trivially re-set from the outside, you can’t pop the latch with a wire, the hinge pin won’t come out in the closed position without extensive violence, and there isn’t a tubular override key that can be trivially impressioned (common problems with similar products), but that lock is pathetic. Amazon actually posted my review to the effect, so… I guess also points for honesty?
TL;DR: I converted my Anycubic Linear Kossel to Klipper. Using a hacked Chromebook as the host, which is great and I recommend hacking surplus education market Chromebooks for many of hobby projects usually done with an SBC. Running in containers as a docker-compose project, which works but is dumb and wasteful. Next I’m going to try to do some systematic performance experiments to it, but that’s for a later post. Details below.
Between having more disposable income and the potential for upcoming trade weirdness, I’ve been pretty fast-and-loose about buying random bullshit that catches my fancy from China of late.
A recent Aliexpress order consists mostly of USB tampering supplies, like an exhaustive set of breakouts for all the common USB connectors, including a couple extra type C breakout boards because I’ve been doing a bit of Chromebook abuse recently and a SuzyQ cable is likely to come in handy.
In that batch, I also picked up a 10 pack of these little HW-398 USB C PD trigger boards (aka Decoy boards). They have a USB C Female connector on one end, + and – pads on the other, and a series of little pads marked with voltages that you can solder bridge to an adjacent resistor to have it negotiate that mode – apparently via any of the PD/QC/AFC protocols for doing so. They seem like they should be useful for powering projects from easily obtained junk, but merit investigation because they also seem a little sketchy and were a whopping $0.63/each.
A couple of the behaviors and design decisions are interesting:
The USB C connector is slightly inset on the PCB, which is probably good for strength, but rather unfortunate for mounting it into a project.
They DO supply 5V when none of the higher values are selected, which is both reasonable and desirable.
There is a little blue indicator LED next to the plug, and it’s marginal at lower voltage.
It appears if you set a voltage not supported by a supply, you get the next voltage down, which I gather is suggested by the standard. This is not unreasonable but has real potential for unwelcome surprises if you don’t protect your design. The most likely issue seems to be a lot of supplies don’t support the optional 12V mode: I didn’t thoroughly test, but I popped one strapped for 12V onto a little Anker 313 whose label says 5V@3A,9V@3A,15V@2A,20V@1.5A and it delivered 9V.
The IC is conspicuously unlabeled. It’s an SOP10 package, and is smart enough to do to necessary USB negotiations (which are actual USB protocol traffic performing a handshake to read and write some control registers at either end). A little googling seems to imply the chip is a Fastsoc FS312 – It’s in the right package, supports PD,PPS, and QC negotiations, which lines up with the product description, and the setting straps are connecting resistors of values 184(180kΩ)=20V, 140(14Ω)=15V, 104(100kΩ)=12V, 513(51kΩ)=9V which matches the datasheet.
I haven’t done any load testing, but I don’t see any sign of regulation on the board, so I suspect the regulation will depend entirely on the supply.
I have no idea if the USB IF considers these legitimate.
It certainly seems like a useful gadget for the parts bin. LCSC doesn’t seem to stock the FS312 IC, which is a shame since now that I know they seem to just work with the very minimal suggested circuit in the datasheet that requires only a capacitor and resistor beyond the chip and connector, I’d start designing them into boards if I could get them stuffed by the usual scumbags.
Since finally being rid of my PhD work, I’ve been hitting a bunch of projects that have been on my TODO list for ages. The oldest so far is this PSU which has been sitting for …decades… in my parts pile with some compatible machines, and I’ve always intended to try rebuilding it. It died with a “ticking” symptom some time in the mid-00s.
I finally got around to it this week, and it wasn’t a bad job. About $10 of parts, a few hours of work, and it’s back in action. Rebuild details below.
If you have an OG Xbox sitting on a shelf somewhere, this is your reminder to pull the clock capacitor and clean up whatever corrosive goo it has already vomited out. I was talking old video games with a student in one of my labs this week and it reminded me that this is one of the many projects I’ve been meaning to get to but haven’t had time for for the last several years.
Microsoft used some cheap 2.5V 1F early super caps which will inevitably fail and spray electrolyte on your motherboard, because they made some cost-cutting choices about the RTC, so now everyone with an old Xbox has to fix it.
Having now collected two graduate degrees myself, and been related many stories by friends and acquaintances doing grad school at UK, some advice for those behind me.
UK’s graduate school is the absolute archetype of the rule of thumb that, at UK, every individual you deal with will be nice and helpful, but as an organization they’re the most useless, obstructive motherfuckers you’ll ever have the displeasure of dealing with. You will never find an individual to punch in the face when something deserving happens, because responsibility has been diluted sufficiently that there isn’t usually an individual bad actor responsible for whatever bullshit is going on; the problem is that there are half a dozen overpaid people with inflated titles not dealing with the thing, and an assortment of folks in lower-titled, public-facing positions having to scramble to make things work around the administrative dysfunction. The fact that UK just went into administrative bloat overdrive by dissolving the faculty senate with no concrete plans to replace their functions, then started hiring random assholes who don’t even have the context to know how things have to work to take over matters the senate used to handle has made it even worse this year than usual.
