#general-chat
1 messages · Page 184 of 1
(also, iirc Python fors get their own scope while whiles don't
I think this is takefor_in? ```python
def takefor_in(pred, seq):
for i in seq:
if not pred(seq):
break
yield i
def elihw(pred):
for i in itertools.count():
if not pred(): break
yield None
for _ in elihw(lambda: x>1): ...```
Oh my takefor_in was simply "repeat the beginning of @sick adder's solution, delete the spaces, and flesh it out enough to make the self-reference joke but not enough to trip on the fact that it's not actually a coherent idea"
... why do you want to replace while here?
You seem a little... salty..
I did not noted the moonwalk haircut in the photo. I think it was supposed to be formal. Really.
Wut?
I squared C
ok so i just need to get myself a good camera.
unfortunately i only have a good camera on on damaged sony phone but the touch pad is dead.

So i managed to use usb otg to connect a mouse
and paired a Bluetooth media controller.
and i used the usb otg port in mhl mode.
What's your aim with this good camera that you need
?
now all i need is a cheap hdmi cap dongle and i have a good camera with nice focus
Well i can capture my screen fine.
Why do you need a mouse for a camera?
but my work bench not so much
so i figure id
put it on a stand and make use of it's optics for recording and maybe if i find the right focus and lens my microscope
so i can record on the device and on my pc at the same time.
nice lil redundancy
Sweet. That sounds cool
Yeah, right
i tought it was gona be voodoo
i saw
WDYM
elon once said AI was a huge risk and a threat to humanity
Look have you watched deep mind beat pro starcraft 2 players?
have you seen what it can do with the human limits turned off?
Maybe even a hundred years later
Yeah, but that's a video game
And AI has limits too
Me too
i would become a tony stark / elon and build robots anyday
Absolutely, me too
Tony Stark more than Elon
Without the dying part
Cause you can't snap in reality
Which is a good thing
IMHO
"Build robot friend, not robot enemy" – LadyAda
Right!!
If you're mean to Baymax, Baymax will kill you
XD
He will hug-smother you into submission
then eat all your Ho-Hos
LOL
Machine was made to serve man, not man machine.
Right... But you can still like your robot butler
And you don't have to be mean to him - robot friend
Just rehydrate them into grapes
drop testing is all part of the fun
Better yet, have the robot do it for you
Cheers.
recently i saw how much you can do with a nvidia 3080 gpu's cuda .cores.
well turns out you can run a slew of operations at a time that are in the domain of super computing.
i understand sadly this tech is also why crypto mining is often done on these too..
and the power we have in terms of what the common man could buy.. is insane
Alas, the nVidia/Apple feud has made nVidia unavailable to me, so I'm looking at other avenues.
Why unavailable??
Because it depends on the CUDA library, which won't build any more.
Speaking of NVIDIA, what do you think about the estimated time of acquisition of ARM?
I see, that makes sense
I wish they'd hurry up
Blasted European antitrusts
I do have some recourse, I have access to an AC922 at work with a PowerPC CPU and twin V100 GPUs.
🎉
Some people (including Apple, naturally) are nervous about the ARM acquisition.
Of course.... but I'm hoping for some epic new stuff from NVIDIA because of it
I'm also hoping it won't be a monopoly
I figure their DGX A100 will be replaced with something much nicer. But Apple would have few options if they were denied ARM licenses.
Competition drives competitive pricing....
I bought a 3080 at MSRP for no other purpose than “massive tax rebate”, but now I honestly feel unqualified to own it haha
I'm hoping!!
hehe
Retro gaming is a bit smoother with more effects on
I can do a standing backflip so anything bouncy just makes it easier
I know I know pics or it didn't happen...
Hey, way off topic, but really not...
What's up with having orders getting flagged when some of us have 2 or more homes in different states?
Superscript, lol. On smartphones you usually press and hold 2 and it gives you the option, on Windows you can open the Character Map and find it or use the correct ALT+ code [which I don't know]... MacOS or Linux probably ALT+ codes
I'm jelly, yo
I'd recommend emailing support@adafruit.com -- they can take a look at your account and should be able to tell you what's going on
Did that, to "Ben" @ support @adafruit.com.
I just get in the habit of writing it as b^x but that gets confusing when some boolean algebra conventions use '^' for XOR
Would be nice to just have superscript and subscript modifier keys built into the keyboard standards...
Soooo yeah definitely adding that to my laptop
I started laying it out with the FPGA board and it's gonna be a chonker but I will love it all the same
The screen is actually almost exactly the same size as the FPGA dev board on its own so I may just do the laptop as an FPGA board project and do a console-style box for Project: Testbed
Because then I can make it all nice and slim, have an access hatch on the bottom to the dev board, and have the panel driven by part of the FPGA design
As it stands adding the 6507, Pico, and RAM boards either makes it super chonk or the screen has to have a weird lopsided bezel to maintain strength
AHH THAT'S WHAT MY KEYBOARD NEEDS
Although I could mount the Pico and 6507 stuff inside the screen panel and make it the same thickness as the bottom panel...
And put the memory module behind the keyboard with the top of the FPGA dev board facing down and access hatches from either side
Yeah I think that is the solution.
Since the thickest part is the memory module and I can still make the screen fairly thin with the Pico and 6507
Yeah it will look like something from my birth year but it's gonna be sweet
And the specs arent going to be much better than something from my birth year either (with the Pico as coprocessor) just simpler construction required
Oh! FPGAs!
Anyone here know how to re purpose one from a device that you don't use?
I have a gameduino but somehow the flash chip got ripped off so it doesn't work anymore and I want to reuse the FPGA for something
I mean my normal approach is to just pore over the chip's datasheet for a few hours to weeks depending on complexity of the given chip but a shortcut would be nice
And yes I have a universal chip programmer
Oh and on the laptop...
I'm already going this far
I have a socketed teensy
may as well make a coprocessor coprocessor
to do the stuff the pico sees as heavy lifting XD
I showed my physics teacher a protoboard I made. “I’ve seen professional boards…not as good as this one”.
“You have…done this before. You’ve done this many times”
Me: “Haha, I’ve done it for years”
“Yes you have. Yes you have”
My heart was melting
Hehe
@hasty quarry got told by the engineering department lead at community College "I've seen many commercial designs not as well considered as this" to something I threw together and printed brackets for overnight. Good feeling ain't it?
(3D printer)
Of course working from 2020 extrusion and staying up all night helped...
Yeeeeah I found out the hard way that's not how it works
For some reason they like some fancy piece of paper about diplomacy that says you learned something you might not have.
I have a fancy piece of paper that was supposed to get me a job with a starting salary of $60k and raises to $120k within 3 years... I've been out of school for years and still waiting to hit that $60k mark
Aaaand that's why I dropped out. If I'm going to get in debt I might as well have prototypes of inventions to show for it.
I think it's sortof a combo of your luck and skill... If you have something of value to offer, a smart company will snap you up, paper or not, but some companies WILL NOT HIRE ANYONE WITHOUT A DEGREE EVER FOR ANY REASON... despite the fact that a kid with a paper might be worthless to them, and they might be passing up someone with loads of experience.
Yep. So I'm working toward my own company with help from people who just want to see it happen. And I'm like... there almost... Just yesterday someone was making sure my crypto wallet was set up so they weren't sending money to the void
(haven't checked it yet today but I don't need it until I have a board design ready to be fabbed)
Though I really should get that debit thingy researched and figured out
So I'm not frantically doing it when I need to order and then getting frustrated because of a waiting period or something
Lemme know if you can hire me XD
If I can it will be as a subcontractor
Not looking to manage a payroll so unless I have an accountant that's how I'd have to do it.
And I won't hire an accountant I don't have complete and utter faith and trust in.
Because that's how embezzlement happens
(or it's ordered by higher-ups but I'm not going to be giving orders like that. Period.)
Trust good, embezzlement bad
So until I have found someone suitable all work for Full Bootstrapping Solutions will be done by contract or volunteer labor.
(or by me which is tantamount to volunteer labor)
(most likely mostly by me)
Oh actually
I was considering doing early contributor stock options instead of pure volunteer work.
Not quite sure how to do that but it would be ideal
Hmm... What sort of stuff are you looking for people to do?
Good question I had only really considered myself as the main labor force so I had just adopted a "figure it all out myself" attitude... So really anything that relates to programmable IO on Pico and implementing various signal interface state machines to a particular spec would be wonderful... Still struggling a lot with that...
Ah... I can't help with that XD
Well any help with design / implementation of hardware helps too I'm just much more comfortable with that area and pretty much have it figured out (just need to buckle down and learn kicad)
(because while I like eagle, kicad is so much more configurable in the ways I need and doesn't limit your board size if you're poor)
I've played around with it a bit, and I like what I've seen it's just a bit overwhelming with the options
A good thing once I'm familiar but an initial learning curve as a stumbling block
And yeah, at least until there's enough funds for prototyping and to pay out what I consider fair for a contract job, all work will have to be for stock options.
Because there simply is no other fair way
I'm good at testing hardware and trying to break things :P
Haha yeah I'm better about fixing them when they inevitably break than actively breaking them myself so that could be helpful
So maybe like bug bounties for being able to show precisely how to reproduce a particular problem
That would be super helpful
Because if I know the source fixing is always easy but finding the right conditions is always hard
(I do usually get it in the end just not my forte)
Loooots and lots of coffee and print statement fueled over nighters trying to find bugs that took 2 seconds to fix ;_;
I wish coffee could fuel me
Oh... Actually... How are you with 3D printers or linear robotics in general?
I may have something...
(a future huge money maker project that I have entirely designed on paper mathematically with some diagrams [can re-derive if needed] but haven't put into cad yet and I have this severe mental block due to the project coinciding with a... Dark time in life...)
Also how are you with openscad?
0% proficient XD
Because final version needs to have no more than about 10-15 maximum configuration variables that set up the rest of the design
I do shapes in TinkerCAD...
And source project I'm deriving from is written in openscad
(actually how I learned the language was by formatting that code to be readable...)
(do not have formatted version but much better at reformatting for readability now)
(so can do again)
Anyone here use Solidworks?
I wish...
My dad works for Dassault Systemes. Tech Support. He's one of the best there
My brother uses it for his job (civil engineer)
Sweet
Used it briefly at work before. Amazingly powerful, but too pricy for personal use, alas.
Funny thing is he is credentialed as a mechanical engineer but has only landed one mechanical engineering job and the rest civil engineering lolol
Is it really good?
My dad got me a license
Close enough right? Civil engineering is just large scale mechanical engineering
Yes, it's pretty much best of class for a general mechanical CAD tool.
Dang
Yeah he knows it bottom to top. He got me a license to try and get me to learn it
But it's...very overwhelming for me
I've heard a lot of good things but been told it has a steep learning curve
Absolutely
(just like any broad spectrum tool I guess)
If he teaches you, then you can make tutorials and become a famous YouTuber :O
Yeah, working through tutorials is recommended, even if you usually learn software by just playing with it. There are some key concepts and techniques you really need to understand to design things well with it.
Or he teaches you, you teach me, and I become the famous YouTuber :P
The thing about those super high end, enterprise softwares is that there's no community, huh
yup
Because no one can use it
And remaking what you learn as a tutorial sets it better in your own brain even if you never release it.
(though sharing is caring)
He was a university professor in his 20s, so I can kinda trust he knows what he's saying there
haha
But how do you teach something you don't know? XD
By exploring it and figuring it out
Well I don't think "learn" implies "learning for the first time"
ah
By teaching, you learn. It's like...the difference between reading and writing. I'm learning japanese, and I could read it here and there. But when I started writing it, hooooly
Any leg up from the point of no knowledge and having to develop from scratch is a boon as far as I'm concerned
But I assume "develop what you want from scratch in abstract and see if there are any existing tools to implement that with"
Works better than just assuming someone else will implement it and gives a nice pleasant surprise when someone has
I'd guess teaching cements what you learn
Well, that's a more reified version of it
Learn, show, do, teach, repeat
But you have to know a thing on some level before you can teach
But I'd say cementing adds to what you already learn. So...you're learning
It's still learning
Yeah, I'm just saying it's different from initial learning XD
Interestingly, that reminded me of the study technique where you learn something, and write about it as if you were explaining it to someone. Then, look it over, and review. Then explain it like it's to a 5 year old
And do it until the latter makes sense
That is when you have mastered the topic
Yeah that's why I like explaining what I'm doing to actual toddlers (lots of niblings running around at any given time)
Even if they don't get it, attempting to simplify often gives great insight
Yeah, simplification is complicated XD
It's quite ironic
If you can't explain something simply, you don't get it
You need some serious connections in the brain
Or you're bad with words, lol
Well yeah, that too
bbl XD
Okay, posted in the main community of my startup: Achievement Get! Company Slogan: with the slogan in quotes. It's official.
It felt too right and I got tingles so I had to.
What's the slogan?
"learn, show, do, teach, repeat"
Woops
Got tingles a second time writing it so that just solidifies it
The show is important. If you don't show your teacher what they taught you, you may have learned it wrong.
Then you do it to get a feel, then you teach someone and move on to the next thing.
Having now attained the skill, you can strive for another. 😄
And because part of the process is to teach it requires new learners and perpetuates the cycle.
But unfortunately what you find in a lot of colleges is professors who went straight from the "learn" step right into the "teach" step skipping important steps in between. This is why it should never be a career goal to be a teacher until after you have been in the field you are teaching to some degree long enough to be competent.
Otherwise you have potential incompetence being perpetuated and amplified and it's where the "memory / computing power is so cheap these days don't worry about efficiency as long as it works" mentality we see in a lot of software design. Sure that's fine assuming only your program is running. But again, sharing is caring.
So any code designed for a multi-application environment should be designed as minimally as possible or you end up artificially requiring better and better devices to accommodate more and more poorly written code.
Assembly for life!
(only half kidding but really if you can make at least parts of your code more efficient using assembly or even c libraries it takes the general strain off of the system and allowing much more parallel productivity)
But the "just use as much of the system resources as you need to make it work and don't fix it after because yay you did it, it works!" thing will not fly on my game console's officially supported releases.
There will be caps. Not even artificial, intrinsic to the system. It will be capable but you will have to be clever about implementation efficiency to make a quality product that the users won't instantly realize was poorly designed
And that's not on purpose, just an observation I'm making as of now due to the architecture haha
But it's perfect for my ethos
If you can't get it running on this hardware with those limitations then you can't release the game...
(or other software)
Some would say THERE WILL BE CAPS
Bahahahaha
Got a good chuckle out of me gg
How's it going @late fulcrum
Made me lose my train of thought with that haha
As bubbles would say: "deeeeeeecent!"
Anyway so I realised there's actually quite a bit to do before I can run code on the laptop
Because I decided to integrate an FPGA dev board and leave room for a socketed Teensy as the Pico's coprocessor later
Almost like how the 386/387 pairing was but if you were already using the 386 as a coprocessor for a 286
Knocking off after a long week. Got some new early experimental pack film in the mail, looking forward to giving it a try.
The old style instant cameras used film that came in packs. There was the "integral" film that most people think of as "Polaroid", and the earlier pack kind which you'd pull out, let develop, then peel apart.
Yes. But this early production stuff is "packs" of one exposure apiece.
Nifty
I've been cycling between my projects so I'll let you know when I'm back on the train to CRTville
(probably a few days before then, want to design and print and assemble a custom 6507 / Pico / FPGA / ESP8266 / eventually teensy 4.1 based laptop first)
Here's a fun meta picture I made, taking a picture of one old Polaroid camera with another old Polaroid camera, loaded with long-expired pack film (hence the weird colors, uneven exposure, and one part where the developing gel didn't manage to go)
That is super cool
Those are some really interesting artifacts
I hear museums sometimes are known for having interesting artifacts on display.
hides from the joke police
♪ Their house is a museum, where people come to see 'em ♫
I was quoting the theme song to The Addams Family TV show. When I was a kid, I wanted to grow up to be a weird mix of The Addams Family, a mad scientist, and an inventor. Years down the road, I haven't really grown up, but I've more or less managed the other three.
ah
my backordered PLDs just arrived!
I might have enough now to implement a 6502 with them...
later goals tho
AYYYYYYYY!
just got my first donation specifically for my company 😄 😄 😄
holy crap things are looking up today
Taking the level of sophistication down a few orders of magnitude... I am in need of a new charger for my laptop. The stock charger is 65W max, USB-C. Are all 65W USB-C chargers (that plug into 110V wall outlets) from a semi-known brand effectively identical, or is there some aspect on which a crappy (but still regulation-compliant) one could decrease the longevity of my laptop?
(Also, is this more a question about the charger itself or about my laptop's battery?)
For some laptops, it might cause a problem, but USB-C has well defined power delivery standards, so any compliant supply and laptop should play well together.
what goes into a power delivery standard?
presumably at least a bound on the deviation of the power from its nominal value, anything else?
USB has a whole protocol where the device and the power supply negotiate voltages and current limits, etc. It's fairly involved.
I'll look it up 🙂
thanks
For some reason a power supply being able to talk to the device about what it knows about its own capabilities just sounds awesome
My conversation led to you discovering that haha
yes indeed
like, it has been my philosophy for a long time but never in such concise wording
of course!
until you plug this power supply into a previous generation device that doesn't provide that info and it refuses to power it at all
Safe defaults sounds good to me
I thought they'd default to the previous generation 5V if they didn't see any of the power delivery protocol.
Today I finished my precalculus test 20 minutes before anyone else (again)
I'm starting to feel bad for my classmates. Like...a guilty feeling
Looked around and saw people suffering, scratching their heads
I'm a new student there, and the first time the teacher pulled me out. The conversation went like "Where do you come from? How?"
Noice... One time my teacher pulled me aside after precalc and was like "you see overarching patterns nobody else can see that we teach in later classes, how do you do it?"
Fortunately for me I had already taken Calc at that point hahaha
But it was about stuff that had been glossed over in the first go around and I was coming to realizations about
Right, I know the feeling
Honestly I thought I went pretty slow this time around
I was tired, I felt like I could've made much quicker work of it
And when I turned it in I expected someone to follow a minute after. But 20 minutes passed
That teacher let me use myself as a reference on my test for a calculus relationship I saw useful for one of the problems but hadn't been taught yet hahahahah
Ah, nice
By "deriving" a formula in calculus, what do they mean? I feel like they don't mean the derivative
She was really cool and I only barely passed the class due to her good graces because I hit a severe depression bout that semester
Do they mean something like how the circle area formula is just an integral of circle slices with slice width approaching 0
And that's how it's "derived"
Ah, so a derivative in calculus is an operation
Yes, to find the rate of change at any point of a function
they mean getting the formula by logical arguments and computations from known results
Yes but it has a formula you can use, whereas the colloquial use of "deriving a formula" means to create a formula from observations or other formulas which describes your system.
Or starting from the known axioms (like sin² + cos² = 1) and doing the algebraic manipulation to derive the formula you need to solve a problem. I did this a lot in school, because I'm rubbish at memorization, and realized that instead of memorizing 50 equations, I could just memorize three axioms and learn how to derive the others when I needed them.
"When you understand something, you don't need to know much about it" - my physics teacher
This goes back to Euclid who derived all of geometry from his axioms, or postulates
well, almost
So "taking a derivative" is a specific operation, whereas deriving something is to create a formula from underlying patterns and knowledge and does not have one set "formula" to accomplish
yes
And the logical question may follow, "if taking a derivative is an operator, does it have an inverse?" and the answer is yes. Integration.
But there are some subtleties
Because of the fact that you lose information when taking a derivative (derivative of a constant is 0 and the operation applies term-wise on a polynomial) you have to put in an unknown constant value when you take an integral
Yeah, I know a function is an integral of its derivative, the concepts are fitting neatly in my head for now
I need intuition
Yeah, that comes with working problems
I've been playing on graph paper on my own time lately
I'm pretty sure when they ask you to derive something they are asking for one of these https://en.wikipedia.org/wiki/Derived_functor
In mathematics, certain functors may be derived to obtain other functors closely related to the original ones. This operation, while fairly abstract, unifies a number of constructions throughout mathematics.
But once you have the intuition you begin to view each axis as built up of layers of each previous axis
left derived or right derived?
And the whole world becomes layers
excellent, another of the order
@wooden schooner are you an algebraist too?
I earn my bread and butter (and money to buy soldering irons) by deriving functors left and right
meaning you're a mathematician working somewhere in the vicinity of algeraic topology or geometry?
All these people getting jobs in mathematics and here I am just having fun with it...
Lol
Not that I think I would enjoy it as a job
It is handy to know shortcuts facilitated by math for faster code...
Which is actually what the calculus class I took focused on
shortcuts are good
but most mathematicians I know went into that because they consider math to be beautiful
and they do have fun with it
Oh I like the beauty for sure, the shortcuts are a bonus
I just think I wouldn't enjoy doing it as a job
Would suck the life out of it.
I'll use math techniques to make my code faster, but the main advantage is that any math-heavy jobs go to me, so I'm really valuable to the company.
Hmm, not CRTs, but giant nixies shaped like CRTs: https://www.youtube.com/watch?v=c1ybZcS1dPg
We have parts for 10 more H tubes, I am now trying to get them to finished tubes so we can look at the problems and make design changes for the next batch. Glasswork coming!
NEWSLETTER
► https://www.daliborfarny.com/newsletter
WEBSITE
► https://www.daliborfarny.com/
INSTAGRAM
► https://www.instagram.com/daliborfarny/
FACEBOOK
►...
I found a nice writeup here: https://learn.sparkfun.com/tutorials/power-delivery-board---usb-c-qwiic-hookup-guide#usb-c-power-negotiation
I remember two events from High School about being an "advanced" student. One was the geometry teacher (her first year only a couple of months in) where she reported to the department head that *I was 'disrupting' in the classroom, because I was sitting in the back row clearly not paying attention and doodling, but when she called on me (to punish me), I always had the right answer". The next day the department head sat in on the class and the next day he was standing at the door and pulled me aside into his office. He had me hand him my geometry text book, then presented me with "This is a final exam for the class, take it now." He watched me finish the hour test in 30 minutes, said I got an A+ in geometry and handed me the Algebra book and took me in to introduce me to the Algebra class that met the same time. That Department Head became a great friend and later told me his analysis was that I knew the subject better than the teacher did.
Me was the opposite, my teachers said I was a gifted student, etc. to everybody and I always tried to obey them(the professors) hard to avoid any problem to me, I never liked to call atention, it bothered when I finished the tests in less than 40% of the time of most of the classmates and got only great grades, I never liked to be out of the normality and everybody noted me getting out of the room first, participating the class more than everybody else, having more academic relationships with the teachers than everybody else, etc.
My father told me to avoid to make any slight problem to any of the teachers all the time and I always did it.
I always avoided to make any problems with everybody.
🪶
Even right or wrong.
I disagreed with my teachers from pre-kindergarten on, and never fit in with the normal kids.
Hehehe I loved all my teachers from kindengarten...
I remember all of them.
One more pretty than the other.
:)
My parents still remember a lot of their teachers, but I just find that weird. I can only call to mind a handful of them.
Yay LOL
Are you me?
Only difference is I didn't do pre-k because apparently you had to pay out of pocket for it
(or rather parents do)
I just found a bunch of FRAM chips on DIP breakouts... Time to make a crazy NVRAM system using the serial ports on some VIAs
Well, I mean, not time yet... Still working on layout still because I keep getting ideas to add ;_;
Buut
Looks like all the stuff I want will fit in a nice compact space
Note that the boards on top will be going behind the monitor
Not too bad
Top and bottom separate
Chips on breadboard irrelevant just haven't removed them yet
Probably will actually flip teensy around
For sd card access...
And yes I have batteries. Will be repurposing a USB power bank.
Probably will use status indicators on the power bank boards to double as inputs to the system to let it know when it is running out
Since I'm pretty sure it has a charge indicator with 4 LEDs that does flash codes
Yeah I can actually see the LEDs... Just bodge some tiny wires in to act as signal inputs to some transistors for sibslling the MCU or something
Or have a PLD that tracks state as a code internally which can be read by the MCU that might be better
This thing gonna be a beefcake
And chonky
Like a cute little chonk pupper
Wait a second Im being silly
I'm literally putting in an PFGA why don't I have that track battery state and store it in FRAM
As well as boot config options
Might build my own battery charge level detector for finer granularity than 25% but use those LEDs for the flash codes
And break the signals I'm using out to bigger LEDs on the front panel
Since this unit will probably be on the back or side
Since I want the batteries to snap into the sides of the base
Or rather, have an assembly that holds them on the side but connected as one unit to prevent charge imbalance
May even reverse engineer this and build it bigger if the SMD chips I'd want to reuse have a low enough pin count for my breakouts
So I can actually tap straight into its onboard systems
Looks like I can get at least 2 of the required chips on breakouts
I can just about do the other 2
One pin too many on each side
I could do the tiny bodge wire thing there too...
I did have a roll of fine shellacced wire somewhere around here
Oh!
It's just one chip with too many pins :D
So only 2 tiny bodges needed
Looks like a 2 layer board too awesomeness
Looks like I can read all the SMD part numbers but like.... Caps... What's the procedure for checking capacitance again? (will look up, but this looks promising)
Worst case I just use original caps very carefully kept separate and not lost.
Might just be able to find which one is the charge controller and rebuild the circuit from the datasheet parameters
Well that one's just a MOSFET
Found it ht4901
That's the one with too many pins lol
Also there's 2 tiny 6 pin chips that are either teeeeeeeny tiny microcontrollers or transistors marked with U as the MOSFETs are...
Anyway easier to RE than a CRT controller lel
Noice they give a complete schematic in the controller datasheet!
I bet it's a direct implementation as this was a cheap Chinese device LOL
And I happen to know this controller has active sensing for low current draw and does a sleep mode
But keeps operating if current keeps being drawn steadily until it's close to dead. It also has the ability to be shut off and turned on by a signal which I can multiplex between software control and a button for the power switch
:D
So all I have to do is reimplement this system in easier to tinker / more robust components and then I can integrate it into my laptop for power / charge controls
Nift
PERFECT! one of the pins that sticks of the end isn't used in the circuit and the one opposite it is ground
😄
I CAN DO THIS WOOOOOO!
And yeah looking at the schematic and at the board at the same time it looks like a direct implementation of the schematic in the datasheet which is awesome but I do only get those 4 charge indicator signals
But I think I can use a high resistance voltage divider circuit to bring the peak voltage up to 3.3v for Pico's analog input range to get a more accurate estimate of charge level
Actually these are 3.6v cells is that too far outside of Pico's direct analog input spec?
I guess there's a reason I'm using El Pico Yolo for this
We shall see
Lol
It might not harm the input if it's just a little out of range, but anything above 3.3V would just read as the maximum analog value rather than provide any useful measurements.
Which is fine if I consider anything above 3.3v "charged"
But I need to see what the spec on 20650s is for that
Generally a lithium cell would be down to probably 25% capacity or so by the time it dropped to 3.3V.
Hmmm... Well then what I can do is use the onboard LEDs to get the 25% granularity and check with analog when in low voltage to trigger a sleep mode or something
Or use a divider like I was saying
Should not need high current if sensing voltage right?
So I can use really high resistances?
Yeah. The main limitation would be the input impedance of the ADC... it needs a little current to take a reading.
(that way it isn't passively discharging too much)
So just kind of find where it reads best with the lowest current I can get away with?
The datasheet will usually spec it.
Ah so then I just throw the book at it and pull out ohm's law in a court of schematics
Find the current it says is recommended and find my resistances from there
Do you know if a 3.6v cell is "3.6v nominal" or "3.6v maximum"?
Nominal. They'll usually have a fully-charged voltage more like 4.0V or so.
Okay so I'll tune my divider to that
Maybe with a trimmer pot so I can fine tune
(in addition to the high resistance on whichever side I put it on)
Basically tune for a range of 3.6 to 4.0 so I can adjust to the cell wear
(for more accurate readout)
Just thinking ahead idk
I could use a special register location that's software controlled eventually that's effectively a DAC but buffered through transistors which have their collectors connected to the battery and their bases connected to the digital signal. Basically a selectable divider...
That is, an analog divide by n
But digitally controlled
Because if a resistor ladder is just that same concept but connected to the stable system voltage why should it not work for fine tuning the charge indicator signal in software
Though... Would need optoisolators also powered by the battery on their output and signaled by the MCU to bring 3.3v signal to whatever voltage battery is at minus diode drop
So its range more or less matches transistor collector range
Then just tune your divider to account if you want high precision but just use that if good enough is good enough
And it's software tunable
But for now just a trimmer pot
;) This is an ongoing project several years in the making
With multiple branching evolutions
Yeah thinking I'll reimplement the battery charge monitor chip's circuit from scratch using the components I don't have replacements for from the original board (coils, chips, etc.)
And then have a signal that's ORed between an IO line and a button for where that button normally goes to the chip
Idea
If I use a 0.7v diode
It should bring the voltage within spec right?
Without drawing passive current
Since it's a constant 0.7v drop
And I don't need a current path from positive to ground
Might even be able to tap that signal directly from a diode I already need to have in-circuit...
checks schematic
Oh never mind it uses that diode on the output rail
But still one diode is better than 2 resistors and a trimmer
Even if I lose that trim capability
Okay there are subtle differences between the schematic in the datasheet and the implementation on this board but I have a solid starting point
And should be easy to clear the board and do inventory of part numbers, values, see if those match part numbers in datasheet schematic etc.
Since it's SMD and no more than 16 pins on a chip
Then I can see if I can get it charging and discharging properly
With indicators and whatnot, then I can see if I can read voltage after diode drop
Since max charge voltage is about 4v then 4v-0.7v=3.3v if my understanding of diode drop makes sense
(and you know basic arithmetic at midnight)
Think I figured out how to build the game console prototype housing...
Had to trim juuuuust a bit off the ends for them to slot in nicely but it has room for several deep
Well
2 and some change
Which means I can fit 8 of these boards in here.
The board after trimming...
Didn't even have to remove any pin holes 😄
And the way these sit, each pair of boards can have a set of screws as posts to keep them aligned with each other because those end holes don't get covered by the rails
Not only that but check this out:
Note the alignment of the closer holes with the screw holes on the chassis
Backplane here we come
Something like this
Each half can be socketed into a pair of boards and removed separately and the same can be done from each end with a gap between the boards inside
smells like specially processed american meat
Good morning
these constructions at NASA are incredible.
I'm sure soon human kind will discover another propellant other than these common gases.
Well, you don't want to use anything that's actually poisonous to release in large quantities to the air, so you're somewhat limited in the choice of chemistries.
Anyone know how well this would work?
(now that some people are awake)
Basically it's selectable parallel resistances just like a normal resistor ladder but with the control coming from a different voltage than the supply so that you can smoothly control the supply using a different control voltage signal
For tuning in battery charge readings
Woah!
I just had an awesome idea
Pitop in a box
(yes that's an adafruit box)
Lol. I was working on a Game Boy type thing with a Zero W and a TFT LCD in a cardboard case at one point
Just whipped up a cute little calculator of center of mass
For physics (yes, I'm enjoying it that much that it's inspiring me)
Noice! Yeah physics is quite enjoyable, glad you're having a good time with it :D
How do I find a triangular prism's center of gravity
That's wonky
Do I have to use one of the triangle's geometric centers? Probably
I bet it's the centroid
It is! Hahaha
If my conversation with my brother about center of mass that I'm remembering from years ago is correct, any smooth density object will have its center of mass at centroid
Nice
Rather, same density throughout.
He gave us an assignment to find the CoM of this wonky system in X Y and Z...gonna juuust program this calculator to let me put objects on specific coordinates
Heh
Yeah I mean that works
Doesn't give you as much intuition as working the problem by hand
Hmmm, true
Reason I didn't learn algebra until college is because I wrote programs to solve my high school math problems
So yes, write the program but use it to guide you rather than give the answer
How would I algebraically find this...hmmm
What was the line formula in 2-point form?
This is beyond my area of comfort haha... Haven't formally learned CoM stuff to a point of being able to teach it
I could make two formulae and solve the system
Thats generally my approach if I have enough info to do so
Lots of rework and substitution
Wait a second
Shouldn't it be (x1+x2+x3 / 3,y1+y2+y3 / 3)
Wow, yeah, that is it
Works with just 2 points (Nvm I'm not sure about that)
Yeah, it seems that's what the centroid really is
You know, with how beautiful math is...if I'd attempted to solve it via equation system, would I have arrived at the same conclusion?
The thing about math is that you always end up in the same place, obviously or not
Yeah so long as the steps you took to get there weren't erroneous
You'll come to the conclusion you're looking for from several paths
The most you'd get is the same solution in different arrangement
Which is fine, rearranging is easy
But I find that beautiful
Really gives some skeleton to what Galileo said about math being the language the universe speaks to us in
If your permutation rules are set up correctly every permutation of a given system is functionaly equivalent
This is where Turing completeness comes in
That is permutation of underlying structures of a system, given the correct permutation to input will give the same output
nods
Also why emulation of other CPU architectures works.
(in code running on a CPU that the program code isn't written in)
Anyone know if modern ARM chips are software compatible with GBA ARM?
Welp, I applied for Grad School
Would be cool to emulate GBA hardware on my console too and run the games directly on the native CPU
Right on! Good luck!
MSc in Computer Engineering, Online at NC State
Was my application 2 months after the July 1st deadline? Yes. Did I apply anyway? Yes because life is about chances
That and federal loan repayment starts in March so I’m hoping to not be in repayment then lol..
Oh snap! It's an arm7tdmi in a SoC I could totally run the code straight on Pico if I emulate the other hardware
I just use the GBA emulator on my Odroid-Go Advance.
I'm making a game console so I'm trying to find ways to make it more capable
Needs moar Z80... Obvs. 😁
Already have it planned to do any 6502-based system's programs...
(on an actual 6502)
(but emulating the rest of the system)
I'm supposed to be designing the precursor project today which is a laptop of similar architecture but not the any-6502-system capability of the final design
(using a 6507 with 6k of 3 banked 2k slots controlled by dedicated hardware with the lowest 2k being controlled by Pico to prototype that system since the final design will have 6502 use Pico's memory for all of its address space but that's hard to debug if the 6502 can't fetch test code)
But yeah it's a really slow day today... Mentally and physically
(first bank is 2k out of 64k SRAM, next is 2k out of 256k flash, and the highest block is 2k out of 32k of EEPROM, each bank is selected by storing values corresponding to the high bits of address for each slot to a special register at the top of the lowest 2k which points to transparent latches with their output on the high address lines of each bank slot)
(well, 3 special registers with 6 planned, want to extend each bank tp 128m eventually because why not)
(current maximum is 512k per bank as it's 2k banks [11 bits from cpu] and 1 or 2 8 bit transparent latches each for high address [8 or 16 bits] which comes out to 19 or 27 bits of total address per bank slot [512k or 128m of address space per, 384m total memory without even needing to talk to Pico] but the Pico will also detect stack pointer rollover/under and keep track of an internal count of what stack page it is on, thus deepening the stack while using it in the standard way)
(and being separated from main memory)
At some point I want to have Pico talk to 4gb of laptop RAM just to have a 6502 system with 4gb of stack space
It would then be the goal to figure out how much you possibly can get away with storing on the stack
Which means lots of recursive functions
Also since Pico would be the mediator, it could operate on the memory space directly without affecting stack operations
So 6502 can dump a bunch of stuff to stack, Pico can scan over it, and if you don't want to have the 6502 read that data back in, just have it set up stack pointer so that it's back where it was when it started.
And the Pico is fast enough that it can buffer in all the stack data before the 6502 would ever have a chance to overwrite it
Tfw you come back to another FBS thesis
It's something I'm known for everywhere I go...
Lol
Too much zipping around in my head to not go crazy if I don't discuss it
That sip of coffee was way too hot...
Oh I know about coffee
Hah
That's quite relatable
Coffee's effect on me is like what ADHD medications do to those who have it
I get 30 times more talkative, and start fixating hard
And I get so much more curious too
One time I spent like 2 hours researching about the ancient prototaxites fungus...don't ask. Look it up though
Yeah... I've uh...
Well let's just say coffee fueled this beauty:
https://i.imgur.com/icRq2Ki.png
No clue why it isn't showing a preview...
There we go
Guess it wanted the link on its own line or something
The other side
Wow, those drill hits are small for the Teensy: what kind of socket are you using?
I used an IC socket
Just bog standard
Didn't realize how small they were going to be until after the pins would not fit directly from the teensy loooool
(made the footprint from scratch)
Let's just say this was 3 days of mania and coffee with a time budget on how long I would have money to get it fabbed...
I made it do what I set out to do but it wasnt to the standard I wanted... Pico came to the rescue with PIO
Turns out just overclocking the crap out of teensy only does so much for vga signal artifacts...
(mostly edge noise that caused line shear)
There were some resolutions it could do cleanly but it just wasn't quite what I wanted...
And the ones it could do cleanly it could do at surprisingly slow clock speeds
But even cranked up to 1.008GHz it didn't completely fix the line shear
On the ones that it couldn't drive cleanly
You may notice a few mistakes if you look hard enough lel
For one
That is the wrong type of usb port 🤣
Meant to be a host-capable port but ended up putting a square one on...
Like I said mania and coffee
Was so focused on my experimental twisted pair traces that I didn't pay attention to what port I was putting on hahaha
Likewise on teensy profile I was so focused on getting pins in the right place I didn't even think about size
But hey it worked! EPILEPSY WARNING
Oh I guess it wasn't flashing that fast...
Still better safe than sorry
That might still trigger someone, so better safe than sorry
Sometimes it's not about speed, but about contrast shifting and patterns
Ah yeah that makes sense, large potential shifts in brain signals and whatnot
Unrelated... I just called the awesome looking cupcake shop I walk by most days, and sadly can't eat their cupcakes because they make peanut butter ones with everything else DX
Very
But it's a tiny shop, so I can't blame them for not having a separate kitchen
Or complex handling procedures
Right it makes sense it's just unfortunate
Yeah... Why does my life suuuuuuck
❤️
Don't worry be happy, find a good cupcake recipe that's easy enough to make at home which you can have
Maybe I'll just make my own cupcakes! With black chocolate and cookies!
You think that's bad?!?
Five guys doesn't deliver to my apartment! 😫
Five Guys will kill me
Yeaaaaah can't escape the peanut allergens there...
I can't go near a Five Guys without feeling sick
Oof
Chic-Fil-A too
No allergies, so.. yay me?
Lucky human
Aye
That sounds like an excuse.. 🤣
I sneeze a lot too but that's because I have a genetic miswiring between my optic nerve and the ones that trigger sneezes
I won't ask you to feed the rabbits then..
Bright light or sudden light shifts can make me sneeze
I mean, I'll still feed and pet them XD just gotta wash my hands
Hehe
Mutant!
(apparently this only affects about 1/3 of the population, I used to think it happened to everyone lol)
I've never met anyone who has photosneezes
Yes this is the light bleaching chemicals in your retina
Yeah there's this big unshielded nuclear fusion reactor just blasting us with radiation constantly and nobody is freaking out
😎
Do you need moonscreen?
Queue Corey Hart
404
REMASTERED IN HD!
Music video by Corey Hart performing Sunglasses At Night.
#CoreyHart #SunglassesAtNight #Remastered
... I never really focused on what the heck he was saying to understand XD
Can't tell you how many times I've heard that song and never knew what he was saying, lol
I know most of the lyrics :P
Ah that song
Never looked them up though
When he goes to the higher/strained parts i can't understand at all
So don't know all
It's about tripping at night
"I wear my sunglasses at night so I can, so I can keep track of visions in my eyes"
Ohhh.... I understood that part and had no idea what it meant XD
I, meanwhile, wear my sunglasses at night so I can see
There's also a part where he says he wears them so he can see the lights before his eyes or smth
Which applies here
Lol
I want goggles that can selectively block out intense light, like headlights... even with the sunglasses, when I'm driving on the highway, they're painfully bright, especially when idiots have their high beams on...
Auto darkening welding helmet with adjustable darkness
Just pull the module from one of those and make a set of goggles around it
Or make one from scratch with an LCD panel and a low voltage solar panel
(how the commercial ones are made)
Needs to be selective -- those darken the whole glass
Ah... Sooo you'd need computer vision and a dot matrix
Basically, yeah, it would need to be able to detect light and block out just that, not my whole vision
I mean you could just have a well characterized camera and just say "if there's a high contrast point here, it's coming along this angle and lines up with this point on the panel"
You know the telescope filters that block out most of the sun so you can view the corona? Same idea, but automatic and shifting
Then you use the camera's sensor to characterize the size and position to block from the image
The tricky thing is making sure it can do it fast enough and accurately enough to use while driving without blinding me
True...
Just like... Get some of that element zero from mass effect, put the earth next to a black hole, and locally make your time run normal
(using the element zero)
lol, I need to play Mass Effect
That way everyone else is going so slow you have plenty of time to react
Today is reeeeally slo brein dai
Blegh
I need to get to work on that laptop but I am really not feeling it
Some days my brain thinks "openscad! What a wonderful idea!" but today is not that day
I was working on GPS and now I'm doing a financial analysis that I told myself I'd do next month
Isn't it crazy how pressing future goals are when you're working on a present one?
That didn't work
There we go
why does ibuprofen have to make me groggy?
took some for my knee cause it was killing me to stand on and now my brain is mush
My brain is just mush and I'm in paaaaain
took a nap because blegh and had a dream where I couldn't focus and felt like crap, then I woke up to a reality of the same description...
What’s up fellas
Sky
Hey I know this probably isnt the best place to ask for help but someone here has to know this, I need to replace these capacitors but I don't know what to buy
I know how to solder, I don't know how to read capacitors or what they're measured in
#general-tech Will get you more attention
If you look at the other side you'll probably find the info you need. There are 2 important things: the voltage, and capacity. Voltage will be a number with V, so like 16V, capacity will be in uF, so like 47uF. Just match those values (especially uF needs to be exact, voltage can go a bit higher) and you should be good
I wish I could piano
Thanks, 35v 47uf
No problem! Thankfully that should be an easy value to find 
I sent that robot idea(that obey commandments from parents in a big list, more than 1000 for example) to researchers/professors at MIT, Stanford,University of Oxford,University of Cambridge,KU Leuven, Politecnico di Torino, Saint Petersburg ETU, Heidelberg University,Saint Petersburg Public University,Jiao Tong University,Osaka University, ENS Lyon...
Only University of Cambridge I remember counting more than 200 people.
I believe it was more than 1000 people, no joke.
I'm glad you're enthusiastic about the idea, but the reason nobody has done it so far isn't that they didn't think of it, but it's just that it's too hard. Nobody knows how to build a robot to follow a general set of commands like that, even if they're an actual university robotics researcher.
As has been said many, many times before... A robot can't just take a list of commandments or rules, it has no way of understanding what those really mean. You need to define every minute detail, and even after you define everything, you might need to specify more. It's not like teaching a person rules
I noodle but I'm by no means proficient... Which is silly there's a baby grand sitting outside my room...
I should be practicing
I mean it kind of is... Ever watch a child develop?
You gotta do a lot of examples and there are many many failed attempts to parse, repeat, and carry out things they are trained for
But eventually the system is "good enough" to perform in its environment as expected by the others in that environment and how they collectively have been trained to behave
Some professors here at Brazil too,etc.
Pretty great work.
Okay well I went and got my notebook from my friend's place and now I'm feeling more like writing code than drawing but at least I have something to write measurements down in now
And I did a quick sketch and description of what I'm trying to draw in openscad
So I know where I'm headed now 😄
The laptop will be made of stacked cup-shaped rectangles and for the very bottom it will just act as a sort of lid / interface cutout around its perimeter for the FPGA board... Then the next one will have mounts on one side for the FPGA board and on the other side for proto board, then a special keyboard panel to cap the bottom section off. For the top it will be similar but no FPGA board and will be capped off by the screen
(just the 2 layers on top not the 3 the bottom has)
i found a pfga i like that can run at 400MHz wit a OC of 500 and 600
i have 4MB OF VRAM configured and created a lib for the esp32 to easily send textures from memory or SD.
I am porting my 3d functions and shared engine to the fpga right now to compare exactly how much more freedom the esp32 can get not having to do any 3d or drawing math and not needing to drive the lcd or tv out
now the thought that comes to my mind.
Letting 3 chips access the same SD card. "not sure how but it seems im gona need to make a controller just for the SD if the FPGA runs out of guts"
But yeeah i have a audio codec, a esp32 and a fpga and they all need to be able to access data from the SD.
so i'm now writing the SD lib for this.
Too bad the kendryte k210 was not friendlier in terms of software support.
would have skipped 10 steps here
yeah there is a huge difference not needing the esp 32 to handle texture and model data to be loaded.
i have to now learn how sd cards work properly.. joy..
if it's done right i should be able to let each device wait or have a interrupt to be able to ask for access.
and i need to also figure out how the audio codec requests data from the sd..
i think i hate myself.. i'm biting off too much to chew and i think it's getting too complex
and complex means problems
There's one huge problem with that analogy: fresh humans have a base level of programming, a robot does not. Humans are designed with intrinsic pattern recognition, and basically immediately can tell what humans are (like their parents) -- a robot can't even identify shapes without a bunch of training. Getting a robot to recognize humans is a process; and even with a lot of training, occasionally a robot may think a lamp is a rocket or something. But the real trouble comes with trying to parse human rules: like "do not harm humans". Harm can mean a lot of things; mental, physical, emotional? If physical, are we talking about causing any sort of cellular damage, or breaking off a limb? Nuance isn't something that's easy to convey to a machine. If we're talking about emotions, a robot has no frame of reference; humans are designed to feel empathy (yes, I know, some are broken and can't do that, but that's a manufacturing defect). The more abstract the rule, the harder it is to get a robot to understand, and the more likely it is for something to go haywire.
@dusk oracle maybe you should have another MCU talk to the SD card and have everything talk to that one?
@delicate streami was planning on such but i haven't seen anything online doing the same
Having one act as an SD controller sounds easier than making a library for 3 things to talk to the same card
What exactly do they need to access? Is it dynamic or static? Could have multiple cards if static
well that the thing i need to now see if i can get a fast enough mcu to feed and buffer every request from all 3 devices.
acting as a memory controler
extreme MCU overclocking
well i was thinking of just a sam at 96 mhz
Are they just going to be reading? Or writing too?
Mostly reading, but yes save states and layouts might be written at times.
Mmmk, need to coordinate on one card then if they need to access the same items
well think about it.
it's a gaming device.
so what will happen most of the time is the device will enter a load state.
the gpu will get commands from the mcu
I wasn't sure what exactly was going on XD
the gpu asks the memory controler
hey give me the texture on index 7
it gives it.
but while this is happening the audio codec will be using most of the 256KB ram as a buffer.
i think 128 to 64k will be enough.
so i have a swamp of work making sure all 3 chips request data and i respond in time.
so i need a SD manual
XD
lol
sure there is something on google
Yeah, the SD protocol is available info
i really wanted to use a SD since i find it's fast enough and provides nearly unlimited space for project this small.
also i added a dirty Z-buffer option
XD
ps1 fans will enjoy it.
the world needs more info of fpga's there's few cheap options
and the power is quite amazing with some.
Who?
I'd imagine it would be rather difficult to teach a computer about abstract concepts.
Yes which is why the area of focus should not be on programming action, rather programming an ability to learn abstract learning
Not that I'm claiming to know how to do that but it should be the goal
That would be massively complex, if doable at all...
But then, I don't know much about AI/ML.
Guys, i have a question
how is it that EE is still probably one of the most coveted college faculty
despite it most of it being useless in the realm of real life implementation?
meanwhile you have IT
which is gradually rising to popularity
and highly coveted in the working space
Why do you think EE is useless?
well, inherently its not
but after job searching, most of the entry level jobs i get are all either technicians or mechanics
and then you'd have the high level jobs of being supervisors, but that's about it
no scientist staff, very few research jobs
meanwhile for IT you got all sorts of opportunities
Research / science is really what happens mostly at universities.
full stack dev, front end, back end dev, AI engineer, and stuff like that
yeah, true
but still, i thought the field for EE would be huge, not just being repairmen or factory technicians
Well, just like anything else, a lot of EE is actually software these days.
but then again, i learn electronics rather than power systems
Lots of companies need software for all kinds of things (pretty much every business needs a website or an app for example), while fewer companies build electronics.
But electronics companies may have 1000s of engineers working for them. Software companies are usually smaller.
i guess so
man, i just dont know if i could survive in maintenance
the reason why id want a more workshop job is because, well
i suck at analyzing problems, ahahahaha
i suck at measuring and calculating voltages or currents across components
That's pretty much an essential skill though.
But it's also something you can learn. I'm mostly a software developer and analyzing why stuff doesn't work is 99% of what I do.
true, but when you get to inductors, capacitors, and transistors, that's the part where i would probably fail without bringing up my notes
i've already forgotten the equations for those
too many situations for that
That's where experience comes in.
i guess
in maintenance tho, how much of circuit calculations do you have to do exactly?
or how much of transient analysis do you have to do?
I used to debug college kids' code when I was in high school, fix it, and leave snarky comments that made it obvious they didn't do their own work XD
If they didn't read through it before turning it in their professor would know, but if they did read before turning it in they could both see what I fixed (made comments explaining what I did) and be able to remove the snarky comments...
Fun times hehe
Still love a good debug puzzle
i bet you quite a few dolars the first robots will begin assisting us in homes.. and they will mostly study us.
and over time the will make our lives so much easier.
and you might even be left in the dark without a robotic assistant.
soon we might forget about thinking as machines interpret our lives and plan our day for us.
it's comming.
machines can be taught context.
google is the leader and have shown some demo's of emulating human conversations.
that are amazingly real and fascinating that they can give it a wiki page and it spits out a profile that it created and tries to emulate a character based on the data on the wiki.
We already have household robots.
alexa and siri are pretty linear
and i expect the first as in humanoid robots to be very linear.
but at some point well have machines on our level or faster.
just look at what boston dynamics has done.
Computers are already massively faster than us.
At certain things, sure.
For sheer processing speed.
ever human now has a gpu "a super computer"
The minute my household robot says it likes Nickelback, I am putting a bullet in its CPU.
what i'm more interested in what limits the first robots will have
My wife asked why I carry a gun in the kitchen..
I said "In case the Decepticons attack".
She laughed.
I laughed.
The toaster laughed...
I shot the toaster.
XD
Why a bullet? Just have another NN trained for positive mention of Nickelback that triggers an EMP inside of the robot and make the robot itself a Faraday cage.
what ur fav band?
I don't play favorites with bands, but if I were pinned against the wall by 6 orderlies and the only way to be released was to name a single band as my favorite right now it would be half alive
I just like music. If I enjoy a song, I listen to it again later
My best "get out of a funk" pair of songs right now are both half alive songs. One is "still feel" the other is "ok ok?".
One to make me feel alive, the other to make me feel okay.
Music is medicine
I'd have to go with Led Zeppelin.
Under the same condition, of being forced to pick. Lol
I can name a favorite band in each genre I listen to, but, to narrow it all down to one single band. Hmm... I'm going to have to go with Bad Religion.
i try listen to everything.
i have alot of stuff i like
i can spot trash a mile away.
It felt right. I do not know how to explain.
Explain jerking, someone please
The derivative of acceleration
I can’t picture it
What is that
Like. Things in place.
Imagine something not moving, and then it starts accelerating. That transition point is "jerking" it into motion.
If it starts moving smoothly, then the jerk is low. If it's abrupt, then the jerk is high.
A negative jerk would be either starting motion in the opposite direction, or starting to slow down. If you imagine a drag race, the car has a positive jerk when it starts off the line, is mostly zero while it's running the race, and then it goes negative at the finish line.
Ah
Truly is the rate of change in acceleration then
That’s difficult to intuitively picture for me, but thanks for this start
To be honest, it doesn't come up directly in physics all that often.
It comes up in linear robotics. If you have an object in motion on a rail, it's experiencing very little force from the motors to keep it moving at that speed (just enough to overcome friction) but if you're trying to speed up to some other speed, you want it to happen smoothly. If you do it too quickly the system "jerks" around.
And takes inordinate amounts of energy per unit time to do so, thus putting undue wear on the rails, bearings, and motors
But if you slowly ramp up how quickly you're increasing in acceleration over time, you have a smooth acceleration curve and an even smoother velocity curve
Erm rather you slowly ramp up the acceleration over time, as slowly increasing rate of acceleration increase is the next level up and is not covered in this chapter...
But really in any system that you have complex stretching / crumpling / warping it comes up at least a little...
Also imagine being the acceleree, (and assume your mass never changes). Newton's second law says your acceleration is proportional to the total force incident on you. So jerk is the rate of change of (total) force you feel. I'm guessing that's where the name comes from
Also if you're ever accelerating in a car, you'll have some degree your brain is automatically accounting for it as your rate of acceleration is not constant throughout the journey
Especially in a manual... Higher revolution speed of the engine in a given gear, higher acceleration within that gear
Which is why you get really good acceleration in low gear as the ratio allows higher revolutions at lower speeds.
So if you’re using something like wheel angles to find your displacement, could you use the 3rd derivative to start the motors to a value which would raise the jerk a bit and keep it there for a sec until the desired velocity is achieved
Intriguing implications
Honestly I have a hard time wrapping my head around that, because don’t you need a continuous function for this to work? It’s like a theoretical thing to me. In reality, wouldn’t you just keep variables for velocity deltas, and work with those to serve that purpose?
Well in linear robotics it's approximated because of stepper motors usually
Into a function resulting in discrete steps over time
To uh, illustrate @EdKeyes's point by exaggeration, if there was a name for the tenth derivative of position, would you want to develop intuition for what that thing means?
You'd get something from it, but it's also ok to not 🙂
Well yeah
If it's useful, which I argue jerk is in that one area at least because otherwise it wouldn't be used in every control software for 3D printers that I've seen...
They'd just control for acceleration and it would be hunky dory
But it has an actual implication for the strain on your machine so it is important
Now that's not to say it is applicable everywhere just that where it is applicable it is pretty important
And you can develop an intuition for it if you have a 3D printer you feel comfortable changing values on... Set jerk to some insanely low number and do a short test print, then bump it up keeping the same speed and acceleration values and keep doing that until it's where it was. You will pretty quickly get a feel for it.
And not an imprecise intuition either, a precisely quantifiable one.
As you know exactly the values you entered and don't have to guesstimate like you do trying to get a feel for it in a car...
You can even make some custom gcode that does the lowering of jerk and increasing it at each layer until the place the defaults were.
With a single wall cylyinder so as not to waste plastic
In fact... You could probably set up a custom gcode macro in cura using spiralised contour mode...
Now I kind of want to explore that but I have lithium charging circuitry to hopefully not explode
Is it really applicable though if you have to take samples granuarly?
My brain can’t apply it
Like I assume you can calculate jerk using the previous velocity readings in your sensors
Or even with just positional readings you’ve got it, don’t you
Your dt is a few milliseconds, and not very accurate, but accurate enough. I feel like I’m not understanding some aspect here
You calculate what steps are required over time to achieve a desired average
And then blindly apply that timing in hopes the system will respond correctly
(in low tech solutions)
And the root of those timings is a quartz oscillator which steps through the program often in combination with counters that trigger interrupts when the next pulse is required
So you're not usually measuring the acceleration at all, just calculating what you would like it to be over time, knowing how far each step takes you, and extrapolating the proper pulse timings to take those steps
That is you project a model and then make granular steps to best approximate that model
Here "pulse" means current spike?
What is so bad about just linearly going where you need to be
Linearly, like at a constant speed?
Yes, to a stepper driver which moves the motor a single step of known angular distance (which is why linear control is harder, you need sensory feedback to know how far you have moved, which again comes in granularly anyway and all computations are done granularly because of boolean logic as its base)
In a physical implementation you can't really do it linearly without an analog computer controlling it with all analog feedback. Have fun tuning that...
(for precise applications at least)
Doing things in discrete known steps and approximating a linear system with them is kind of the best way we have for precision systems these days...
Especially when you start to get into microstepping which is a whole other territory of approximately the linear model.
(basically combinations of field strengths between steps to allow smaller increments than whole steps of the motor)
But for that the theory is the same you've just given yourself finer granularity
As its usually a setting that affects the control chip and you just have to send some number of steps to it (usually some power of 2) to do the same amount of motion as a single step.
But then take everything I say with a grain of salt. I'm not "qualified" I'm just invested in the topic.
nice, thanks. Trying to figure out what concepts apply here. So to check my understanding:
- By "linear" you mean that the force exerted by the motor is linear in the current
- This simple control algorithm is approximately solving for the signal needed to bring about desired trajectory of motion
or did you mean "linear" in some other sense as @late fulcrum suggested
By linear I mean a smoothly changing function. We can't do that with digital computers so we approximate it. The simplest way to do that with no feedback is a stepper motor using known discrete angular distances between each step
And then to apply those discrete steps to match whatever curve you're approximating as closely as possible
So the ideal function is linear mathematically (a smooth input change smoothly changes output) however the resolution of the system doesn't permit this on a fundamental level so a discrete (no intermediary states on output) approximation is used instead
