|
Hey guys. I'm an idiot who puts to much effort into dumb things, and lately, I've been curious about building a custom wireless charging pad. I don't really know anything about electronics at all, other than a course or two I took in high school 15 years ago. I essentially want to re-create the wireless charging pads that cell phones use these days. Here's the question: if I built one sort of large - let's say a diameter of 16" - would the entire surface area within the coil provide an electromagnetic field? And if so, would it be able to supply power to ~40 small receiver coils? (Each ~1" in diameter) Each receiver coil is only attempting to power a small LED akin to these guys https://www.superbrightleds.com/cat/through-hole/ 
|
#
?
Feb 8, 2019 15:42
|
|
|
|
| # ? May 29, 2024 06:41 |
|
|
The Wonder Weapon posted:Hey guys. I'm an idiot who puts to much effort into dumb things, and lately, I've been curious about building a custom wireless charging pad. I don't really know anything about electronics at all, other than a course or two I took in high school 15 years ago. So a wireless charging pad is effectively acting like a transformer with air as its core material. To be at all efficient, the coils need to be resonantly matched with each other, which usually means they have to be around the same size and shape, close together and relatively parallel to each other. I don't think you'd get much out of your single large coil trying to match to a bunch of tiny coils...
|
|
#
?
Feb 8, 2019 15:50
|
|
|
Shame Boy posted:So a wireless charging pad is effectively acting like a transformer with air as its core material. To be at all efficient, the coils need to be resonantly matched with each other, which usually means they have to be around the same size and shape, close together and relatively parallel to each other. I don't think you'd get much out of your single large coil trying to match to a bunch of tiny coils... Dang, ok. I suppose there's the possibility of using a series of small inducer coils on the bottom half, but at that rate it seems like other solutions would be easier. Thanks!
|
|
#
?
Feb 8, 2019 16:29
|
|
|
The Wonder Weapon posted:Dang, ok. I suppose there's the possibility of using a series of small inducer coils on the bottom half, but at that rate it seems like other solutions would be easier. I mean you could get your big coil working with the tiny coils, it would just be horribly inefficient so you'd basically need to run a whole lot of power through it to get the things to light up at all, kinda like how you can light a fluorescent light with the energy radiated by a tesla coil  Other solutions would probably be easier though, yeah
|
|
#
?
Feb 8, 2019 16:32
|
|
|
Make your mini coils and toss them on an induction cook top to test.
|
|
#
?
Feb 8, 2019 23:55
|
|
|
CopperHound posted:-I just started trying to use KiCad. It is quite impressive for free software. I think I made a layout that will work, but I don't know Contrasting opinion: For a 2 layer board, your goal is to have the entire back side be one big ground plane. If you have to drop a signal onto it, make it as small an interruption as possible (via down, cross under the top layer trace, via up). The top layer will be a pour for your VCC plus signal routing. High frequency current wants to flow back directly underneath the trace it came out on, try to make that path exist on the ground plane and also don't send it across any sensitive analog bits. Beyond that: - if a trace can be big, make it big so the PCB works even if the tolerances the board house claimed are lies - vias make good test points or points for cut-and-jumper mods. Bring NC pins out to vias so you can get at them unless there's a compelling reason not to. - thermal relief pins unless they need lots of current (kicad should default to doing this) - take your Gerber's at the end and sanity check them in a 3rd party viewer to make sure they look right - print out your copper or soldermask layer on paper at actual size and put your parts on it. Did you screw up a footprint? Can you solder it? Do the non-footprint aspects of parts bump into each other?
|
|
#
?
Feb 9, 2019 07:32
|
|
|
The Wonder Weapon posted:Dang, ok. I suppose there's the possibility of using a series of small inducer coils on the bottom half, but at that rate it seems like other solutions would be easier. If you're only trying to light up a handful of dinky 20ma LEDs, "gently caress efficiency" is a workable approach
|
|
#
?
Feb 9, 2019 07:36
|
|
|
Foxfire_ posted:If you're only trying to light up a handful of dinky 20ma LEDs, "gently caress efficiency" is a workable approach I mean the efficiency is so incredibly bad you'd need to drive the primary coil with quite a bit of power to get them to even glow, it's definitely doable but idk if it's a great choice for a first project. Source: my sense of disappointment when I tried basically this same thing years ago and it didn't work at all
|
|
#
?
Feb 9, 2019 08:13
|
|
|
Foxfire_ posted:Contrasting opinion: Nit: signals want to follow the path of least impedance which often but not always means they want to reduce current loop area and will attempt to go underneath the source trace but if you have weird impedances in the way it'll go around them and might result in crosstalk
|
|
#
?
Feb 9, 2019 21:38
|
|
|
With my 2-layer boards I tend to put a ground pour on both the top and bottom, lay out power first so I can keep it short and direct, then signals, then via stitch the top and bottom ground planes obsessively around every trace so current can take as direct a path as possible, is this dumb? I have one of those infrared board heaters so I can heat up the whole thing and make it actually solderable so that's not really a problem. I think I might just like to have an excuse to add a bunch of vias because it's weirdly therapeutic...
|
|
#
?
Feb 9, 2019 23:23
|
|
|
Shame Boy posted:With my 2-layer boards I tend to put a ground pour on both the top and bottom, lay out power first so I can keep it short and direct, then signals, then via stitch the top and bottom ground planes obsessively around every trace so current can take as direct a path as possible, is this dumb? I have one of those infrared board heaters so I can heat up the whole thing and make it actually solderable so that's not really a problem. I think I might just like to have an excuse to add a bunch of vias because it's weirdly therapeutic... The point of via stitching is to reduce impedance so I mean it's difficult to see downsides except due to the lower thermal impedance it'll be harder to solder. Also might wear out pcb drills faster/ slow manufacture. I just have the cad program do it automatically and also via fence high speed traces which is mostly to avoid making unintentional antennas. Malcolm XML fucked around with this message at 00:17 on Feb 10, 2019 |
|
#
?
Feb 10, 2019 00:13
|
|
|
Malcolm XML posted:The point of via stitching is to reduce impedance so I mean it's difficult to see downsides except due to the lower thermal impedance it'll be harder to solder. Also might wear out pcb drills faster/ slow manufacture. Yeah this is about what I figured but I realized I was basically just cargo-culting things I'd seen on other boards so I wasn't sure I bet there's some kicad script or plugin that could auto-stitch for me that I might want to look into, hmm...
|
|
#
?
Feb 10, 2019 00:43
|
|
|
Stabby McDamage posted:OSHpark is good, but if you want to compare, I always use https://pcbshopper.com/ to price compare. You put in board stats and they give you a bunch of vendor prices and ship times.  This is my first time trying a build with smd I'm getting some extras just in case I sneeze and loose a bunch of 603 sized bits. Arg! I just spotted some more routing that is going to drive me crazy, but too late now:  and I wrote SPI instead of ISP... e: for reference, that board is about the size of two AA batteries. CopperHound fucked around with this message at 07:51 on Feb 10, 2019 |
|
#
?
Feb 10, 2019 07:47
|
|
|
Don't worry too much, that looks a shitload better than my first ten PCBs
|
|
#
?
Feb 10, 2019 08:19
|
|
|
Yeah man, you good. Be proud.
|
|
#
?
Feb 10, 2019 11:21
|
|
|
Shame Boy posted:Yeah this is about what I figured but I realized I was basically just cargo-culting things I'd seen on other boards so I wasn't sure I just did this by copy and pasting vias but kicad 5 now has this apparently https://www.youtube.com/watch?v=wwM_SEhOxls
|
|
#
?
Feb 11, 2019 04:34
|
|
|
Malcolm XML posted:I just did this by copy and pasting vias but kicad 5 now has this apparently Yeah I know about being able to place vias directly now and I've used it a ton already, it's great
|
|
#
?
Feb 11, 2019 07:03
|
|
|
Meanwhile while I'm waiting for that 4-layer board to ship I've started a new design for a GPS breakout for the uBlox NEO-M8T GPS timing module, since it has a lot of neat features that are available on pins which apparently no goddamn breakout board actually breaks out. This is the first time I've ever done a board that involves RF stuff, and boy does the hardware design manual have some scary paragraphs in it:quote:Any I/O signal line with a length greater than approximately 3 mm can act as an antenna and may pick up arbitrary RF signals transferring them as noise into the GNSS receiver. This specifically applies to unshielded lines, in which the corresponding GND layer is remote or missing entirely, and lines close to the edges of the printed circuit board. This'll be... fun... I'm guessing the general rule to this sort of thing is just "follow the datasheet / app notes exactly", right? Shame Boy fucked around with this message at 16:40 on Feb 11, 2019 |
|
#
?
Feb 11, 2019 16:37
|
|
|
Yep. What footprint is the module? All the gps modules I've seen have been BGA or similar, which as far as I'm concerned are only theoretically possible for a human to solder Edit: BGA not VGA lmao Splode fucked around with this message at 22:31 on Feb 11, 2019 |
|
#
?
Feb 11, 2019 21:20
|
|
|
Splode posted:Yep. What's VGA? I assume it's something-Grid Array, but google just turns up video connectors... Anyway the one I'm messing around with is just those cut-in-half vias, and the pin pitch is 1mm which is well within my ability to hot air gun reflow solder: 
|
|
#
?
Feb 11, 2019 22:25
|
|
|
BGA, not VGA, my bad. Notable for having pins under the IC, so you can't check your connections without an x ray machine
|
|
#
?
Feb 11, 2019 22:31
|
|
|
Check your connections by turning it on 
|
|
#
?
Feb 12, 2019 01:23
|
|
|
Turn it on before you put it on the hotplate When it starts working, you know it's fully soldered
|
|
#
?
Feb 12, 2019 01:32
|
|
|
ante posted:Turn it on before you put it on the hotplate So stupid it might actually.
|
|
#
?
Feb 12, 2019 08:57
|
|
|
What's the deal with 3-terminal feedthrough capacitors? The GPS module's app notes uses them a bunch for decoupling the power supply pins and this is the first time I've ever seen them. Why would you want to do that rather than just a normal capacitor?
|
|
#
?
Feb 12, 2019 18:55
|
|
|
I imagine the idea is have the capacitor as close as possible to the path of the current. In practice though, I can't comprehend the difference of a 3 terminal capacitor vs running the power traces in and out the solder pad for an smd cap. E: https://www.murata.com/en-eu/products/emiconfun/emc/2011/09/28/en-20110928-p1 Looks like the chip versions actually feed power through the capacitor structure, so I guess you don't get the parasitics of a branched cap. E2:... And trying to understand inductance in high frequency application is reminding me why I dropped out of an EE major. CopperHound fucked around with this message at 20:33 on Feb 12, 2019 |
|
#
?
Feb 12, 2019 20:25
|
|
|
Classic feedthrough capacitors are designed to remove RF from DC lines; the classic type is screw mount and used to get DC signals in and out of shielded boxes. This is a different application from power delivery, the goal isn't necessarily to get an ultra low ESR capacitor to handle transient current, but to reduce the inductance to ground through the capacitor to a minimum (often at the expense/benefit of adding some inductance in series with the signal). Many SMD types include some ferrite material over the lead-structure, forming a T-filter as well as just being capacitive - usually the larger and more expensive types. The classic Murata 3-terminal feedthrough type is not a terrible bypass capacitor, but I'd guess they use them to filter as much RF from the supply (both coming in and going out) as possible rather than for their amazing energy storage.
|
|
#
?
Feb 12, 2019 20:50
|
|
|
CopperHound posted:I imagine the idea is have the capacitor as close as possible to the path of the current. In practice though, I can't comprehend the difference of a 3 terminal capacitor vs running the power traces in and out the solder pad for an smd cap. Yeah there was a blurb in the application note that they had "low ESL" but I'm not sure if that's just the particular part they're recommending or some feature of all feed-through capacitors CopperHound posted:E: https://www.murata.com/en-eu/products/emiconfun/emc/2011/09/28/en-20110928-p1 I figured based on the schematic symbol that it was some kind of tubular capacitor with one conductor running through the center, sounds like that's about right
|
|
#
?
Feb 12, 2019 20:51
|
|
|
Put a new omron switch into my mouse today! Easy as it gets, 9 screws and a couple solder points, works perfect.  That's 90 bucks worth of electronics I repaired instead of tossing in 1 week.
|
|
#
?
Feb 13, 2019 01:26
|
|
|
Salt Fish posted:Put a new omron switch into my mouse today! Easy as it gets, 9 screws and a couple solder points, works perfect. Yeah I recently replaced all the buttons in my old-rear end Logitech mouse that dates back to 2000 and it's good as new again, I love being able to do that
|
|
#
?
Feb 13, 2019 05:08
|
|
|
Shame Boy posted:What's the deal with 3-terminal feedthrough capacitors? The GPS module's app notes uses them a bunch for decoupling the power supply pins and this is the first time I've ever seen them. Why would you want to do that rather than just a normal capacitor? I’m saying a lot of the same things as longview did in my post below, but in a slightly different way. If you calculate the transmission of an AC signal from one side to another of an ideal shunt capacitor, it has perfect rejection at sufficiently high frequencies. The capacitor effectively shorts all of the AC signals to ground, which is one goal of power supply bypassing. If you calculate this transmission for a shunt capacitor with a non-zero ESL, it only has the great rejection at the capacitor self resonant frequency. Past the self resonant frequency, the capacitor acts more like a shunt inductor, which eventually looks like an open circuit to ground at sufficiently high frequency, and so does a poor job of rejecting ac signals. I think the major feature of the three terminal capacitor is that they tend to have lower ESL than similarly valued and sized two terminal caps. They are a more ideal capacitor, and so they are more effective at bypassing high frequency interference/noise on power supply lines. This might be important when designing sensitive radio receivers.
|
|
#
?
Feb 13, 2019 12:04
|
|
|
Thanks for the info about feed-through capacitors, it was very interesting. I have another more ~controversial~ question: This GPS module has a USB interface for reconfiguring it, and I want to add a micro USB port to my design to facilitate this, as well as allow everything to be powered by a USB charger or battery pack. The last time I used a USB port, the datasheet of the device (a microcontroller I forgot the specifics of) said you should connect the shield ground to the board ground via a 4.7nF capacitor in parallel with a 1M resistor, because otherwise the USB cable "becomes a giant antenna". However, as far as I can tell the datasheet for the GPS module doesn't actually say what I should do with the USB connector's shield ground. I checked StackExchange and the answers are basically a slap fight: https://electronics.stackexchange.com/questions/4515/how-to-connect-usb-connector-shield Seems like several people are just pointing out that many datasheets say you should do it, while a bunch of other people say you should just connect it directly and that "A chip datasheet is not a good source for that type of information".
|
|
#
?
Feb 13, 2019 19:20
|
|
|
Without reading the entire stackexchange thread: A core part of designing for reduced EMI is always knowing where a signals return path is. In the case of USB power the return is a tiny black wire inside the shielded cable. If an alternate return path is available that can form a loop that radiates more than the tiny area between two lines in a twisted pair. If you DC ground the shield, it will obviously be a far better conductor and will carry most of the current, AC coupling avoids that by only allowing AC current to take the shield as a return path. For a standalone device the difference is minimal (the cable shield is pretty close after all) - the issue is if your device chassis is connected to something else that's grounded. AC current can flow from the DC+ line in the cable and return via some other piece of chassis that's far away making a loop antenna. The correct way to do it according to old school EMI practice would be to sit around like a classic Greek philosopher for a while and decide that you understand how current will flow, and design a complicated single point grounding system based on your inherent knowledge of all things electrical. Typically the computer in this case will have the single DC- chassis ground connection, your USB devices should then exist as shielded boxes where the electronics are not connected to the shield at all (no capacitor even, if you wanted to actually follow the old school design rules instead of just pretending). Actually doing this kinds of works but my experience with these systems is that they a) radiate a fair bit, b) have a lot of common mode currents flowing all over the place inside shielded cables, and c) anything analog has a tendency to be noisy as hell because the electronics ground is surrounded by a (relatively) floating chassis that couples in to anything high impedance. People fix this by putting capacitors to chassis ground all over the place, effectively AC grounding the DC- and electronics ground to chassis anyway. Since the capacitors have a bit of ESL/ESR they don't actually work as well as just connecting every mounting hole to chassis all over the place. A modern take for high end is that all signals shall have a known single return path and all subsystems that want to ground the DC- have to use an isolated DC/DC converter (with EMI filtering). DC+/- are a differential power pair and the output of the DC/DC can be connected however you want it to be since you know the return path to the source. DC- connected to the ground plane and all mounting screws used for chassis grounding. This + common mode filtering or full isolation on all signal interfaces (with shielded wires all over the place) tends to work a lot better in my experience. My take on how to do this cheaply is also apparently pretty common for automotive electronics: DC- is connected to chassis ground all over the place, but all units get power via a common mode choke and twisted pair (sometimes just the twisted pair). All signal lines should also have a defined return (i.e. be differential like CANBus). The theory is that the DC return being kind of vague is actually not a problem for EMI because DC isn't the problem. Using a CM choke and twisted pair wiring will give a lower impedance return path for higher frequencies through the twisted DC- line than chassis paths, preventing massive loops forming. One possible exception to this is shielded Ethernet wires; they may come from far away (like a different building with a ground fault) and have large common mode voltages relative to local ground. Using a high voltage capacitor to ground for the shield can avoid large AC/DC currents flowing through the shield - in this case the cable would be hazardous to handle as well so it's a bad situation anyway but it will save the product. So my USB solution (if I ever make something that uses bus power) would be to use two common mode chokes, one for the data lines, and one for the DC+/- lines. DC- after choke is hard grounded along with the shield. longview fucked around with this message at 21:21 on Feb 13, 2019 |
|
#
?
Feb 13, 2019 21:18
|
|
|
longview posted:Without reading the entire stackexchange thread: Thanks for the effortpost. That common-mode choke idea makes a lot of sense, especially since the example design already calls for one for the data lines so it's not that big a deal to throw in another.
|
|
#
?
Feb 13, 2019 22:08
|
|
|
Ethernet connections are often passed through signal isolation transformers, always if it's a PoE device, which helps with this stuff. I've never even considered that directly grounding the shield in a usb cable might not be correct. It's never caused me any issues but I've never designed anything that uses highspeed (or whatever the fast one is called) usb
|
|
#
?
Feb 13, 2019 22:17
|
|
|
I think it's even in the Ethernet spec that it's supposed to be galvanically isolated. They even make modular jacks that have the magnetics integrated into them!
|
|
#
?
Feb 13, 2019 22:23
|
|
|
I get a little wound up over the whole "DC MUST NEVER BE CONNECTED TO CHASSIS" concept that is repeated a lot, it feels very dogmatic sometimes and many engineers refuse to think about what they're actually accomplishing by using isolation. That being said I've had some good discussions with EMC engineers on this and we've at least principally agreed that the automotive style of power delivery is fairly sensible and DC isolation isn't actually required most of the time (it can have other benefits though). One thing I forgot to mention is other schools of thought that say you shouldn't actually connect the shield at both ends, this can also be sensible in some cases (a typical case happens to be Ethernet again). Floating the shield at one end with ethernet means it's much harder to get huge ground loops, and it will shield reasonably well against some types of interference (mainly capacitive coupling). Usually not recommended for general use, but can also be useful for specialized systems (like maybe very high impedance cables for photomultipliers or something that might also have a driven shield). If floating the shielding solves an EMI issue in a larger system it's probably an indication of poor grounding or high leakage currents in some other device. This could quickly lead into "Why unbalanced audio is trash for idiots rant #45" so I'll stop now. E: BattleMaster posted:I think it's even in the Ethernet spec that it's supposed to be galvanically isolated. They even make modular jacks that have the magnetics integrated into them! The center tap on the transformer for each pair is terminated together via 75 ohm-ish resistors and then to chassis ground via a high voltage 1nF cap. This works fine, but there can be cases where grounding the shield on shielded cables can cause large currents to flow, this is enough of a problem that one of the IEC codes for building power/grounding has a special section on it. Their recommendation is that in cases where large shield currents can flow due to ground potential differences at either end, an additional gently caress-off huge ground wire should be pulled to equalize the potential, as well as shield bonding to PE at building entry points. Cable TV systems in Norway at least were notorious for having the shield of the coax (an exposed metal piece) sitting at 90-120V AC to everything else that's grounded in the building because lazy installers/random idiot who wants cable TV never bothered to properly ground the coax shield when it enters the building. TVs were (still are, mostly) double insulated so it was usually fine but when stereos and other things were connected to them there was a lot of potential for trouble. It's a bit off topic from the USB question but I thought it was worth mentioning. longview fucked around with this message at 22:39 on Feb 13, 2019 |
|
#
?
Feb 13, 2019 22:25
|
|
|
Unbalanced audio IS trash for idiots, that poo poo should be banned. If you don't connect your shield at both ends does it even work? I'm afraid I'm very much an EE in the school of learn by doing, I actually studied mechatronics so my electrical theory isn't great.
|
|
#
?
Feb 13, 2019 22:39
|
|
|
Analog audio and digital circuits between separate chassis are different signal levels and things though, right? I get the shield causing capacitive coupling.
|
|
#
?
Feb 13, 2019 22:43
|
|
|
|
| # ? May 29, 2024 06:41 |
|
|
A shield that's floating at one end is sometimes called an electrostatic shield, it works well enough if you have say a high voltage AC line running parallel to a shielded signal line. For an unshielded line (say it's a two-line signal), the capacitance between the two cables will cause a small leakage current that is coupled to both wires approximately equally. For a differential signal that is common mode and can be filtered or rejected at the end, in an unbalanced system the coupling will often be mode converted to differential at the receiver and the noise is added to the signal. With an electrostatic shield, all the coupling is mostly to the large low impedance shield, and the leakage current goes back through it without a large voltage developing across it. The internal wires are coupling to the shield, which is at a mostly stable voltage despite the leakage from the high voltage cable. I've never seen it actually measured in an EMC lab at higher frequencies, but it can definitely work for some types of noise.
|
|
#
?
Feb 13, 2019 22:48
|
|