|
01011001 posted:quick and dirty pcb for giant babbys based on my half-baked memories of my packaging class 2 years ago: -dont do your etching in an aluminum pan liner because ferric chloride+aluminum->exothermic reaction and the resulting heat makes the next bit of the reaction go faster and faster and pretty soon you've got a tub of boiling chemicals spitting crap all over and emitting a cloud of toxic fumes and you have to run the gently caress out of the garage and stand in the driveway waiting for things to clear out while wondering if the chlorine gas is gonna gently caress up your tools and/or car (spoiler: it did not)
|
# ¿ Mar 2, 2014 17:17 |
|
|
# ¿ Apr 24, 2024 07:11 |
|
is it a cmos chip? theres a weird thing with some cmos families where if you dont wire all the inputs to hard +v or ground they draw shittons of current and burn up. try wiring all the inputs to some sane stable state and see what happens
|
# ¿ Mar 28, 2014 03:20 |
|
Jonny 290 posted:is it wrong that i want one of the old rear end fluke bench dmms with the big LED displays some old dude dropped this off at our hackerspace's pile of interesting junk, i was lucky enough to be there and scooped it up instantly it didnt measure voltages anywhere near correctly anymore and some of the buttons didnt work but the nixie tubes still function a-ok. the tubes conveniently run at 170V dc which is the same as standard US line voltages and just draw a couple of microamps so it should be pretty easy to gin up a stupid hipster nixie clock or similar atrocity of technology gotta finish my cnc machine first tho before ADD-ing off to another stupid project
|
# ¿ Mar 31, 2014 01:27 |
|
also get yourself some kind of magnifier stand and a set of fine-tipped tweezers or you will go nuts for SOICs i usually pre-tin one of the corner pads then place the chip with the tweezers and push down while reflowing the solder to get a connection. flux helps a lot. then tack down the opposite corner to keep the chip from twisting while you do the rest of the pins when you're done use a q-tip soaked with acetone to clean up the flux and you're set
|
# ¿ Jul 19, 2014 18:10 |
|
movax posted:i saved connectors for last and suffered as a result, having to cram them in: the connectors J5 & J8 are real close and pointed at each other, the cable for one of them is gonna have to get bent weird unless they're both supposed to get routed up. flipping one or both of them around would fix that e: on second thought it depends on how your cables are polarized i guess, but still a thing to think through before sending the boards out PDP-1 fucked around with this message at 04:19 on Jul 23, 2014 |
# ¿ Jul 23, 2014 04:06 |
|
01011001 posted:labview is fine for simple purposes but i would probably go insane if i had to make complicated vi's regularly install the .net extensions for labview and then you can code in VS/C# like a civilized human for the main part of your application and just use their only slightly awkward interface stuff to access the hardware. wrap the labview interface up inside your own class with sane public methods/properties and hide their old school style callback methods behind some events/async stuff and you can wind up with a pretty nice code interface for doing hardware control you still gotta install their ~1GB omnidriver on every system your program will be used on and each system is supposed to be using at least one NI hardware item for license purposes, but it's still better than using labview G for anything more complicated than a 'push button, read voltmeter over GPIB' app
|
# ¿ Jul 27, 2014 23:28 |
|
Fanged Lawn Wormy posted:so i'm making this arduino resistor calculator for leds the E96 series of 1% resistors has 96 values per decade and those values could be stored in an int16 so you'd need 192 bytes of rom data which seems pretty reasonable even for small microcontrollers to handle. there's no reason to ever bother going to tighter tolerance resistors if you're just trying to power a led. heck, you could even go to the E24 series of 5% resistors since they'd be good enough for dumb current limiters and you'd save some rom/typing if you care about searching through that table fast (or just want to doodle around to learn stuff) you could use a binary search of the sorted value list where you'd start at the middle value, decide if it was greater/less/equal to what you wanted. if the middle value were greater than what you wanted you;d jump to the middle of the lower half of the list and check that value, etc. etc. until you found your desired value in log2(N) branches on average. the only remaining part is to figure out which decade you're in, something like decade=floor(log10(exactValue)) seems like it ought to work
|
# ¿ Mar 9, 2015 02:18 |
|
Olivil posted:- 256 kb+ of ram and 16 kb of rom iirc the 6502 can only address 64k directly so you might have an issue here some of the apple // series came with 128k but you could only use it in 64k chunks and had to switch between chunks by poking at a special asic that was memory mapped to the $c000 page. it had some special functions for copying blocks of data between lomem and himem and i think some special way of storing state to the stack of one memory bank when moving to the other and then restoring that state when you came back
|
# ¿ Mar 9, 2015 03:10 |
|
Sagebrush posted:often enough, but only for a few minutes. i keep my iron at about 325C (620F?), cooler than average, because at higher temperatures i think the flux burns off too fast. but it's too cold to solder anything with a good heat sinking effect like the ground leg of a TO-220, so i have to raise it temporarily to about 375 (700) for that stuff. i usually run at 700F for drat near everything, the ability to work quick on a wide range of stuff is great and i've never damaged any parts at that temp or had a problem with pads lifting off unless it was one of those old timey phenolic boards where the layout was obviously done by a photo process where someone drew the traces out by hand. modern components are pretty robust to both heat and static so don't bother overthinking and just solder that poo poo up in other idiot spare time project news i just got gifted this 1940's era Philco tube radio: i pulled the electronics rack and antenna out, they don't look terrible but def need some work. i've never worked on tube base circuits before but according to a book i got there's several distinct architectures of tube radio with only minor variations between manufacturers and i've pretty much narrowed down which generation this radio came from. it looks like the first step is gonna be to identify the power supply components, disconnect them from the rest of the circuit, and see if they withstand being plugged into the wall without tripping a breaker. then you start working forward from the RF amp/mixer stage and on toward the IF bit by bit until you can play glorious AM radio stations through the busted up grody speaker.
|
# ¿ Jan 23, 2016 06:12 |
|
solderchat: yes, dump flux on everything you want to solder particularly if it's too small to use regular rosin core solder (i.e. anything below about a 0.050" pitch). follow up the soldering with a cleanup step using q-tips and acetone to remove the excess flux tho. finally, using a magnifier of some kind (2x or 3x is plenty) is the most overlooked way to turn soldering small stuff from hard/impossible to easy/doable if you're working in the sub 0.1" pitch range. 1940's tube radio chat: i pulled the electronics chassis apart today and cleaned out most of the dust and dead insects. the mechanical parts are amazingly intact outside of a few bits of 70 year old string connecting the knobs to the display. squirted some contact cleaner on all the switches and lubed up the knob mechanisms and the physical parts seem to basically 'work' now. packed it up to take to work tomorrow so I can hopefully crank up the AC power through an autotransformer to see if the filament power supply and internal DC supplies still function worth a poo poo.
|
# ¿ Jan 25, 2016 05:39 |
|
i told one of our old grognard EE guys that i was working on restoring a 1940's tube radio and today he showed up with this thing that he's letting me keep for a week: it's a tube tester. i have no idea what the gently caress all those dials and knobs do yet, or what the Phasotron needle gauge thinger in the upper right is, but there's a little book attached to the cover that kinda-sorta explains it along with a bunch of warning labels that suggest doing things wrong will let the magic blue smoke out of either the tube being tested or the operator.
|
# ¿ Jan 28, 2016 04:49 |
|
Olivil posted:this is a very good tube tester it's a TV-7D/U, ebay shows them going for $650-$850 but it is built ridiculously well and does the job it sets out to do so i suppose it'd be worth it if you had a lot of tubes to test i downloaded the manual which contains a lot of military filler text and some helpful hints on how to properly destroy the device to keep it from falling into enemy hands. fortunately it also explains how to use the device which turned out to be pretty simple - insert tube, look tube up in attached booklet, set knobs to the positions indicated and push a series of buttons to test the tube's functions. if the needle gauge doesn't go above a listed minimum value for each test the tube is bad. turns out i have a few bad tubes but the old engie who lent me this tester said to let him know if i had any failures and he'd check his stocks to see if he had some before i order more. he's one of the last of the EE generation that grew up totally on analog stuff and he's cool as heck Silver Alicorn posted:
this is nifty, actually all of this old technology is nifty - it's like steampunk in that everything works via gears, pulleys, and springs while being unlike steampunk in that it was designed by normal people trying to make poo poo work within the constraints of the technology they had available instead of nerds gluing gears that do nothing to their hats. p.s. sorry to hear about your bike sagebrush
|
# ¿ Jan 29, 2016 05:25 |
|
Jonny 290 posted:i always start out ok in eagle for a couple traces and then lines start getting weird and i dont know how to undo and then it turns to garbage and this was a stupid idea a common mistake here is to try to use the delete key to remove traces - you can use delete for stuff that you draw directly onto the board layout like milling cuts or silkscreen text but traces have additional logic behind them that tie them to the schematic so you gotta use a thing called the rip-up tool to remove them. also this guy has a series of ok tutorials on eagle for anyone who wants to get started. if you're using the newest version of eagle the icons will look different but are still mapped 1:1 with what he shows https://www.youtube.com/watch?v=1AXwjZoyNno longview posted:they're the best option for hobbyists, especially since it's hard to mess up the ordering and there's no confusing options like what sort of minimum annular ring dimension your board has been routed for on my last order there was a gently caress-up where pins that weren't connected to anything didn't have any physical copper on the board despite the layout calling for an isolated copper pad it didn't end up mattering to me since the unconnected pins were just unused outputs or whatever, but the guy was still cool about it and shipped a second run of corrected boards for free a+ will osh park again
|
# ¿ Feb 7, 2016 15:59 |
|
Cocoa Crispies posted:yes i am obsessing over coffee these days check out the maxim MAX31855 series if you haven't already - you can get versions to match any common type of thermocouple and they handle all the compensation/curve-fitting bullshit internally for you and poop out the measured temperature data on a SPI port to like 0.25C resolution this lets you skip the step where you sperg about measuring how hot something is and dive right into the part where you sperg about your control loop algorithms
|
# ¿ Feb 22, 2016 05:25 |
|
Bloody posted:diptrace is much better than eagle what makes diptrace better? i haven't used it and since all specialty niche software sucks in some way or another once i browbeat eagle into suiting my needs i haven't bothered checking other packages out.
|
# ¿ Mar 25, 2016 05:34 |
|
Jonny 290 posted:the russians....used through hole! the russians also used to mark components with printed values on one side of the device - so if you had a thru-hole 1k resistor it'd be stamped with '1k' on the side and you'd have to be careful to install it with the label side up or lose the component value info. despite the many flaws of using color bands to encode part values the color band system was still better up to the point everything moved to surface mount
|
# ¿ Mar 25, 2016 05:39 |
|
longview posted:
that looks badass as gently caress, and good job so far on getting it working again and all, but what the heck do you plan to do with it when it's done? ham radio type thing? this makes me embarrassed that i haven't made more progress with my 1940's philco radio but as someone who grew up in wisconsin and now lives in florida this is the time of year when everyone from back home remembers i exist and all fly down to stay at my place like it's some kind of free hotel. 3+ weeks of near constant company really punches a hole in your spare time project's free time. tubes are all tested tho and only one bad one out of eight which isnt bad considering they're all older than my dad
|
# ¿ Mar 28, 2016 02:12 |
|
in other news - because i can never seem to finish one dumbass project before getting an idea for another - i've been kind of wanting to try out a sous vide setup for making yummy steaks but have been put off until now by the $200 price point for a specialty gimmick steak cooker. this morning i realized that i already own a crock pot which is really just a big tub that makes stuff inside it hot, so i ordered a SketchExpress brand PID controller, a 10A SSR, and a type K thermocouple off amazon for like $20. i'm gonna basically put the crock pot knob on the high setting and let the PID loop control the temperature. all the hardware should be here by tuesday, delicious steaks and/or food poisoning should be here by wednesday
|
# ¿ Mar 28, 2016 02:19 |
|
'controlled outlet' is a good description for the architecture i'm planning to use too - pretty much gonna cut the end off an old IEC cable and run that into the PID/SSR with an outlet box as the output and then nail all of that down to a chunk of scrap lumber. plug the whole thinger into a GFCI protected outlet and see if i like it enough to bother building a better version or just buying a commercial unit. good to know someone else got theirs working and was happy with the results tho
|
# ¿ Mar 28, 2016 02:51 |
|
not for this iteration, no. i just figured i'd use a vegetable steamer rack to stand the steaks off from the bottom a bit and maybe put it off to one side of the pot to let convection do the circulating for me. this is very much a quick knock-off project to see if i like the whole sous vide thing enough and would use it enough to bother with buying a semi-pricey kitchen dohickey that only does one thing. if the steaks are awesome enough and the process is painless enough that i'd want to do it more than once a month i'll just buy a commercial version. also gonna just use ziploc freezer bags during the initial trials instead of a proper vacuum sealer. same philosophy applies - do a quick hack and if the results look promising go ahead and invest in doing it right
|
# ¿ Mar 28, 2016 03:29 |
|
the parts for my podunk sous vide system came today and got screwed to a board for test. works great so far and locked right onto the set temperature with hardly any tuning needed: time to buy some meat and start cooking with science
|
# ¿ Mar 30, 2016 04:38 |
|
Glorgnole posted:here's what it looks like right now: this is awesome, you are awesome
|
# ¿ Mar 30, 2016 05:04 |
|
Star War Sex Parrot posted:are the function generators built into DSOs (like Rigol) usually significantly better or worse than an equivalent generator that could be purchased with the difference from buying a DSO without one? it seems to be about a $300 or so difference i use a tektronix scope with a built-in function generator at work and it's ok. it has one trick that regular function generators can't do without some futzing around which is that you can record a trace through one of the oscilloscope probes and play it back out the function generator port which seemed real useful at the time the salesguy showed it to us but i haven't had a reason to use in the year since we bought it. the advantage of a stand-alone is that you'd probably get two output channels for the money (the in-scope ones are one channel only) and that's actually a pretty useful thing to have. do you think that foam will crush or start to break cell walls and get waterlogged at depth?
|
# ¿ Apr 2, 2016 13:48 |
|
longview posted:would make a nice wristwatch a year or so ago i did a thing with some of those large 7-segment displays and the ones we used were 15-20mA draw @ 9v supply with about 4 individual LED's per segment*. i ended up driving them with a 5v microcontroller/logic system terminated in a line driver chip that could handle the (resistor limited) current, but i had to make use of the forward voltage drop of each segment being 4*1.3=5.2V to dump the 9v power supply on the display down to something the 5v logic could handle. it wasn't the cleanest, most elegantly engineered solution ever but it's been running for a year or so and hasn't fried out yet so by physics lab "just make a one-off that works and do it fast" standards it's been a total success. * a minor exception is the decimal point display which is just one LED, but for our purpose it that segment was fixed in place so i just used a resistor in series to limit the current and hard-wired it
|
# ¿ Apr 12, 2016 05:31 |
|
BedBuglet posted:Wait, no, a ferrofluid clock. this looks all kinds of amazing and i want one. unfortunately there's zero technical details on how to make it work - i'm assuming there are a bunch of c-shaped magnetic cores on the back with windings on one leg to produce a field that draws the ferrofluid in to the open side of the C to complete the flux path. also you'd need to have additional magnetic display 'segments' to pull fluid up from the bottom reservoir to the digits. oh man that would be an actually interesting thing to design and build
|
# ¿ Apr 13, 2016 04:16 |
|
a couple of years ago i made something like that - it was a pumpkin for our lab's halloween party that could answer spoken science and math problems by using voice rec to transcribe a question, then send that question to the wolfram alpha web api, and finally speak the plaintext version alpha came up with using text-to-speech. you could ask it what the square root of 2 was, or how many protons in a nitrogen atom, what the population of arizona was, etc. people seemed to have fun with it despite having a semi high miss rate with the voice recognition, and it got a lot worse once the room filled up and there were 37 different conversations going on in the background. having a fairly directional microphone that was aimed in a way where it wasn't picking up a ton of background noise helped that google api might work out a lot better than the freebee microsoft robotics stuff i was using, but if you can limit your range of recognized words down from 'the entire english language' to 30-40 keywords you'll get a much better hitrate (and also preempt that one guy who figures out what you're doing and just yells BOOBS and BUTTS all night long)
|
# ¿ May 13, 2016 23:48 |
|
BOOBS!
|
# ¿ May 13, 2016 23:49 |
|
big scary monsters posted:if i am an idiot who knows next to nothing about electronics and microcontrollers (and i am) but want to make a thing with leds that i can individually turn on and off and vary the brightness of, how is getting an arduino and neopixels from adafruit? are they terrible or anything, because it looks like a really easy way to start and it's cheap enough anyway that whatever premium i'm paying to be a ~~*maker*~~ doesn't seem that bad arduino + neopixel would be a fine way to make controllable lights, you'll also need an extra 5V power supply capable of supplying current to however many LEDs you plan to use. you can talk to the arduino with matlab or anything else that can control a COM port on your PC. this is a bit of an odd way to do things since modern computers don't have physical COM ports anymore, but the drivers that come with the arduino simulate having one and use a USB port to do the actual data transmission. it's just a convenient abstraction since COM ports basically give you a set of outgoing and incoming data streams to write/listen to which is exactly what you wanted to do anyway. just keep the code on the arduino side as simple and direct as possible and do as much of the calculation work on the PC side as you can. arduinos are cool and good but have the processing power of a mid-1980's home computer so you want to keep its workload light. the arduino thread is a pretty good place to read more or get specific questions answered
|
# ¿ Jul 16, 2016 05:16 |
|
if you get an actual clone knock-off they can be a cheap drop in replacement for a real arduino but i've seen ones with funky USB bridge chips or non-standard pin layouts that cause more headaches than the $5 savings is worth. that goes double for people who are just starting out and don't have the experience to handle unexpected weirdness. if it's your first time to teh arduino rodeo just buy an official branded card and leave the knock offs for later when you have working code and want to make 10+ of something
|
# ¿ Jul 16, 2016 05:40 |
|
also the knock offs come with that magical "it will be there in 6-27 weeks" low budget overseas shipping time
|
# ¿ Jul 16, 2016 05:42 |
|
echinopsis posted:my power bill last month was $451 and I still buy fire wood on top of that your new idiot spare time project should be getting one of those IR cameras that uses a phone for the output screen and one of those watt meters that tells you how much each appliance is using and finding out why you're burning off so much power. i live in loving florida and it's been the hottest year since forever and my combined electric/water/sewer/trash utility bill averages $120 USD/month for a full house. unless you're trying to climate control an uninsulated airplane hanger with a server farm inside your utility bill is nuts.
|
# ¿ Aug 3, 2016 03:45 |
|
Panty Saluter posted:we used 1174 kwh last month, most of which was a/c to keep 866 sq ft at 75F. granted we have computers running a lot of the time but this is also the month in which the a/c unit hosed up so it was "running" but the fan was not always blowing. i suspect a lot of that was pissed away on the malfunctioning a/c, since a typical bill even in the texas summer is 30% lower than that a couple of months ago I woke up at 3AM in a sweaty puddle and discovered that while my house was sitting at a balmy 88F the furnace was still kicking out heat at full blast. I shut it off and called up the furnace guys the next morning and they told me that "it sounds like your control board is busted, we can fix that for $500". that seemed insanely expensive so I opened up the furnace panel and there was exactly one thing that could reasonably be identified as a "control board" so I pulled it out and got the part number. the local appliance repair place sold me one for $27.50 and it's been working great ever since. anyhoo I guess the point is that most appliances are dirt simple and if you have the skills to feel comfortable with working on electrical stuff safely it'll save you buckets of money. also gently caress that repair place forever for trying to charge a ridiculous markup on what should have been a one hour repair visit.
|
# ¿ Aug 3, 2016 03:59 |
|
i got nothing useful to say about your actual questions, but drat whoever routed that pcb did an amazing job with that parts density. like just fitting all that crap on the available space much less actually routing the traces is impressive.
|
# ¿ Sep 18, 2016 04:00 |
|
Jonny 290 posted:4 oz of 99.9% ferric chloride came in today. I don't have drills yet but i can etch pro tip: don't try to do your etching in an aluminum pan like teenage me did. ferric chloride reacts exothermically with aluminum so at first you see small bubbles start to form, then larger bubbles and hey what smells like chlorine gas, then spontaneous boiling starts to happen and clouds of chlorine start to evolve and you just gotta gtfo until the reaction is over and everything has aired out for a while. e: https://www.youtube.com/watch?v=nV0yLE0YxUM&t=90s PDP-1 fucked around with this message at 01:38 on Sep 20, 2016 |
# ¿ Sep 20, 2016 01:34 |
|
meatpotato did a good writeup, i'd just add that usually the protection diode on the 7805 is done like this: the idea is that if the input is shut off and goes to 0V there can still be voltage on the output side which will reverse bias the 7805 in a way it wasn't designed to handle. the protection diode provides a safe path to discharge that output-side charge meatpotato posted:In general, digital inputs should be connected to something that sets the voltage high or low. Letting them float can cause oscillation (I think...) and that creates a whole bunch of issues including noise and high power consumption. yeah, digital inputs typically require only very tiny current flows to change state. if you have floating (aka not connected to anything) inputs strange poo poo can happen like a nearby signals inducing capacitively coupled currents on the floating node causing it to flip between 0 and 1 seemingly at random. some CMOS chips have pathology where unconnected inputs can float up to a voltage that is between 0 and 1 and both the upper and lower mosfets turn on at the same time causing the chip to suck current and heat up. just always tie your unused digital inputs to something and you'll be good
|
# ¿ Sep 25, 2016 23:44 |
|
i remember when they did that 'flat icon' look in visual studio back in the windows 8 days and it was annoying as heck. an IDE like that is one of the most complex UI environments there is with literally hundreds of functional elements on the screen, so changing all those icons at once plus making them monochrome really slowed me down. hell, just making things monochrome by itself was a terrible idea - color is an information channel y'all and picking out the blue thing from a field of not-blue things is always gonna work better than picking one grey thing out from a hundred other grey things thankfully much of that design philosophy went away along with the careers of the senior management people who greenlighted that shitpile
|
# ¿ Sep 26, 2016 01:16 |
|
WrenP-Complete posted:Well that's the idiocy I'm up to, currently trying to figure out if there is a quick and dirty way to combine answers to questions to yield a libertarian subscore. screw identifying libertarians, figure out how to get hot chicks to go on dates with you bill gates supposedly wrote a date matching program when he was in high school and hardwired it so he would always show up as the top match for certain girls he liked. now he's a zillionaire
|
# ¿ Sep 26, 2016 01:23 |
|
ahmeni posted:YOSPOS already tried this and it went uhh.. poorly link? or better, a short summary of the high/low lights?
|
# ¿ Sep 26, 2016 04:06 |
|
Sagebrush posted:like seriously look at this poo poo i love digikey but one time i ordered a set of soic-to-dip converter boards that came shipped in an anti-static plastic bag with another anti-static sealed bag inside and the inner bag contained both a moisture absorbent gel pack and one of those 'has this device ever seen significant humidity' cards. it shipped inside of a cardboard box with some of that expanded paper padding. c'mon folks, it's a bit of copper trace on an FR4 board there is nothing static or humidity or vibration sensitive about this thing and i know your profit margin must be thin as hell how do you even make money on shipping stuff like this. while i can appreciate that they can't rely on the average pick and pack worker to understand and make decisions on the best way to send one of their 250k stocked items it seems like overpacking that kind of stuff would be a huge profit loss issue and something they'd want to tighten up on. that said, i'd love to see a documentary video on how the hell they manage to stock that number of random three cent parts (plus a ton more actually expensive low-volume parts) and manage to store/ship that poo poo out while maintaining a profit.
|
# ¿ Oct 4, 2016 04:07 |
|
|
# ¿ Apr 24, 2024 07:11 |
|
while i'm on a rant about parts suppliers, i gotta say that for low-volume stuff search engine quality is often as not the factor that makes me choose one vendor over another. digikey is kind of my go-to supplier but maybe that's just because i'm used to their part search gizmo. mouser is better for overall search capability if only because they allow you to specify minimum and maximum range values while digikey has a hot mess of 0.99, 1.00, and 1.01 valued parts that I have to choose from. newark's search engine is the worst, at one time i searched for resistors and it told me they didn't have any. they've improved since then but it's still a bit of a fight to constantly exclude poo poo that will ship from england for an extra $20 fee. i work at a science lab where we typically are only building one or maybe three of something so i'm sure we're a weirdo outlier in terms of valuing search time over absolute cost but cripes guys it's 2016 and finding stuff in a database shouldn't be this complicated.
|
# ¿ Oct 4, 2016 04:38 |