|
poeticoddity posted:Fair enough. I assumed there had to be something better I just didn't know how to search for. EMI testing is pretty crazy stuff and I'm not surprised there isn't a lot of layman information laying around. You may be able to remove enough noise using copper tape to line whatever specific enclosure you're using, but an EEG is fairly sensitive. Which makes me think about it; are you physically connected to the EEG? If so, are you galvanically isolated?
|
#
?
Jun 18, 2013 06:53
|
|
|
|
| # ? May 8, 2024 06:08 |
|
|
EDIT ^^ Metal tape is an even better idea than wire mesh  ^^poeticoddity posted:Fair enough. I assumed there had to be something better I just didn't know how to search for. I mean there's probably plenty of fancy things you can do to it, but wire mesh is cheap - if you can measure its emissions you might as well try it and see if that's all you need.
|
|
#
?
Jun 18, 2013 06:53
|
|
|
Delta-Wye posted:EMI testing is pretty crazy stuff and I'm not surprised there isn't a lot of layman information laying around. You may be able to remove enough noise using copper tape to line whatever specific enclosure you're using, but an EEG is fairly sensitive. Which makes me think about it; are you physically connected to the EEG? If so, are you galvanically isolated? I'm a grad student in a psych lab. I've taught myself to design and build equipment for behavioral tasks (visual psychophysics), but I've never actually touched any of the imaging equipment. The hope is to take a task that's currently a behavioral measure and redesign the equipment I built to allow the operator of the EEG to look for a signal indicating the same phenomena we're testing behaviorally at the moment. I would assume that the only electrical connection between my intended equipment and the EEG would be if they were both attached to the same computer (I have no clue that's relevant), and that's not actually necessary if it's problematic. If possible I'd like to get a serial connection to a computer (ideally by wire, but I'll build an IR transmitter or something if I have to) which I know could be one source of EM noise. The others would be the power line to the device, the microcontroller, and the connections to some beefy LEDs. When this thing's built, I'm up for finding the best shielding solution through trial and error (  - coworkers being impressed you even attempted to build something is awesome!) but I've had no idea where to start before tonight. Thanks, everyone!
|
|
#
?
Jun 18, 2013 07:16
|
|
|
Anybody know a website that lists alternatives for parts? I'm looking for a an ALPS SKQUAAA010 4 way tactile switch for the HeaterMeter I'm building. Mouser wont have any in stock for a couple months. And the other sites I found have ridiculous minimums (23, 100, 1000) or only have them in stock in the UK and will cost $20 if freight (Newark). I'm looking for an alternative to that switch with the same design as the board is already made. They have a surface mount one available at Mouser, but it needs to stick up some so you can access the button through a hole in the case. Is their a surface mount to through-hole adapter or something similar?
|
|
#
?
Jun 18, 2013 13:46
|
|
|
poeticoddity posted:I'm a grad student in a psych lab. I've taught myself to design and build equipment for behavioral tasks (visual psychophysics), but I've never actually touched any of the imaging equipment. The hope is to take a task that's currently a behavioral measure and redesign the equipment I built to allow the operator of the EEG to look for a signal indicating the same phenomena we're testing behaviorally at the moment. Might be a dumb question, but do you know if you even need special shielding?
|
|
#
?
Jun 18, 2013 16:06
|
|
|
poeticoddity posted:I'm a grad student in a psych lab. I've taught myself to design and build equipment for behavioral tasks (visual psychophysics), but I've never actually touched any of the imaging equipment. The hope is to take a task that's currently a behavioral measure and redesign the equipment I built to allow the operator of the EEG to look for a signal indicating the same phenomena we're testing behaviorally at the moment. poeticoddity posted:I would assume that the only electrical connection between my intended equipment and the EEG would be if they were both attached to the same computer (I have no clue that's relevant), and that's not actually necessary if it's problematic. If possible I'd like to get a serial connection to a computer (ideally by wire, but I'll build an IR transmitter or something if I have to) which I know could be one source of EM noise. The others would be the power line to the device, the microcontroller, and the connections to some beefy LEDs. Here is a page with someone doing it in a guitar cavity http://www.strat-talk.com/forum/stratocaster-discussion-forum/109690-best-procedure-shielding-cavity-copper-tape.html It's the same idea but backwards, you're trying to keep the noise in, not keep the noise out, but the effect is the same. SnoPuppy posted:Might be a dumb question, but do you know if you even need special shielding? Maybe not, but it couldn't hurt to slap some tape on it. EEG's are retardedly sensitive but if you're not real close it may not be a big deal. They are amplifying tiny signals which means induced noise becomes a large issue as it also gets amplified, but they are fairly low frequency signals and I think the probes often have filters on them to reduce the already-present noise like 60Hz hum and the like. You're fundamentally right though; it doesn't hurt to keep EMI in mind, but it may be a nonissue from the start. There is no reason to make the problem harder than it has to be
|
|
#
?
Jun 18, 2013 16:36
|
|
|
bobua posted:Never heard of the stuff, why do you ask? I thought that it might be the tubing that you need to use. It's a flexible conduit used for both mains and low voltage cabling. It most commonly comes in the color blue, hence the nickname. It was just an idea to get around your need to cut and terminate the cable by using wider tubing. You said that your tubing didn't come in bigger sizes, but I know that ENT comes in up to 3" diameters.
|
|
#
?
Jun 18, 2013 17:42
|
|
|
ShadowStalker posted:I'm looking for a an ALPS SKQUAAA010 4 way tactile switch for the HeaterMeter I'm building. http://onlinecomponents.com has them in singles but they'll charge you $7 in shipping I reckon.
|
|
#
?
Jun 18, 2013 20:47
|
|
|
SnoPuppy posted:Might be a dumb question, but do you know if you even need special shielding? I don't know that for certain, but I know some of the research EEGs have to be shielded from all of the elevators in the building. Delta-Wye posted:The galvanic isolation will be done on the EEG unit. I wasn't 100% sure you weren't making an EEG or something. From what I understand, the sensor unit needs to be isolated by the processing unit/computer; I've always seen the probe unit battery powered and connected over fiber optics. If you're building something else in the same room don't worry about that though. Thanks for the advice. Part of the issue with this is I'm going to be applying power to the LEDs at frequencies well within the ranges that EEG is designed to pick up, so I'm going to try to minimize the impact of that in particular.
|
|
#
?
Jun 18, 2013 21:54
|
|
|
poeticoddity posted:I don't know that for certain, but I know some of the research EEGs have to be shielded from all of the elevators in the building. I was playing around with a biofeedback (binaural audio and CEV blinking lights) setup and was sort of just ignoring the noise issues for the time being. What sort of LEDs are you driving? Any particular driving pattern? There may be a way around instead of over 
|
|
#
?
Jun 18, 2013 22:13
|
|
|
I've got a couple of these: http://www.ebay.com/itm/1m-RGB-LED-Light-Strip-5V-WS2801-Bonus-32-LEDs-Waterproof-Addressable-Color-/160810882132 I'm correct in my understanding that I can just clip it off wherever I want (between discrete sections, each containing the ws2801, a few resistors, a diode, and solder pads at each end) and make my own lengths, right? They even put a line on the circuit board (strip?) between each chunk, with solder pads available, you just gotta clear the clear plastic off. Just looking for a minor sanity check before I do something crazy. What I'm actually doing: going to clip it off into two sections 3-chunks-long and two sections 2-chunks-long for installation in a little project. Once clipped, I'll solder a bit of ribbon cable between the chunks so they can be spaced a little further apart around the inside perimeter of a frame, and then hook one end up back to the controller, that's all, nothing special really.
|
|
#
?
Jun 18, 2013 22:20
|
|
|
Bad Munki posted:I've got a couple of these: http://www.ebay.com/itm/1m-RGB-LED-Light-Strip-5V-WS2801-Bonus-32-LEDs-Waterproof-Addressable-Color-/160810882132 It would appear so as long as you remember to reseal everything and adhere to the datasheet listed on the seller's page. Looks completely modular to me. Edit: also be careful soldering around that IC, heat and all. Another option you might want to consider would be electroluminescent lighting. FireWhizzle fucked around with this message at 22:55 on Jun 18, 2013 |
|
#
?
Jun 18, 2013 22:53
|
|
|
That'd be an option, but I want full RGB control (and the addressable part is a fun bonus that I'll be able to use.) Also, I don't know that I really have to reseal anything unless I want to maintain the waterproofness of the strip, which I absolutely don't care about in this case. As for heat, the IC is a fair distance away from the pads, I don't think it'll be a problem. I'll remove the plastic coating around the pads, feed the leads of a 4-conductor ribbon cable through the holes in the pad (it even has holes already!) and be good to go.
|
|
#
?
Jun 18, 2013 23:01
|
|
|
poeticoddity posted:Can anyone recommend either a specific resource or a set of search terms for learning about keeping your electronic equipment from interfering with EM sensitive equipment? Every combination of terms I've put into google has resulted in links to forum posts about surviving an EMP and other The operative term here is Electromagnetic Compatibility - EMC for short. It's a vast and somewhat idiosyncratic EE sub-discipline. Often referred to as "voodoo." It covers everything from shielding, compliance with legal standards (e.g. FCC), surge protection, noise suppression, crosstalk, etc.
|
|
#
?
Jun 18, 2013 23:22
|
|
stuff, which so isn't what I'm looking for.
|
Delta-Wye posted:I was playing around with a biofeedback (binaural audio and CEV blinking lights) setup and was sort of just ignoring the noise issues for the time being. What sort of LEDs are you driving? Any particular driving pattern? There may be a way around instead of over These are going to be moderately high power LEDs (several 3.8V 150mA LEDs) with a 50% DC between 0.5 and 200 Hz on top of ~4kHz PWM. Boy With Stick posted:The operative term here is Electromagnetic Compatibility - EMC for short. It's a vast and somewhat idiosyncratic EE sub-discipline. Often referred to as "voodoo." It covers everything from shielding, compliance with legal standards (e.g. FCC), surge protection, noise suppression, crosstalk, etc. Awesome! I knew there was a term I was missing. Thanks for the help, everyone.
|
|
#
?
Jun 19, 2013 01:15
|
|
|
Bad Munki posted:I've got a couple of these: http://www.ebay.com/itm/1m-RGB-LED-Light-Strip-5V-WS2801-Bonus-32-LEDs-Waterproof-Addressable-Color-/160810882132 I had a similar strip from Adafruit and hacked it up / soldered it together without any issues. I blew the first ic out in the finished product when I accidentally hooked up the power backwards, but it was easy enough to replace that one "pixel" and it didn't fry anything further down the strip.
|
|
#
?
Jun 19, 2013 02:04
|
|
|
CE self-certification and most UL certifications require a bunch of EMC testing - commonly radiated immunity (from 4 directions typically), conducted immunity, surge, air and contact discharge, and emissions, etc. You may want to look into the emissions parts of those standards to the extent that time and budget allow for ideas on how to test your device's emissions in a more rigorous and less risky way than "turn it on and see if it fucks up the EEG".
|
|
#
?
Jun 19, 2013 02:08
|
|
|
Boy With Stick posted:The operative term here is Electromagnetic Compatibility - EMC for short. It's a vast and somewhat idiosyncratic EE sub-discipline. Often referred to as "voodoo." It covers everything from shielding, compliance with legal standards (e.g. FCC), surge protection, noise suppression, crosstalk, etc. A buddy of mine who is EMC @ Rockwell Collins corrects me every time I call what he does voodoo and recites Maxwell's Equations 
|
|
#
?
Jun 19, 2013 02:08
|
|
|
armorer posted:I had a similar strip from Adafruit and hacked it up / soldered it together without any issues. I blew the first ic out in the finished product when I accidentally hooked up the power backwards, but it was easy enough to replace that one "pixel" and it didn't fry anything further down the strip. Diodes, dear watson! Diodes! (Seriously, use a half wave on any diy project, you'll thank yourself later)
|
|
#
?
Jun 19, 2013 03:08
|
|
|
In this case the voltage drop would have been problematic, because it was battery powered and I didn't have the overhead. But yeah... you live you learn right?
|
|
#
?
Jun 19, 2013 03:14
|
|
|
armorer posted:In this case the voltage drop would have been problematic, because it was battery powered and I didn't have the overhead. But yeah... you live you learn right? A schottky would give the protection with less of a drop, but there are even better circuits for this. https://www.youtube.com/watch?v=IrB-FPcv1Dc I've done the p-channel protection before and it works well.
|
|
#
?
Jun 19, 2013 04:08
|
|
|
Anyhow, now I have this: https://www.youtube.com/watch?v=5v189R0jAL4 The video makes it look like the hot spots are pretty intense, but they're really not. Also, the final installation will have the LEDs in a channel set back a bit, illuminating at a right angle to the bezel a white matte reflector, which will make for a really nicely diffused backlight. And those LEDs are all individually addressable, of course, which makes I can make silly spinning rainbows and poo poo. 
|
|
#
?
Jun 19, 2013 04:15
|
|
|
poeticoddity posted:These are going to be moderately high power LEDs (several 3.8V 150mA LEDs) with a 50% DC between 0.5 and 200 Hz on top of ~4kHz PWM. Why not increase the PWM frequency to something above the EEG sensor bandwidth? I suspect EMI could also be improved by putting the switching very near the LEDs with a large amount of capacitance on the power supply lines, feeding the drive signals over coaxial cable. E: misread slightly, the 0-200 Hz signal would still be present even with the HF PWM, but if the 4 kHz signal is an issue that could be worked around. Putting drive electronics near the LEDs making the power-path for the LEDs very small and low inductance should help reduce EMI. longview fucked around with this message at 09:06 on Jun 19, 2013 |
|
#
?
Jun 19, 2013 09:04
|
|
|
movax posted:A buddy of mine who is EMC @ Rockwell Collins corrects me every time I call what he does voodoo and recites Maxwell's Equations I can personally attest that there are certain amazing people to which these kinds of issues are transparent and methodical. You can show them any device with its enclosure and tell them where an emission source is on a PCB, and they'll be able to tell you right off the bat what modes are going to be a problem and how to get rid of them. They're able to do all the curls and divergences in their head. But they tend to be physicists rather than engineers.
|
|
#
?
Jun 19, 2013 13:27
|
|
|
Newark has some kind of Schrodinger's cat inventory system going on: The superposition is collapsed by hitting the place order button.
|
|
#
?
Jun 19, 2013 13:57
|
|
|
Bad Munki posted:Anyhow, now I have this: That's pretty cool, but I kinda like the hotspots in this case, to be honest. It's a cool effect!
|
|
#
?
Jun 19, 2013 14:31
|
|
|
Sure, if that's what you're going for. In person, though, when that whole bezel lights up so perfectly evenly, it almost looks like an electroluminescent pad but with full color capabilities. And it's sooooo briiiight. Also, I made it rainbow spin late last night, which looks sick. That is, not rainbow cycling, but a rainbow wheel marching around the perimeter. I'm extremely happy with how it came out so far, and I can't wait for lunch so I can work on the frame a bit more today. It was silly that I had to chop up the strip in the first place, but I needed just a little more gap between the edges in order to get the spacing right, and to make that hard 90° corner. Anyhow, next step is to get the web service up and running that will authenticate/authorize pin codes, and then bang out the arduino software that will read the keypad, submit it to the web service, and fire the relay that'll open the electronic strike on the makerspace door. Gonna be all pretty, I'll have it kinda breathe blue or something while it's idling, and when they put in a code, it'll rainbow spin (thanks, Apple!) while it's pinging the web service, and then alert red or alert green based on the results. e: Oh, also, I know my soldering skills are absolute poo poo, but holy cow it was a pain in the rear end to solder those things at first. I ended up getting a fairly good method and rhythm by the end, but figuring out how to get it going was a pain in the butt! I think only one of my solder joints needs any TLC, I think it might be jumping across leads a little bit. Fortunately that's just the very last segment that has any trouble, so everything else before those last two LED chunks still works. Bad Munki fucked around with this message at 14:42 on Jun 19, 2013 |
|
#
?
Jun 19, 2013 14:38
|
|
|
poeticoddity posted:These are going to be moderately high power LEDs (several 3.8V 150mA LEDs) with a 50% DC between 0.5 and 200 Hz on top of ~4kHz If you're really concerned about noise, I'd avoid the PWM all together and use a DAC adjustable current sink. It will be a little more complicated and won't be efficient, but it should be very quiet. But even with the PWM, as long as your cables are shielded you should be fine. You might also look into slowing down the edges of the drive transistor as much as possible. The frequencies you call out all have very long wavelengths, but the ns edges thay make up the PWM will have much higher frequency content that is more likely to radiate.
|
|
#
?
Jun 19, 2013 17:09
|
|
|
If anyone is interested, I finalized that selector valve circuit diagram for my truck! I got it down to just one diode, and that diode is just there for futureproofing those relays as they wear out. I actually prefer this method to my previous one since powering both SPDT relays at once more closely resembles the original DPDT switch wired into a 4-way. I'm also working with another F-150 owner to update his webpage regarding this modification. He did his selector valve the manufacturer's way, which involved repurposing wires everywhere.
|
|
#
?
Jun 19, 2013 23:39
|
|
|
longview posted:Why not increase the PWM frequency to something above the EEG sensor bandwidth? I suspect EMI could also be improved by putting the switching very near the LEDs with a large amount of capacitance on the power supply lines, feeding the drive signals over coaxial cable. For the first time ever, I'm actually planning on mounting the control hardware onto the same board as the LEDs (instead of having cabling between them) to reduce the EMI. Glad to know I'm on the right track, design wise. SnoPuppy posted:If you're really concerned about noise, I'd avoid the PWM all together and use a DAC adjustable current sink. It will be a little more complicated and won't be efficient, but it should be very quiet. I think the PWM noise will be easier to filter out than the noise from the >200Hz flicker I need to create. (I imagine a high-pass filter should cut that out pretty nicely.) The DAC is tempting, but that would require another set of signal lines, another two ICs, and decent bit of programming I'd like to avoid on this project. Related to this project, but unrelated to the shielding issue: The LEDs I want to use have 3 diodes (3 x 3.8V, 50mA diodes per package). I know there are major issues with wiring them up in parallel without individual current limiting resistors. If I have a current limiting resistor on the positive side for each diode to keep it within safe amperages, but have the grounds for each diode connect and run through a suitably hefty trimmer resistor for fine adjustment, will that cause issues?
|
|
#
?
Jun 20, 2013 00:18
|
|
|
poeticoddity posted:For the first time ever, I'm actually planning on mounting the control hardware onto the same board as the LEDs (instead of having cabling between them) to reduce the EMI. Glad to know I'm on the right track, design wise. So I read back some of your posts but may have missed some things. Whats the relationship between this device you're making and the other EEG equipment you're worried about? Part of the same device, electrically connected too it, same power supply or just physical proximity? Shielding and good power supply bypassing should get you a pretty close. If you're communicating between devices opto-couplers or other isolators can be used. As for the DAC suggestion, one thing you could do would be to still use a PWM, but use it to generate just a voltage reference instead of switching actual power. The RC filter on the PWM can be directly at the PWM output. Between the low power and short traces this will drastically reduce noise. This PWM generated analog reference signal can then control an analog current source (as simple as an op-amp and sense resistor). That said, I'm not sure noise is an issue, but on the other hand a simple current source requires very little and does the best job regulating the LED anyway. Also what voltages are available? What are you driving the LED's from? As for you're last question, in general it's fine. All it does is slim down the voltage that's dropped across the individual resistors which leaves you more susceptable to variations between diodes. For example if you're dropping 1V across the individual limiting resistor and one diode is 0.1V more than another (3.9V instead of 3.8V) current (thus brightness) will go down by 10% - because the resistor is setting the current. On the other hand if you're running from 12V and dropping 8V across the individual resistor the same 0.1V variation is more like a 1% difference in current/brightness. The trimming resistor has the same effect as a lower supply voltage. You're immune to this if you have individual current sources or if you can run the LED's in series. asdf32 fucked around with this message at 03:49 on Jun 20, 2013 |
|
#
?
Jun 20, 2013 03:36
|
|
|
asdf32 posted:So I read back some of your posts but may have missed some things. Whats the relationship between this device you're making and the other EEG equipment you're worried about? Part of the same device, electrically connected too it, same power supply or just physical proximity? The device will be in the same room (within 3-4 feet, if possible) with the subject, but it doesn't directly interface with any of the EEG hardware. It may also need to be connected to the same PC (ideally via USB, but if that doesn't work, I'll figure something else out). I need to flicker LEDs at 0.5 to 200 Hz (with a 50% duty cycle) and simultaneously adjust their brightness, preferably with PWM. I think some small design changes and some shielding will probably get me most or all of the way there, from what everyone's suggested, but I'll look into the RC filter. I really appreciate the suggestions from everyone (this is way more effort from you guys than I was expecting). I feel a lot better about this.
|
|
#
?
Jun 20, 2013 03:48
|
|
|
How many LEDs are we talking about here? Or more specifically, how much power? Because at this low of a frequency, with only an indirect path for conducted EMI, I wouldn't image you'd need any shielding unless you were flashing like 100W. Hell, the EE's where I work doing even do any EMI testing under like 1MHz or something. EDIT: You mentioned shielding from elevators---that makes sense because they (a) use a shitton of power and (b) radiate over a extremely broad spectrum. Slanderer fucked around with this message at 04:14 on Jun 20, 2013 |
|
#
?
Jun 20, 2013 04:11
|
|
|
Slanderer posted:How many LEDs are we talking about here? Or more specifically, how much power? Because at this low of a frequency, with only an indirect path for conducted EMI, I wouldn't image you'd need any shielding unless you were flashing like 100W. Hell, the EE's where I work doing even do any EMI testing under like 1MHz or something. Tentatively 4 3-diode LEDs (each diode 3.8V 50mA). I figured this project would be a bit on the tame end of EMI problems, but I wanted to keep that in mind while I was finalizing the design because if this device works well in a laboratory environment, and the test we want to run with it is suitably diagnostic, future versions may be sent out to hospitals for use with whatever EEG they have there instead of the specific EEG my device was fine-tuned for.
|
|
#
?
Jun 20, 2013 07:15
|
|
|
CapnBry posted:Did you check OctoPart? Yeah, I checked Octopart. Onlinecomponents has a $20 minimum order. Mouser ended up getting some in stock today. Now I just have to wait for the parts to show up to put this thing together and test it out on my Big Green Egg
|
|
#
?
Jun 20, 2013 18:25
|
|
|
poeticoddity posted:I really appreciate the suggestions from everyone (this is way more effort from you guys than I was expecting). I feel a lot better about this. There's nothing the blue smoke thread likes more than over engineering some poor sap's project. Question out of the blue, but by chance is anyone participating in Seattle's Startup Weekend? Anyone participate in something similar somewhere else? I'm not even sure what to expect.
|
|
#
?
Jun 20, 2013 20:37
|
|
|
Well, I finished the frame for my little keypad doodad. Need to put a case on it to hide the LED channels & bezel mounting screws and house the rest of the guts and all, but that won't really change the dimensions at all, just make it prettier. https://www.youtube.com/watch?v=OO0Iv_TMvPU As always, looks better in person: it's really almost like a rainbow el-pad.
|
|
#
?
Jun 23, 2013 21:17
|
|
|
That looks great! Very fun. Separately: what are my options with "dead" batteries? If I understand it correctly, they still have a lot of energy left, but at a decreasingly lower voltage than labeled. Could I use a "dead" 9V that still outputs ~7V with a 1.8V regulator to power some very low-power electronics, like an MSP430 based circuit? I think the limit is how far above a voltage regulator the input voltage has to be. Are there any other ways to get that stored current out?
|
|
#
?
Jun 23, 2013 23:21
|
|
|
Kire posted:That looks great! Very fun. If you're not familiar with it already, you should check out the Joule Theif circuit. (There are dozens of websites covering the theory, design, etc.)
|
|
#
?
Jun 23, 2013 23:34
|
|
|
|
| # ? May 8, 2024 06:08 |
|
|
There really isn't a lot of energy left in an alkaline when it's reached 0.8V per cell, but a 7V 9V battery would have almost 1.2V per cell, and that's a reasonable percentage still left over. A single microcontroller with no fancy LEDs or other power sucking accessories could run for a very long time off that. A Joule Thief is good for powering LEDs but I think that circuit has a pretty large idling current so it might be more efficient to power a microcontroller using a high quality LDO regulator with very low I_q, it depends on how much current is needed.
|
|
#
?
Jun 24, 2013 09:27
|
|