4 ms·

The lack of a voltage regulator is quite bad; LEDs reduce their internal resistance as they get hotter, causing them to increase thermal output. That feedback

by zaarn 7y ago

The lack of a voltage regulator is quite bad; LEDs reduce their internal resistance as they get hotter, causing them to increase thermal output. That feedback loop, if not controlled for in some way, can easily destroy the lamp. You also still need a DC converter unless you prefer a 50Hz hum in your LEDs (which sucks).

robomartin 7y ago
Not sure what you mean. LED’s don’t have internal resistance. Can you ckarify?
- zaarn 7y ago
  LEDs have an internal resistance, they're not superconductors. When you switch them on, they have a specific value on that resistance. Due to their internals, as they heat up, this resistance goes down, causing current and thermal output to go up.
  - robomartin 7y ago
    They have something that is sometimes referred to as "dynamic resistance", which is the inverse of the slope of the forward portion of the V-I curve, and yes, it's mostly constant (at constant temperature). For example, a typical lighting class LED will have a dynamic resistance in the order of 0.2 to 0.3 ohms. However, in as much as this varies a bit with temperature the change is mostly negligible. I am also not sure about your assertion that this dynamic resistance decreases with temperature. I took a few minutes to review the datasheets of a dozen OSRAM and Seoul Semiconductor 1W class LED's we use. In all cases the slope of the V-I curve increases slightly with temperature, which means the dynamic resistance decreases slightly with an increase in temperature. Can you explain the effect you are referring to in greater detail? Always interested in learning something new. Thanks.
    - zaarn 7y ago
      Dynamic and internal resistance probably mean different things, afaict, but essentially the internal resistance of an LED is a function of it's temperature. An exponential function in fact; the internal resistance of an LED is an exponential function with the temperature as the inverse in the exponent. Your dynamic slope would also be sufficient to get runaway meltdown. Essentially, imagine it like this; The LED has a starting temperature of 20°C and no power applied. When the appropriate voltage is applied, the LED heats up to 40°C. The dynamic resistance has a small slope, so as a result the dynamic and internal resistance both decrease. The current increases, the voltage is the same, the power output of the package goes up. The LED now heats to 45°C. The process repeats but because the internal resistance is exponentially inverse related to temperature, it speeds up. The LED is now 55°C. It repeats again and the LED reaches 75°C. Resistance drops again at the higher temperature, the current increases more. 115°C. At the next step the LED will reach 195°C, due to the exponential increase and likely the solder points will liquify. If the LED still works it'll hopefully melt itself from the PCB before it catches fire. This can happen incredibly fast, you can counteract it by running the LED well below maximum spec but then you're one power surge or heat surge away from meltdown. Similar to burning wood, the LED's meltdown requires a sufficient amount of extra heat to trigger, otherwise the exponential heat output is too small and is dissipated. This is why you need a current controller or atleast limiter so that the LED cannot pull more current than it's designed for. 0: https://www.researchgate.net/publication/49586062_Internal_Resistance_of_an_LED_as_a_function_of_Temperature https://www.researchgate.net/publication/49586062_Internal_R...
      - robomartin 7y ago
        Thanks, I’ll dig further and talk to the LED applications engineers to see what they have to say. I can’t find anything other than dynamic resistance in the datasheets or manufacturer application notes.
- gingabriska 7y ago
  The amount of current LED consume varies (due to varying internal resistance) with its temepture assuming voltage is kept same. Now, if your LED heats up, it's internal resistance goes down, it sucks more current and it heats up even more and this shortens its lifespan or can kill the LED instantly. A quality driver should handle grid fluctuations, limit the current (by varying the voltage) and LED should have enough cooling (to not disturb its internal resistance much or go above its optimal operating temperature) but often due to space and cost constraints, it doesn't happen. Some even use bad capacitors (low capacity, dries faster, low voltage rating so blows up when power surge happens) this induces flicker. Some use crap LEDs array which does not lit evenly, test for it by lowering the voltage if you see LEDs with uneven brightness across the array or grid, you've bad LEDs. Low quality components, insufficient design, poor QA, inability handle power surge (both grid and local ones) are primary reason for LED flickers and short lifespan Why they don't fix it? To minimize cost, which helps bottom line and boosts profit. Customer's experience with LED will vary (depending on if they've surge protectors or not or grid fluctuations, these are local to you and others might have different situation), so you won't instantly know from reviews or feedback online - which LED is good for you unless ofc you reverse engineer and figure out how it works even then you might not be able to figure out quality of components and degradation with time. Marketing/Sales can fix (hide?) lots of bad reviews and profits can be boosted regardless of underlying quality. Many customers won't pay more because they can't differentiate between quality. And everyone seem to agree that LED as a technology might not last forever, so why bother boosting lifetime/quality of current bulbs when they'll be replaced and an opportunity to make more money will come along? And when a new better bulb technology will come along, they all will blame their cost cutting induced quality / life issue on LED technology and then they'll move on doing same with the new technology.
userbinator 7y ago
Practically all LED lamps have current regulators, either a resistor or capacitor, and in practice this works quite well and is reliable; the flicker is due to a lack of a reservoir capacitor, either because the manufacturer was being cheap, or the lamp doesn't physically have enough room for one.
- zaarn 7y ago
  A simple resistor or capacitor is not a current limiter unless the current it delivers when shorted would be below the maximum the LEDs can handle. A resistor as a current limiter only works reasonably on low wattage LEDs (ie your cheap 0815 plastic LED) and with a stable power supply. 220V grid voltage isn't stable and the LEDs aren't low power anymore. You need an active limiter that regulates the current that flows through the LEDs (or monitors the voltage across a shunt in series with all LEDs), otherwise you risk collective meltdown.
Bjartr 7y ago
Fewer components is good, but too few (and certain switch wiring) and your led bulb can't be turned off due to the capacitance of the house wiring https://youtu.be/1uEmX5XClPY https://youtu.be/1uEmX5XClPY
- zaarn 7y ago
  There are necessary components in devices. I doubt you'd argue "fewer compoments is good" when someone omits the seat belts from a car. LED lights need some components to work properly. One is a AC/DC power supply that converts the AC power to properly smoothed DC. This is the largest and most complicated part of the device. The second is a current regulator that regulates the amount of current going through the LED. Third is a capacitor that smoothes out the flickering of the grid and power supply. In some cases you can merge the DC power supply with the current regulator. If you have a current regulator, your capacitor can be a lot smaller since you already compensate for flickering below the bandwidth limit of that regulator. The largest part is the DC converter, the cheaper bulbs and filament ones basically abuse the LEDs as a DC converter via their diode properties. You can still add a current regulator but it's harder to do and reducing flickering becomes very hard without introducing a lot of capacitance at the AC side. Having a proper DC converter also avoids lighting the bulb on capacitance since you can define drop off voltages and burn some wattage in there that will eat any capacitive voltage.
crispyambulance 7y ago
I've had more success with the LED filament style than the bulbs with the high current converter embedded in the base. The filament versions do have a simple bridge-diode-rectifier + cap. So the flicker should be twice the line frequency ~120Hz in US. see this for teardown... https://www.youtube.com/watch?v=NffhdAz9pc4 https://www.youtube.com/watch?v=NffhdAz9pc4 I think the high-current, low-LED count bulbs have more thermal issues and problems with dimmers than filament LED bulbs. Ultimately, however, I think the best approach is to wire up for LED lights using DC lines and a centrally located converter per room/floor. These tend to be used for more expensive lighting, but hopefully they'll gain more traction in the future. Retrofitting a converter to each bulb just seems wrong.
- zaarn 7y ago
  Filament LEDs at 120Hz have quite a few problems; there is a large number of people who suffer from medical issues if expose long term. Reducing the flicker or bringing it beyond the 100 kilohertz range is worth it. If you use a DC converter, you can pick between high and low LED counts, depending on how much you trust the converter and what it can handle (though a 20W DC converter isn't that large to begin with if you start at 24V LED COBs. The current limiter is a single chip with a precision resistor. I also agree that the best approach is DC wiring. I've been working on realizing that in my own home, 24-48V DC, with solar option, and using the ground wire in the existing cables for communications, so I can tell the LED to dim digitally instead of having to dim the entire power line. It's not terribly expensive either, a 100W DC converter runs you about 100€, reusing wiring is free and a DC LED controller for up to 20W is less than 10$ to DIY.
  - quotefoundslash 7y ago
    > Filament LEDs at 120Hz have quite a few problems; there is a large number of people who suffer from medical issues if expose long term. No one cares. Epilepsy or migraines? Good luck anywhere in public. LED floodlights with unnatural spectrum aimed directly at eye height when browsing any store. Despite use of fresnel with associated hideous fringing the edge of illumination is very abrupt with no chance to adjust. Always 2-10x brighter than needed. Exposed LED strips everywhere with no shade or proper diffuser leading to an overbright spot surrounded by darkness. Many flickering. Due to LED ramp-up/-down the strobing is instantaneous. They are all above and all around and to your side. Of course everyone still carries on with distressing over-illumination. But most establishments gleefully ensure there is always at least one of these directly in your vision, not just illuminating the area. They see interrogation scenes in movies as optimal settings for work, study, play, and relaxation. Accessibility? Equal opportunity? Hahahahahahhahahahahahahahahahahhaha. Lighting is more and more inaccessible. Good luck even if you manage to convince them to let you be a weirdo in the dark. Everything is aggressively recessed in fixtures. The lack of caring could be excused as laziness and ignorance, except that everyone laughs at you and makes faces when you bring it up. It doesn't matter if you speak to your immediate manager at work, or customer service at a business, or whether you speak to the appropriate facilities or HR person. It doesn't matter whether you're brief and to the point or whether you lay out the details and provide references. It doesn't matter how polite or demanding you are. It's not that it was intended to cripple the weak and exclude invalids from society. But they'll happily take it as a bonus. Speaking of bonuses: they can give you the side-eye as "environmentally unfriendly" if you dare to raise any issue with LED/fluorescent. (Of course the same people usually oppose nuclear power, proving their bad faith.)

by zaarn 7y ago

robomartin 7y ago

Not sure what you mean. LED’s don’t have internal resistance. Can you ckarify?

zaarn 7y ago
LEDs have an internal resistance, they're not superconductors. When you switch them on, they have a specific value on that resistance. Due to their internals, as they heat up, this resistance goes down, causing current and thermal output to go up.
- robomartin 7y ago
  They have something that is sometimes referred to as "dynamic resistance", which is the inverse of the slope of the forward portion of the V-I curve, and yes, it's mostly constant (at constant temperature). For example, a typical lighting class LED will have a dynamic resistance in the order of 0.2 to 0.3 ohms. However, in as much as this varies a bit with temperature the change is mostly negligible. I am also not sure about your assertion that this dynamic resistance decreases with temperature. I took a few minutes to review the datasheets of a dozen OSRAM and Seoul Semiconductor 1W class LED's we use. In all cases the slope of the V-I curve increases slightly with temperature, which means the dynamic resistance decreases slightly with an increase in temperature. Can you explain the effect you are referring to in greater detail? Always interested in learning something new. Thanks.
  - zaarn 7y ago
    Dynamic and internal resistance probably mean different things, afaict, but essentially the internal resistance of an LED is a function of it's temperature. An exponential function in fact; the internal resistance of an LED is an exponential function with the temperature as the inverse in the exponent. Your dynamic slope would also be sufficient to get runaway meltdown. Essentially, imagine it like this; The LED has a starting temperature of 20°C and no power applied. When the appropriate voltage is applied, the LED heats up to 40°C. The dynamic resistance has a small slope, so as a result the dynamic and internal resistance both decrease. The current increases, the voltage is the same, the power output of the package goes up. The LED now heats to 45°C. The process repeats but because the internal resistance is exponentially inverse related to temperature, it speeds up. The LED is now 55°C. It repeats again and the LED reaches 75°C. Resistance drops again at the higher temperature, the current increases more. 115°C. At the next step the LED will reach 195°C, due to the exponential increase and likely the solder points will liquify. If the LED still works it'll hopefully melt itself from the PCB before it catches fire. This can happen incredibly fast, you can counteract it by running the LED well below maximum spec but then you're one power surge or heat surge away from meltdown. Similar to burning wood, the LED's meltdown requires a sufficient amount of extra heat to trigger, otherwise the exponential heat output is too small and is dissipated. This is why you need a current controller or atleast limiter so that the LED cannot pull more current than it's designed for. 0: https://www.researchgate.net/publication/49586062_Internal_Resistance_of_an_LED_as_a_function_of_Temperature https://www.researchgate.net/publication/49586062_Internal_R...
    - robomartin 7y ago
      Thanks, I’ll dig further and talk to the LED applications engineers to see what they have to say. I can’t find anything other than dynamic resistance in the datasheets or manufacturer application notes.
gingabriska 7y ago
The amount of current LED consume varies (due to varying internal resistance) with its temepture assuming voltage is kept same. Now, if your LED heats up, it's internal resistance goes down, it sucks more current and it heats up even more and this shortens its lifespan or can kill the LED instantly. A quality driver should handle grid fluctuations, limit the current (by varying the voltage) and LED should have enough cooling (to not disturb its internal resistance much or go above its optimal operating temperature) but often due to space and cost constraints, it doesn't happen. Some even use bad capacitors (low capacity, dries faster, low voltage rating so blows up when power surge happens) this induces flicker. Some use crap LEDs array which does not lit evenly, test for it by lowering the voltage if you see LEDs with uneven brightness across the array or grid, you've bad LEDs. Low quality components, insufficient design, poor QA, inability handle power surge (both grid and local ones) are primary reason for LED flickers and short lifespan Why they don't fix it? To minimize cost, which helps bottom line and boosts profit. Customer's experience with LED will vary (depending on if they've surge protectors or not or grid fluctuations, these are local to you and others might have different situation), so you won't instantly know from reviews or feedback online - which LED is good for you unless ofc you reverse engineer and figure out how it works even then you might not be able to figure out quality of components and degradation with time. Marketing/Sales can fix (hide?) lots of bad reviews and profits can be boosted regardless of underlying quality. Many customers won't pay more because they can't differentiate between quality. And everyone seem to agree that LED as a technology might not last forever, so why bother boosting lifetime/quality of current bulbs when they'll be replaced and an opportunity to make more money will come along? And when a new better bulb technology will come along, they all will blame their cost cutting induced quality / life issue on LED technology and then they'll move on doing same with the new technology.

userbinator 7y ago

Practically all LED lamps have current regulators, either a resistor or capacitor, and in practice this works quite well and is reliable; the flicker is due to a lack of a reservoir capacitor, either because the manufacturer was being cheap, or the lamp doesn't physically have enough room for one.

zaarn 7y ago
A simple resistor or capacitor is not a current limiter unless the current it delivers when shorted would be below the maximum the LEDs can handle. A resistor as a current limiter only works reasonably on low wattage LEDs (ie your cheap 0815 plastic LED) and with a stable power supply. 220V grid voltage isn't stable and the LEDs aren't low power anymore. You need an active limiter that regulates the current that flows through the LEDs (or monitors the voltage across a shunt in series with all LEDs), otherwise you risk collective meltdown.

Bjartr 7y ago

Fewer components is good, but too few (and certain switch wiring) and your led bulb can't be turned off due to the capacitance of the house wiring https://youtu.be/1uEmX5XClPY https://youtu.be/1uEmX5XClPY

zaarn 7y ago
There are necessary components in devices. I doubt you'd argue "fewer compoments is good" when someone omits the seat belts from a car. LED lights need some components to work properly. One is a AC/DC power supply that converts the AC power to properly smoothed DC. This is the largest and most complicated part of the device. The second is a current regulator that regulates the amount of current going through the LED. Third is a capacitor that smoothes out the flickering of the grid and power supply. In some cases you can merge the DC power supply with the current regulator. If you have a current regulator, your capacitor can be a lot smaller since you already compensate for flickering below the bandwidth limit of that regulator. The largest part is the DC converter, the cheaper bulbs and filament ones basically abuse the LEDs as a DC converter via their diode properties. You can still add a current regulator but it's harder to do and reducing flickering becomes very hard without introducing a lot of capacitance at the AC side. Having a proper DC converter also avoids lighting the bulb on capacitance since you can define drop off voltages and burn some wattage in there that will eat any capacitive voltage.

crispyambulance 7y ago

I've had more success with the LED filament style than the bulbs with the high current converter embedded in the base. The filament versions do have a simple bridge-diode-rectifier + cap. So the flicker should be twice the line frequency ~120Hz in US. see this for teardown... https://www.youtube.com/watch?v=NffhdAz9pc4 https://www.youtube.com/watch?v=NffhdAz9pc4 I think the high-current, low-LED count bulbs have more thermal issues and problems with dimmers than filament LED bulbs. Ultimately, however, I think the best approach is to wire up for LED lights using DC lines and a centrally located converter per room/floor. These tend to be used for more expensive lighting, but hopefully they'll gain more traction in the future. Retrofitting a converter to each bulb just seems wrong.

zaarn 7y ago
Filament LEDs at 120Hz have quite a few problems; there is a large number of people who suffer from medical issues if expose long term. Reducing the flicker or bringing it beyond the 100 kilohertz range is worth it. If you use a DC converter, you can pick between high and low LED counts, depending on how much you trust the converter and what it can handle (though a 20W DC converter isn't that large to begin with if you start at 24V LED COBs. The current limiter is a single chip with a precision resistor. I also agree that the best approach is DC wiring. I've been working on realizing that in my own home, 24-48V DC, with solar option, and using the ground wire in the existing cables for communications, so I can tell the LED to dim digitally instead of having to dim the entire power line. It's not terribly expensive either, a 100W DC converter runs you about 100€, reusing wiring is free and a DC LED controller for up to 20W is less than 10$ to DIY.
- quotefoundslash 7y ago
  > Filament LEDs at 120Hz have quite a few problems; there is a large number of people who suffer from medical issues if expose long term. No one cares. Epilepsy or migraines? Good luck anywhere in public. LED floodlights with unnatural spectrum aimed directly at eye height when browsing any store. Despite use of fresnel with associated hideous fringing the edge of illumination is very abrupt with no chance to adjust. Always 2-10x brighter than needed. Exposed LED strips everywhere with no shade or proper diffuser leading to an overbright spot surrounded by darkness. Many flickering. Due to LED ramp-up/-down the strobing is instantaneous. They are all above and all around and to your side. Of course everyone still carries on with distressing over-illumination. But most establishments gleefully ensure there is always at least one of these directly in your vision, not just illuminating the area. They see interrogation scenes in movies as optimal settings for work, study, play, and relaxation. Accessibility? Equal opportunity? Hahahahahahhahahahahahahahahahahhaha. Lighting is more and more inaccessible. Good luck even if you manage to convince them to let you be a weirdo in the dark. Everything is aggressively recessed in fixtures. The lack of caring could be excused as laziness and ignorance, except that everyone laughs at you and makes faces when you bring it up. It doesn't matter if you speak to your immediate manager at work, or customer service at a business, or whether you speak to the appropriate facilities or HR person. It doesn't matter whether you're brief and to the point or whether you lay out the details and provide references. It doesn't matter how polite or demanding you are. It's not that it was intended to cripple the weak and exclude invalids from society. But they'll happily take it as a bonus. Speaking of bonuses: they can give you the side-eye as "environmentally unfriendly" if you dare to raise any issue with LED/fluorescent. (Of course the same people usually oppose nuclear power, proving their bad faith.)