swimmer4uus
Member
What spacing are the LEDs at, and whats the circuit for driving them?
Diy led
- Thread starter swimmer4uus
- Start date
scsinet
Active Member
Originally Posted by swimmer4uus
http:///forum/post/3220153
The resistor itself would work, if LEDs where capable of handling varrying current. They can't. LEDs must be provided a very specific current and voltage. I really don't feel like replacing strings of LEDs due to thermal runaway. It's not that you can't run the LEDs with just the resistor, its that you just can't run them very long without falure, or massive inefficiency.
Yeah I understand how LEDs work. I'm curious why the current would be varying? Why would that be? In an LED circuit, the resistor sets the current. Unless the supply voltage varies, the current won't.
It should not be inefficient as long as the voltage drop across your resistor is not too great. Let's use an example LED rated at 3.6 forward volts at 700ma typical forward current. Powering one of these from a 24 volt supply would place a large voltage drop across the resistor, which needs to be dissipated as heat, so yes, inefficient. The resistor would have to be rated at (a whopping) 30 watts at least.
However, connecting 6 of them in series, would raise the forward voltage requirement to 21.6 volts, making your voltage drop only 2.4 volts.
Applying the basic resistor calculator formula, R=(Supply Volts - LED Volts) / I:
R = (24 - 21.6) / I
R = 2.4 / .7
R = 3.3 ohms (closest value). A 3.3 ohm resistor won't cause too much of a power loss. A small(er) 5w resistor would do the job here.
Even an LM317 is nothing more than a transistor, it will end up dissipating it's voltage drop as heat. There are only two ways to avoid the efficiency loss... series connect the LEDs to close the gap between the power supply voltage and the total forward volts, or use a PWM driver (which may be what those buck-pucks are).
My point was that an LM317 doesn't buy you anything other than varying the voltage. It will not give you any efficiency gain nor change the life of the LEDs.
http:///forum/post/3220153
The resistor itself would work, if LEDs where capable of handling varrying current. They can't. LEDs must be provided a very specific current and voltage. I really don't feel like replacing strings of LEDs due to thermal runaway. It's not that you can't run the LEDs with just the resistor, its that you just can't run them very long without falure, or massive inefficiency.
Yeah I understand how LEDs work. I'm curious why the current would be varying? Why would that be? In an LED circuit, the resistor sets the current. Unless the supply voltage varies, the current won't.
It should not be inefficient as long as the voltage drop across your resistor is not too great. Let's use an example LED rated at 3.6 forward volts at 700ma typical forward current. Powering one of these from a 24 volt supply would place a large voltage drop across the resistor, which needs to be dissipated as heat, so yes, inefficient. The resistor would have to be rated at (a whopping) 30 watts at least.
However, connecting 6 of them in series, would raise the forward voltage requirement to 21.6 volts, making your voltage drop only 2.4 volts.
Applying the basic resistor calculator formula, R=(Supply Volts - LED Volts) / I:
R = (24 - 21.6) / I
R = 2.4 / .7
R = 3.3 ohms (closest value). A 3.3 ohm resistor won't cause too much of a power loss. A small(er) 5w resistor would do the job here.
Even an LM317 is nothing more than a transistor, it will end up dissipating it's voltage drop as heat. There are only two ways to avoid the efficiency loss... series connect the LEDs to close the gap between the power supply voltage and the total forward volts, or use a PWM driver (which may be what those buck-pucks are).
My point was that an LM317 doesn't buy you anything other than varying the voltage. It will not give you any efficiency gain nor change the life of the LEDs.
swimmer4uus
Member
Our power sources are AC (alternating current). They follow a sinusodal wave form under normal conditions. LEDs need DC
And yes the LM317 will not give any of those, it just gets the job done, in the cheapest form.
And yes the LM317 will not give any of those, it just gets the job done, in the cheapest form.
scsinet
Active Member
Originally Posted by swimmer4uus
http:///forum/post/3220225
Our power sources are AC (alternating current). They follow a sinusodal wave form under normal conditions. LEDs need DC
Umm... that's what your power supply is for, dude. 24 volts DC out...
What "job" is the 317 doing that you are referring to?
As a side note... and operating LEDs off AC is also fine with just a resistor provided the supply voltage does not exceed the peak inverse voltage of the LED... but this is a moot point since we aren't using AC to drive the LEDs here.
http:///forum/post/3220225
Our power sources are AC (alternating current). They follow a sinusodal wave form under normal conditions. LEDs need DC
Umm... that's what your power supply is for, dude. 24 volts DC out...
What "job" is the 317 doing that you are referring to?
As a side note... and operating LEDs off AC is also fine with just a resistor provided the supply voltage does not exceed the peak inverse voltage of the LED... but this is a moot point since we aren't using AC to drive the LEDs here.
swimmer4uus
Member
Oh dang.....I see your point now....time for further research. Good point
scsinet
Active Member
I'm waiting to see how this turns out. I know the electronics, drives, , MCUs, etc, but what I don't know are the LED counts, quantities of each color, optics, etc. So I'm hoping to learn that from what you're doing. Post questions if you have them. If I can help you with the electronics, what you learn on the other stuff will help me when I take this on.
swimmer4uus
Member
It never hit me before. I started with the design of someone else. Power chord-> power supply-> Drivers-> LEDs. Was pretty simple. Just never thought about how the power supply already puts out a constart current, and voltage/current division takes care of everything else. The ONLY thing I've really come up with is the driver is there to absolutely make sure that nothing's wrong. Almost like a back up if anything happens. I mean, for 2 bucks I think I'm going to put it in there, but who knows! And for mine, I'll probably end up with 3 total banks. I too think 300 3W LEDs is a little much. We'll see. Should be ordering everything within 2 weeks....
mastertech
Member
Originally Posted by swimmer4uus
http:///forum/post/3220200
What spacing are the LEDs at, and whats the circuit for driving them?
a base line for everything we are testing is one led about every 2 1/2 inches.
we are all using 1000ma leds turned down to about 3/4 power or less.
we are not seeing a signifigant amount of par increases past that point.
we dont have a real amp reading for that potentioneter setting..... yet.
waiting on a few more local members before i will be able to measure current draw vs par readings for the cree xr-e leds.
http:///forum/post/3220200
What spacing are the LEDs at, and whats the circuit for driving them?
a base line for everything we are testing is one led about every 2 1/2 inches.
we are all using 1000ma leds turned down to about 3/4 power or less.
we are not seeing a signifigant amount of par increases past that point.
we dont have a real amp reading for that potentioneter setting..... yet.
waiting on a few more local members before i will be able to measure current draw vs par readings for the cree xr-e leds.
swimmer4uus
Member
Well, I guess that's a good thing then because I'm building only 3 banks the first time around. One of them might not even go on the 90 Gallon
swimmer4uus
Member
So I posted this in another forum, and questioned the need for drivers...here's the explination I got.
"Drivers are necessary to regulate and deliver a constant current to the diodes.
Let's assume your supply will deliver EXACTLY 24V at 6A (for the math's sake). That would mean you could run 6 strings of 8 LEDs. The problem is that no power supply will deliver EXACTLY what it states. It will always be a little less, and how much less, it will be very difficult to determine. So the problem right off the bat is the LEDs will not get exactly what they need. Another problem is that a power supply will fail quickly if you try to run it at full capacity for extended periods. Even the highest quality supplies will perform better for longer if run at a lower load than full capacity. The third problem is you have no choice but to run EXACTLY the amount of LEDs to match supply output, assuming the supply outputs what the manufacturer claims it will.
While Buckpucks might not be necessary, it is necessary to have some sort of driver to deliver the power to the spec that the LED requires. It gives you more flexibility of when and how to incorporate more LEDs into the system."
Drivers are the security. Peace of mind. And, I don't think I could justify spending this much, without having some reliability in the circuit.
"Drivers are necessary to regulate and deliver a constant current to the diodes.
Let's assume your supply will deliver EXACTLY 24V at 6A (for the math's sake). That would mean you could run 6 strings of 8 LEDs. The problem is that no power supply will deliver EXACTLY what it states. It will always be a little less, and how much less, it will be very difficult to determine. So the problem right off the bat is the LEDs will not get exactly what they need. Another problem is that a power supply will fail quickly if you try to run it at full capacity for extended periods. Even the highest quality supplies will perform better for longer if run at a lower load than full capacity. The third problem is you have no choice but to run EXACTLY the amount of LEDs to match supply output, assuming the supply outputs what the manufacturer claims it will.
While Buckpucks might not be necessary, it is necessary to have some sort of driver to deliver the power to the spec that the LED requires. It gives you more flexibility of when and how to incorporate more LEDs into the system."
Drivers are the security. Peace of mind. And, I don't think I could justify spending this much, without having some reliability in the circuit.
scsinet
Active Member
Sounds like marketing fluff to me.
LEDs are not nearly as fragile as these driver manufacturers want you to think, especially in this application.
All LEDs will suffer reduced life if driven at their absolute maximum ratings for extended periods of time, so in an aquarium situation, you want to derate the current and voltage to some degree. For example, the Cree XR-E LEDs from what I've come to understand are traditionally driven at 700ma. However, they can handle up to 1000ma.
If the power supply you are using is a CNC power supply, it is built to handle extreme variations in current draw with little fluctuation in output voltage. Generally speaking, as you load a switching power supply, the voltage fluctuation (if any) will be down, not up. So if you plan to drive a string of (6) LEDs from your 24 volt supply, at 700ma, the Cree data sheet tells you to set up for 3.5 forward volts. So 3.5 * 6 = 21, so applying the above formula gives you 4.28 ohms, so we'll go a bit on the safe side and use (2) 2.2 ohm resistors in series for 4.4 ohms. Now, if the supply voltage drops as load increases, ohms law dictates that the current will fall, along with the forward volts. A drop in current or voltage will not damage the LEDs.
Accounting for a rise, the cree LEDs can handle 3.7fv at 1000ma, allowing for a 5% increase in voltage, which is well beyond what any regulated power supply will do.
So I think you can save your money and DIY your drivers. You can use an LM317 if you desire as an additional regulation step, but in doing so with a 24 volt power supply you'll need to drop an LED from the string. The LM317 requires a minimum of a 3v differential between the supply voltage (Vin) and the regulated output voltage (Vout). You need to account for slight power supply variations, which puts your effective output on a 24v supply at 20v max. Next you need that resistor, which has to have some voltage drop across it to maintain current regulation, so that takes off another couple volts. So your maximum fv can only be about 18 volts at this point, meaning you can use 4-5 LEDs at best per driver.
LEDs are not nearly as fragile as these driver manufacturers want you to think, especially in this application.
All LEDs will suffer reduced life if driven at their absolute maximum ratings for extended periods of time, so in an aquarium situation, you want to derate the current and voltage to some degree. For example, the Cree XR-E LEDs from what I've come to understand are traditionally driven at 700ma. However, they can handle up to 1000ma.
If the power supply you are using is a CNC power supply, it is built to handle extreme variations in current draw with little fluctuation in output voltage. Generally speaking, as you load a switching power supply, the voltage fluctuation (if any) will be down, not up. So if you plan to drive a string of (6) LEDs from your 24 volt supply, at 700ma, the Cree data sheet tells you to set up for 3.5 forward volts. So 3.5 * 6 = 21, so applying the above formula gives you 4.28 ohms, so we'll go a bit on the safe side and use (2) 2.2 ohm resistors in series for 4.4 ohms. Now, if the supply voltage drops as load increases, ohms law dictates that the current will fall, along with the forward volts. A drop in current or voltage will not damage the LEDs.
Accounting for a rise, the cree LEDs can handle 3.7fv at 1000ma, allowing for a 5% increase in voltage, which is well beyond what any regulated power supply will do.
So I think you can save your money and DIY your drivers. You can use an LM317 if you desire as an additional regulation step, but in doing so with a 24 volt power supply you'll need to drop an LED from the string. The LM317 requires a minimum of a 3v differential between the supply voltage (Vin) and the regulated output voltage (Vout). You need to account for slight power supply variations, which puts your effective output on a 24v supply at 20v max. Next you need that resistor, which has to have some voltage drop across it to maintain current regulation, so that takes off another couple volts. So your maximum fv can only be about 18 volts at this point, meaning you can use 4-5 LEDs at best per driver.
scsinet
Active Member
Check out the STSC1 constant current LED driver chip. If you are able to tackle surface mount soldering, this may be a very viable solution for something more than a simple resistor.
http://www.st.com/stonline/products/...3415/stcs1.pdf
Digikey has these in low quantities for 2.98/each, PN 497-5961-1-ND. By the time you add a surfboard and the supporting components you could probably build them for $10 each. While that's also a big cost, one could argue that having the PWM functionality built in is useful. You could simply bus all of the PWM lines for the blue and white together. For now, simply adding a switch from that to the Vcc would give you simple on/off functionality, but then when you wanted to add microcontroller control, it would be incredibly easy to interface it into a PIC (or a Basic Stamp if you can't write assembly) and use the built-in PWM function to acheive intelligent dimming.
I'm getting excited here. I smoked a JBJ 12g nano ballast the other day... now I'm thinking about tearing it apart and modding a small scale setup and using the PWM control and computer interface... might be a good project to get some advanced control threads off the ground here... and let me contribute to your project...
http://www.st.com/stonline/products/...3415/stcs1.pdf
Digikey has these in low quantities for 2.98/each, PN 497-5961-1-ND. By the time you add a surfboard and the supporting components you could probably build them for $10 each. While that's also a big cost, one could argue that having the PWM functionality built in is useful. You could simply bus all of the PWM lines for the blue and white together. For now, simply adding a switch from that to the Vcc would give you simple on/off functionality, but then when you wanted to add microcontroller control, it would be incredibly easy to interface it into a PIC (or a Basic Stamp if you can't write assembly) and use the built-in PWM function to acheive intelligent dimming.
I'm getting excited here. I smoked a JBJ 12g nano ballast the other day... now I'm thinking about tearing it apart and modding a small scale setup and using the PWM control and computer interface... might be a good project to get some advanced control threads off the ground here... and let me contribute to your project...
mastertech
Member
Originally Posted by SCSInet
http:///forum/post/3220493
Check out the STSC1 constant current LED driver chip. If you are able to tackle surface mount soldering, this may be a very viable solution for something more than a simple resistor.
http://www.st.com/stonline/products/...3415/stcs1.pdf
Digikey has these in low quantities for 2.98/each, PN 497-5961-1-ND. By the time you add a surfboard and the supporting components you could probably build them for $10 each. While that's also a big cost, one could argue that having the PWM functionality built in is useful. You could simply bus all of the PWM lines for the blue and white together. For now, simply adding a switch from that to the Vcc would give you simple on/off functionality, but then when you wanted to add microcontroller control, it would be incredibly easy to interface it into a PIC (or a Basic Stamp if you can't write assembly) and use the built-in PWM function to acheive intelligent dimming.
I'm getting excited here. I smoked a JBJ 12g nano ballast the other day... now I'm thinking about tearing it apart and modding a small scale setup and using the PWM control and computer interface... might be a good project to get some advanced control threads off the ground here... and let me contribute to your project...
Good link and idea. i was quite the dabbler in solid state electronics and have been interested in building my own drivers, pwm signal generators and timer circuits but.... i have two little boys now and i cant devote that much time and energy into a project of that manner. so i used avail componants. but i would sure like to tag along and help out with ideas and support if you were to take on such a project. that would be rightous!
http:///forum/post/3220493
Check out the STSC1 constant current LED driver chip. If you are able to tackle surface mount soldering, this may be a very viable solution for something more than a simple resistor.
http://www.st.com/stonline/products/...3415/stcs1.pdf
Digikey has these in low quantities for 2.98/each, PN 497-5961-1-ND. By the time you add a surfboard and the supporting components you could probably build them for $10 each. While that's also a big cost, one could argue that having the PWM functionality built in is useful. You could simply bus all of the PWM lines for the blue and white together. For now, simply adding a switch from that to the Vcc would give you simple on/off functionality, but then when you wanted to add microcontroller control, it would be incredibly easy to interface it into a PIC (or a Basic Stamp if you can't write assembly) and use the built-in PWM function to acheive intelligent dimming.
I'm getting excited here. I smoked a JBJ 12g nano ballast the other day... now I'm thinking about tearing it apart and modding a small scale setup and using the PWM control and computer interface... might be a good project to get some advanced control threads off the ground here... and let me contribute to your project...
Good link and idea. i was quite the dabbler in solid state electronics and have been interested in building my own drivers, pwm signal generators and timer circuits but.... i have two little boys now and i cant devote that much time and energy into a project of that manner. so i used avail componants. but i would sure like to tag along and help out with ideas and support if you were to take on such a project. that would be rightous!
swimmer4uus
Member
So how would one use the STSC1 module? I see the diagram for a 500mA LED, but what changes would be needed for 750mA? Or even 1000mA. At the moment I'm not that interested, but if you could really lay it out very simply, and the components needed, I might take a shot at a couple of these.
scsinet
Active Member
I've not worked with this chip personally, I am only going by the datasheet.
According to the datasheet, current to the LEDs is set by using an external resistor. However, this resistor, as opposed to a traditional LED/Resistor pair, does not pass the full current of the LED, it simply instructs the logic inside the chip to regulate LED current to a given value.
Section 7.1:
The current is set with an external sensing resistor connected to the FB pin. The feedback
voltage is 100 mV, then a low resistor value can be chosen reducing power dissipation. A
value between 1 mA and 1.5 A can be set according to the resistor value, the resulting
output current has a tolerance of ± 10 %.
For instance, should one need a 700 mA LEDs current, RF should be selected according to
the following equation:
RF = VFB / ILEDs = 100 mV / 700 mA = 142 mΩ (or 0.142 ohms)
This above section tells you exactly what to do for a 700ma drive.
Applying the above formula for a 1000ma drive...
RF = Vfb / Ileds = 100 mV / 1000 mA = 0.1 ohms.
The trick from what I can see is coming up with the resistors. Resistors between 0 and 1 ohms are out there, but they aren't easy to find. I only found a few resistors sized at 0.1 ohms at Digikey that could be ordered in small quantities. You'd almost certainly want to use a high precision resistor for this, such as a 0.5% - 1.0% tolerance.
According to the datasheet, current to the LEDs is set by using an external resistor. However, this resistor, as opposed to a traditional LED/Resistor pair, does not pass the full current of the LED, it simply instructs the logic inside the chip to regulate LED current to a given value.
Section 7.1:
The current is set with an external sensing resistor connected to the FB pin. The feedback
voltage is 100 mV, then a low resistor value can be chosen reducing power dissipation. A
value between 1 mA and 1.5 A can be set according to the resistor value, the resulting
output current has a tolerance of ± 10 %.
For instance, should one need a 700 mA LEDs current, RF should be selected according to
the following equation:
RF = VFB / ILEDs = 100 mV / 700 mA = 142 mΩ (or 0.142 ohms)
This above section tells you exactly what to do for a 700ma drive.
Applying the above formula for a 1000ma drive...
RF = Vfb / Ileds = 100 mV / 1000 mA = 0.1 ohms.
The trick from what I can see is coming up with the resistors. Resistors between 0 and 1 ohms are out there, but they aren't easy to find. I only found a few resistors sized at 0.1 ohms at Digikey that could be ordered in small quantities. You'd almost certainly want to use a high precision resistor for this, such as a 0.5% - 1.0% tolerance.
swimmer4uus
Member
Alright. I'll take a look see at that circuit, and will make a parts list. I have a couple 1W LEDs mounted on starboards. They are rated for 700mA I believe. Was thinking if I did the LM317 route, I could put a couple of the 1watters on the end of the series to kinda "makup" for the 3V loss at the driver.
swimmer4uus
Member
Also, I'm having issues finding out how I'm supposed to impliment the STSC1 into a circuit to drive the LED. I see the one in the data sheet, but I'm definantly not familiar with chip circuit drawings
scsinet
Active Member
Well unfortunately this chip is surface mount. I like chips that you can get in DIP varieties because I can prototype on perf-board and breadboard, but this chip isn't available that way. There are two ways to get around this that I know of... one is to make a PC board for the circuit you need, but this is probably more involved than you want to get. The other is to use a "surfboard" which are little boards designed to accept a surface mount chip, that carries the lines down to a row of pins, and you can plug the whole thing into a perf board for prototyping. This is probably the easiest way, but surfboards add a few bucks... normally not a big deal until you need to build a whole bunch for a small amount of money.
I bet if I do some more searching I can find a constant current LED driver chip that's more flexible for what we're doing. Let me noodle on that.
I'm not sure what you are saying from the LM317 perspective. Basically what I was saying is that 3 terminal regulators, like the 317, require a minimum differential between the input voltage and the regulated out. You cannot feed 24 volts into them and get 24 regulated volts out, it just won't happen. The maximum output voltage you can expect from a 3-term regulator is 3 volts less than the voltage in. For example, if you are using a 7805 (a 5 volt, fixed output regulator), you need to use an 8 volt supply at the minimum. All 3 terminal regulators of this type have this characteristic.
So if you have a 24 volt supply, you can expect a 317 to give you 21 volts maximum. All a 317 does for you is regulate voltage, it does not regulate current. You still need a resistor in series with the LEDs to regulate the current. In order for a resistor to regulate current, you need to establish a voltage drop across it. So when you figure a 3 volt drop in the regulator, and another 3 volt drop (or so) across the resistor, you've dropped your effective maximum LED drive voltage from 24 volts to 18. This means you can not use as many LEDs in a drive chain.
IMO it's a moot point. Your power supply is already regulated. I don't understand why another regulator is needed. All it's going to do is burn energy and produce heat. I wonder if dimming is the aim... the LM317 is an adjustable regulator... it's simple to vary the output using a couple extra supporting components and a potentiometer. Perhaps the individual who pointed you in this direction was using the 317 to dim the LEDs. IMO this is not workable for you. For one thing, because of the way the potentiometer fits into a typical LM317 adjustable output circuit, there isn't a practical way to "bus" the numerous 317s you'd be using together to run off a common dimmer control. For another thing, the 317 is not a PWM device, it's basically resistive... as you dim, it just burns the excess energy as heat, making them incredibly inefficient. For this amount of power and this scale, PWM is the way to go to achieve dimming.
I bet if I do some more searching I can find a constant current LED driver chip that's more flexible for what we're doing. Let me noodle on that.
I'm not sure what you are saying from the LM317 perspective. Basically what I was saying is that 3 terminal regulators, like the 317, require a minimum differential between the input voltage and the regulated out. You cannot feed 24 volts into them and get 24 regulated volts out, it just won't happen. The maximum output voltage you can expect from a 3-term regulator is 3 volts less than the voltage in. For example, if you are using a 7805 (a 5 volt, fixed output regulator), you need to use an 8 volt supply at the minimum. All 3 terminal regulators of this type have this characteristic.
So if you have a 24 volt supply, you can expect a 317 to give you 21 volts maximum. All a 317 does for you is regulate voltage, it does not regulate current. You still need a resistor in series with the LEDs to regulate the current. In order for a resistor to regulate current, you need to establish a voltage drop across it. So when you figure a 3 volt drop in the regulator, and another 3 volt drop (or so) across the resistor, you've dropped your effective maximum LED drive voltage from 24 volts to 18. This means you can not use as many LEDs in a drive chain.
IMO it's a moot point. Your power supply is already regulated. I don't understand why another regulator is needed. All it's going to do is burn energy and produce heat. I wonder if dimming is the aim... the LM317 is an adjustable regulator... it's simple to vary the output using a couple extra supporting components and a potentiometer. Perhaps the individual who pointed you in this direction was using the 317 to dim the LEDs. IMO this is not workable for you. For one thing, because of the way the potentiometer fits into a typical LM317 adjustable output circuit, there isn't a practical way to "bus" the numerous 317s you'd be using together to run off a common dimmer control. For another thing, the 317 is not a PWM device, it's basically resistive... as you dim, it just burns the excess energy as heat, making them incredibly inefficient. For this amount of power and this scale, PWM is the way to go to achieve dimming.
swimmer4uus
Member
Yes, I've been pondering this for a while, and it seems I'm going to have to go another way other than the 317
swimmer4uus
Member
while going to bed last night, I though about different configurations of arrays, while still using my original intended power supply. So now, I reconfigured the array, into 5 by 5. Making each bank 50 LEDs instead of 48. Each series will be 5 3w LEDs. This will allow the cheaper, yet less efficient, LM317 to be utilized. Yes, I know it's not optimal, but it's MY build. hehe
It's officially a build thread now. Ordered the heatsinks today. 3 total. As soon as I get them here, I can start mocking up the bracketry, LED placement, blah blah blah. I'll also try to outline some physical wiring placement, to make them an easy plug and play type system.
It's officially a build thread now. Ordered the heatsinks today. 3 total. As soon as I get them here, I can start mocking up the bracketry, LED placement, blah blah blah. I'll also try to outline some physical wiring placement, to make them an easy plug and play type system.