Underdriven 12V LED chips in a 120VAC system

[primuspaul]

If I want to connect a bunch of 12V LED chips in series to work from 120VAC, with the only other components being diodes (regular ones, not LEDs), would I get it to work if wired like this?


What would the heat output be for each cell? VAC has 170 peak I think, but overall the heat output would be 12V equivalent on each chip, or to put it in other words, rated power, right?

If so, I will probably add an LED or two in there since I want to underdrive them, reducing brightness and heat in order to extend life and reduce waste heat (and also having to spend less $ on heat sinks).

Would that work? What kind of diode ratings am I looking for on those 4 diodes?

Would I need a capacitor in there somewhere to deal with the peaks or it wouldn't matter if I'm underdriving with 11-12 LEDs instead of 10?

Is any waste heat being created by any components, like the 4 diodes (with the exception of the inescapable generation of heat by the LEDs)?

 

[kellys_eye]

Standard practise is to use a diode bridge with suitable regulator after (or dropper resistor in the LED chain) or even a capacitive dropper. Some form of current limiting is essential.

 

[duke37]

D1 and D3 will steer the supply to the two loops. D2 and D4 are not required.

A fat resistor should be included to control the current. If you wish to drive the leds with 20mA average, then a resistor of about 100V/20mA = 5kΩ. from the power supply. Power = I*I*R = 2W, use a 5W.

Any high voltage diodes will do. The 1N4007 is very capable and is common.
A capacitor is not necessary if the flicker is not a problem.

 

[dave9]

It might help if you linked to these 12V LED chips so we know how they arrived at that spec or claim. Some might be 1 to 3 LEDs in series with a resistor built in. Some might just be 3 in series, maybe even additional parallel series of three, but need a resistor or other current limiting in the circuit to run off 12V successfully.

There is no LED that runs off exactly 12V without other components involved and we don't know what current they're supposed to be nor your needed light output if you were to under drive them.

Why on earth wouldn't you supply all the relevant info up front? A datasheet for the LEDs would always be something to post.

As far as the diodes go, you don't even need them at all if the sum of all the LED dies reverse voltage tolerance exceeds that of the supply voltage. That # is not 12V/LED.

 

[primuspaul]

D1 and D3 will steer the supply to the two loops. D2 and D4 are not required.

A fat resistor should be included to control the current. If you wish to drive the leds with 20mA average, then a resistor of about 100V/20mA = 5kΩ. from the power supply. Power = I*I*R = 2W, use a 5W.

Any high voltage diodes will do. The 1N4007 is very capable and is common.
A capacitor is not necessary if the flicker is not a problem.

Two of these in place of D3 and D1?

https://www.ebay.com/itm/30Pcs-Lot-...-SMT-Rectifier-Diode-2-Pin-Black/202062312825

Also, could you explain a bit more about the current limiting? I redesigned the circuit below to use 2 diodes and 12 LEDs instead of 4 diodes and 10 LEDs. Now the voltage per diode should be a cool 10V per light chip:

Do I still need a resistor? I'm just wondering how it works since I have operated 12V LED chips with voltage-driven 12V power supplies in the past with no problem. Basically, if I understand correctly, your standard $0.99 "wall wart" is a voltage-driven power supply as opposed to the LED-specific current driven units on sale as LED drivers on eBay. I needed to make a low-powered LED lamp, so I took the 120VAC-->12VDC wall wart and connected it to 2 LED chips in series. Since it's a voltage driven unit, it split the voltage in half between the LED chips and they ran at reduced light output and did not appear to accumulate any heat even with absolutely no heat sink. So why would I still need a resistor with 12 LED chips as above? Sorry if I sound like I have very limited understanding of electricity. This is just a hobby for me.

Re: flicker, how much flicker are we talking about if I use the design above without a capacitor? Doesn't the voltage flip 60 times per second? How would a person even notice that? What sort of capacitor would I need? Would a supercap work (I happen to have a few 2.7V 500F 35*60MM Super Capacitors)?

 

[primuspaul]

It might help if you linked to these 12V LED chips so we know how they arrived at that spec or claim. Some might be 1 to 3 LEDs in series with a resistor built in. Some might just be 3 in series, maybe even additional parallel series of three, but need a resistor or other current limiting in the circuit to run off 12V successfully.

There is no LED that runs off exactly 12V without other components involved and we don't know what current they're supposed to be nor your needed light output if you were to under drive them.

Why on earth wouldn't you supply all the relevant info up front? A datasheet for the LEDs would always be something to post.

As far as the diodes go, you don't even need them at all if the sum of all the LED dies reverse voltage tolerance exceeds that of the supply voltage. That # is not 12V/LED.

I don't have the sheet for the LEDs I have, but I think they're identical to these:

Specifications:
1.Model Type: High Power Integrated Chip
2. Product name: Integrated LED Bead
3. Shapes: Tetragonum
4. Integrated way:
5. Power: 10W/20W/30W/50W/100W
6. Forward Voltage: 10W:9.0-12V / Others: 30-36V
8. Luminous Flux: 90-110LM/W
9. color-rendering index: 60-85
10. Beam Angle: 120 degrees
11. Working Temperature: -20 centigrade to 60 centigrade
12. Color Temperature: Warm White: 3000-3200K
White: 6000-6500K
13. Lamp Body Material : Pure Copper Bracket
14. Life Span: 120,000 Hours
https://www.ebay.com/itm/10pcs-High...lb-Chip-DIY-10W-20W-50W-100W-12V/272389345999

I'm not sure how much light I need. I'm just going to try them at 10V a piece and see if it's enough. This is mostly to experiment and see if such a project is possible with minimal power conversion/heat.

 

[(*steve*)]

Read this.

If you can get yourself to the point where you understand all of it, you're at a point where you can think about the project you're contemplating.

 

[Audioguru]

The LEDs are very cheap. Cheap things usually have poor quality. It looks like they have 16 LEDs in series/parallel.
The spec's say that some of them have a forward voltage of only 9V. If most of them are 9V then they will blow up in your circuit that has nothing to limit the current.

 

[primuspaul]

The LEDs are very cheap. Cheap things usually have poor quality. It looks like they have 16 LEDs in series/parallel.
The spec's say that some of them have a forward voltage of only 9V. If most of them are 9V then they will blow up in your circuit that has nothing to limit the current.

I will obviously test these before I install them. And if I need to, I can always add 2 more LEDs and get 8.5V

I had a chinese floodlight that burned out after a few months. Turns out they had one of those LEDs being powered by a driver rated at 24-36V. Am I correct in saying that LEDs pretty much always die due to excessive heat? And that thing still lasted a few months. Are you saying running a 9V with 10V for even a few seconds will kill it instantly? I would think it would need at least a few minutes to build up enough heat, right?

 

[Audioguru]

I have never bought cheap Chinese high power LEDs so I do not know how much they lie about the power rating.
Instead I buy high quality Western LEDs that meet or exceed the detailed spec's in their datasheet.
I think the chip heats almost immediately then it takes time for the frame to heat up.
LEDs have very thin wires inside that burn out with too much current. They do not use thick wires that block the light.

 

[dave9]

The only thing that makes sense out of that is that you are buying those generic 10W LEDs which are configured as a 3x3 Series Parallel array.

They are horrible LEDs, having resistive leaks and meh, I have some lying around and will buy more of something else before I use them.

The likely reason you had success using a 12VDC adapter previously was that it's inherently a current limit by itself and just happened to have a voltage not too high before current limiting kicked in. I have tried a setup like that with the generic so-called 12V LEDs, WITH a resistor current limiter and they lasted about 2 weeks continuous runtime before the dies cascade failed due to resistive leakage defects on the package.

Anyway, this isn't the way to do it. Get a 100W (or lower depending on heatsinking capacity) LED driver off ebay (if you have sufficient heatsinking, if not there is no reason to use so many LEDs even if they are cheap) and get a proper 110VAC to (whatever current) LED driver.

Here is an example, then just put your 12V LEDs in series according to the charge on that page, and pick the # of LEDs and output current appropriate to the heatsink capacity... but do realize that with these leaky generic LEDs, you're going to need about twice the heatsink capacity for the same light as if you'd used efficient, major brand LEDs. They're a false economy unless you have huge chunks of scrap aluminum to mount them on and don't pay the power bill.

https://www.ebay.com/itm/3W-5W-10W-...-Supply-85-265V-Constant-Current/322532619842

 

[primuspaul]

The only thing that makes sense out of that is that you are buying those generic 10W LEDs which are configured as a 3x3 Series Parallel array.

They are horrible LEDs, having resistive leaks and meh, I have some lying around and will buy more of something else before I use them.

The likely reason you had success using a 12VDC adapter previously was that it's inherently a current limit by itself and just happened to have a voltage not too high before current limiting kicked in. I have tried a setup like that with the generic so-called 12V LEDs, WITH a resistor current limiter and they lasted about 2 weeks continuous runtime before the dies cascade failed due to resistive leakage defects on the package.

Anyway, this isn't the way to do it. Get a 100W (or lower depending on heatsinking capacity) LED driver off ebay (if you have sufficient heatsinking, if not there is no reason to use so many LEDs even if they are cheap) and get a proper 110VAC to (whatever current) LED driver.

Here is an example, then just put your 12V LEDs in series according to the charge on that page, and pick the # of LEDs and output current appropriate to the heatsink capacity... but do realize that with these leaky generic LEDs, you're going to need about twice the heatsink capacity for the same light as if you'd used efficient, major brand LEDs. They're a false economy unless you have huge chunks of scrap aluminum to mount them on and don't pay the power bill.

https://www.ebay.com/itm/3W-5W-10W-...-Supply-85-265V-Constant-Current/322532619842

Underdriven, those cheap LEDs produce 0 heat. Heat produced plummets like a rock when you drop the voltage. Are you saying that, even underdriven and constantly running cool, they're still going to fail early?

Also, won't the constant current driver up the voltage in order and therefore keep the LED brightness the same? I tested a constant current driver by connecting first one LED then two in series and brightness did not change. That means they're still producing lots of waste heat, which is what I want to avoid.

 

[primuspaul]

I have never bought cheap Chinese high power LEDs so I do not know how much they lie about the power rating.
Instead I buy high quality Western LEDs that meet or exceed the detailed spec's in their datasheet.
I think the chip heats almost immediately then it takes time for the frame to heat up.
LEDs have very thin wires inside that burn out with too much current. They do not use thick wires that block the light.

Too much current or too much heat? So the wires can burn up even if sufficiently cooled with a heat sink?

 

[Audioguru]

The internal wires and chip heat almost instantly. The frame attached to the heatsink heats up slowly then reaches its maximum temperature that is far less than the internal wires ands chip. The cheap LEDs do not have a detailed datasheet that states the thermal resistance that is stated on the datasheets of all half-decent LEDs.

 

[dave9]

You wouldn't be able to depend on a datasheet for them. The resistive shorts are random defects, result in excess heat loss which varies per specimen so it's measuring current and Vf for best guess thermal dissipation, or measuring heatsink temperature.

The only thing positive about the resistive shorts in the dies is that when one die in a series burns out, the next two in the series continue to stay lit, lol. Some people speculate that they are just the result of dumpster divers grabbing up defective product and that's how they end up on eBay so cheap.

 

[dave9]

Underdriven, those cheap LEDs produce 0 heat. Heat produced plummets like a rock when you drop the voltage. Are you saying that, even underdriven and constantly running cool, they're still going to fail early?

No, running cool enough you might get reasonable life but why do that? I mean you can get something like 5W Cree LEDs for under $1 each now and they are over twice the brightness per current of the defective generic 10W LEDs. At some point that has to count for something whether it be the power bill or the size, weight, and expense of the heatsink, or a larger more costly driver to handle it.

Also, won't the constant current driver up the voltage in order and therefore keep the LED brightness the same?

You pick the appropriate constant current driver, whether it be (just examples) 300mA, 600mA, 1A, etc. then since it is constant current you have the voltage range it can handle, so for example you could have 300mA @ 20V for two of those LEDs in series, about 6W minus whatever is generated as light, or you could have 600mA @ 40V for 4 in series, 24W. Run enough of them at low current and you still need a suitably sized heatsink, but at least the thermal density is spread out more.

A lot of those constant current drivers can also be modified by just changing a resistor or two in the feedback loop if you want the current reduced. Increased can be more of a problem unless you're sure the only differences between that version and the next higher current version is the feedback resistor values.

I tested a constant current driver by connecting first one LED then two in series and brightness did not change. That means they're still producing lots of waste heat, which is what I want to avoid.

It can depend on the driver design but if it works correctly then two in series will mean each of the two in series is as bright as one in series was, so double total brightness. With the cheap drivers this often isn't quite true, that the more you put in series the more the current droops, so you could have a driver spec'd at 600mA but it's only driving a few in series at 540mA (random example).

You can't get around the waste heat issue with the defective generic LEDs. Merely having enough forward voltage to light up the LED dies will cause the loss through the resistive faults in the dies. The resistive faults remain to produce waste heat even after the die they're in has burnt out.

Then again if I were stranded on a deserted island and it's all I had, with an adequate heatsink I'd run them at a couple watts each. Only when I saw a a ship and needed a way to signal it.

 

[primuspaul]

No, running cool enough you might get reasonable life but why do that? I mean you can get something like 5W Cree LEDs for under $1 each now and they are over twice the brightness per current of the defective generic 10W LEDs. At some point that has to count for something whether it be the power bill or the size, weight, and expense of the heatsink, or a larger more costly driver to handle it.



You pick the appropriate constant current driver, whether it be (just examples) 300mA, 600mA, 1A, etc. then since it is constant current you have the voltage range it can handle, so for example you could have 300mA @ 20V for two of those LEDs in series, about 6W minus whatever is generated as light, or you could have 600mA @ 40V for 4 in series, 24W. Run enough of them at low current and you still need a suitably sized heatsink, but at least the thermal density is spread out more.

A lot of those constant current drivers can also be modified by just changing a resistor or two in the feedback loop if you want the current reduced. Increased can be more of a problem unless you're sure the only differences between that version and the next higher current version is the feedback resistor values.



It can depend on the driver design but if it works correctly then two in series will mean each of the two in series is as bright as one in series was, so double total brightness. With the cheap drivers this often isn't quite true, that the more you put in series the more the current droops, so you could have a driver spec'd at 600mA but it's only driving a few in series at 540mA (random example).

You can't get around the waste heat issue with the defective generic LEDs. Merely having enough forward voltage to light up the LED dies will cause the loss through the resistive faults in the dies. The resistive faults remain to produce waste heat even after the die they're in has burnt out.

Then again if I were stranded on a deserted island and it's all I had, with an adequate heatsink I'd run them at a couple watts each. Only when I saw a a ship and needed a way to signal it.

Where are you getting brand name 5W LEDs for under $1 each?

 

[(*steve*)]

I have never bought cheap Chinese high power LEDs

I have, and I think I might post about the obvious problems.

My success rate in getting obviously bad items is about 75%. However, my success rate in getting refunds for the bad ones is currently running at 100% (eBay is good for something). As @Audioguru will remind me, I can't be sure that the other 25% are actually good.

 

[primuspaul]

Two of these in place of D3 and D1?

https://www.ebay.com/itm/30Pcs-Lot-...-SMT-Rectifier-Diode-2-Pin-Black/202062312825

Also, could you explain a bit more about the current limiting? I redesigned the circuit below to use 2 diodes and 12 LEDs instead of 4 diodes and 10 LEDs. Now the voltage per diode should be a cool 10V per light chip:
View attachment 39306
Do I still need a resistor? I'm just wondering how it works since I have operated 12V LED chips with voltage-driven 12V power supplies in the past with no problem. Basically, if I understand correctly, your standard $0.99 "wall wart" is a voltage-driven power supply as opposed to the LED-specific current driven units on sale as LED drivers on eBay. I needed to make a low-powered LED lamp, so I took the 120VAC-->12VDC wall wart and connected it to 2 LED chips in series. Since it's a voltage driven unit, it split the voltage in half between the LED chips and they ran at reduced light output and did not appear to accumulate any heat even with absolutely no heat sink. So why would I still need a resistor with 12 LED chips as above? Sorry if I sound like I have very limited understanding of electricity. This is just a hobby for me.

Re: flicker, how much flicker are we talking about if I use the design above without a capacitor? Doesn't the voltage flip 60 times per second? How would a person even notice that? What sort of capacitor would I need? Would a supercap work (I happen to have a few 2.7V 500F 35*60MM Super Capacitors)?

So what's the verdict? Will the above design work properly (without shortening the life of the LEDs) assuming the LEDs are properly 12V-rated?

 

[(*steve*)]

It depends on what you mean by "properly 12V rated", "work properly", and "shortening the life".

I would tend to say no, but, based on your experience reading the LED resource, what do you think?