If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below. |
|
|
Thread Tools | Display Modes |
#21
|
|||
|
|||
green led
T. Ment wrote:
The cathode of the LED is connected to ground, so I will see 0 volts there. The LED needs 2.2v to light up, so the anode will measure 2.2v, Without the resistor, which is definitely required in this situation, the anode will measure 2.8V. The 2-pin header on the mobo provides 5V. |\ | 5V ----------| |------- GND : : |/ | : : : : : : ^^^^ : 5V 5V 2.2V 0V *drop* From the eBay auction page, there is no mention of an internal resistor inside the LED packaging to let that part work with 5V on its anode. If the 5V supply side could output an unlimited current (which it obviously cannot), there is no drop on the wire between the 5V supply to the anode side of the LED. You are running the LED *way too hot*. I'm surprised it hasn't already burned out. You're putting 5V across the LED, not 2.2V. If you bought more than one, toss the old one and start anew with an unused one. Voltage regulation cannot handle all that load, so the supply will drop and possibly damage parts there. With the 2.2V drop for the LED, you *must* have a resistor between it and the 5V supply. |\ | 5V -----^v^v-------| |------- GND : : |/ | : : : : : ^^^^ : ^^^^ : 5V 2.8V 2.2V 2.2V 0V *drop* *drop* ^^^^^^^^^^^^^^^ 5V *drop* The eBay auctions list 20 mA for the LEDs current. I'd stick with that for now. You can use alligator clips to temporarily hold the resistor in place between the anode lead of the LED to the PWR LED wire going to the 5V mobo header until you get the right current for the brightness level you want. Starting with the 20 mA spec for LED current, the resistor has to be: 2.8V / 20mA = 2.8V / 0.02A = 140 ohms For 5% resistors, you'd have to choose between 130 and 150 ohms. Since you want it to be less bright (but remember you're currently over-amp'ing the LED by putting 5V directly on its anode), go with the 150 ohm resistor so there's more of a voltage drop on the resistor meaning less remaining voltage across the LED which means less current through the LED. Less current through the LED means less brightness. It doesn't matter which side of the LED the resistor is inserted in the circuit. https://en.wikipedia.org/wiki/LED_circuit "An LED has a voltage drop specified at the intended operating current" For starters, I wouldn't wander too far from the "intended" load of 20mA. The LED you bought says foward voltage is 2 to 2.2 V. The above figured a 2.2 V drop across the LED, but going down to 2 V by increasing the resistor (so it drops 3V at 20mA, or a 150 ohm resistor) means less current through the LED for less brightness. If you put that LED into a 5V circuit with no drop-down resistor, you are running it at 5V (or whatever the supply can manage to provide), not at 2.2V. The high brightness you don't want is because you're frying the LED. With a drop-down resistor and running the LED with 20mA will likely get what you feel more comfortable for brightness. Trying to go down to 4mA through the LED probably means not enough to overcome the breakover voltage needed to get the diode to work and then also illuminate the LED. Once the breakover voltage is reached, the resistance of the diode severely drops, so little changes in additional voltage change the current a lot. From the wikipedia article: The voltage versus current characteristics of an LED are similar to any diode. Current is approximately an exponential function of voltage according to the Shockley diode equation, and a small voltage change may result in a large change in current. If the voltage is below or equal to the threshold no current flows and the result is an unlit \ LED. If the voltage is too high the current exceeds the maximum rating, overheating and potentially destroying the LED. If the 2-pin LED mobo header connects to a voltage regulator that provides a constant current, maybe your old LED is okay. I haven't looked at the mobo circuit diagrams to see what the LED mobo header connects to. For the old LED that failed, was there a resistor for that one? I don't remember ever seeing an inline resistor in the 2-wire cable from the mobo header to the LED. It is possible, the old LED has an internal resistor, but I doubt it. If the mobo didn't use a current-limiting voltage regulator for the power LED, I'm sure what you would've seen when you applied power was a fleeting flicker of the LED as it lit and then burned out. https://os.mbed.com/users/4180_1/notebook/rgb-leds/ Even in the eBay auction, the current remains the same but the breakover voltage changes (higher for blue and white). With less voltage across the drop-down resistor but the same 20mA current, ther resistor would need to be smaller. For their blue or white LED with a 3V breakover voltage and 20 mA going through it, you'd need 3V / 0.02A = 150 ohm resistor. Yes, you could try reducing the current to reduce brightness but the breakover voltage for the LED doesn't change, and with less current through the LED then the higher the resistor to maintain its same voltage drop. Get too far from the 20mA spec and it's likely the LED will fail to light up. Even if the breakover voltage were applied, your 4mA target current seems just way too tiny for the LED to work. The breakover voltage might be maintained across the LED but remember it's heat that is lighting up the LED, and insufficient current means insufficient heat, and the brightness change with current change could be exponential, so wandering too far from 20mA through the LED could mean the LED would quickly become too dim to of any value as an indicator of power to the computer (at which point you might as well not even bother attaching the LED as a power indicator). That's also the output side of the resistor. The supply side is 5v. In my circuit. those two will never change, no matter what resistor value I choose. The only thing that can change is the current flowing through the LED. So I don't understand why they are talking about resistor voltage drop above. That seems irrelevant. Or maybe I still don't get it. Rather than go through all the testing of inline or drop-down resistors and possibly causing damage to the current-limited voltage regulation on the mobo, or frying an LED, or both, why not just go buy a dimmer LED? The eBay auction you used doesn't list the brightness (at the spec'ed current and for *slight* variations of voltage across it beyond the breakover voltage). Also, it might only appear bright because you're close to the LED and it was designed for a wide dispersion angle instead of more straight ahead. I remember using a case holder for an LED where you push the LED into the inner shell that has a front-side flare (that you would normally put first into the hole and then slide in the LED) and a ferrule that slide over the inner case that pinched the LED (they have a flare end) and wedge the two halves into the case hole. Looked similar to: http://tinyurl.com/yxghsrgh As shown in the lower right inset, the LED stuck out too much. I took a short black ferrule that had the same inside diameter as the diameter as the LED and glued it to the LED case holder that was on the outside of the case. The LED was no longer shining into my eyes from the light at its sides. It might still light up the room too much and bother you when the room was dark, like when your sleeping, but then you should be closing your eyelids when sleeping. Might not work for your Power LED, though, as those are usually on a PCB that sits behind the case with a plastic lightpipe that brings the light to shine out a hole in the case (and why I suggested using something to diffuse the light, like self-stick rubber/vinyl feet that you can buy at the local hardware store, or a recessed black rubber bumper to glue on the front to keep the LED shining more straight out. https://images.homedepot-static.com/...67-64_1000.jpg https://www.redco1.com/images/site/A...r_bumper_1.jpg If there is a clear lightpipe from a PCB where is the LED, that brings the light to a hole in the case. See if you can remove that piece of plastic off the PCB. Then the LED will be farther behind the hole scattering its light everywhere inside (so don't bother if you have side panels with plastic windows). I suspect the voltage from the LED mobo header is a constant 5V. It is possible to effect a dimming of an LED by pulsing it; i.e., changing its duty cycle. At 100% duty (constant high voltage), it is brightest. At a 50% duty cycle, it would be dimmer. Even dimmer at lower duty cycles. The cycle length (Hz) is so high that you'll never notice the flickering even at low duty cycles. These are called PWM dimmers. https://www.newegg.com/p/1Z5-00NP-01...1337-_-Product That looks small enough that you could apply some double-sided sticky foam tape to the backside of the PCB (to insulate it) and glue it somewhere inside the case where you could reach it. Attach the Power LED to the LED+ and LED- pins and the + and - pins to the 2-wire cable that goes to the mobo's LED header. Then touch the foil to alter the duty cycle to change the dimming on the LED. What I can't tell from the product's description if it works on 5V or 12V input, or handles a range. The jumpers probably take care of selecting input voltage and output current. If it's for 12V input, like for RGB lighting strips, well, hook it a 12V pin on Molex connector from the PSU instead of the mobo's PWR LED header. After all, the 12V lines on the PSU go dead when the computer powers down (it's the 3.3V line that stays up for soft-power control). |
#22
|
|||
|
|||
green led
On Thu, 27 Jun 2019 13:05:02 -0500, VanguardLH wrote:
If the 2-pin LED mobo header connects to a voltage regulator that provides a constant current, maybe your old LED is okay. Doesn't matter. I got a pack of 25. I can start over with a new one. I haven't looked at the mobo circuit diagrams to see what the LED mobo header connects to. For the old LED that failed, was there a resistor for that one? The original case LED is identical in appearance to the ones I bought. They both have the two legs, one longer. There is no resistor in the connecting wires. They both insert in a small plastic housing the same way. And then you push the little connectors on the ends of the wires inside the little housing to make contact with the LED legs. It's designed so the LED is easy to replace. The new one is just too bright. I have a second identical case, the two sitting side by side. In one case the original LED still works. It's not a matter of viewing angle. The replacement is LED much brighter than the original. It drowns out the HDD LED below it. You say a resistor too high will drop the voltage too low for the LED to light. But I thought Paul suggested 4mA. Much of the discussion is over my head, so I will just have to try some resistors and see what happens. But that may take some time, I have none handy right now. |
#23
|
|||
|
|||
green led
T. Ment wrote:
On Thu, 27 Jun 2019 13:05:02 -0500, VanguardLH wrote: If the 2-pin LED mobo header connects to a voltage regulator that provides a constant current, maybe your old LED is okay. Doesn't matter. I got a pack of 25. I can start over with a new one. I haven't looked at the mobo circuit diagrams to see what the LED mobo header connects to. For the old LED that failed, was there a resistor for that one? The original case LED is identical in appearance to the ones I bought. They both have the two legs, one longer. There is no resistor in the connecting wires. They both insert in a small plastic housing the same way. And then you push the little connectors on the ends of the wires inside the little housing to make contact with the LED legs. It's designed so the LED is easy to replace. The new one is just too bright. I have a second identical case, the two sitting side by side. In one case the original LED still works. It's not a matter of viewing angle. The replacement is LED much brighter than the original. It drowns out the HDD LED below it. You say a resistor too high will drop the voltage too low for the LED to light. But I thought Paul suggested 4mA. Much of the discussion is over my head, so I will just have to try some resistors and see what happens. But that may take some time, I have none handy right now. At 4mA, the LED will still light. Those LEDs will light at 1mA! The voltage will still be at the 2V range. Even if you dropped to 100uA, there will still be around 1.9V or so. But you would need to darken the room, to see a dull green glow. ******* On high power LEDs, running them at extremely low current flows, causes the color to shift. The LEDs "look unhealthy" in such cases. And the manufacturer mentions this, as a limit to running them too low. It's a cosmetic issue. For color purity on white LEDs, it doesn't pay to go too low. The color comes from the phosphor on a white LED, which is part of the problem. For high power LEDs, you make sure to buy the right LED for the application. (Don't buy a 200W array when a 3W array is the right one.) That should not be an issue in this case. Green, red, and yellow indicator LEDs are "pure" GaAs devices doped with particular things, and no phosphor is involved. On the "pink" LEDs, those are phosphor based. The white LEDs use phosphor too. The "original colors", before phosphors came along, are pretty good (200-300 nm "line width", they're not "lasers"). ******* I built a logic probe once, with darlington transistor pairs for high impedance. And the LEDs on that ran at 1mA to save power. The display of ones and zeros was plenty bright for home lab work. Paul |
#24
|
|||
|
|||
green led
On Thu, 27 Jun 2019 15:47:44 -0400, Paul wrote:
At 4mA, the LED will still light. I don't know if my motherboard has a resistor, or what value. And unless you care about fire, wild guesses are more fun than math. 1k ohm sounds about right to me. That should be an easy one to find. |
#25
|
|||
|
|||
green led
T. Ment wrote:
On Thu, 27 Jun 2019 15:47:44 -0400, Paul wrote: At 4mA, the LED will still light. I don't know if my motherboard has a resistor, or what value. And unless you care about fire, wild guesses are more fun than math. 1k ohm sounds about right to me. That should be an easy one to find. You know it has a resistor. A resistor is the cheapest and most appropriate thing to add. Some companies *do* add over-the-top LED driving circuits, with a ton of analog devices in the path. They do this to control RF emissions from where the LED pokes through the faceplate on equipment. The computer industry is not burdened by such foolishness :-) You got your resistor. It cost a nickel. End of story. That's how the thinking goes. Everything is cost driven. Paul |
#26
|
|||
|
|||
green led
On Wed, 26 Jun 2019 16:25:17 -0400, Paul wrote:
And this is the way to buy resistors. An "assortment pack" gives resistors at from $0.03 to $0.05 a piece. This one happens to be sold out, but it's just an illustration. https://www.radioshack.com/collectio...nt=20332255301 You will find 30 each of 1k, 10k, and 100k; ten each of 1, 10, 100, 120, 150, 220, 330, 470, 560, 1.5k, 2.2k, 3.3k, 4.7k, 15k, 22k, 47k, 220k, 470k, 1meg, and 10meg; Five each of 2.2, 15, 22, 33, 39, 51, 68, 82, 180, 270, 390, 510, 680, 820, 1.2k, 1.8k, 2.7k, 3k, 3.9k, 5.1k, 5.6k, 6.8k, 8.2k, 12k, 18k, 27k, 33k, 39k, 51k, 56k, 68k, 82k, 120k, 150k, 180k, 270k, 330k, 1.5meg, 2.2meg, 3.3meg, and 4.7meg. Notice I'm getting 180, 680, 820, the usual kinds of standard values. $15 is more than I want to spend right now. Ebay has cheap assortments, if you can wait 4-6 weeks for shipping from China. USA assortment sellers are pricey, like Radio Shack. But I found 1/4 watt packs, choose your ohms, shipping included, cheap: https://www.ebay.com/itm/1-4W-25-Wat...SJJa BvQvMz9g $1.55 is my kind of purchase. If 1k ohms is the wrong value, it won't cost too much to buy different ohms. Even three wild guesses are less than $5. |
#27
|
|||
|
|||
green led
T. Ment wrote:
On Wed, 26 Jun 2019 16:25:17 -0400, Paul wrote: And this is the way to buy resistors. An "assortment pack" gives resistors at from $0.03 to $0.05 a piece. This one happens to be sold out, but it's just an illustration. https://www.radioshack.com/collectio...nt=20332255301 You will find 30 each of 1k, 10k, and 100k; ten each of 1, 10, 100, 120, 150, 220, 330, 470, 560, 1.5k, 2.2k, 3.3k, 4.7k, 15k, 22k, 47k, 220k, 470k, 1meg, and 10meg; Five each of 2.2, 15, 22, 33, 39, 51, 68, 82, 180, 270, 390, 510, 680, 820, 1.2k, 1.8k, 2.7k, 3k, 3.9k, 5.1k, 5.6k, 6.8k, 8.2k, 12k, 18k, 27k, 33k, 39k, 51k, 56k, 68k, 82k, 120k, 150k, 180k, 270k, 330k, 1.5meg, 2.2meg, 3.3meg, and 4.7meg. Notice I'm getting 180, 680, 820, the usual kinds of standard values. $15 is more than I want to spend right now. Ebay has cheap assortments, if you can wait 4-6 weeks for shipping from China. USA assortment sellers are pricey, like Radio Shack. But I found 1/4 watt packs, choose your ohms, shipping included, cheap: https://www.ebay.com/itm/1-4W-25-Wat...SJJa BvQvMz9g $1.55 is my kind of purchase. If 1k ohms is the wrong value, it won't cost too much to buy different ohms. Even three wild guesses are less than $5. I prefer carbon composition (as they're flat to 1GHz), but for this project, the metal film will suffice. The metal film might be a bit more brittle as well. If you need to bend the legs on a resistor, you can use needle nose pliers, hold the leg away from the body of the component, and bend the leg ninety degrees, such that the stress doesn't go onto the meniscus during the bending operation. The carbon composition ones, in my opinion, can take a bit more stress. The metal film ones use more "glassy" materials. You can place resistors in series and in parallel, to create values other than the "mono-value" you get in a bulk bag. Two 1K in parallel is 500 ohms. Three 1K in parallel is 333 ohms. I could combine a composite 500 ohms and a composite 333 in series and make a 833 ohm resistor. See the fun you can have ? Paul |
#28
|
|||
|
|||
green led
On Fri, 28 Jun 2019 17:07:32 -0400, Paul wrote:
I prefer carbon composition (as they're flat to 1GHz), but for this project, the metal film will suffice. Flat? I know the world is not flat, but with resistors, I have no idea. I can get 10 carbon or 20 metal for the same price. I thought quantity 20 was a better bargain. But I don't need more than 10, so if carbon is better I'll get carbon. The metal film might be a bit more brittle as well. If you need to bend the legs on a resistor, you can use needle nose pliers, hold the leg away from the body of the component, and bend the leg ninety degrees, such that the stress doesn't go onto the meniscus during the bending operation. Good idea. The carbon composition ones, in my opinion, can take a bit more stress. The metal film ones use more "glassy" materials. When I google "carbon vs metal film resistors" the fist hit says: Metal film produces less thermal noise than carbon. Metal film resistors also typically have a much lower inductance/capacitance than carbon so they(metal film) work better at higher frequencies. Carbon has no real advantage except that they are cheaper. Not that I would know, but that's what somebody said. |
#29
|
|||
|
|||
green led
T. Ment wrote:
When I google "carbon vs metal film resistors" the fist hit says: Metal film produces less thermal noise than carbon. Metal film resistors also typically have a much lower inductance/capacitance than carbon so they(metal film) work better at higher frequencies. Carbon has no real advantage except that they are cheaper. Not that I would know, but that's what somebody said. I was told that from an RF perspective, a 75 ohm carbon composition (terminator) remained 75 ohms with no significant L or C up to 1GHz. After which, it is likely that all "leaded" components would have strays associated with them. Using a network analyzer, you can extract a ladder network equivalent model of a passive component. A "resistor" with legs, at a high enough frequency, might have an equivalent ladder network with say, a dozen passive components making the equivalent of the circuit. I've never worked on any circuits that needed a network analyzer for characterization. Everything I've worked on at those frequencies was "a transmission line", a chip produces the signal, and I make the best damn possible transmission line possible for the signal. This includes single layer routing with no vias on the PCB, to prevent reflections, no sharp edges on the tracks and so on. And I had a layout guy with a fetish for details who would go overboard on meeting the requirements. Because I've also had layout guys who were weasels, and didn't read the notes provided with a design, I've also had randomly routed PCB boards, and... they still worked :-/ So much for theory. In a "competition" (two boards with same sort of topology), the weasel-routed one out-performed the fetish-routed one! The fetish-routed one had tighter layout, with hundreds of tracks in "bundles", with crosstalk galore. You learn a lot when using "a million monkeys typing", sometimes earning yourself a "great novel" for your trouble. Paul |
#30
|
|||
|
|||
green led
On Fri, 28 Jun 2019 19:08:01 -0400, Paul wrote:
You learn a lot when using "a million monkeys typing", sometimes earning yourself a "great novel" for your trouble. IDK about that. I don't believe in the unproven, unprovable, and unscientific theory of evolution (hoax). But I like the warm tan color of the carbon resistors, so I'll get those instead of the cold blue metal. |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
When a Green drive is not Green, especially | Flasherly[_2_] | Homebuilt PC's | 4 | March 4th 15 05:27 PM |
P5Q SE2: Green LED Lit, But Nothing Else? | (PeteCresswell) | Asus Motherboards | 5 | February 15th 09 09:48 PM |
MPG's go green? | J. Clarke | Nvidia Videocards | 3 | January 6th 06 12:50 PM |
MPG's go green? | [email protected] | Nvidia Videocards | 2 | January 5th 06 04:22 AM |
MPG's go green? | Dommy | Nvidia Videocards | 2 | December 30th 05 01:02 AM |