NordVPN is trying to get me to come back at $2.88 a month (for two years, so $70) now. Fuck you, Nord. That price isn't terrible, but you'll need to support port forwarding before I would even consider it.
Made another interface. This one omits the power switch, and the aux out terminals on the radio it's bolted to are already switched. This one also got a 1/4" TRS jack for headphones.
Still no luck in fixing the tuner. I have two of them, but the working one is a couple revisions newer and no aid in troubleshooting beyond confirming my bench test methods are vaid. Things I've done so far: -Test all the small signal diodes (and falsely replace two that tested bad in circuit) -Shotgun replace the 4 discrete signal transistors -Replace and socket the OP amps -Remove, test, and reinstall in a socket the 7x TVS diode array chip -Remove, test (as best I can with a multimeter) and reinstall the double balanced mixer -Confirm the MCU and controls are at least partially working (at boot time, it energizes all the inductor relays to shunt them and is able to operate status LEDs; if any control chip is missing, it does not do this) -Confirm the high power RF path is good -inspected for any visually bad passives in the low-power RF/sensing area
Last thing I can think of before writing it off: -Shotgun replace and socket the 4 CMOS logic chips in the low power RF section This feels like a longshot, but I haven't any other ideas besides shot gunning all the passives in that area.
It's possible some input pins on the MCU are bad, in which case it's dead. I don't have a scope or schematic of the board revision I have in hand to directly test that besides by ellimination. If replacing those cmos chips doesn't work, my next best idea for it is to make a primitive manual control to allow me to control the tuning elements manually. Perhaps just controls for each section of elements to step up, step down, and reset. As I recall, they have 6, 7, and 5 elements each with values in 2-powers, with the 7 element one being inverted. That plus a controller box is a lot of damn work though, especially if I want track positions of stuff. Closed loop control with a serial link would be ideal, but need coding on both ends and be far too complex for me.
If replacing the CMOS doesn't work, I'll probally just scrap it.
I know my tek scope model is prone to that, but mask rom's failing aren't usually something I hear about. Maybe they were OTP ROM's (EPROM's without a quartz window).
Bit rot in the ROM chip or MCU is also possible, but still a write off. No way I'll find the code for em, and the OEM isn't going to hand it over or sell me flashed chips. For over half of what a new one costs (and more than I paid for this unit) they'll service it. I suspect service means stick a new current revision board in it. The case is also modded, but that's not part of the actual problem. To be fair, the current revision is all SMD for the logic, and on all the boards every through-hole is plated with pads on both sides. I lifted a few pads already, but only the dummy ones...
This post has been edited by Wayward_Vagabond: Jan 19 2022, 05:20
Damn, the latest update of G HUB seems to have broken whatever I did so that a middle click would never accidentally turn into a middle double click. And of course at these times I don't remember how I got things to work in the first place. That's one of the reasons I would not be good with a system requiring frequent reinstall and/or linux.
Damn, the latest update of G HUB seems to have broken whatever I did so that a middle click would never accidentally turn into a middle double click. And of course at these times I don't remember how I got things to work in the first place. That's one of the reasons I would not be good with a system requiring frequent reinstall and/or linux.
with linux you can write scripts so you don't have to remember how to do things.
And it could just be that the actual switch in the mouse is dirty. I've been able to take apart and clean those to fix them before in my ($20) mouses.
This post has been edited by dragontamer8740: Jan 19 2022, 17:38
That is retarded to not know if its true. 5G has been rolled out elsewhere without this issue? Seems they have been reporting on the risk for a couple of years.
Found a bizarre bug, which I am just going to assume is an FVWM bug, where changing a window title too soon after changing the window's class in FLTK would make the window manager crash, but only in OpenBSD.
Fun stuff. The system logs don't show anything.
I am assuming FVWM because it's running git master and my debian systems are running an older git revision. ---- Also, I did a doodle during my (two hour long) SW engineering class and feel like I learned more doing it than from the lecture. Specifically a way that I might be able to draw eyes (and noses) more easily. It's pretty different doing line art from doing pixel art.
More on topic, though, I hate software engineering mindsets. They feel so intellectually disinterested - enough that they have a class on software testing even though anyone who actually has dove into the field with open source stuff has encountered it before (and I have had instances where I've had to write tests before). And most of the things they talk about just feel like they should be a thought exercise that takes at most a day or two depending on your exact situation and team size, instead of a full semester long course. Especially given the huge amount of pre-existing information on the topic out there.
Sure, I guess some people are motivated most by money - but I think that's a damn shame. I wish people could find things they actually cared about instead. Software engineering sounds so depressing as a job.
Notes removed for privacy or something I guess. This started as a single eye doodle and just kind of grew from there. So it's not perfect by any means. But it's the most I've done in months. Drawn in [xournal.sourceforge.net] Xournal. [i.imgur.com] (IMG:[i.imgur.com] https://i.imgur.com/P2zuucK.png)x ---- Also, the GNOME guys were too busy fawning over bad UX to consider checking for the word "stylus" when determining whether digitizers were touch screens or not ([gitlab.gnome.org] source). That issue has plagued me for years on both my thinkpad and my USB tablet. So I have added an additional patch to my increasingly annoying patch set for GTK3 (in addition to fixing the file chooser for type-ahead find). And compiling GTK3 once again.
This post has been edited by dragontamer8740: Jan 20 2022, 07:36
I always wait way too long to do my updates; since I have a multiarch setup with qemu usermode emulation, there are 2.8k packages I'm installing right now on my laptop.
Update: still going, four hours on. Almost done now I think.
I need to clean out packages I don't use much.
This post has been edited by dragontamer8740: Jan 21 2022, 05:45
That is retarded to not know if its true. 5G has been rolled out elsewhere without this issue? Seems they have been reporting on the risk for a couple of years.
5G isn't composed of a single band, but multiple. Often times, you'll see ISPs setting up equipment to interface with one to three different bands which are commonly:
Sub-GHz - Long range, low data rate.
GHz range - Short range, high data rate.
High GHz (for 5G) range - Very short range, very high data rate.
To add to the confusion, the GHz range is split into multiple ranges such as low-GHZ ranges below 802.11b/g/n Wifi (2.4GHz), ranges that include 802.11a/ac WiFi ranges (5GHz) and a bit lower, as well as the high GHz range of 6GHz. (Early 5G models were of the layer-cake model with low, mid, and high bands depending on deployment areas)
Now the current problem is that deployed 5G near airports were using what's called the "C-band" (3.7-4.2GHz), which is close to what airline altimeters use. They should be separated by a guard channel, which is just empty radio space. However, the very real danger here is that the 5G band transmissions used by ISPs will bleed across that guard band and into the range that flight altimeters use. This is something that is power-based. The more powerful the ISP transceiver, the more it will bleed into surrounding channels above and below their official transmission range.
Apparently, this is what is happening and one does not easily just lower the transmission power on the transceivers as that'll affect service. Might even not be feasible, depending on how the transceiver equipment is built.
Basically, the interference would cause issues similar to jamming. One can see this with 802.11b/g/n WiFi where one router can interfere with another, because the signal of one bleeds into the other, causing signal interference and loss.
This is especially dangerous in altimeters as it can cause a crash if that altimeter is not correct. Altimeters are often used in the landing phase, and especially so in bad weather conditions. Let's say that your altimeter is reporting, "Hey, I'm currently 200 feet above the ground" while your plane is descending. Next moment it gets interference from an nearby ISP 5G transceiver that's blasting out frames at full power. At that moment, it could just lose measurement, but still report "Hey, I'm still 200 feet above the ground." Now take into account that your plane is still descending. Because of that faulty report, computer error may cause the plane to overshoot the runway. Human error may cause the plane to descend even faster from the faulty report, because obviously you're not descending fast enough since you think you're still at 200 feet in altitude. Keep in mind that these are actions made in less than a second and that's all it takes for a plane to hit the ground too hard, collapse the gears, before doing a belly slide and igniting fuel.
This post has been edited by chriseras: Jan 21 2022, 10:19
Nice writeup. I was going to mention there are times where even in 802.11 stuff is supposed to automatically hop off of certain higher channels in the presence of other broadcasters. But you did a much better job explaining the "why" than I was going to.
This post has been edited by dragontamer8740: Jan 21 2022, 10:19
Adding to my previous post, here's also two examples of how nerve wracking landing a plane can be in inclement weather conditions, especially with one that's worth millions of dollars, weighs thousands of pounds and can't just accelerate or maneuver on a dime.
Props for the detailed writeup. My main remark is just "The telcos have nearly acheived regulatory capture of the FCC."
I'm declaring the tuner a write off, and removed all the relays, and RF parts from it. My working theory was brain damage to the MCU- either bitrot in it or it's ROM chip, or fried input pins. 7400 logic isn't that sensitive, and it wasn't a low hanging fruit type fix. I did too much damage trying to remove the logic chips and clean the plated through-holes but I suspected it was futile at that point.
I may move the good one into it's case, as I fitted a nice circular control connector and N for the RF- something to use it remotely and keep the connections weather proof. The good one is stock- a single cable some 8' long that terminates in an PL-259 and open wires.
This morning, while I was playing, I felt a shake in my mouse. 1st reflex was obviously to check if it wasn't my hand that was shaking (IMG:[invalid] style_emoticons/default/laugh.gif) Then I figured the shaking came from nearby ongoing construction. I wonder how much it's dangerous from my hardware that the shaking could be felt that much (IMG:[invalid] style_emoticons/default/unsure.gif)
This morning, while I was playing, I felt a shake in my mouse. 1st reflex was obviously to check if it wasn't my hand that was shaking (IMG:[invalid] style_emoticons/default/laugh.gif) Then I figured the shaking came from nearby ongoing construction. I wonder how much it's dangerous from my hardware that the shaking could be felt that much (IMG:[invalid] style_emoticons/default/unsure.gif)
I can confirm that if it shakes as much as if it would be in the back of a van, there will be damage
Woa, Norton Crypto was more shit that expected. For a paid software, you'd think the crypto was a free benefit or something but... they take a huge fee 15% fee lol.
it also does not have any optimisation, so it will chew too much power relative to the mining rate lol. Without temp / power control it will cause problems on laptops for noob users.
This post has been edited by EsotericSatire: Jan 22 2022, 00:16
So, after updating my shit, I decided to play some music today. My file structure is a mess (things scattered across multiple drives) so i usually use 'locate' (actually a faster variant, 'plocate') to find where things are.
But guess what. Some asshole decided that that thing they did years ago (2016, I think) where they changed the behaviour of GNU 'ls' (so that it lies to the user about what files' names are) was a great idea to implement in 'plocate,' too. Apart from that, everything seems to have been relatively smooth, but there's always something like this that I discover.
The old way, you could copy and paste into a GUI and get sane behaviour, and just be an intelligent ape with a brain when pasting file paths into a terminal by putting quote marks around it as necessary. Whoever it is that thinks I want machine-readable output instead of human-readable output, but only when NOT working through a pipe, is an asshole.
A terminal is supposed to be a human interface device, especially in its original, truest incarnation, where it was a physical piece of hardware. It is supposed to facilitate the users' ability to see what is happening on a computer, and also to allow for user interaction. So making it harder for a human being to read the output to a terminal is incredibly stupid.
I swear, stuff like this makes me think eugenicists might have been right about a few things.
This behaviour should be an option, but lying should not be a goddamn default for something that's supposed to facilitate enumeration of files on a filesystem.
TL;DR: We're raising a generation of retards. </rant>
OpenBSD gets more and more attractive by the day. And just ditching all GNU tools in general.
Edit: this looks interesting. [github.com] https://github.com/DiegoMagdaleno/BSDCoreUtils ---- Also, Debian (really a debian package maintainer) is trying to axe "which", which has only been in Unix systems since 3BSD in the late 70's. Sure, it's not part of POSIX. But 'command -v' doesn't perform quite the same way 'which' does regarding aliases, shell functions, and the '-a' parameter, and I think you'll find the only Unix systems today that might not have 'which' are HP-UX and AIX. Might have mentioned the 'which' thing before, but I forget. But anyway, the guy who's trying to deprecate it is an asshole, too. Hope Debian decides to put 'which' in its own package that is installed in the base system, to free it from that specific maintainer.
'which' is not in the BSDCoreUtils repo above, so I have ported OpenBSD which to my machines myself. Getting a system that doesn't work against its user is getting harder yearly, thus reinforcing my belief that Linux would be better off more obscure.
Update:
Since updating my laptop seems to have stopped setting trackpoint sensitivity through hwdb (the default is not sensitive enough on my machine; additionally, there's differences between the correct levels for NMB and Chicony-made boards).
So I've made an ugly shell function in /etc/rc.local to set it for me.
CODE
# function to set the sensitivity of my trackpoint. # hwdb seems to be ignoring it at boot nowadays. # checks if device(s) are actually trackpoints first. trackpoint_calib() { # this has to be a huge function because I can't just set a variable inside a # while loop in shell and have it stay set outside of the loop (to check if # any trackpoint was found). # so instead i do a while loop inside a shell escape to set a variable. TPSENSPATH="$(/usr/bin/find /sys/devices/platform/i8042 -name sensitivity | sed 's!/sensitivity!!' | while read -r candidate; do grep -q TrackPoint "$(/usr/bin/find "$candidate" -name name | head -n 1)" if [ "$?" -eq 0 ]; then if [ -e "$candidate"'/sensitivity' ]; then # department of redundancy department redundant check echo "$candidate"'/sensitivity' fi fi done)" # end of shell escape if [ -z "$TPSENSPATH" ]; then 1>&2 echo "WARNING: trackpoint(s) not found. Might be a problem in /etc/rc.local." else echo "$TPSENSPATH" | while read -r line; do echo 227 > "$line" done fi }
This post has been edited by dragontamer8740: Jan 22 2022, 23:49
Glued the cracked hinge back together on my first NDS. the lid wasn't detached or anything, but there was a big crack in the smaller of the two hinge sections that extend out of the lid, and it felt wrong when opening/closing and the lid wobbled worryingly.
Need to get a new shell, really, but I don't trust the chinese plastics either, and it's cheaper to limp along with this for a while longer.
Also I'd probably have to find a way to get the nintendo badge off the lid of the current case if I wanted it to look right, since they never clone the typeface quite correctly.
Thought: I honestly have no idea how people scratched up so many NDS touch screens so badly. The one I've had since it was new has perfect screens, even if the rest is beat to hell. My siblings' also looked good in that respect, except it had a plague of dead pixels on both screens so I had to steal screens from another unit to fix it.
Never bought a used one that didn't have deep scratches on its touch screen, though.
I have four fat NDSes, and zero lites, because the original is better in every way - except for parts availability.
It's pretty fucking annoying how any search for nds gets 90% results about the lite.
This post has been edited by dragontamer8740: Jan 23 2022, 08:15
Damn, the latest update of G HUB seems to have broken whatever I did so that a middle click would never accidentally turn into a middle double click. And of course at these times I don't remember how I got things to work in the first place. That's one of the reasons I would not be good with a system requiring frequent reinstall and/or linux.
I "think" I fixed it. It probably was from the LUA script (very hard to find in this shitty program) no longer working, or the delay suddenly turning out to be too low, anyway after doing a bunch of things, including changing the delay, I think it work again.
What is the last thing you thought?, Tech Edition
Jan 18 2022, 16:15
