|
Thanks. Done on schematic.
|
#
?
Apr 1, 2020 21:49
|
|
|
|
| # ? May 9, 2024 19:45 |
|
|
Sagebrush posted:What are you trying to drive with the relays? <2A DC I would probably go with a logic-level MOSFET. Screw them onto an aluminum plate and that'll be plenty of heat sinking and they're cheap and silent and switch instantly and last forever. One small tip on top of this advice: if your microcontroller is 3.3V, then those transistors won't carry much current when switched at that voltage. That's why I keep IRLB8721PbF - they will carry 10A when switched with 3.3V logic, more when switched with 5V logic.
|
|
#
?
Apr 2, 2020 20:01
|
|
|
i’ll take “mistakes even my dumb rear end was warned not to make” for $200, alex https://twitter.com/mountain_ghosts/status/1245754158910705668?s=20
|
|
#
?
Apr 2, 2020 20:48
|
|
|
Ambrose Burnside posted:i’ll take “mistakes even my dumb rear end was warned not to make” for $200, alex The A350 was the same... https://interestingengineering.com/airlines-advised-to-turn-old-a350s-off-and-on-again-every-149-hours-as-software-bug-fix The F-22 had a date overflow error caused by the International Date Line: http://content.time.com/time/nation/article/0,8599,1597043,00.html 
|
|
#
?
Apr 2, 2020 21:02
|
|
|
I've heard stories of floating point being used the same way. It's fine if you don't leave it on for too long.
|
|
#
?
Apr 2, 2020 21:20
|
|
|
Matt Parker's latest book has a lot of number overflowing incidents. His book is itself an overflow error.
|
|
#
?
Apr 2, 2020 21:21
|
|
|
Cojawfee posted:Matt Parker's latest book has a lot of number overflowing incidents. His book is itself an overflow error. Much love for Matt Parker. Link for the lazy: https://smile.amazon.com/Humble-Pi-When-Wrong-World/dp/0593084683?sa-no-redirect=1 Youtube channel: https://www.youtube.com/channel/UCSju5G2aFaWMqn-_0YBtq5A
|
|
#
?
Apr 2, 2020 22:50
|
|
|
What PCB/schematic software do you recommend besides Eagle? Just figured out the $100 package is out. Not trying to drop $500 to get a 4-layer PCB. Super frustrating; hope I can export my files. KiCad? Seems to accept eagle imports.
|
|
#
?
Apr 3, 2020 00:32
|
|
|
I like Upverter a lot, I've talked about it a fair amount in this thread. It seems to be dying, though. I've started dipping my toes into KiCad.
|
|
#
?
Apr 3, 2020 00:41
|
|
|
I don't suppose anyone knows how to add progressive unlock to a vehicle's power lock system that's Type C while still maintaining the existing door switches, would they? Progressive is where the driver's door unlocks on the first press and the others unlock on the second press. Type C is a reverse polarity system for the 2 door solenoids. I got an Avital 2101L. It has 2 SPDT relays inside along with a "second unlock" wire that is supposed to be used for a negative pulse to another, not included relay. I checked and the 2nd button press doesn't activate the first unlock relay a second time. I figured out how to do it without progressive unlock, but separating the existing wiring to make both doors lock and unlock on the existing door switches while having separate functionality on the remote press is making my head spin. Switching logic isn't my strongest suit. Here's the diagram for the existing locks. Mine doesn't have illumination. In my sketches I keep ending up with dead shorts to ground by backfeeding the existing switching. Can anyone help?
|
|
#
?
Apr 3, 2020 01:03
|
|
|
Is there some sort of handy reference website for deciphering discrete component ID codes where colour bands aren't used, ideally sth that also has a little bit of plain-english commentary about the practical role/utility for a particular component if it's noteworthy or unusual? Stuff like this (very handsome + well-packaged A++ would pay extra again) Elenco diode kit:  I've seen a million 1N4004s and their ilk, for example, on circuit schematics so I know the role they play, as well as the general purpose of zener/schottky diodes and the like, but JEDEC codes seem incredibly opaque and useless ("1 means diode, N means N, everything else means [consult ten million datasheets]") alongside, idk, EECA schemes, where you can actually determine most of the critical parameters just based on the code, similar to what you'd expect from resistors. or is there a reliable rule for knowing the practical difference between a 1N4001/4007, say, or other comparable components where the package is identical?
|
|
#
?
Apr 3, 2020 03:06
|
|
|
https://www.kb6nu.com/ever-wonder-why-transistors-have-2n-part-numbers/ Yeah good luck that Part numbering is the wild west out there
|
|
#
?
Apr 3, 2020 03:14
|
|
|
oh good, the north american standard is some hideous grandfathered-in disaster from the pre-digital age where the only characters that actually mean anything dont even make sense in 2020. the 1 prefix in diodes mean the vacuum tube in question has no filament or heater. absolutely. why not somehow not surprised at all tbh
|
|
#
?
Apr 3, 2020 03:26
|
|
|
Anyone know where I can search rocker switches by mounting hole size? Looking at the data sheets of hundreds of switches is getting time consuming and monotonous. Grainger has the best system but doesn't carry what I need. mount hole is 15/16" by 1-1/2" or roughly 24x38mm e: i'm willing to go with something that's not a perfect fit and glue it in place as long as the face covers the 24x38mm hole GnarlyCharlie4u fucked around with this message at 05:18 on Apr 3, 2020 |
|
#
?
Apr 3, 2020 05:14
|
|
|
Digikey has that, I think most of the big distributors should have hole size as a filter option
|
|
#
?
Apr 3, 2020 05:16
|
|
|
taqueso posted:Digikey has that, I think most of the big distributors should have hole size as a filter option Yup I'm dumb. They refer to it as "Panel Cutout Dimensions"
|
|
#
?
Apr 3, 2020 05:20
|
|
|
Okay dumb question #2: Can I use a single DPST switch for both positive and neutral legs of a 120vac circuit? Example: positive to one pole and neutral to the other pole of the same switch For reference this is going to be a power switch for a Class D amp with built in power supply. https://www.parts-express.com/icepower-50asx2se-class-d-audio-amplifier-with-power-supply-module-2-x-50w--326-212 I mean a DPST switch should have both poles isolated from one another so this should be fine right?
|
|
#
?
Apr 3, 2020 05:58
|
|
|
ante posted:I like Upverter a lot, I've talked about it a fair amount in this thread. The Eagle license thing's frustrating, esp since googling Eagle subscription shows a $100 std package, including links and ads on AutoDesk's site. The only way to get more than 2 layers now is $500/year for Fusion. The cloud-based save thing's a put-off. Getting the impression Autodesk doesn't have aficion for software design/utility; they're more about selling and locking people into a product. Dominoes fucked around with this message at 07:28 on Apr 3, 2020 |
|
#
?
Apr 3, 2020 06:34
|
|
|
GnarlyCharlie4u posted:Okay dumb question #2: Don't assume that the poles are isolated, and then be careful that they have the right creepage distances. You're a little bit off-label in usage there, so it's possible that that is no bueno.
|
|
#
?
Apr 3, 2020 06:44
|
|
|
Eagle is dead since Autodesk took over. Either go Kicad or """obtain""" Altium
|
|
#
?
Apr 3, 2020 11:56
|
|
|
ante posted:Don't assume that the poles are isolated, and then be careful that they have the right creepage distances. You're a little bit off-label in usage there, so it's possible that that is no bueno. Yeah gently caress it, it's not worth it. I'd just as soon only use one pole of the switch than risk it. Also: while these aren't the switches I was after I just want to let everyone know that if they are in the mood for a fat chunky clunky rocker, they can consider this; https://switches-sensors.zf.com/us/product/rocker-yr-series/ ZF bought out Cherry brand in '08 and then divested 5 years ago but kept manufacturing the switches under their own name. really satisfying click. GnarlyCharlie4u fucked around with this message at 14:56 on Apr 3, 2020 |
|
#
?
Apr 3, 2020 14:52
|
|
|
GnarlyCharlie4u posted:Yeah gently caress it, it's not worth it. I'd just as soon only use one pole of the switch than risk it. https://www.homedepot.com/p/Leviton-20-Amp-Industrial-Grade-Heavy-Duty-Double-Pole-Toggle-Switch-White-382-01222-02W/301447177 It's even in a standard form factor!
|
|
#
?
Apr 3, 2020 14:55
|
|
|
Ambrose Burnside posted:I've seen a million 1N4004s and their ilk, for example, on circuit schematics so I know the role they play, as well as the general purpose of zener/schottky diodes and the like, but JEDEC codes seem incredibly opaque and useless ("1 means diode, N means N, everything else means [consult ten million datasheets]") alongside, idk, EECA schemes, where you can actually determine most of the critical parameters just based on the code, similar to what you'd expect from resistors. or is there a reliable rule for knowing the practical difference between a 1N4001/4007, say, or other comparable components where the package is identical? Try figuring out why various MOSFET's are coded the way they are. I mean, BC548, that's a good name for a transistor, I can remember that, FQP30N06L (neato ones), what even the gently caress is that.
|
|
#
?
Apr 3, 2020 14:58
|
|
|
I'm still alive and so is the garden timer with the string light solar cell attached. Given that the two AAA batteries I used were low capacity Chinese NiMH batteries meant for garden lights, I believe the solar cell is working. At least to some extent.
|
|
#
?
Apr 4, 2020 00:37
|
|
|
Would anyone mind skimming this PCB pic? Device purpose: Connect hydroponics sensors and actuators to a Raspberry Pi Compute Module. Intent: Get this manufactured to JLCPCB, to demonstrate I can design and order PCBs. Ie, not too concerned about optimization, but that the basics work. (I won't be upset if HDMI doesn't work because the EM field is misbehaving). 4 layers. Top is Red, Bottom is Green. Top and bottom have traces. Middle layers are solid ground to accommodate HDMI. Top is filled with ground (Not evident in pic). The misc vias are to spread the ground around through the layers. My understanding of how this works is Chinese Room level. Looking for things that will cause this to fail or be dangerous, or bad habits. Much appreciated. Let me know if a schematic or closeups would help.  Higher-res Trace thicknesses range from .28mm to 1mm. Ports on top are a a mix of analog/digital sensors that work in diff ways, with Pu-resistors. Ports on the bottom-center are a breakout to a relay board. Bottom left is a temp usb port for wifi dongle. Bottom right is μHDMI and accessories for it. The capacitors near the Dimm-pins are bypass. Missing trace near the bottom is fixed. Top left is power input at 5v, and regulation to 3.3v and 1.8v. Dominoes fucked around with this message at 02:18 on Apr 4, 2020 |
|
#
?
Apr 4, 2020 02:06
|
|
|
Yeah, there's some weirdness going on there. The first thing is to do some serious detangling. For example, take a look at the bottom left there. You could do that whole section without a single via. From looking at it, you might not understand that through-holes are like a copper tube that goes the board. That header 2 pin going to ground is the same situation - You don't need the via, it will just be connected naturally to the pour, with the through-hole. I see some unconnected ratsnests. I'm also not sure about that DDR2 footprint. Is it on the top? Will it interfere with the other components? Usually a good 4-layer stackup is signal-gnd-vcc-signal. You don't get any benefit from a double ground. Are those some 0402 or smaller components I see? Maybe stick to 0603 if you can get away with. Keep your traces consistent sizes. I usually do maybe 8mils for data if I can get away with it, 12 mil for power.
|
|
#
?
Apr 4, 2020 02:37
|
|
|
Thank you very much! - USB (btm-left) rerouted without Vias, using the hole-goes-through technique you mentioned - Removed all traces from ground holes - Replaced 0402 capacitors with 0805 to match the resistors. - Probably going to cut down the fat 1mm 5v traces to .5mm to match the 3.3v ones I'm not really sure on the DDR2, but if it's like the Pi IO board I have (The footprint looks like it), the SO-DIMM module sits horizontal, ie flat over the board, with some give. It'll get in the way of tall components, but might be ok with SMT resistors/capacitors? Or maybe it's a bad idea and will short something out :/. The IO board has a few (resistors? caps?) in that space, but it's empty other than that and traces. I need to look more into the 4-layer thing. I had a reference design I looked at where there were 2 ground layers in the middle; I don't understand what I need to do to make the HDMI work, but haven't fully explored the resources. Tangent: What's the possibility of any given board you handle using led solder? Want to know if I should be regularly washing my hands when handling boards. Getting worried since some internet searches imply it's widely believed that lead in solder's OK, and leadless solder sucks. Calling bullshit. I'm new and don't have an issue with the Tn-Ag stuff, and pb is not worth the risk. It's a great material for many things, but there's a reason it has its stigma. I'm sure the plumbers, paint makers, and gas manufactures said similar things about risks being low, and unleaded varieties being inferior. Dominoes fucked around with this message at 03:39 on Apr 4, 2020 |
|
#
?
Apr 4, 2020 03:32
|
|
|
Dominoes posted:Thank you very much! The chances of any of the high speed lines working on your board are slim to none, including the USB. I only hope that it's USB 1.0 speeds, as that's a very resilient protocol and can handle a lot of interference, so you can get away with a lot of crimes. None of your data pairs are run together, I should be able to spot your data busses at a glance if they're routed correctly. You need to look into the extremely specific rules for routing USB and HDMI data. It will dictate the distance the traces should be apart, how far apart the ground plane around them is, and how wide they need to be (this depends on the thickness of the copper layers on your PCB). It'll dictate maximum lengths too. You must avoid cutting the ground planes under these data busses. You must keep them far away from any other sources of noise, and keep them far away from your analogue signal inputs as they'll in turn put noise on those signals. Here's a picture of the pi, and you can see what I'm getting at:  1. These squiggles are a result of strict rules about trace length matching. For high speed signals, this can be for a number of reasons, but for many it's because the signals are so fast, if they're not the same length the data will be jumbled as signals won't arrive at the correct times. How close the lengths need to match is dictated by the data used, it should be in the standards for that signal. 2. Again you can see it here, for specifically HDMI data lines. you can also see that each trace is actually a pair of tightly together tracks. These are the data pairs I'm talking about, and they have to be at a given distance together, and another given distance away from ground, and another given distance from other data pairs. 3. Here you can see the same thing going on for USB. It's not as strict as HDMI in many cases, though USB 3.0 is about as strict. I don't have the data on hand for precisely what rules you need to follow for HDMI, but the information is available. If you ignore these rules your data will get mangled and your board will not work. As for leaded solder, if the board isn't marked with RoHs compliance markings don't take your chances and wash your hands. Pretty much anything sold commercially will be lead free, but older stuff, or stuff used in automotive, medical or aerospace is exempt. This is for good reason! While it's true that anyone semi-competent can solder with lead free solder without too many headaches, lead-free solder develops tin whiskers that can cause failures due to random shorts. It's also brittle, and thus prone to failures due to vibration. The quantities of lead involved mean it's a problem for factory workers, and it makes e-waste more toxic. It's not going to have any real impact on you as a hobbyist, but feel free to avoid it. But the problems with it are real, and those exemptions for automotive, aerospace and medical industries should give you food for thought. Don't take all this stuff so lightly. This is electronic engineering. It's a profession, not a hobby. Splode fucked around with this message at 04:31 on Apr 4, 2020 |
|
#
?
Apr 4, 2020 04:29
|
|
|
I'm using the USB so I can SSH into it while I figure out which WIFI chip to use and how to set it up; they're USB 2.0 ports. Thank you for the detailed info on signal trace routing. Those highlighted pictures are a good indication I'm missing large chunks of the picture; going to pause, do some reading, and redo those. I didn't even realize it was a factor for USB, and made no attempt. I'm hoping the short HDMI length will mitigate my low skill. I appreciate those dets on the lead. Just took an inventory of the boards around me; mix of seeing the RoHS label and not. Lead really is such a nice material... So many applications, very cheap etc. I agree that I'm probably very low risk of lead exposure due to much lower time around it than professionals. Using lead-free only myself. We all have risk/reward scales at various levels, and this one seems like in most cases, not worth the risk. Going to hand-wash every time, and not buy the leaded solder. I'll post an update after sorting out the signal traces. edit: The USB signal paths were truly terrible. Dominoes fucked around with this message at 06:46 on Apr 4, 2020 |
|
#
?
Apr 4, 2020 06:04
|
|
|
Excellent post, Splode. I typically use leaded solder for all of my personal poo poo, and hand-built prototype professional poo poo. It's easier to use, and has lower temperatures. I wash my hands every working with the stuff, every time. RoHS isn't about health while handling stuff, it's about keeping lead out of the waste chain, which totally matters when you're building 100-100,000,000 of something. When you're building one or two, ehhh, drop in the bucket. For USB 2.0, I've never had anything fail. However, I always do keeps my leads really really short, and straight from A to B, no weird jogging - basically the width of the microcontroller or USB bridge or whatever IC I'm using. Less than 2cm or so, and I don't expect any problems. I've never done any HDMI stuff, but I do hear it's nightmarish.
|
|
#
?
Apr 4, 2020 06:24
|
|
|
I love lead, covered my fuel injector workbench with it at my last job. I hope the guy who relieved me appreciates it, he's probably stuck there forever with the virus bs lol. Always had to wear ppe anyway because of the hydraulic oil we used to test the things, itchy poo poo
|
|
#
?
Apr 4, 2020 06:28
|
|
|
I'm glad that was helpful and not discouraging! short tracks will let you get away with murder most of the time, depends on the speed. I've done some terrible crimes with USB 2.0 myself and gotten away with it (only just barely though, often if you sniff the USB traffic you'll see it's losing a lot of packets: the digital side of USB is very, very resilient) When you get up to the WiFi stuff, if you're putting a module or IC on your board, post here again. RF stuff is it's own can of worms with it's own set of problems. I'm no expert but I can at least help you avoid the common mistakes (We had a client who tried to get us to put a GPS antenna in the middle of a ground plane, in a metal enclosure.)
|
|
#
?
Apr 4, 2020 06:37
|
|
|
I should probably move the USB to the other side of the board, near the chip connections. Failing that, a careful, more direct routing. Appreciate the wisdom again; will post back once I figure out what chip to buy; trying to find one that will have minimal driver issues.
|
|
#
?
Apr 4, 2020 06:49
|
|
|
Also pro tip from my own fuckups: add some mounting screw holes so you can actually attach the board to a thing in a way that's not double-stick tape.
|
|
#
?
Apr 4, 2020 07:21
|
|
|
guess who hosed around and made a magnesium-copper pulvermacher chain battery
|
|
#
?
Apr 4, 2020 08:04
|
|
|
Shame Boy posted:Also pro tip from my own fuckups: add some mounting screw holes so you can actually attach the board to a thing in a way that's not double-stick tape. And leave enough room around the holes for the screw heads (d'oh!)
|
|
#
?
Apr 4, 2020 09:20
|
|
|
Splode posted:And leave enough room around the holes for the screw heads (d'oh!) Also make the hole larger than the screw nominal diameter, for M3 at least 0.2mm larger, maybe 0.5mm if you're hand-drilling the mounting holes later.
|
|
#
?
Apr 4, 2020 09:55
|
|
|
Dominoes posted:. In tyool 2020 you should be regularly washing your hands regardless.  Lead solder in consumer products is banned in enough large markets that the likelihood should be low. But if you're worried about it, it doesn't hurt to assume it's there (either because enforcement is difficult or because the product has carved out an exception) and act accordingly. Also the absolute quantities are so much lower that a favorable comparison to those industries isn't unfair, but I'm not going to tell you to hand wave your health.
|
|
#
?
Apr 4, 2020 15:34
|
|
|
Dominoes posted:Tangent: What's the possibility of any given board you handle using led solder? Want to know if I should be regularly washing my hands when handling boards. Getting worried since some internet searches imply it's widely believed that lead in solder's OK, and leadless solder sucks. Calling bullshit. I'm new and don't have an issue with the Tn-Ag stuff, and pb is not worth the risk. It's a great material for many things, but there's a reason it has its stigma. I'm sure the plumbers, paint makers, and gas manufactures said similar things about risks being low, and unleaded varieties being inferior. Anecdotal for sure, but I have seen with my eyes the differences between eutectic Pb-Sn and Lead-free solder. I work in aviation, and I have experienced equipment failures from it. I open a box up, and the 1978 circuit board has a 2016 repair where all the new stuff has tin whiskers and is cracked off the board. I fix the new repair with leaded solder and return it to service (then wash my hands). If your equipment isn't routinely cycling from 20 to -40 to 90C and from 15->2psi->15 every few hours while experiencing constant high-frequency and frequent low-frequency high-G vibration for years on end, lead-free is probably going to work fine for you.
|
|
#
?
Apr 4, 2020 16:05
|
|
|
|
| # ? May 9, 2024 19:45 |
|
|
Touch solder, shake hands, touch face. Got it. Still working through the diff pair routing dets, but have a start. Switched from μHDMI to normal, since I don't think soldering the back row of pins will be feasible using an iron (They're under the unit). Moved the pins even closer to terminals. Showing track length of 5.2 ±1mm. 0.2mm traces and intra-pair gaps. Routed the USB on one of the middle layers to get it straight and mostly the same len. HDMI:  USB:  babyeatingpsychopath posted:Anecdotal for sure, but I have seen with my eyes the differences between eutectic Pb-Sn and Lead-free solder. I work in aviation, and I have experienced equipment failures from it. I open a box up, and the 1978 circuit board has a 2016 repair where all the new stuff has tin whiskers and is cracked off the board. I fix the new repair with leaded solder and return it to service (then wash my hands). Dominoes fucked around with this message at 16:25 on Apr 4, 2020 |
|
#
?
Apr 4, 2020 16:17
|
|