|
catbread.jpg posted:Alternately, depending on whether your measurement circuit is end-to-end resistance based (and depending on how it treats the polarity of low vs high resistance), or 3-terminal voltage based with an input-wiper-ground, you could: B isn't possible as I can't control the voltage coming out of the lead. A is brilliant though. The manual says 0-1k is "no throttle" and 4k (or maybe 4.5k, can't remember off the top of my head) and up is "full throttle". Awesome idea, thank you. I'll go see if I can find a 0-15k+ plus pot tonight.
|
# ? Aug 5, 2009 23:31 |
|
|
# ? May 8, 2024 17:38 |
|
Slung Blade posted:Yep, exactly. Old joysticks are either 0-5k or 0-15k; My throttle is a 0-5k 90 degree pot.
|
# ? Aug 6, 2009 04:00 |
|
Another good option if you find the right pot but its value is too big is to parallel it with a resistor (assuming you're using it with the wiper tied to one of the two ends). So if you had a 0 -> 10k trimpot and you paralleled it with a 10k resistor you wind up with a 0 -> 5k response. You can do a similar trick with two resistors if the wiper is connected to something independent.
|
# ? Aug 6, 2009 04:58 |
|
Poopernickel posted:Another good option if you find the right pot but its value is too big is to parallel it with a resistor (assuming you're using it with the wiper tied to one of the two ends). So if you had a 0 -> 10k trimpot and you paralleled it with a 10k resistor you wind up with a 0 -> 5k response. This is true, but you should keep in mind that you will mess with the resistance curve of the pot. If you have a linear pot, by adding a resistor in parallel you get a more logarithmic output. This may or may not matter for your application, but it's something to remember.
|
# ? Aug 6, 2009 05:37 |
|
Fantastic thread here guys! I have a question that hopefully someone here will be able to answer: I'm working on a project that requires me to control the output of somewhere around 100 3-color LEDs. The LEDs are the kind sold here (they are red/green/blue and draw around 20mA at 2-4V). I must be able to switch on/off each color of each LED individually, giving me 8 possible states per LED (pwm control would get me a lot more and would be ideal, but I don't really need it). The LEDs would be controlled by a PC, via USB or serial. I've been searching on various electronics websites and most of what I've come up with are LED controllers that come with the LEDs attached, or that control a board of LEDs, neither of which work for me (the LEDs need to be completely separate from the controller and from each other). The only thing I've found so far is this 64-LED controller, which is a pretty cool gizmo, but is expensive ($92) and not big enough for my application (it's designed for 1-color LEDs, so I'd only end up with a 21-LED controller). My question is twofold: 1) Is there some product out there that does exactly what I need? 2) If there isn't, then what is the best way to go about designing my own? I saw that some of you guys have designed LED controllers, but is there a relatively cheap way to make one for so many LEDs? (I have been working on this and have some Multisim pics that I'll upload if you're interested). I would really appreciate any help on this. Also, sorry for the overly-wordy description.
|
# ? Aug 6, 2009 06:10 |
|
Corrupt Politician posted:Fantastic thread here guys! I have a question that hopefully someone here will be able to answer: A PIC or AVR with a pile of IO expanders seems like the the obvious choice. http://www.maxim-ic.com/products/interface/io_expanders.cfm How quickly does it need to refresh? I don't think you could get 30fps out of the thing, but I haven't crunched the numbers because edit: Oh hey they have a bunch of LED-specific driver chips too http://para.maxim-ic.com/en/search.mvp?fam=disp_driv&397=LED&tree=master Mill Town fucked around with this message at 08:31 on Aug 6, 2009 |
# ? Aug 6, 2009 08:28 |
|
You can use this example to do what you want: http://www.arduino.cc/playground/Main/DirectDriveLEDMatrix Use some latching chips to expand it past the normal pin capacity, and treat each colour like a separate grid, I guess. Does that make sense, or should I elaborate a little?
|
# ? Aug 6, 2009 11:26 |
|
I've got an idea for home automation that involves having speakers and microphones in several rooms, each of which is tied to a central computer. A person could speak a command into a microphone and the computer would relay its response back to the speaker paired with the microphone. And I would like to do this wirelessly code:
Basically I don't know what's out there for a project like this. I'm much more in tune with the software side of my dumb ideas than the hardware. Please educate me
|
# ? Aug 7, 2009 01:19 |
|
Some of the literature for the XBee I've seen indicates that it can handle around 9.6kbps of throughput. That's enough to send voice. Many Ventrilo servers use less than that. It won't be very high quality, though.
|
# ? Aug 7, 2009 02:33 |
|
Why not a circuit that switches on an FM transmitter and signals the receiving computer to switch to the correct frequency? Do as much of the audio in analog as possible and just use the controllers for advanced following and multiplexing. I don't know what the FCC has to say about unlicensed FM transmitters, though.
|
# ? Aug 7, 2009 02:42 |
|
gently caress the FCC. Actually, that's a good idea. It's even legal if you just use walkie talkie(49MHz) or CB(27MHz) frequencies.
|
# ? Aug 7, 2009 03:17 |
|
Corrupt Politician posted:Fantastic thread here guys! I have a question that hopefully someone here will be able to answer: You can use shift registers I think. 4 8-bit shift registers will give monochromatic control of 256 LEDs. You can either replicate this three times, or use 6 8-bit shift registers and some creative wiring. Or, you could do one grid with the 4 8-bit shift registers, and turn on the three different colors individually for one third of the time. Hopefully that helps. I'm really rusty on this stuff.
|
# ? Aug 7, 2009 03:26 |
|
Corrupt Politician posted:2) If there isn't, then what is the best way to go about designing my own? I saw that some of you guys have designed LED controllers, but is there a relatively cheap way to make one for so many LEDs? (I have been working on this and have some Multisim pics that I'll upload if you're interested). Actually, this project is a prime candidate for an FPGA-based solution. You can buy FPGAs with hundreds of output pins, all independently controllable. With some careful work you could probably PWM every single pin as well (maybe with one PWM circuit shared between every 16 LEDs or something similar). You could easily cook up a simple interface (maybe 7-bit LED address, 8-bit data, and a strobe pin) which you could load using your desired microcontroller of choice. Looking around on Digikey, you can buy a Xilinx Spartan-IIE with 146 I/O pins for about $14. Three of those, an oscillator, some current-limiting resistors and a cheap microcontroller are all the circuitry you would need. Double-check the total drive capacity of one of those FPGAs though - I know each pin is rated for something like 30mA but it would be smart to check the overall power-sink capabilities. I would recommend using an open-drain topology where the anodes of each LED are tied to the desired voltage rail and each cathode is connected (via current-limiting resistor) to the FPGA's output pins. Obviously this method is somewhat hardware intensive, but you would get the best results by far (full PWM, any framerate you want, no worries about brightness). Poopernickel fucked around with this message at 03:45 on Aug 7, 2009 |
# ? Aug 7, 2009 03:41 |
|
Poopernickel posted:Actually, this project is a prime candidate for an FPGA-based solution. You can buy FPGAs with hundreds of output pins, all independently controllable. With some careful work you could probably PWM every single pin as well (maybe with one PWM circuit shared between every 16 LEDs or something similar). And all you have to do is learn VHDL! For someone with little experience, IO expanders or shift registers are going to take a lot less time to master.
|
# ? Aug 7, 2009 04:36 |
|
Mill Town posted:And all you have to do is learn VHDL! verilog
|
# ? Aug 7, 2009 05:29 |
|
catbread.jpg posted:
Catbread, thank you man. It worked perfectly.
|
# ? Aug 7, 2009 05:29 |
|
Poopernickel posted:Actually, this project is a prime candidate for an FPGA-based solution. You can buy FPGAs with hundreds of output pins, all independently controllable. With some careful work you could probably PWM every single pin as well (maybe with one PWM circuit shared between every 16 LEDs or something similar). Jeez, 30 mA per pin? I wouldn't be surprised if the total current sinking ability was not too many multiples of that.
|
# ? Aug 7, 2009 06:57 |
|
Poopernickel posted:verilog Until you publish a study of suicide rates in Verilog and VHDL programmers vs the general population, I'm not going to believe that one's any better than the other.
|
# ? Aug 7, 2009 07:02 |
|
catbread.jpg posted:Jeez, 30 mA per pin? I wouldn't be surprised if the total current sinking ability was not too many multiples of that. I just checked the Spartan-II datasheet. It is capable of sourcing or sinking 24mA per pin with no listed global maximum. I'd imagine your standard Xilinx FPGA is actually capable of sinking quite a bit of current. I mean, an FPGA in open-drain mode (or any open-drain logic chips, for that matter) has FETs inside which provide the ground outputs. So that means you're talking about a DC resistance of (for really lousy FETs) maybe 500mOhm. Compare that against the 100 Ohm resistors being used for current limiting and it's pretty much negligible in terms of how much power the internal FETs need to dissipate. Even if you multiply that amount by 100 I/O pins, it's still pretty negligible. I think that every single output could be active and the chip would still be cool enough to touch comfortably. After all - there's a fair amount of surface area to dissipate heat uniformly. Poopernickel fucked around with this message at 08:36 on Aug 7, 2009 |
# ? Aug 7, 2009 08:28 |
|
Does anyone know where I can get some thin surface-mount coincell holders? I'm looking at the CR2016 specifically, but it seems everyone wants you to stack two of them in a CR2032 holder. Also, to go with this, I'm looking for some thin, bright, cheap SMT LEDs. So far this is the best deal I've found, but it's a bit thick. Is there anything comparable that's a bit thinner? http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=160-1414-1-ND
|
# ? Aug 7, 2009 08:45 |
|
Mill Town posted:Also, to go with this, I'm looking for some thin, bright, cheap SMT LEDs. So far this is the best deal I've found, but it's a bit thick. Is there anything comparable that's a bit thinner? Honestly, the best source I've found for cheap LEDs is ebay. They'll take a little longer to ship, and the product description isn't very nice, but they are absolutely dirt cheap. I mostly buy through-hole LEDs, though, so your mileage may vary.
|
# ? Aug 7, 2009 13:25 |
Mill Town posted:And all you have to do is learn VHDL! here's a datasheet for one I've used in the past: http://www.allegromicro.com/en/Products/Part_Numbers/6278/6278.pdf
|
|
# ? Aug 7, 2009 19:11 |
|
ANIME AKBAR posted:there are much easier ways than learning to use fpgas and using discrete components. One is ICs that are essentially shift registers whose outputs are configurable current sources. You can daisy chain any amount of them on the same serial bus, so you could operate them all with a single microcontroller. Yeah, that's the same thing I linked in my first answer, just by a different manufacturer.
|
# ? Aug 7, 2009 20:13 |
|
Thanks for all the advice. By the way, I'm a college student studying ECE, so I don't worry too much about the learning curve. Supplying power won't be a problem because I'm going to use a MOSFET to switch each LED on and off, so the FPGA/microcontroller/whatever I use will not be supplying the current. My original plan was to use a bunch of demuxes so I only needed an 8-bit output on the microcontroller I was using (I was planning on using the MSP430, since it's the only one I've programmed before). The MSP430 would continually refresh the state of the LEDs one at a time (each LED would have a D-Latch to keep it in the same state until refreshed). However, this would require an irritatingly large number of discrete components and connections between them, so I thought you guys might have a better way to do it. I had actually thought of using an FPGA to do it earlier. (I do know a bit of VHDL, and no, it doesn't make me want to kill myself... though the horrendously buggy Xilinx GUI does.) However, the development boards are expensive (>$200) and the chips themselves are all surface-mount, which I know wouldn't scare a lot of you guys, but I suck pretty bad a soldering (I'm much more of a theory-type guy than hands-on, though I'm trying to remedy this). I mean, how am I supposed to mount something that looks like this: Though if someone can convince me that it isn't as hard as it looks, I'll consider it. I hadn't seen those LED driver chips that you guys sent links to. I might use the Allegro one. I'd need something like 15-20 of them, if I'm using the 16-bit part. That would save the need to use latches. One question that none of these solutions seem to address is how I control all of this with a PC. Is there some sort of board I can get that interfaces CMOS with USB/Serial ports? Ideally, I would like to do away with the microcontroller completely and just use a PC to do all the work. Is this feasible/realistic?
|
# ? Aug 7, 2009 22:33 |
Corrupt Politician posted:Thanks for all the advice. By the way, I'm a college student studying ECE, so I don't worry too much about the learning curve. For soldering, what you need is a skillet: http://www.sparkfun.com/commerce/tutorial_info.php?tutorials_id=59 As far as connecting PC<->Device, I would recommend using a USB UART chip like the FT232 series: http://www.sparkfun.com/commerce/product_info.php?products_id=718 You can use the included schematic if you want to design your own board. They also have 'cheaper' FPGA dev boards that would save you some soldering: http://www.sparkfun.com/commerce/product_info.php?products_id=8595 I don't consider $80 cheap, but its better than $200 by a fair margin. IMHO this is totally doable with an MSP430 (especially since you don't seem to want brightness control). 18 pins + Charlieplexing could take care of your I/O requirements, if you could switch fast enough, and I'm sure there are other options as well. EDIT: Looking at the numbers and thinking of firmware, it might makre more sense to do 3x11pins, so you have a charlieplexed array for each color - the circuit board is going to be a bitch at this point. Delta-Wye fucked around with this message at 23:20 on Aug 7, 2009 |
|
# ? Aug 7, 2009 23:16 |
|
Delta-Wye posted:For soldering, what you need is a skillet: http://www.sparkfun.com/commerce/tutorial_info.php?tutorials_id=59 quote:Cons: That's hilarious.
|
# ? Aug 7, 2009 23:57 |
|
Just to emphasize this - you won't need external MOSFETs if you go with an FPGA and set the pins up as open-drain. As for assembly, it's totally doable if you know how to solder a little bit already. If you can do SOIC-8s then you can probably handle a TQFP. I've done them by hand before. A TQFP-144 takes me probably between a half-hour and an hour to hand solder. Youtube has a _ton_ of soldering technique videos which will show you the best ways to solder. For me, when I'm doing a TQFP I usually use the following approach: 1) Align chip on the pad 2) Tack down one corner with a glob of solder. It can touch a few pins, but you want to keep the chip right on location while you do this. 3) Heat and adjust if needed to fix the chip's alignment. 4) Tack down the opposite corner. 5) Squirt flux all over your pins and poo poo. 6) Put a little bit of solder on the tip of your iron. It really doesn't take hardly any solder at all for a single TQFP pin. Emphasis on 'a little bit' of solder. 7) Touch the iron to a pin. Solder that motherfucker. 8) Do the pin next to it. Do it now. Don't refill your soldering iron tip. 9) Once you run out of solder, apply more to the tip and do it again. 10) Have some solder wick on hand for cleaning our solder bridges. You will probably bridge a lot of pins the first time and then a few pins the next time. 11) When you finish the first side, next do the opposite side. 12) Clean off the globs of solder you used for tacking. 13) Two more sides and you're done! Another good route to do the solder-paste & heat method, with a toaster or a frying pan or something like that. Essentially making a lovely reflow oven. If you want to go this method (great results quickly with some practice), I'd recommend using your regular toaster oven. You can buy these little dots that change color at a particular temperature (for soldering). Put one on top of your TQFP, stick it in the toaster on medium, and wait until the dot changes color to indicate it's hit the reflow temperature. Then you can turn the heat off, open the door, and wait for it to cool. This method doesn't require any manual dexterity but it does take practice. I've never used a skillet so I can't comment on how well it works or doesn't work.
|
# ? Aug 8, 2009 09:58 |
I'm right now in the process of assembling about 25 boards using the skillet/stencil method. I'm finding it's not quite as easy as I expected, but that's probably because I have around fifty 0603 components per board. Simply placing them takes hours. And I'm using lead free solder paste, which I'm starting to regret, since it doesn't flow nearly as nicely as leaded solder does. The skillet I used literally couldn't put out the heat necessary, so I ended up finishing it off with my hot air station. I was going to make a long post about it, but I don't have a working camera, so it would be kind of pointless. And if you're not familiar with soldering, using paste or a skillet isn't any easier. It's only useful when you're producing lots of boards, or if you have components that are truly impossible to hand solder (BGA, DFN, etc). I've soldered 144 pin EQFP cyclone II FPGAs (the same package in that picture, I believe) by hand before and they weren't so bad, so long as you have a decent microscope. Took maybe 15 minutes in all. ANIME AKBAR fucked around with this message at 13:52 on Aug 8, 2009 |
|
# ? Aug 8, 2009 13:49 |
|
One addition - here is the website I used to learn basic SMD soldering: http://www.curiousinventor.com/guides/Surface_Mount_Soldering/
|
# ? Aug 8, 2009 16:45 |
|
Poopernickel posted:I'd recommend using your regular toaster oven. Get a separate toaster oven, eating soldery food will give you lots of cancer
|
# ? Aug 8, 2009 20:07 |
|
Poopernickel posted:I've never used a skillet so I can't comment on how well it works or doesn't work. We use a regular hotplate at the lab. It works really well with the caveat that you need to move the work around to compensate for cool spots like the center of the element.
|
# ? Aug 8, 2009 22:10 |
|
What's a quick and easy program to use to simulate logic gates? For my course I'm learning how to use Altera Quartus II, but I'm not sure if it can do GUI simulations of logic gates. I can do the waveform simulations, but would like some visual feedback (on the actual wiring diagram) of what's happening. The Falstad Circuit Simulator worked ok for this, but is just a bit clunky. Any idea's?
|
# ? Aug 11, 2009 02:55 |
|
Quartus can do it. Create a block/schematic file.
|
# ? Aug 11, 2009 04:17 |
|
I've got the Block/Schematic file already created. Basically, what I was hoping to do (if it's possible) was to click on the Input Pin's to toggle them high/low, to see how that effected the schematic. Currently I use a simulation waveform, which I can set the timings/values for the input pins, which then gives me the outputs. I'm using the Web Version and currently just getting my head around the software to see what is/isn't possible.
|
# ? Aug 11, 2009 04:35 |
|
You want software that can generate gate logic based on inputs/outputs that you specify? I know that there is stuff on the web that can give you an equation that's usually pretty similar to something you could come up with by hand. It's not very hard to draw that out with the proper symbology. A better solution is to do it yourself/learn how to do it yourself, though.
|
# ? Aug 11, 2009 04:59 |
|
This is basically what I want: Click here for the full 894x497 image. I've added the Green/Red lines in paint. Just to be able to actually set inputs on the schematic high/low and follow the logic through the diagram. I simulate the logic with the waveform, but at the moment I'm finding it a little clunky, so was hoping for something a little more visual. Thanks for your help btw
|
# ? Aug 11, 2009 05:10 |
|
I think Logisim does what you want.
|
# ? Aug 11, 2009 10:14 |
|
I came across a circuit to be used for touch sensing. Could anybody here walk me through how/why this is supposed to work ? The input is a 9v square wave. The potentiometer (top right) is supposed to set the trigger touch capacitance (lower right). Output is connected to a FET gate. What's that diode doing there?code:
|
# ? Aug 11, 2009 23:20 |
Well, the lower impedance the capacitance, the more the base of the PNP will be pulled down and the more current will be delivered through the collector into the output. So increasing capacitance should raise the output voltage. The diode is likely keep the collector-base junction from being forward biased when the square wave is low.
|
|
# ? Aug 12, 2009 06:19 |
|
|
# ? May 8, 2024 17:38 |
|
Hello people smarter than I am, I have a question about three-state logic chips. Say I had a few registers with three-state outputs (in my case, 74hc574's) connected to a bus. What I want to do is have the output enable pins connected to some sort of decoder, so that only one of the outputs will be enabled at any given time. What I'm wondering, though, is if there will be some in-between state as one register is disabling and another is enabling, where the two will short together and damage each other. If I had an oscilloscope I'd test it myself, but I am a hobbyist with no money so that's out of the question for now. Thanks!
|
# ? Aug 12, 2009 20:31 |