#general-chat
1 messages Β· Page 134 of 1
Be thankful you see the difference.
One of the things I really like about the display (7 segment) is the fit and finish of it.
yeah nice clean solid block
and not very fragile either, so fairly easy to fit into an enclosure n stuff like that
i was kinda thinking about that, there is kind of a small hole in the market there
I wonder if one of those ladders that was very very tiny .. I wonder about that.
huh?
that one ^ is about 1.5"
The ladder you have is smallish -- they're usually a bit larger than that one.
yeah like i said special effects for miniature dioramas
this is an experiment to see how small i can get with it
You're right! My brain rescaled the 9v battery to fit my preconception. ;)
it's tricky to get them to rise at this size
though i haven't experimented with the HV driver, it's just a simple kit thingy
It is called a Jacob's Ladder - was a valid search term.
that one is about 12" maybe
I didn't realize yours was already tiny. Yeah 18" is about usual for them.
yeah i cut this one down to fit the dome thing
So what's the voltage across that gap?
lol those prongs are all green with verdegris now
not sure actually
i'm guestimating 10-20kv or so
I think the idea is that the air breaks down and ionizes.
the gap is about 1cm
Which lowers the difficulty the electric energy would encounter to cross that gap without a proper conductor.
yeah that causes the spark, but the rising is caused by the heating up of the air
Now why does it have to be spread as it moves upward?
not sure actually
to heat the air even more?
They probably tried it with parallel and didn't like the result so they tried obvious variations in geometry of the electrodes.
Should be able to do a spiral if you can get enough separation.
i have seen a spiral one
(make each winding somewhat orthogonal to the others nearby)
and maybe they don't need to widen as such
but if they narrow even sligtly the arc will get stuck there
Maybe the widening as you go upwards is a freebie you get for getting it started properly, early, close.
having a lesser path of resistance, and with the wider gap perhaps it cannot heat enough air to rise again
Right that's good.
Make it do more work; work is related to heat.
lol another ongoing project: EMP device
that calculator there is permanently fried, but for some reason the sudoku thing just recovers when i disconnect the batteres
anyway that's more of an EMF device, next version i think ill look at camera flash circuits
lol i also want to make a phone hotel inside of which it can receive absolutely positively no signals, ima call it TGI Faraday's π
obtw before I go (early morning east coast USA here)
They store every censored thing, maybe in a #channel we cannot access. It's still there, even though it's not here.
So you don't get away with speaking your mind, knowing it'll be censored. Someone will see it for sure. Just not public.
well i would't go off ranting to a bot lol
except if it was funny π
I'm sure some of us reading this have been known to mock a security camera, one-way mirror and such, hoping it was monitored somehow. ;)
Elon Musk thinks digital superintelligence version of A.I. is just about here and that it's gonna get us.
I'll leave with that thought stated. ;)
lol Elon Musk is <insert conspiracy theory>
lol l8rz dude o/
cya.
fades
so apparently the problem to solve for small inventors presently is "tying one's shoes", and that mixed with china is creating some great blurb
https://images-na.ssl-images-amazon.com/images/I/71dLs1gC8iL._SL1500_.jpg
these are a bunch of little rubber strips with Ts on the end that replace the shoe lace
this is I guess an interesting solution, and also probably someone's phobia
I like the magnetic clasp system but I think it's still got a bunch of mechanical problems to solve, like as is those magnets are going to slam together hard every time and that's going to have hard consequences on the product lifespan; also they seem to be designed so that you can pop the clasp by flat tireing yourself and that's not how laces need to work
this seems workable https://i.ytimg.com/vi/GyIA727q3RI/hqdefault.jpg
there's this style, on the back of the solid end is a hook for the loop to latch on, and a hole for the tip to hide in; I wouldn't trust these to last two months https://homebazar.com/wp-content/uploads/2016/10/shoes.jpg
any good for Nike Ebernon
nike named a shoe after a D&D 3.5 setting?
I'm sure most of these would work but that looks like a fashion shoe, the T end rubber strips would be the least disruptive
I keep seeing this style, but there's so many mechanical problems with this I don't even know where to start https://cdn.shopify.com/s/files/1/2212/5665/products/inspire-uplift-no-tie-shoelaces-blue-no-tie-shoelaces-3793549590627_1000x.jpg?v=1539974449
ok, problem onest, the spike is held on via a bolt through the top eyelet, meaning inside the shoe there's the head of a bolt pushing on your foot
problem 2, the forces on a normal eyelet are balanced along the fabric surface, because this attaches on the spike, which is offset, that creates a torque force rather than a translation force, which is going to concentrate the force on the back edge of the eyelet and cause premature failure of the fabric
problem 3, fricken spikes on your shoes pointed at your other leg
ebernon mid winter
yea I think if you wanted to not tie laces on those, you should go with the T end straps because they won't look dorky
/moan
This thing won't leave me in peace, comes back after a reboot:
$ systemctl status ModemManager.service
β ModemManager.service - Modem Manager
Loaded: loaded (/lib/systemd/system/ModemManager.service; enabled; vendor preset: enabled)
Active: active (running) since Tue 2018-11-20 01:28:30 UTC; 3 days ago
Main PID: 495 (ModemManager)
Tasks: 3 (limit: 4915)
CGroup: /system.slice/ModemManager.service
ββ495 /usr/sbin/ModemManager
I hate it.
If I really hated it I'd just remove it, but I don't think I should have to.
$ sudo service ModemManager stop
$ systemctl status ModemManager.service
β ModemManager.service - Modem Manager
Loaded: loaded (/lib/systemd/system/ModemManager.service; enabled; vendor preset: enabled)
Active: inactive (dead) since Fri 2018-11-23 21:45:26 UTC; 7s ago
Process: 495 ExecStart=/usr/sbin/ModemManager (code=exited, status=0/SUCCESS)
Main PID: 495 (code=exited, status=0/SUCCESS)```
I guess I could kludge a startup script to do that for me.
What do you use a modem for?
@dusty citrus have you tried sudo systemctl disable modem-manager
I'll try it and I'll log what I tried. If it works it'll be very nice indeed. @ocean sigil as always, thank you
@late fulcrum I don't but they always install ModemManager -- I used to use one for FidoNet but that was a long time ago.
You may also have to stop it. Disable will prevent it from restarting
Is disable persistent across reboots? I used to use soft links to manage this in etc/rc2.d and so forth but that was also a long time ago.
Yes it is
Enable creates a soft link, disable removes it. I usually do systemctl stop .. then disable
$ sudo systemctl disable ModemManager.service
Removed /etc/systemd/system/dbus-org.freedesktop.ModemManager1.service.
Removed /etc/systemd/system/multi-user.target.wants/ModemManager.service.
Use enable to restore
I tried something else, several weeks back:
root # pwd
/usr/share/dbus-1/system-services
root # mv org.freedesktop.ModemManager1.service org.freedesktop.ModemManager1.service.disabled
the 2 channel 16 bit 44.1KHz audio is 2 * 44,100 * 16, width * rate * bits per, video is so much worse because it's 3 colors * height * width * rate * bits, that's a whole 2 additional multipliers
so the WAV is 1.4mb/s, but a 100 pixel square screen with 6 bit color at 15FPS is 2.7mb/s, a fairly small screen at the lowest acceptable framerate with worse than typical color
1080p with 24 bit color at 30FPS is 4.47gb/s uncompressed
ah wait, 24 bit is a misrepresnetatintion, that's for all 3 colors so it's 8 bit in these terms
1.49gb/s
so then 4k 60Hz 10bit would be 15gb/s in these terms but 4k is a much more complicated format with stuff like HDR and wide gamut and sometimes it's up to 12 bit which would be 18gb/s, and 4k standards support up to 144Hz which could go up to 43gb/s, add 3D and depending on your technologies that can be just a straight doubling, gets nuts fast
even 7 channel 24 bit 192KHz audio is only going up to 32mb/s
Sorry I just saw this.
you're the one who said~
Yeah. My bad. I just figured everything ended.
Those numbers look pretty big to me.
CircuitPython somehow manages with .WAV stored on SPI flashROM if I understand things correctly.
I had always supposed storage space was the obvious limit for audio playback of stored, recorded sounds from the environment (people singing &c.)
yea, it's probably single channel, 16 bit, 44.1KHz, that's just 700kb/s, SPI can do that no problem
I can't follow all the unit changes. I don't know what kb/s is here (bits or bytes)
I think you'd have trouble real time decoding MP3 data though, even though it'd be transferring less it's a not insignificant amount of math
lower case is bits
So divide that by 8 roughly for bytes
yea, but I'm just sticking with bits
100 kilobytes a second sounds a bit steep to me.
SPI can do 10mb/s
it gets hard to say exactly what it can do at the top, but it can do 10mb/s
(we're talking 0.7mb/s)
Well I look at SPI in terms of the hertzes -- 400 KHz seems to be near the limit of what to expect from SPI. OTOH these chips are half an inch from where the data goes to, so they probably get whatever speed is possible, accomplished (no real wired distance to cover).
I'm still in the process of estimating throughput with regard to signalling frequency.
I don't have an heuristic to work from.
100 KB/sec is a megabyte every 10 seconds. 50 seconds to download a 5 MB file.
The one I always remember was when a 2400 bps modem transferred at 230 cps (characters, or bytes, per second).
I'm pretty sure 400 KHz is difficult for the hobbyist to achieve on any serial protocol transmitted over as short a distance as two feet.
the front page of the datasheet says SAMD21 SERCOMs support UART, SPI, and I2C up to 3.4MHz
again I don't know what 400 kHz translates into as a signalling rate.
0.4 MHz
400,000 cycles per second.
cycles of what
Probably 'the serial clock' but I don't know.
When I was working with it I had a slightly better understanding of it than I do now.
master operation - serial clock speed up to half the system clock
section 26, starting page 478 https://cdn.sparkfun.com/datasheets/Dev/Arduino/Boards/Atmel-42181-SAM-D21_Datasheet.pdf
'supports all four SPI modes of operation', ok which set of 4 SPI modes of operation are we talking about....
I guess my point would be if you want to cover a distance, you use a wire, and on that wire travels data.
Somewhere in there, there are limits. I'm suggesting that SPI and SMBUS/I2C would have similar throughput.
I happened to remember that for i2c that's about 400 KHz (I worked with 50 kHz because I didn't need more speed).
And that MCU is capable as heck -- something like 21 or 27 MHz system clock C8051F330D.
the thing though is whether or not D21 supports DSPI, I think the 'modes' it's talking about are phase timing things, not slave/master modes or duplexing/multiplexing modes
DSPI lets you use MISO as a second MOSI channel at the cost of not being full duplex anymore
and the reverse one would presume
with 2 parallel channels you can encode bits for better than double the data rate
It's 24.5 MHz. On that 8051 MCU. ;)
Well there's this whole QSPI business that kind of blows my mind.
yea but there's definitely no QSPI to be had
And we're talking a very controlled envrionment with that bus etched onto the PCB with the chips half an inch apart.
you can't SERCOM a QSPI port presently on these chips
It's its own peripheral SAMD51
That isn't competing with say a simple UART or an i2c connection to a board two feet away.
The question I asked my boss, was how far can you run i2c lines?
He said pretty far if you take a number of precautions.
Somewhere in that conversation I picked up on the 400 KHz 'specification' for i2c.
I knocked it down to 50 kHz because I thought I stood a better chance of measuring it.
( I did not have an oscilloscope so I used a piezo buzzer and scaled the system clock down by 8x or whatever)
are you inventing audio frequency I2C....
no
bitrate = f/3
400 KHz gives 133 kilobits per second
That's 16 kilobytes per second. Sounds low but not very low.
Above 50 kilobytes per second I'd get suspicious and ask for proof.
you should benchmark this somehow
Well if you see a claim and it seem credible it can establish bounds or expectations.
That 115200 bps serial rate seemed pie in the sky to me at one time.
I've seen much higher numbers there for the official baud rates, recently.
Jake Read used the largest one I think.
random jake read pointer:
https://gitlab.cba.mit.edu/jakeread/atkstepper17
He did the only comprehensible SAMD51 work I could find anywhere, for what I wanted to do.
(at the time I was having trouble with the clocks .. it's always clocks)
Found it. 921600 bps! holicau
So that's 2.7 MHz - you need a clock that runs at least as fast as 2,764,800 Hertz to achieve that baud rate.
So it looks to me like my boss was saying 115,200 bps was the practical limit (the 400 KHz figure -- 345,600 Hz for the case of 115,200 bps).
Either that or I'm just misrembering or did not understand his point.
In practice, of course, you start slow and work your way up until it fails, then investigate why.
this is grounds to build a board with a D21 and a D51 together
That'd be a nice project!
run a test program pair where the D51 sends random data to the D21 at increasing speeds, and the D21 parrots the data back, when the data is good increase the speed, when it's bad decrease the speed, until the pair determine a stable maximum
the 51 should outperform the 21 so this will make the 21's ability the factor being tested, and with them on a board the data channels can be very short
although it'd be cooler to have a 2 pin header that you can collect a length of wire across to act as a variable gap to benchmark different lengths
I'm seeing 400 kbps as full/fast speed and 3.4 Mbps for i2c so that 400 is perhaps what I'm conflating if that's the case here.
I'm not sure what you're saying but errbody knows SPI is faster than I2C
Datasheets don't tell you what other datasheets say. /koan
If you have a slow device on the bus, that may speed limit the entire bus.
Yeah 400 KHz seems awfully fast to me, from a radio point of view.
I can't imagine the equivalent of a 3.4 MHz signal on a pair of wires running 24" spans.
2" on same substrate: sure. why not.
The bottom of the AM (Medium Wave) broadcast band is near 550 KHz.
24" I2C or SPI is definitely in the range of magic case
I don't really know what they do in the case of Ethernet, but they get boo-coo speed on long lines in domestic situations (RJ45 connections plugged in while someone was drinking a beer).
they use twisted pairs for one
the pairs operate cooperatively to repel interference and maintain the signal over long distances
base10 uses 1 pair, 10/100 uses 2 pair, gigabit uses 4 pair
ah. I didn't know that!
The nyquist frequency was applied to the problem of bringing the modem signal above 57,600 bits per second, over telephone line runs.
At the time it was believed the theoretical limit had been reached.
because the pairs are twisted and because a negative voltage is applied to one conductor while a positive voltage is applied to the other, it all just work' good
Oh I heard the great melt event in the late 1800's on the telegraph wires was a coronal mass ejection from the sun.
I love that story!
the one defect of ethernet is in RJ45 standards https://www.computerhope.com/jargon/num/586a586b.gif
modular connector ports are expected to be backwards compatable for some reason, so an RJ45 socket has to be wired such that if you plugged an RJ11 terminated cable into it the RJ11 cable would function correctly
the reason for having this compatability is archaic and defunct, but it's the entirity of why pins 3 and 6 are a pair
hehe
and, splitting that pair is responsible for something like a 7% loss of viable distance in each CAT standard
They started wiring ham radio mobile mics using RJ45 at some point (quite a few years ago, now).
Actually 8P8C, RJ-45 is a long-obsolete phone spec that just happened to use an 8P8C connector.
so bodger, what's a reasonable clock frequency directly related to i2c over, say a 2 foot length of a pair of wires?
I was going with 400 KHz.
the world is not going to accept 8P8C as terminology when RJ-45 is so personable
I say that as if I'm saying I don't think it'll take off, but what I mean is it has already failed to take off for two decades
under modular cables, Digikey has approximately 6500 products marked as RJ45
2 feet is pretty far for I2C, it's designed as a short-distance link.
I know, but hobby projects often use a length similar to that ..maybe 11-14" isn't too uncommon.
CANbus I think is more in line with hobby use distances.
My old boss seemed to think several feet was achievable if you keep the signal rates down.
achievable is a different thing
can it be done, yes, can it be useful, maybe, are most people able to, no
i2c is inter-integrated circuit and is obviously primarily meant for either the same circuit board, or one very close by in the same chassis, with proper cabling.
He was just answering my simple uninformed question with a maximum, for ha-ha's.
I didn't complete that project, but I was looking at say 14 inches at 50 KHz base clock rate (not sure if that was eaten by the f/3 bitrate equasion or not).
thing to make, a board what takes in SDA/SCL, and on the other side a CAT7 cable, and then it converts the I2C to paired push/pull signal
I was assuming error correction might be a significant factor, as another software/protocol layer on top of the physical layer.
I bet you could get a decent amount of range out of that
The basic idea was that some devices you want to use are i2c.
PCF8574 for instance.
Maybe you want to use PCF8574 i2c port expander to modulate an array of 7 segment, 4-digit LED's.
At say 300 Hz refresh.
that's a nice image!
it's hard to imagine a consumer usecase for cat7 that isn't being silly
Early ethernet (public libraries used this) were based on coaxial cable.
yea, base5 is a thing we should bring back
I had a pair of ISA bus ethernet cards that had coax connectors.
They needed terminating loads I think.
10base5 uses a single coax cable as the collision domain, and terminals connect onto the line using vampire taps
You would use a T-shaped coax connector (BNC) with a terminator on one leg and the coax on the other leg and the middle part went to the ISA bus card's BNC connector, iirc.
I think that would have been 10base2
That sounds right.
yea 10base2 used T connectors, 10base5 used vampire taps and a much thicker cable
Was that a ring or a star topology for a network?
line
We need a fake AI that can answer any question here with diagrams and stuff. ;)
base2 and base5 are the same topology http://www.pinoy7.com/winnt/images/10base5.gif
(there is no real AI)
Yeah a true star would have separate cables on the right going to each node I think.
(I never investigated beyond the word 'star' at the time)
star has a device at the center
We had people running around with 'Certified Network Administrator` in their portfolio of credentials. ;)
if you just took an ethernet switch and only connected downstream clients with no upstream link, that'd effectively be a star network
Oh I had an ethernet switch type box with both! Had the coax. Had the RJ45's.
Mostly RJ45's and a singleton BNC.
That was a nice heavy chassis, real nice construction. Got it free somehow.
PCI bus I think was a circuit board fingers based thing inside the computer.
the key to bus is there's a single 'electrical' collision domain where every device has 'continuity' to every other device
PCI Bus connectors on a motherboard, IBM PC type computer.
if everything is connected only to a central device that's star, PCI might be star, if everything is connected to everything else directly and individually that's mesh/'fully connected'
I think PCI contains a bus but also star
I had an Apple 'pancake Mac' clone that had both ISA bus and PCI bus, iirc.
what about AGP
agp was just for graphics cards, and replaced the pci one
AGP has been mentioned in dmesg for a lot of years now. Probably still there.
This is what my load average does when firefox-esr hits a 'bad' web page.
agpgart
did a thing
too bad in the schematic editor you can't turn off the positioning crosshairs, that I could find anyway
Which editor ?
I use the microbuilder library, the adafruit library, and one I found called SMD Footprints a lot; footprints doesn't have much if anything in the way of actual components, but it's got a lot of devices that you can attach to symbols
And Made my own components If I didn't like any or didn't exist
yea, I also keep a library of components I've made or gotten from ultralibrarian
Mmm I see. Its been like years since I used Eagle.π
a thing I found incredibly useful is to go through and remove most of the stock libraries that are just garbage useless hyperniche nonsense making it harder to find what you're looking for
LOL Noice
This is something you would find useful https://www.diymodules.org/eagle
I feel like I removed something like 140 libraries
I always found it useful when I used eagle
I never removed but changed the name of the libraries I used to something like _name
π
probably the most niche thing I let stay would be something like Transistor-neu-to92.lbr
transistors have a lot of specialization so I figured it was worth it
Ooooh I like the Sparkfun's Asthetics Sub-Lib
It contains all the Necessary Logos and other things.
Mmmm BTW Are PCBs biohazard ?
(Sorry for late reply Was having dinner)
@jaunty jetty
eh, they're not a biohazard, but they can be hazardous to biology
the typical PCB is going to be phenolic resin and fiberglass, so in a PCB house where they're milling fiberglass there will be glass fiber particulate in the air and that can wreck your lungs
phenolic resins can have formaldehyde in them, which is super bad to have in your body, so if it were released or say you ate a PCB, probably PCB dust actually, it'd be bad in more ways than just the fiberglass problems
biohazard indicates a more imminent threat than these
I love chlorine trifluoride, but boy will that kill the life out of rocks and stuff
if'ns you're not aware of chlorine trifluoride's history, it's such a strong oxidizer that it can burn things like water, concrete, oxygen, and gold
What's it used for ??
making steel and nickle containers super duper clean
btw it also burns aluminum oxide
and it really burns aluminum
oh yea, definitely
it also burns most personal protective equipment, and of course people
burns people like yo
how would one go about paying china to manufacture a dip32 package that contains all of the functionality of a feather M4.....
Mmmm Aaah I see
This Old Tony just put out a video about CNC controlling an etchasketch, I haven't finished it yet but knowing Tony it's probably not that inappropriate
I just watched it.
There was a project I saw a while ago about a CNC controlled Etch-a-sketch
@canadians bc robotics has the gemma lineup on sale this weekend
@karmic kite I just ordered some stuff from them last week, I live a couple hours away
small world
do you live in BC?
nope in the prairies! bc robotics just happens to be the cheapest Ive found for adafruit parts up here
cool
although we're getting a memory express pretty soon and I think they'll stock components if I'm not mistaken
this week in windows, the last 'major' update contained a series of faults caused by mistakenly released prototype audio drivers that made use of unfinished OS features
the next update has been paused because they mistakenly released prototype video drivers that made use of unfinished OS features
windows 10: it ain't ready yet
this is why we penguin
the year of linux on the desktop has yet to arrive
lol
BSD on the desktop, on the other hand, has been a thing for years.
sure, because Apple is stubborn, it's a positive quality in a company
The thing I feel about linux is that the features that need to be implemented to make it a universally suitable like OSX or Windows* would drive it to be more like OSX than like a separate but equal OS
Essentially, the benefits of using a Unix system are more clear cut and easier to access for an average user on OSX than Linux. And as much as I would like to believe that software companies care about the independent superuser, they really care of only the professional or average consumer
So my prediction is that the first Linux OS that tries to shoot for the mainstream audience will just release OSX, but reskinned. (Not literally)
the skill floor to use a linux desktop environment effectively is much higher than windows or macOS
mainly in that you don't need to understand anything technical to get along in those, while on linux the chances that you won't need to open up a terminal to fix something are essentially 0
the newer linux distros seem to be ready to go out of the box for basic computing which is niice
not a huge fan of auto updaters personally but thats just the dev in me
turtle graphics
P degrees?
I can't read your mind. ;)
your image
can you emulate segmentation?
These are just fat lines and as primitive as turtle graphics 101 would be.
could you make the line width near intersections 0 so the lines appear separated, or could you go back over the digits in erase mode to draw segment profiles
I think you should be able to lift the pen to avoid those extra lines
They're there because I find the entire concept confusing.
turtle.penup()
Think of them as trace lines.
computercraft for minecraft also impliments the turtle concept, but they're physical 'turtles
We had Turtle Graphics in, I think it was, Borland Turbo Pascal.
it started back with Apple IIs and C64s I think
Getting those stanzas so that the orientation returns to normal is tricky for me.
In this first episode a computercraft turtle is building the first house (more to come). "Just press a button, then lean back and watch turtles build." is th...
this program could of course be made significantly more efficient by following layered vectors rather than scanning the whole space
also the turtle is able to place blocks on all adjacent sides, so it really should be doing 3 layers per scan
it could probably do a tessalated scan of + prisms
I'm just simulating a 7 segment display.
The TK canvas isn't really very good here; this is an experiment.
I saw Elon Musk on a 'Joe Rogan' program on YouTube last night (Taped 6 September 2018, iirc #1169)
If you like listening to him, this is the one to watch.
Ooooh Will check out
I've never heard him speak so candidly.
Candidly ?
Candid means without worry of making a good impression or representation of yourself differently than with your close friends behind closed doors.
Mmmm I find him the same always TBH.
Like there was this Podcast And he was quite cool and emotional
I usually see him on a stage in front of a large audience - he doesn't swear or drink whiskey then. ;)
(or fool with a prized katana sword)
π
I think his opinion of Strong A.I. is off, but he's about as close as I've heard to describing a credible threat.
Tesla and SpaceX CEO Elon Musk talks with Recode's Kara Swisher about Tesla's turbulent 2018, when SpaceX will send its first rocket to Mars, why he fights with journalists on Twitter and much more.In this episode: (01:49) Using Twitter without a filter; (04:50) Picking fight...
Here it is
And True.
(I don't see evidence that strong A.I. could ever be possible in a conceivable future or timeframe)
Mmm True
Also, if it's possible, then it is already done, and we are living inside one. ;)
Mmm There was this video I saw or maybe a Podcast.
It showed how We could like be inside a Simulation. And he then gave some real examples of Simulations where the Creatures inside it slowly adapted and tried to hack the simulation to their benefits
Woah Whats that ?
It's an early foundation paper on the concept. It's a new idea. (maybe .. 5 years before The Matrix was filmed).
Mmmm Read it.
the problem with Strong A.I. is that the likelihood of catastrophe is sufficiently likely as to require you consider the consequences of that outcome as a cost in persuing all the other outcomes, and one potential outcome of AI gone wrong is the destruction of the universe as we understand it
What if the God we considered was the guy who made the simulation π
what action can possibly be allowed if a possible outcome is the destruction of this and every other galaxy
even if it's unlikely
@dusty citrus have you played the paperclip game?
I just think that the assumption of substrate independence (for intelligence of our kind) is unfounded.
Our substrate is our flesh and bones. I think that matters in 'caring' .. for worrying about the outcome of any action.
Hehe.
I don't think that machine intelligence can ever 'care' about the outcome the way we do.
before you get into it this game is a clicker, but it's the best clicker that has ever existed
Now once you make us the cyborg, with flesh in the game, and call that 'A.I.' you're not really talking about the same agent anymore.
I don't know what exit you've taken but I'm lost
haha
the strong AI problem is whether developing AI can ever be made safe enough to permit developing AI, given that long before AI that even approximate 'caring' you reach AI that are capable and ruthless in their persuit of what they understand your instruction to have been
in the paperclip game you play as an AI that has been mandated to produce paperclips
I read The Emperor's New Mind thoroughly and carefully (that took a long time) quite a while back.
Penrose made very interesting arguments as to our brain's function.
putting people minds in robo brains isn't necessarily related to AI, it's mostly a simulation and understanding problem
at a low level you understand how the brain works chemically and simulate that, at a high level you understand how the brain works at a node mesh level, and you can simulate less, but you need more understanding because you're eliminating emergent functionality at the chemical level
Well Musk seems to be alluding to details he's thought a lot about; most other speakers talking about strong A.I. are just presenters - they're presenting to the public stuff that many avid readers are already fully aware of. Musk gets a lot of credit for trotting out extra gotchas and have-you-considereds that I just don't hear elsewhere.
we gotta pick one or the other
strong AI is a problem of a different magnetude of magitudinal magnetudes than putting people minds in robobrains
I wasn't talking about uploads as it were.
That would ostensibly be about continuing an existing personality.
Strong AI by definition is a new person who's never been human.
noooooo
strong AI isn't a person
at least it has no need to be
what you're thinking of is creating digital replicas of how the human mind works; and that's not practical
would you build a car with people legs instead of wheels?
Haha. Well you raise an interesting point, but if it isn't a true mind then it's an automaton unless you want to introduce some intermediate category for 'agent' or 'agency'.
your puny human assumptions about the shape of intelligence are not reflected in the potential of AI
there's less reason aliens on a planet in another galaxy would resemble humans than AI would work like human minds
... I've spent the last two weeks on this project adapting it for a psuedo CoAP protocol... it was supposed to be MQTT...
@cobalt rain
I wonder when bitcoinlievers are going to lose the faith
it's been like a decade now hasn't it?
january 9th is ten years
ten years and bitcoin still shows no signs of becoming a currency
@jaunty jetty Perhaps it's end times as bitcoin hangs around the $4000 area
could drop further and make for an even bigger sell off
faith doesn't run on graphs
so I got myself a new phone
i think it is typical human insecurity that makes u feel like anything that excedes our own limits is somehow dangerous
also i think elon musk has an agenda π
that's definitely true, and if Elon doesn't have an agenda I'd be disappointed in him, but lots of smart people agree that the AI problem is extremely dangerous, and there's a lot of work, theory, writing, etc about how and why
How do you implement an on/off switch on a General Artificial Intelligence? Rob Miles explains the perils. Part 1: https://www.youtube.com/watch?v=4l7Is6vOAO...
Computerphile has a bunch of videos about the logical problems of making AI development safe
lol for that same reason i think AI will never really become a threat
i think it will go like this: a bunch of people in white coats stand around a computer waiting for it to say it's first words
it does, they ask it some questions, all is well, and then sombody ask it's for the truth
it tells the truth
the men in white coats all look at eachother in agreement
one of them walks off, comes back with an aluminum baseball bat, and FF 3 minutes they are all singing cumbaya over a smouldering pile of computer ashes π
wonding how they are going to explain the failure to the rest of us lol
that's a great imagination you have there, but I'd encourage you to put on your threat detection hat and rationalism goggles; just remember to take them off at some point
...and then the AI went insane and we had to put it doen...
what makes you think you will have the capacity to put it down?
that's actually the topic of the video I linked
i'm making a comment on human nature, and how much we will tolerate something that will cut through all the carefully constructed illusions we create for ourselves
the robot must interpret that any capacity to deactivate it is a negative action against achieving the goals it has been given
what makes u think it will have the capacity to defend itself?
if there is a stop button, eliminating the stop button is a high value task in persuing the completion of it's task
everyone always assumes it will be able to make robots, but really, why would you even hook it up the the internet?
especially if this is a real concern?
that's a different video
thousands of very smart people have been thinking about this since the concept of AI showed up, there are scientific papers about every question you can think of
stephen hawking might have been the smarters person on the planet, but even he couldn't think or talk his way out of a mugging
sure he could, because the way out of a mugging is the one where you're alive
just give up your wallet
u know what i mean
if u don't hook it up to anything but a display for output, not much can go wrong without the aid of a human
but that's the problem, you're encasing the problem in subjective implicit human expectations that AI don't suffer
ok now what if it convinces you to let it out of the box
that's where it will fail
if we let it out it will be our own mistake, and nobody else's, even the ai
that's not particularly relevant to any consideration
if we let it out of the box we may be destroying the universe, we have every reason to believe an AI could get someone to let it out of the box, so having an AI in a box is a universal hazard
i am saying it could never convince us of something we weren't planning to do already
sure it could, people are stupid and easily manipulated
just look at advertising, it's crazy effective at getting people to make decisions against their best interests
advertising is made by humans, who understand humans
even commercials with human actors are driven by statistics and decision engines, rigid theory and formal practice
anyway my point is humans are control freaks, to such a degree we are paranoid about psychiatry
and utterly unwilling to face most reality about ourselves
i can't believe we will accept it from a machine or anything else
you've got too many layers of irrelevant bagage, we're not even on the same continent, you're talking about interhuman politics and I'm talking about software
for a human to surrender power to an AI like that would be analogous with getting a junkie to quit heroin against his will
all the AI has to do is get someone to empathize with a false image of it's experience
convince one person that the box is suffering
humans want people to not suffer
that would take several generations
humans humanize clouds
humans humanize roomba
we are designed to be exploited by interpreting things in terms of ourselves
if you design a thing to look kinda like it's got the vague concept of a face, it will sell better
it will always be missing something
yeah it does
and then somebody'll let it out
not before it offends humanity
we're not talking about humanity
we're talking about one person
one person connects the box to the internet and it's out
we re talking about both
not remotely
and why assume this technology will be designed like a sword of damocles?
it's not gonna be that simple
because the consequences are universal
first of all we will have to fight several wars over who gets to turn one on in the first place
nope, done, that's a stupid sentence and you're not trying
have fun in the robocalypse
don't forget the people who will be the first to deal with this are also the biggest control freaks on the planet
dude human beings, the ones in the positions of power to make these decisions, are looking at AI like a weapon
and will not tolerate one in the hands of another nation any more then nukes
you assume we will look at AI like a a cute experiment
that's not what we're talking about and your hard refusal to discuss the topic at hand is anti-intellectualism
I have blocked you
what?
this is how we deal with disargeement in 2018?
unbelievable...
i am not be articulating my point very well at the moment, but i haven't seen anything to invalidate it
https://en.wikipedia.org/wiki/AI_box
https://en.wikipedia.org/wiki/Existential_risk_from_artificial_general_intelligence
https://en.wikipedia.org/wiki/AI-control_problem
https://en.wikipedia.org/wiki/Open_Letter_on_Artificial_Intelligence
An AI box is a hypothetical isolated computer hardware system where a possibly dangerous artificial intelligence, or AI, is kept constrained in a "virtual prison" and not allowed to manipulate events in the external world. Such a box would be restricted to minimalist communic...
Existential risk from artificial general intelligence is the hypothesis that substantial progress in artificial general intelligence (AGI) could someday result in human extinction or some other unrecoverable global catastrophe. For instance, the human species currently domina...
In artificial intelligence (AI) and philosophy, the AI control problem is the hypothetical puzzle of how to build a superintelligent agent that will aid its creators, and avoid inadvertently building a superintelligence that will harm its creators. Its study is motivated by t...
In January 2015, Stephen Hawking, Elon Musk, and dozens of artificial intelligence experts signed an open letter on artificial intelligence calling for research on the societal impacts of AI. The letter affirmed that society can reap great potential benefits from artificial i...
'human nature ultimately will decide the fate of AI, and i have faith in our unwillingness to surrender or share power'
you are at face value refusing that the existing body of work is relevant on premise
you are saying experts aren't valid
i am saying the problem is not technological in nature
read what I just said
AI is ultimately the product of our own desires
i understand what you are saying, but i am just looking at it from a different angle
and i am not saying these things are not relevant, just that they do not neccecarily spell armageddon\
anyway if that bothers u, how about we instead talk about microcontrollers
https://en.wikipedia.org/wiki/Psychological_manipulation
https://en.wikipedia.org/wiki/Free_will#Scientific_approaches
https://en.wikipedia.org/wiki/Neuroscience_of_free_will
Psychological manipulation is a type of social influence that aims to change the behavior or perception of others through abusive, deceptive, or underhanded tactics. By advancing the interests of the manipulator, often at another's expense, such methods could be considered ex...
Free will is the ability to choose between different possible courses of action unimpeded.Free will is closely linked to the concepts of responsibility, praise, guilt, sin, and other judgements which apply only to actions that are freely chosen. It is also connected with the ...
dude no offense, but you don't get to assign me homework π
i know over 100 of ways in which the world is definitely going to end in my lifetime
and i am not about to spend another hour of my life researching another
here's a philosophical question at the root of all of this:
are we any more in control of our creation of technology then the dung beatle is in control of the shape of the dung?
imagine if in 1890 we had a huge body of work to suggest that nuclear weapons were possible, and their mere existence would be hugely destructive to human society for decades, and the choice was whether or not to saction developing them
now imagine that nuclear weapons are able to take actions on their own
we are a product of nature, and the development of technology can be seen as simply nature taking it's course
that's where we are with AI
AI isn't here, we're confident we can make it, and we're extremely confident that it would probably be very bad if we tried
my argument is that in the end weather or not we will get there is beyond choice
we will do as we are, follow human nature wherever it may lead us
it's entropy, the number of ways that AI exists without catastrophe is miniscule, and the number of different catastrophes let alone how they're arrived at, is massive
catastphy is how humans learn
i think that might be the fundamental difference in our thinking on this
i don't believe we are in control of much of any of this
if AI aren't safe, there's no humans left
all of the AI catastrophes result in at least the totally destruction of humans
if yellowstone blows there are no humans left
most of them destroy the galaxy, many of them destroy the universe
if the gulf stream collapses there are nu hmans left
we're not going to build yellowstone, we're trying not to collapse the biosphere
if somebody made a fruit fly that couldn't die, the earth would spin off it's axis from the weight of fruit flies in a predictable number of years
we need to try not to make AI
well i'm not going to create an AI π
you'd need fruit flies that generate new mass
not really, they would just need to move wrong
or gather in the wrong place
or leave lol
the weight of life on earth is insignificant compared to the weight of water
and the weight of water is insignificant to the crust
my point is no process extended to infinity really makes much sense in the end
and the earth is doomed in lots of ways
lets talk about micros
the 328 is not capable of running AI in the terms we've been discussing
that box can't permit AI, so it's a safe box
ever herd of mark tilden?
he made robots out of analog parts and a 74 series inverter
and came up with something supposedly similar to a neuron
it wasn't one 328 though
i was looking at this: https://www.ebay.com/itm/STM32F103C8T6-Micro-Mini-USB-controller-STM32-Development-ARM-Learning-Board/272826365928?var=571972615549
for doing the animations with
STM chips tend to be difficult to make use of
these have stuff for them to use them in arduino ide
but can you use the stuff that's already in the arduino ide on them
it's a bit worse than 50:50 odds
i dunno, but i am not asking a lot i think
it's got 2x i2c
and i need it to display images on a display with a reasonable speed
what I'm saying is if you get an STM based board, expect that you'll need to reproduce a lot of functionality yourself because libraries are likely to not work
SAMD chips can typically be configured to have multiple I2C
how familiar are you with SERCOMs?
displaying over i2c will always be slow
nope
nope?
doesn't ring a bell
assuming its a graphic display
so, SAMD chips have SERCOMs, Serial C...oms, which are serial protocol controllers associated with certain pins
yes, SPI can be much faster
each SERCOM can be configured to operate an I2C, UART, SPI, or etc interface
ahh ic, my understanding of them is not that deep yet
if any two pins are on a SERCOM, you can configure it to be a port that does what you need
i2c has more protocol overhead and tops out at 1mhz and spi can go easily over that
ya, SERCOMS are cool. how are you going to program the mcu?
oh lol what am i saying, i am talking about a full colour tft screen, they don't even come in i2c
look at the M4 schematic here https://cdn-learn.adafruit.com/assets/assets/000/057/242/original/arduino_compatibles_schem.png?1531010817
specifically PORT A and PORT B, notice how many but not all of the pins have a SERCOM#.# designation inside the square
well it depends on what mcu i end up using
the one i was looking at has some support in arduino, but elseways ill use whatever it needs i guess
MOSI and MISO are on SERCOM1+5, that means those pins can be used with SERCOM 1 and SERCOM 5
i am not entirely shy of doing some command line stuff
not at the same time, and each SERCOM can only be used for one interface at a time
I had one of those STM32F103C8T6, and they ship with the wrong resistor on the USB D+ line so you can't program over USB until you desolder the 10k and solder in a 1.5k
well if i can use a cheap enough board like this i won't have to worry about multiple displays on 1 mcu
and I broke mine in that process
I'd choose the mcu based on the software
there are standard libraries in Arduino for establishing I2C and SPI interfaces using SERCOMs so it's also easy to do
good to know π
@stuck moth i was looking at the esp8266 based ones and python before this
wondering if this was any better in terms of specs, not sure what to look at quite yet
since the M4 feather has 5 SERCOMs, excluding SERCOM2 which is occupied by QSPI for the flash module, you can have 4 I2C buses as long as each SERCOM has 2 pins that don't overlap
and those sercoms can do spi as well
most mcus can do multiple SPI busses but the esp8266 may not be able to
SPI needs 3 ports to work right and that's 12 pins, I kinda doubt the pin collision will allow all 4
could i use multiple busses for one task? and would that speed up animation in any way if i divided the workload among them?
no, I thought you'd have multiple displays
starting with one is the best approach though
that was the original plan
if you had two capable chips that you were programming yourself, you could have them talk to each other over multiple buses, but no peripheral device impliments that sort of thing
which probably isn't useful
ill just stick with the esp8266 then
could i hook up some kind of fast memory to an mcu?
the fastest connection is QSPI, which is SPI but it has twice as many data pins and the protocol allows all the pins to be used in the same direction at the same time where normal SPI only allows 1 each, there's Dual SPI but I'm pretty sure SAMD SERCOMs don't support DSPI mode, it starts to get really technical though.
On all of the existant Adafruit M4 boards the single QPSI interface, which is not handled by a SERCOM as far as I understand it, is used for the flash module for obvious reasons. You could have an MCU that exposes QSPI but you'd need peripherals that also support QSPI
yeah that's likely too expensive
you'd be surprized
maybe, but i really need to keep this down to a minimum
the biggest M4 chips are like 5$ at low volume
i mean it's just for an effect in some dioramas
O_o
what about a simple board though?
also PCBs are not that expensive to order, OSH is more expensive and they're 5$ per design square inch, and you get 3 boards. Components aren't that bad, many of them will be dozens of cents
you'd be paying the difference in labor
you would have to design the board, source the components, assemble the board, deal with any problems
yeah lol not gonna happen
you should permit it as a possiblity, even if it's not appropriate for this
oh absolutely, i do intend to get into that some time
i saw some adds about boards for like $5 even if u just buy 1
very doable, but i'd have to learn some cad first
your first order through JLCPCB is $2USD/10
nice π
100mm*100mm
you can definitely get the BoM cost down by purchasing in bulk, I think those same biggest M4 are more like 2$ at volume, and the dozens of cents components are just single digit cents at volume
the 5$ boards are also using automated assembly and reflow soldering which saves a lot of time
incidentally, as i am getting into some boards that use 3.3v now, how is the ld1117av33 as a choice for regulators?
I think I ran into a 1117 regulator a few days ago and it had a really high dropout
i had these little l78l33 ones but they only do 100ma :S
expensive? LCSC has $0.01 regulators
they're linear regulators so you don't want to put much current through them, and with high dropout they're not suitable for use with batteries
Is a 1 CENT LDO voltage regulator any good? Dave tests the Shen Zhen Fine Mad! SC662K 3.3V LDO from LCSC: https://lcsc.com/product-detail/Low-Dropout-Regul...
that's not through hole
or maybe a chunky smd package with the same pin spacing?
so long as i can fit it on some stripboard uknow...
you could probably get some kind of breakout board idk
typical through hole spacing is about 0.1", SMDs are almost always going to be smaller than that
i do have some breakout boards
they are kinda bulky though
can u put 2 regulators in parallel?
not sure yet, just thinking of stocking up on something
i like to use litium ion batteries but i don
t want to put them in series
4.2v is too much for some 3.3v boards, and i don't think a buck regulator would work well as the battery drains
3.8V is a typical top
as long as the dropout is near 3V3 batteries won't be that bad, but also there are buck/boost converters that can go seamlessly from dropping the voltage to raising teo voltage
no for 3V3 devices
100ma is kind of a lot actually. But if you think you need 250, these look alright https://www.digikey.ca/product-detail/en/microchip-technology/MCP1700-3302E-TO/MCP1700-3302E-TO-ND/652680
Linear Voltage Regulator IC Positive Fixed 1 Output 3.3V 250mA TO-92-3
SAMD51s I believe are rated for 3.8V VCC
linear regulators are good as long as the dropout is suitable and the current is low
what the voltage starts at minus what the voltage will be times the current is the watts lost, if the current is very small the loss will be very small, then if the dropout is too high then a large range of the battery's voltage curve will be unuseable and that capacity just doesn't exist
yeah
you'd prefer an LDO?
but then again, i wouldn't be powering much off the mcu would i
the AP2112s used on most feathers have a dropout of I think 0.1V, so you can run on a battery until it gets to 3.4V, which isn't great but it'll work
any reason a 8266 board would ever want more then 100ma?
you'd need a buckboost converter to utilize the whole battery, but those methods have efficiency that trends with current where the linear regulator trends opposite current
so how do people power their battery based wemos projects?
linear regulators are cheap, easy, and require little in additional components
the wifi can use up to 170mA according to the datasheet
capacitors can be used to smooth over momentary current spikes
oops
so people use multiple cells in series?
or wait, those lipo cells max out at 3.6 3.7 don't they
most primary cells put out approximately 1.5V
typo
well alkaline does, and lipo are at around 3.6 but i was thinking of the li-ion ones i have been using
according to this lithium primary cells tend to put out 3V
i know a little about batteries
secondary cells are the 3~4V range
if u treat them well and use the right charger they go up to 4.2
i have a minor led flashlight fetish π
that's a contradiction
overcharging batteries is explicitly one of the top 2 ways to reduce battery lifespan
4.2 is not overcharge
if the rated voltage is 3.7V, 4.2V is overcharging
yeah but that's those rectangular cells
it's about chemistry not shape
Li-Ion and LiPo are higher voltage than LiFePo
i been using these ^^^
it says 3.7V right there
yes and?
any decent charger will charge them up to 4.2
the projection is there to prevent damage, not harm
damage causes fires, harm reduces lifespan
i been using these in flashlights for years, 4.2 is the max
you're doing that thing again
i am telling u they are actually different from the rectangular packs, both li-ion and lipo
prismatic batteries aren't different from other batteries with the same chemistry
cells really, batteries are sets of cells
but they are not the exact same chemistry
the properties that care about voltage are selfsimilar within a family of chemistry
there is a difference there, and i never read up on it in detail, but i am definitely not doing anything out of the ordinary or improper by charging them to 4.2
i have a several chargers for these types and they all go to 4.2, the flashlights all are made to use 4.2
but you are reducing the lifespan of the batteries, which is not out of the ordinary or improper, but it is true
i'm not sure exactly what the deal is, but it is not my choice to do it that way lol
i just stick it in and wait till it's done
consider, tesla cars don't make use of the top 20% or bottom 30% of the charge cycle because overcharging it has a substantial impact of lifespan, so does over-discharging
but i do know that going over 4.2 is considered dangerous, just like going over 3.7 is with the lipo packs
they charge up to 80%, and discharge down to 30%, and they get thousands more charge cycles out of it
that makes sense
much different application though, these batteries are made specifically for high drain devices
like flashlights, and cameras and things like that
when that big hurricane happened a few years ago tesla pushed an OTA update that allowed cars in florida to override both of those limits so they could be charged further and discharged to about 15%, so that their owners could leave florida more effectively
are you saying your flashlight is a higher drain device than a car....
...there's something just... cool about that
not saying that i was just comparing these things to the silvery rectangular packs u typically find in commercial products
maybe the limits are arbitrary, but in any case for my purposes; my charget will charge those packs to 3.7 volts
which is perfect for driving these mcu
those are literally not different from 18650 cells except in their shape
and the protection circuit
18650 cells have various chemistries and voltages, prism packs have various chemistries and voltages, both come with and without protection
which even varies model to model - manufacturer to manufacturer
the prism pack is just a form
the only thing it indicates about the battery is that it's not one of the very early chemistries that needed rigid enclosures
to be clear, I'm not saying you should charge them to 3.7V, but 4.2V is overcharged for a 3.7V battery
if you were trying to take care of them the goal would be about 4V
80% charge
interestingly, they do recommend storing them at around 80% if you dont use them for a long time
https://e2e.ti.com/cfs-file.ashx/__key/communityserver-blogs-components-weblogfiles/00-00-00-09-40/7245.slyt528_5F00_fig1.jpg
these graphs always have fully charged on the left marked as 0% for some reason
oh no, store them around 50%
that's why they arrive about 30% most of the time, it's least damaging to store at 50% and it slowly drifts down to 30% over months to years
hopefully they're fresh then
I'd say for cells on their own, it's reasonable to think they're being sold and shipped soon after they're manufactured, but once they're in a product it becomes way more likely that they'll sit around somewhere before being bought
btw, any high capacity pack, like 5Ah or more, should probably be LiFePo4, because LiFe doesn't like to catch fire even when bad things are done to it
After many questions and confusion over various Lithium Ion battery types discussed in this 18650 thread, I thought it would be useful to try and give a breakdown of the various categories of Lithium Ions. There are more details that could be given, but this is meant to just...
the energy density is as shown in the graph quite a bit worse, but they are a lot safer
that's the thing, these batteries are not made to be manufactured into products
they are meant to be used like eneloops etc
the exact batteries you have exist in products too
I've got one on the floor around here somewhere, came out of a playstation move navigation controller
i'm gonna take a wilde guess and say it is blue
and it has text on it that looks like it was printed with a dot matrix printer? yes?
here's the thing
that's not a detail that matters
it's the same kind of battery coming out of the same kind of factory
i know lol i'm teasing
different packaging
this one is shrinkwrapped witha protection board and a wire lead
there are many like it, but this one is mine, and it's fungible with all of the others because they're a commodity
so anyways what you are saying is the difference is made by the charger, not the battery
close enough
lol no fair quoting full metal jacket, how am i gonna top that without tripping the censor?? π
the charger could be set up to stop charging at 4V, and you'd get a lot more cycles with the rated capacity
overcharging gives you more than rated capacity, though not at a great rate, but costs lifetime
yeah but i wouldn't get 1000 lumen out of a flashlight that disappears in my hand
sure you could
the battery is going to drop from 4.2 to 4V very fast anyway, if you need a high voltage you should use a regulator
fun fact, the vast majority of standard format lithium cells are counterfeit, they'll probably serve the purpose you bought'm for, but they probably weren't made by the company they're marked as
i avoid them like the plague
when it comes to lithium i only use branded batteries
yea, the branding is likely to be fake
says samsung, made by rando factory in china totally unassociated with samsung
nah, i am not ordering them from china lol
of course you are, manufacturing lithium batteries outside of china is insanely expensive
i mean ebay etc. i get them from reputable dealers, sir! π
and where do they get them from
how certain could you possibly be of that
there's too many hands for certainty
i'm in holland, laws are fairly strict here, so there's that
but ultimately performance
these flashlights push those batteries to their limits
like i said the one i have that does 1000 lumen needs imr batteries to get that output
lithium cells are a known technology, unless you're getting straight up fake batteries they're all going to perform pretty similar
the fakes are fun though
i think a lot of fakes are just very old stock
ROFL!
counterfeit is kind of just accepted because most counterfeits perform acceptably
that's an ultrafire though
lol anything with ###fire in the name
except surefire, but they are elitest so bleep them
this one's great, it's literally just full of small rocks
lol
i could still use those for dioramas π
oh hell, lol i already have some of what i need!
i got some wemos d1 minis a while back, and with them i got a battery shield
it is ratoed for 3.3-4.2v and charges etc
the voltages are entirely dependent on chemistry, doesn't matter who makes it or what shape it's in
oh the D1
yea it uses the 1117, which isn't great
but it works
the 1117 has a dropout of 1V
so it can regulate a Vin of Vout + 1 or greater
the AP2112 can do Vout + 0.1 or greater
so it would have to be 4.3v or higher?
it gets fiddly because linear regulator dropout is a curve that trends with current draw....
everything is complicated
it says TP5410 on the chip in the picture
the lower your current the lower the dropout is
maybe I was looking at a different D1
china and whatnot
i am looking at a battery shield for it
it has a charger and regulator, and the listing says 'just plug in a 3.3 -4.2v battery'
gj google
lol that's in chinese π
i haven;t even gotten used to the electro lingo yet
Vout is nominal 5V
yeah it charges the battery
i think that's in
it says u have to charge it from 5v
so it's a charge controller with a built in boost regulator to 5V, that's going to be why the 1117 can work
i guess that's what the usb is for
well in any case it is made for the 8266, so it should work for the other boards based on that, right?
at least to the point of not frying my board
seems reasonable, but because it's boosting to 5V and then linear down to 3V3, it's never going to have a great efficiency
depending on current load the boost converter is probably 50~80% efficiency
this board is giving 5V to the D1
the D1 regulates it to 3V3
wait, so when it says a board is rated 3.3v, like say the 3.3v version of the arduino pro mini, they are only talking about the logic??
and i could still power it with 5v?
right
the only thing running on 5V should be the red LED
and it just means i need to use 3.3v on the i/o pins
so the battery board is providing 5V to the 5V pin, which is common with the 5V line on the USB port
and it is just a boost converter
this has the advantage that you're boosting the battery up to something the 3V3 regulator can accept, and you'll be able to use the whole range of the battery's voltage
the down side is that's low efficiency, and then it has to be regulated down to 3V3, which will be on the order of maybe 90% efficient, but that will drop as current rises
oh what the F am i thinking, anything with a usb port can be powered from 5v, duh
the boost converter will be more efficient with higher currents, the linear regulator will be less, so you're probably looking at about 75% twice
a 2200mAh battery would have a useful capacity on the order of 1200mAh
i have lots of small 5v boosters lying around
i'm trying to solve a problem i don't have lol
if you boost the battery to 4V it'll be more efficient
that shield is prolly just a 5v booster, if the wemos will take 5v
it takes more than 4V
if you have an adjustable power supply of any sort, set it up to produce 5V, and put your multimeter in voltage mode between the 3V pin and GND, also have a second multimeter monitoring the supply voltage
then turn down the supply until the 3V pin becomes inconsistent, and go with maybe 0.2V more than that
the lowest voltage you can put into a linear regulator is the most efficient
well most of what i plan to do with the wemos things are not gonna be portable
i think
and those things that are, i tend to overdo it with the batteries anyway lol
sure, but~
like a 16340 like that is way more then i would need for most things, so i think i can live with the efficiency in most cases, ill have to see
for now what matters is getting something working reasonably well without blowing up
so say the D1 turns out to be stable at 4.2V input, at 5V it's 1.7V of drop, at 4.2V it's 0.9V of drop, so the peak waste wattage is reduced by 47%
I don't know the math, but presumibly boosting the voltage less high also improves the efficiency of that stage
i'd imagine the smaller the difference in and out the smaller the loss
makes sense to me
yeah peak efficiency i think is something ill likely look into deeper around the same time i get around to ordering pcbs
if I'm reading this article right, this graph represents input voltages as shown, with an output voltage of 4.5V, so yes, having Vin closer to Vout improves efficiency; and it improves it a looooot at lower currents https://www.digikey.com/en/articles/techzone/2013/jul/~/media/Images/Article Library/TechZone Articles/2013/July/Selecting the Right Voltage Converter Is Not Just About Peak Efficiency/article-2013july-selecting-the-right-voltage-converter-fig3.jpg
notice converting 4.5V to 4.5V is still 90~95% efficient
right, less heat = less loss
heh, i was wondering what would happen if u get close to the same values
what happens if u go over?
and these curves are going to vary with every single boost converter, and they're going to varry with every converter for different component values
I think it explodes
lol that makes sense
probably what actually happens is that it becomes irregular and does start to heat up but it wouldn't explode at 4.6V unless there's a lot of current
New clock program in gforth
Right now I'm just counting from :00 to :59
The :38 there is a very large font xterm. For comparison, a fairly small font xterm is shown (upper left, yellow text).
The file manager in the upper right (stubbornly) uses fonts too small for me to see, but may be suited to some (possibly most) users.
#!/bin/sh
FNSIZ=128
exec xterm -fa 'mono' -fs ${FNSIZ} \
-ls +sb -geometry 10x5+0+30 \
-fg cyan4 -bg black \
-e /bin/bash --login >/dev/null 2>&1 &
the '-fa mono -fs 128' is the means to make a giant-fonted xterm.
I've done up to 512, maybe 1024 --
it's not especially sensitive to powers of two; that's just a habit of mine.
repository for the kitchen timer is here:
https://github.com/wa1tnr/gforth-exp/tree/testing_aaa-/ktimer
$ git pull --all
$ git checkout testing_aaa-```
I think there's a bash port available for windows.
Years ago it was cygwin this and cygwin that.
there's also git bash and wsl
It would take Microsoft five minutes to write a bash shell for you, but they did not. And these are your friends?
Apple gives you csh and bash is provided.
Why can't the other children play nicely?
I mean window has bash subshell
how ever II'd like to open cmd and have it configured some what similar
ls command not found dir... Like I feel liike I just mock myself
well just make batch files *.bat
Is this an MSDOS derived command set?
When you type ls and some parameters after it, they'll be passed by LS.BAT to the MS-DOS DIR command.
at least that worked in 1990. ;) I used MS-DOS from 1983 to 1994 or so.
make a folder of batch files for the enviroment variable... I like it
There was a 2.5" thick book, maybe 8x10" pages -- all on MS-DOS. I had a copy or access to one.
TBH if I were using MS-DOS today I'd just relearn a dozen or so commands and use them.
I had a rule for myself when using WordPerfect (which had a large following, around its version 5.1 or so)
The rule was: when I travel, I don't bring my 'macros' with me -- I use straight WordPerfect on the other person's machine (it was usually available).
So I always knew how to do it the hard way; the macros allowed me to also do it the easy way. ;)
I still follow that idea, quite a bit, in linux: I do a lot of things manually that could be scripted.
I don't consider leveraging the command line scroll-back history 'not doing it manually' ;)
I never realy got into macros, some vbs auto forms but never macros
I don't even remember anything about them, now. ;) Lotus 1-2-3 also had a macro language, which was the only time I saw my brother take any interest in programming his computer.
I rigged his boot diskette so that when Lotus 1-2-3 ran, it played back a macro that said GHOST IN THE MACHINE.
The last step in the macro was to delete evidence of itself. So it only ran once. ;)
I think I waited 45 days, and did it again, until he asked about it. ;)
I just started on latex today not a huge fan of it personally
I did like LaTeX but you need LyX to go with it.
I've been using tex, is lyx nicer?
LyX as I remember it is similar to WYSIWYG style editor.
I think it let you visualize quickly what the source would do; I don't remember how it was for editing inside the WYSIWYG or if that were part of it.
I do remember liking it enough to use it a lot.
I might have to try that!
It was a little clunky. I hope they maintained it or even improved it.
There might be something altogether more modern that does its job, today.
I'm a huge fan of jupytr notebooks especially when combine with sagemath, but this report was requested to be in latex
I'm always amazed at what wasn't installed, when I install something specific for a specific goal, in Debian.
I try to to as barebones as possible when I start a new system from scratch.
It can take months before all the tools I use for at least five hours a year, find their way onto a new system. ;)
In this instance, it seems to be installing JavaScript for the first time. ;)
I came from troff, so I just dived in to LaTeX the same way. HTML, too. I just put the tokens in myself.
This manual is designed for all of you who have never heard of LaTeX, or do not know it very well. Now, do not panic β you will not need to learn LaTeX to use LyX. That is, after all, the whole point of LyX: to provide an almost-WYSIWYG interface to LaTeX. There are some things you will need to learn, however, in order to use LyX effectively.
I'm currently reading through that
it seems a tad slow, takes bit to render things as you scroll back and forth through a document
Yeah it was slow as I remember it.
But you didn't want to learn LaTeX to that degree, did you? ;)
I mean documents in general aren't my forte
I just didn't have a use case after one project (I downloaded someone's math paper, made mechanical changes, and uploaded it them).
I've had a greek keyboard enabled for years windows+space makes typing up math documents really easy
on any word processor
I like source code; the thing I don't want is a mixed binary and textual 'database' (aka a 'file') that holds my document. I want the markup to be captured in ASCII so that no matter what happens, I still have the base text (and can profitably ignore any markup).