|
Delta-Wye posted:Is it something like bank 1 works, but bank 2 doesn't? It was 7 on bank 1 worked, the 8th on bank 1 did not work, and all the ones on bank 2 did not work. I tried another board and all the relays on both banks did not work. I believe I have narrowed down the issue though. I thought minimum load requirements was related to how much power is being supplied, but apparently it's how much power the device is using, right? I was testing the relays on devices that were drawing only 33 mA. When I used a much more powerful device, the relays worked as they should. So does minimum load correspond to the amount of current that the SSR can leak, and therefore anything that can run on that amount of leaked current would stay on? Edit: After a bit more reading, I think what I just said is totally wrong, so I have no idea why one device worked when the other didn't. Hidden Under a Hat fucked around with this message at 17:47 on May 9, 2012 |
# ? May 9, 2012 17:40 |
|
|
# ? May 8, 2024 07:02 |
|
Hidden Under a Hat posted:Here are the pictures of my device: http://imgur.com/a/U3Fif/embed This might be a dumb question, but do those black and white terminal strips act as a bus? Like all the screws on each one are connected together internally?
|
# ? May 9, 2012 18:17 |
|
SolidElectronics posted:This might be a dumb question, but do those black and white terminal strips act as a bus? Like all the screws on each one are connected together internally? Yes, those are jumpers that connect all of the same colored terminal blocks together.
|
# ? May 9, 2012 20:07 |
Base Emitter posted:On the position sensing thing. It's not passive and it requires line of sight for a minimum number of sensors (redundancy helps), but: ANIME AKBAR fucked around with this message at 05:32 on May 10, 2012 |
|
# ? May 10, 2012 05:27 |
|
I've got some time sensitive projects I'm working on right now, and I decided to start using the Launchpads. The kind-of meh documentation aside, paying $4.30 per dev board is so cheap, I don't mind leaving it in the project at the end. You can saw off the flash-emulator portion of the board to save space if you want, but four bucks for a complete platform to support a MCU isn't bad at all. I ordered like 10 of them to just keep around for instances like this. Though, can anyone comment on mspgcc? I hate Eclipse with a vengeance and would not mind at all just using Notepad++ as my IDE. The upside of using Eclipse/CCS is that debugging is amazingly easy, whereas with mspgcc, I get to use lovely product of FOSS, GDB.
|
# ? May 11, 2012 19:51 |
movax posted:I've got some time sensitive projects I'm working on right now, and I decided to start using the Launchpads. The kind-of meh documentation aside, paying $4.30 per dev board is so cheap, I don't mind leaving it in the project at the end. You can saw off the flash-emulator portion of the board to save space if you want, but four bucks for a complete platform to support a MCU isn't bad at all. I ordered like 10 of them to just keep around for instances like this. I personally think CrossWorks is the best MSP430 dev environment around, followed by IAR and Eclipse and lastly mspgcc. There are some benefits to mspgcc, and the coding/compiling is fine (i use gcc a ton) I just hate the debugging. Eclipse is a bloated nightmare. IAR is clunky. But CrossWorks? Just right. Some initial googling indicates there may be launchpad support in CrossWorks but I've never used it for that. They have a free 30-day demo, and it's not terribly expensive if you aren't developing commercially (and maybe you are, but even then I don't think it's terribly expensive for a professional tool.)
|
|
# ? May 11, 2012 20:00 |
|
I use MacVIM + mspgcc on Lion exclusively for Launchpad development. It works really well and if you're already familiar with the GCC toolchain and comfortable debugging with GDB it'll be right up your alley
|
# ? May 12, 2012 04:25 |
|
Hey guys, I'm just trying to learn electronics poo poo (actually officially 'started' about a week ago). I'm learning about the basic parts and still trying to understand and find out the math but I realized something I wanted to make. I play Magic the nerdenning and in it I have to carry around a big thing of dice because you have to pretty much, and it can be a pain finding the sides on a dice and remembering which dice were moved and whatnot. I wanted to try to make some small seven or fourteen segment displays I could use as counters to make it easy to read with three buttons and a single pole single throw for the on/off switch. I also realize that normal throughhole technology may end up bigger than the cards at 63 x 88 mm (2 15/32 by 3 7/16 inches) so I wanted to also think about surface mount as an option powered by a coincell, and hoping that I could get this smaller than a miniature milk duds box (1 inch by 1 inch by 1 1/2 inch) I'd also like to start with a prototype that can simply act as a life counter for the game (fourteen segment display same as the above pretty much) 0. How crazy am I 1. Where can I find a sample counter seven (or fourteen) segment display schematic. 2. Where can I find surface mount technology that doesn't want me to buy ten thousand seven segment displays 3. Am I going to have to manufacture my own SMD boards or is there some trick I'm missing or am I just going to go with nothing but prototype boards
|
# ? May 14, 2012 06:54 |
|
Forer posted:Hey guys, I'm just trying to learn electronics poo poo (actually officially 'started' about a week ago). I'm learning about the basic parts and still trying to understand and find out the math but I realized something I wanted to make. I play Magic the nerdenning and in it I have to carry around a big thing of dice because you have to pretty much, and it can be a pain finding the sides on a dice and remembering which dice were moved and whatnot. 0. Maybe a little, but I'm not sure it has anything to do with this project. 1. I don't have a schematic for you, but I bet you can find one with some googling. A 7-segment display is just a set of LEDs with a common anode or cathode. If you look at a datasheet for one, there is a schematic showing what is inside. You will need to decide if you want to use a microcontroller or not. If you use a microcontroller, pick one with powerful enough outputs and you can just drive the 7-seg directly (though current limiting resistors). If you want to avoid the microcontroller (probably a good idea for the purpose of learning digital electronics), you will probably want to use a BCD to 7 segment decoder IC plus an up/down counter IC. Start by making a counter that counts from 0-9 with a single digit. After that is working, figure out how to extend that to two digits. You should read datasheets like crazy, put together a block diagram and then post back with more specific questions. 2. Digikey, Mouser, or Newark, or you can order direct from Kingbright 3. I would actually not use surface mount for this. You don't gain much by going SMT with the 7-segment displays, other than there aren't pins sticking through the board. The SMT displays are actually larger than the TH displays because there is additional space for the pins on the outside of the part. On the TH parts, the pins are under the display. I think you will be able to hit your target size with through hole. With TH, you can wire-wrap this thing and it is simple enough it won't drive you completely crazy. You will be able to do it on a breadboard and then use the same components when you move to perfboard or a real PCB. That said, there are some super-cheap proto PCB places you could use (look back in this thread) and get some boards for not too much money. I would just caution against adding the complication of SMT before you need to. If you go to school, you might have access to a PCB milling machine.
|
# ? May 14, 2012 15:53 |
taqueso posted:3. I would actually not use surface mount for this. You don't gain much by going SMT with the 7-segment displays, other than there aren't pins sticking through the board. The SMT displays are actually larger than the TH displays because there is additional space for the pins on the outside of the part. On the TH parts, the pins are under the display. I think you will be able to hit your target size with through hole. You wouldn't save much space going SMD with the displays, but you may with the other parts. I would recommend getting a breadboard and a list of parts you'll think may be handy (look around and ask questions here to get a good shopping list) and experimenting to see if you can get it to work. If you can get it to work on a breadboard, you'll have some insight on whether you want to just solder it up on perfboard, switch to SMD, or throw it all away and never think about it again.
|
|
# ? May 14, 2012 16:42 |
|
Forer posted:I wanted to try to make some small seven or fourteen segment displays I could use as counters to make it easy to read with three buttons and a single pole single throw for the on/off switch. Someone asked almost the same question a while back and I pointed them to some posts on my ammunition counter project here and here. If you scroll down far enough I went from using DIP chips and perfboard to an SMD microcontroller and prototype PCBs, so you can see the size difference. If you use BatchPCB as a protoyping service small boards can be very cheap; ~5 bucks shipped each for 2" x 2" 2-layer boards. I don't think you'll be able to use a coin cell as their maximum continuous current draw is pretty low; I would suggest just two AA batteries, or a single 3V CR123 lithium battery if it needs to be smaller.
|
# ? May 14, 2012 18:40 |
|
Quick question, does the 3db loss per adapter rule of thumb hold up for audio connections?
|
# ? May 14, 2012 18:50 |
|
Pudgygiant posted:Quick question, does the 3db loss per adapter rule of thumb hold up for audio connections? No, you won't see anywhere near that kind of loss at audio frequency. Contact resistance will make up the entirety of loss. Hell, I've never even seen a cheapo microwave adapter with that kind of loss.
|
# ? May 14, 2012 18:56 |
|
sixide posted:No, you won't see anywhere near that kind of loss at audio frequency. Contact resistance will make up the entirety of loss. Hell, I've never even seen a cheapo microwave adapter with that kind of loss. That's what I was taught for both satcom waveguide and networking stuff like punch blocks, wasn't sure if it held true for actual electronics.
|
# ? May 14, 2012 19:10 |
|
Forer posted:Hey guys, I'm just trying to learn electronics poo poo (actually officially 'started' about a week ago). I'm learning about the basic parts and still trying to understand and find out the math but I realized something I wanted to make. I play Magic the nerdenning and in it I have to carry around a big thing of dice because you have to pretty much, and it can be a pain finding the sides on a dice and remembering which dice were moved and whatnot. I think I'd made this suggestion before, but consider modifying one of these guys instead of making something from scratch. You'd need to write new firmware (and wire up the connections for an ISP), and you could hook microswitches into I/O header. If you really want to go full-custom, then you can still use the eagle files as a starting place. You'd just need to extend the width of the board, and put your switches below the 7 segment, and then put the coin cell on the back. The sparkfun / opencircuits eagle parts library should have everything you need. The only word of caution is that the sparkfun device doesn't use current limiting resistors on the LEDs, which you might want to use.
|
# ? May 14, 2012 19:43 |
|
sixide posted:No, you won't see anywhere near that kind of loss at audio frequency. Contact resistance will make up the entirety of loss. Hell, I've never even seen a cheapo microwave adapter with that kind of loss. It's in the ballpark for 2.4ghz stuff
|
# ? May 14, 2012 20:53 |
|
3 dB is also the rule of thumb I was taught for connectors when calculating TV-distribution and that's only around 400 MHz. For audio signals you'd have almost exclusively ohmic losses, I suppose for really long lines with low impedance termination it *might* become relevant to calculate losses.
|
# ? May 14, 2012 22:25 |
|
What I was getting at was, a lot of people on head-fi recable their headphones with a female minijack on the cup so the cable is detachable and easily replaced. It got me curious how much, if any, loss that introduced into the system. edit And when you're working with something like an Andrews dish with no ODU, that 3db adds up quick on a 30ft waveguide with multiple turns. A 100W SSPA for a 4W signal isn't unheard of. Pudgygiant fucked around with this message at 22:45 on May 14, 2012 |
# ? May 14, 2012 22:42 |
|
Otto Skorzeny posted:It's in the ballpark for 2.4ghz stuff I'm having trouble imagining 3dB of loss at such low frequencies. Hell, worn out BNCs are better than that. Headphones are typically "high impedance" (relative to speakers). A typical 500mOhms (~250 mOhm signal and return) of contact resistance will be pretty minor. longview posted:3 dB is also the rule of thumb I was taught for connectors when calculating TV-distribution and that's only around 400 MHz. F connectors aren't so hot, so I can believe that as an upper bound. Better than that garbage connector they use in Europe, but still pretty bad. sixide fucked around with this message at 22:55 on May 14, 2012 |
# ? May 14, 2012 22:52 |
|
Back when I was a kid we had a Tandy TRS-80 Color Computer and it had a neat little game where you went around from room to room using diodes and resistors and whatnot to solve puzzles (or something...the details are kind of hazy) all as an excuse to learn about electronics. Does anything like that exist out there today for the PC?
|
# ? May 14, 2012 23:08 |
|
Thermopyle posted:Back when I was a kid we had a Tandy TRS-80 Color Computer and it had a neat little game where you went around from room to room using diodes and resistors and whatnot to solve puzzles (or something...the details are kind of hazy) all as an excuse to learn about electronics. Was it Rocky's Boots? That was a great game. AFAIK there is nothing similar today.
|
# ? May 14, 2012 23:11 |
|
taqueso posted:Was it Rocky's Boots? That was a great game. AFAIK there is nothing similar today. Yes! That was great and it's a shame if nothing similar exists today.
|
# ? May 15, 2012 01:25 |
longview posted:3 dB is also the rule of thumb I was taught for connectors when calculating TV-distribution and that's only around 400 MHz. That's insane. My VNA at work registers about -1.5dB of loss at 1.5GHz when you hook the two ports directly to each other with BNC coax (and that's including N-to-BNC adapters, so four connections total). Are you guys using steel connectors or something? Thermopyle posted:Back when I was a kid we had a Tandy TRS-80 Color Computer and it had a neat little game where you went around from room to room using diodes and resistors and whatnot to solve puzzles (or something...the details are kind of hazy) all as an excuse to learn about electronics. ANIME AKBAR fucked around with this message at 01:43 on May 15, 2012 |
|
# ? May 15, 2012 01:34 |
|
sixide posted:F connectors aren't so hot, so I can believe that as an upper bound. Better than that garbage connector they use in Europe, but still pretty bad. We typically used F connectors (this is in Norway) that you just screw on, more recent installs use a kind you crimp on that at least stays on the drat cable when you pull on it. The 3 dB rule is a worst-on-worst number I think, because it would really suck to get to a customer and find you're a few dB short of a good signal because you used too low numbers on the interconnections. Keep in mind we're talking sub mV signal levels (typical 80 dBµV max IIRC).
|
# ? May 15, 2012 08:58 |
longview posted:3 dB is also the rule of thumb I was taught for connectors when calculating TV-distribution and that's only around 400 MHz. I always thought the 3 dB rule was because you were using a splitter, and dividing the power in half. Am I way off base? I've actually never thought of it beyond that.
|
|
# ? May 15, 2012 09:05 |
|
I am conscious and wary of the deadly effects of high voltage and will exercise all proper work habits when touching this thing. OK beards, dig deep. I picked up a mid-sixties all tube AM/FM/record player stereo console from the flea market today. AM is fine and strong, and the turntable, as skate and skiptastic as it is, sounds great, but we have really low volume on FM. If you dime the volume, you get a 50/50 mix of hum and music at very low volume, like, if the air conditioner kicks in it's gone. Output is a pair of 6L6's so there should be excessive power available, and Rush Limbaugh rattled the windows earlier, so the finals are good. We're tuning around fine, the sensitivity is okay and it's picking up the stations I expect it to. What would you look at? FM side tubes? Coupling caps? Switch? Audio tracing tube radios sketches me out. I'm assuming I should put a billion volts of blocking cap in line with any audio tracer. I don't think it's a bad switch contact, or any of the FM front end or IF tubes, but I could be wrong. It's a JC Penney "Two Sixty" and I'm sure it was just a commodity item, so I am not expecting a schematic, but there is a tube diagram on the chassis, so I at least have a toehold. Jonny 290 fucked around with this message at 23:37 on May 15, 2012 |
# ? May 15, 2012 23:34 |
|
Today I wired up a piezo to electrify an acoustic guitar. Every diagram I could find shows to wire them the say way so that's how I did it but the off doesn't work on my pot and turning the knob towards off turns the volume up instead of down. Here is the diagram I followed: Here is my pot: (my soldering is messy I know) http://www.radioshack.com/product/index.jsp?productId=2062298&filterName=Type&filterValue=Potentiometers Blue wire on left #3 goes to red on piezo. Blue wire in center #2 goes to positive on jack. Brown wire #1 goes to negative on jack and black on piezo. Do I just need to switch the blue wires?
|
# ? May 16, 2012 01:16 |
wormil posted:Today I wired up a piezo to electrify an acoustic guitar. Every diagram I could find shows to wire them the say way so that's how I did it but the off doesn't work on my pot and turning the knob towards off turns the volume up instead of down. Just swap the two outside wires. Leave the middle lug as is; this will change pots operation in relation to clockwise vs. counterclockwise. Is it a linear taper or audio (log) taper pot? EDIT: Had some trouble parsing your post, but did you say it was an on/off pot too? I think the two lugs on the back of the pot are the on/off switch posts. Do you have a multimeter to check continuity through them? If that's the case, you just need to pick a signal (middle lug would be my choice) and connect it to one of the lugs on the back and the other lug to the middle position so when the switch opens, it gets disconnected. Delta-Wye fucked around with this message at 01:58 on May 16, 2012 |
|
# ? May 16, 2012 01:56 |
|
Delta-Wye posted:Just swap the two outside wires. Leave the middle lug as is; this will change pots operation in relation to clockwise vs. counterclockwise. Yeah, it's an on/off volume pot. Linear I believe. Not many details on the package: 10K-Ohm Audio Control Potentiometer with SPST Switch Just found this reference quote:Wiring details: Okay, I think I'm getting it. Switch the outside wires and move the #2 (middle) to lower right lug on the back? I found this pic on Instructables using the same pot, also wired to a piezo and jack. Although he has his ground wired the same as mine. But I also found this in the comments: quote:Great info, but the photo of the wiring is incorrect. The side mount should actually be the middle mount on the bottom with the others. Not sure if that makes sense. Link to photo and comments: http://www.instructables.com/id/Electric-Cigar-Box-Guitar/
|
# ? May 16, 2012 04:04 |
|
wormil posted:Yeah, it's an on/off volume pot. Linear I believe. Not many details on the package: 10K-Ohm Audio Control Potentiometer with SPST Switch "Audio" usually indicates "audio" (log) taper.
|
# ? May 16, 2012 04:47 |
|
hey guys, I'm looking through how to do that counter idea and I'm looking for the IC's and voltage stuff I'd need. I'm trying to think about what I'm powering it with and the only thing I can think of is a stack of two cr2032's to keep it small, and do I just need a 4033 for the IC and a seven segment display along with some resistors, an spst switch and wiring?
|
# ? May 16, 2012 11:34 |
|
I have some more issues with solid state relays. Right now, I have a 12 VDC, 13 watt solenoid valve being controlled by a DC solid state relay on a controller board. Valve: http://www.coleparmer.com/Product/Solenoid_Operated_Two_Way_Pinch_Valve_Normally_Closed_10_mm_Tube_OD_12_VDC/EW-98305-10#Tech Relay (MPDCD3): http://www.crydom.com/en/Products/Catalog/m_p.pdf Controller: http://www.controlanything.com/Relay/Device/SSR2x I have ten of these setups, and several of them have the issue in which the valve does not shut off when the relay is deactivated. Recently I had the same issue with AC solid state relays, but I eventually figured out it was because my device required less current than the leakage current of the relay. However the specs for the DC relay don't seem to list a leakage current, and even if it did, the valve needs about 0.9 A to run which seems to me like it would be way above an leakage current. I tried treating the issue like a leakage current problem though by hooking up an LED with a 1 A current driver in parallel with my solenoid valve so that both would turn on when the relay was activate, and this actually worked. Both the valve and the LED turned off as they should when I deactivated the relay, whereas if only the valve was connected it would stay on. The manufacturer of the controller suggests using a snubber circuit if I'm experiencing the problem of devices staying on, so is what I did basically a snubber circuit? I've been reading up about them but I'm not sure I understand them yet. I'm still trying to wrap my head around solid state relays and inductive loads, so hopefully someone can explain this to me. Edit: After doing a little more research, it sounds like what I want is a unidirectional TVS diode wired in parallel with my valve but in the opposite direction. Does that sound right? If so, based on the voltage/amperage requirements of the valve (12 VDC, 0.9 A), what would be the breakdown voltage and reverse standoff voltage I'd want to aim for in the TVS diode? Hidden Under a Hat fucked around with this message at 14:39 on May 16, 2012 |
# ? May 16, 2012 14:13 |
wormil posted:Yeah, it's an on/off volume pot. Linear I believe. Not many details on the package: 10K-Ohm Audio Control Potentiometer with SPST Switch If it is an audio pot, switching the wires will change the characteristics, but I suspect it's already backwards and you'll be fixing it. If the potentiometer is a SPST, I think the two poles on the back will be connected together if it's 'on' (closed) and will be disconnected if it's 'off' (open). For a normal non-switched pot, you would connect to the middle lug directly for your output, but by going middle lug->sw1 and then sw2->output, you can break the circuit by putting the pot in the off position. FYI http://sound.westhost.com/pots.htm check out figure 4. The brain doesn't recoginize volume changes on a linear scale, so you need a logarithmic scale in volume applications. A linear pot doesn't care which order you connect it in (it has the same relative sweep in both directions) but a audio will be different. Try it and see though, can't hurt anything. Just keep an
|
|
# ? May 16, 2012 15:45 |
|
Theory question before I get started on this. I'm gonna try and roll my own power over ethernet in my house, mostly just for the learning experience. My ap needs 5 volts and pulls 1.5amps, then I have an little controller(havent picked it out yet, maybe just an arduino or something) that we'll assume needs 3.3v at .5 amps, ethernet cable is cat6 and 20ft. I'm thinking of sending something like 10v+ over the line just using a old wall wart that has enough amperage, then putting a 5 volt and 3.3 volt regulator on the other end for the two devices. Is there anything I should look out for on this? With this sort of voltage, would I have to worry about interference or anything?
|
# ? May 17, 2012 19:52 |
|
I would recommend going for a much higher voltage. Typical UTP is ~24 AWG, low voltage distribution is not advisable with that much resistance. Typical PoE is more than 40V, though I don't know much about it.
|
# ? May 17, 2012 20:01 |
|
What would I see? Heating the wire or unreliable voltages? I'm kinda curious about it from an academic standpoint and not just getting this one instance to work.
|
# ? May 17, 2012 20:13 |
bobua posted:What would I see? Heating the wire or unreliable voltages? I'm kinda curious about it from an academic standpoint and not just getting this one instance to work. Your distribution wire and load form a voltage divider, such as this: where R1 is the ethernet wire and R2 is the load. Looking at a standard wire gauge chart, http://www.powerstream.com/Wire_Size.htm, 24 AWG wire is approximately 25.7 ohms per 1000ft. You mentioned 20ft, and you will have a return path for a total of 40 ft or about 1 ohm. At 1.5A, you will drop 1.5V across the length of the ethernet (just in resistive losses), and waste about 2.25W. Power dissipated in a wire is related to current. P = I V V = I R Therefore, P = I^2 R If you need to deliver power to the load (for your AP, P = 5 * 1.5 = 7.5 W) you are much better transferring the power with higher voltage and less current. For instance, if you wanted to deliver 7.5W @ 40V, you would only be pulling ~200mA through the transmission line. If you had a very efficient (and 95%+ DC-DC converters are available) you would take in whatever your input voltage is, step it up on the injector, and step it down before putting it into your AP. Redoing the losses calculation from earlier, 1 ohm w/ 200mA results in a voltage drop of .2V and a power loss of .4W
|
|
# ? May 17, 2012 21:32 |
|
Amazing post. Never even thought about it as a voltage divider, and wouldn't have thought about the return path even if I did. Thanks a lot.
|
# ? May 18, 2012 04:52 |
|
sixide posted:"Audio" usually indicates "audio" (log) taper. Delta-Wye posted:FYI http://sound.westhost.com/pots.htm check out figure 4. The brain doesn't recoginize volume changes on a linear scale, so you need a logarithmic scale in volume applications. A linear pot doesn't care which order you connect it in (it has the same relative sweep in both directions) but a audio will be different. Try it and see though, can't hurt anything. Just keep an Thanks guys for all the help. I went wrong in two ways, first the plan I was following assumed I would be using a 500k linear pot but I bought a 10k taper. Second, I should have busted out a multimeter and put more brainwork into this but I was caught up in the other parts of the project. Based on your link and some other reading, I'm wondering if I wouldn't have been better off with a linear pot. Also I'm wondering how the 10k pot will affect the sound, based on this article: http://www.premierguitar.com/issue/webexclusive/200611_web_guitarpots.asp Less resistance mean less treble. Guess I will have to try it and see. I can always change it out.
|
# ? May 18, 2012 06:26 |
|
|
# ? May 8, 2024 07:02 |
bobua posted:Amazing post. Never even thought about it as a voltage divider, and wouldn't have thought about the return path even if I did. A few more details: The circuit is actually a 3 resistor divider: +V path, load, return path. I just added R1 and R3 together because 1) that was the picture I found and 2) it doesn't change the calculations. In reality if you were dropping 1.5V you would be dropping .75V before and after the AP. From the AP's viewpoint, it sees (Vin - drop) - (Gnd - drop) or Vin - 2drop, which is the same as just doing it as I did previously. If you were doing Vin = 10V, it would be roughly 10-1.5 = 8.5V or so. All theoretical calculations of course, but it doesn't paint a pretty picture. There is no reason that couldn't be regulated down to a stable 5V (hell, I suspect this idea would work even as I advise against it), but that regulation step will add additional inefficiencies. We're at 8.5V being delivered to the load, and need to regulate it down to 5V. If you use a linear regulator (LM317, 7805, etc) it will drop additional voltage to maintain a steady input to your AP; think of it as an additional resistor, whose value changes to maintain a constant voltage drop across the output. Unfortunately, that means it burns power according to Ohm's Law (8.5V-5V)*1.5A = 5.25W. To check the efficiency of the system take a ratio of useful power to total power (% = Pdelivered/Psource). Pdelivered = 5V * 1.5A = 7.5W. Psource = 10V * 1.5A = 15W. As a sanity check, 2.25W in the wire and 5.25W in the regulator = 7.5W wasted, wasted + delivered or 7.5+7.5=15W. This PoE system has an efficiency of 50%. To answer your direct questions: quote:What would I see? Heating the wire or unreliable voltages? I'm kinda curious about it from an academic standpoint and not just getting this one instance to work. You may also see unreliable voltages; I've been modeling all of this as purely resistive load at 1.5A max; your AP will draw varying amounts of current, largely dependent on what the radio is doing at that moment. If the radio goes from idle to burst transmissions, it is suddenly going to draw a lot of current. The wire isn't going to be purely resistive, but have some sort of inductance as well. The inductance will cause voltage drops as the current changes rapidly and the inductance resists the sudden change in current; some capacitors on the load end should filter this out, but you may see some fluctuation in output voltages anyways. I've been kind of getting a kick out of this, so I thought I would do some 'high' voltage (it's high for the sort of digital work I do, but most electricians would scoff at the idea that 40V is high ) calculations using the standard PoE voltage levels. Wikipedia lists a common voltage level as ~44V. There are a bunch of unknowns, but we'll assume we can supply that. There are dedicated devices, but I will be doing this with generic step up/step down parts instead of specific PoE devices (just for show, if I was really building this...). 10V in, expected load of 1.5A@5V. First, step 10V up to 44V using a MAX863. Transmit through cable, Step ~44 down to 5V using a MAX5090. Working backwards from the source, the MAX5090 has an expected efficiency of ~88%. This means for the 7.5W at the load, we will need 8.5W at the input to the step-down stage. 8.5W/44V = 0.193A. Pulling 0.193A through 1ohm worth of cable drops .2 V and burns 40mW of energy. So far so good! Now the step-up stage, which needs to provide 8.5W. The datasheet for the Max863 indicates somewhere in the ballpark of 90% efficiency, which means it needs 8.5/.9 = ~9.5W of input power. At 10V, that is around 950mA. % = Pdelivered/Psource = 7.5 / 9.5 = 79% efficient. Seems better Delta-Wye fucked around with this message at 06:50 on May 18, 2012 |
|
# ? May 18, 2012 06:48 |