Led amperage basic question

[SemiMan]

Perfect, thanks. What about if I use 3 1 amp bucks in parallel to drive a 3amp led, then dynamically switch the circuit to exclude 2 of them and drive a different 1amp led? Would that work? I understand each buck might have a +-5% margin on it's output, so it would increase the margin the more bucks I add, but just wanted to know if the theory is sound. Thanks.

You described a 5V, 4A buck regulator. You would not use that to drive an LED. That is a constant voltage output product and you need constant current for the LED. One of several things may happen if you did this. The unit would limit the output to 4amps (unlikely), the unit would shut down for any variety of reasons (possible), the unit would drive considerably more than 4 amps trying to get up to 5V and then either blow the LED, fail, or engage a thermal shut-down from over heating.

Most LED drivers, i.e. buck-pucks can be paralleled for more current. No guarantees. Best to ask the supplier.

Semiman

 

[meyerovb]

I'm trying to understand how paralleling buck regulators would work. If I parallel 3 1amp bucks (that support being paralleled), would that give me a 3amp current to my led?

 

[ianfield]

I'm trying to understand how paralleling buck regulators would work. If I parallel 3 1amp bucks (that support being paralleled), would that give me a 3amp current to my led?

Assuming each of the bucks had a higher voltage than required for the LED - you would be using a current limiting resistor in series with each buck output, as well as protecting the LED from drawing too much current, the resistor for each buck would help share the current draw for each regulator.

Its important to take into account; the smaller the voltage headroom you have, the smaller the resistor to get the required current, so regulation accuracy is more important. As the LED Vf is pretty much fixed, any voltage deviation is largely expressed across the resistor - small changes in voltage across a low resistance can mean significant changes in current.

 

[ianfield]

PS - parallel operation of buck modules designed for driving an LED directly should probably be OK, as they should be rigged for constant current not constant voltage - that is an imprtant point to make sure of, and no balancing resistors are needed.

If the regulators are constant voltage; due to manufacturing tolerances, one may have slightly higher voltage than the others - it will then try to to supply more than its share of the load.

 

[meyerovb]

Let's say I'm only referring to 3 constant current 3.7V1A bucks being driven by a single 12V5A ac-dc switched regulated wall plug psu. In that case, can I just wire all 3 bucks to the psu, then wire the single xm-l to all 3 bucks? Will that drive the led at 3A without overdriving? Will the circuit need more components (like a resistor) in between somewhere?

 

[ianfield]

Let's say I'm only referring to 3 constant current 3.7V1A bucks being driven by a single 12V5A ac-dc switched regulated wall plug psu. In that case, can I just wire all 3 bucks to the psu, then wire the single xm-l to all 3 bucks? Will that drive the led at 3A without overdriving? Will the circuit need more components (like a resistor) in between somewhere?

As long as your buck modules truly are constant current output, there shouldn't be any problem connecting the outputs in parallel to make up the full rated forward current of your LED.

Adding resistors in that case would simply waste the power dissipated in those resistors.

 

[SemiMan]

meyoerovb,

Do not refer to it as a 3.7V, 1A buck regulator. Refer to it as a buck regulator perhaps with a regulation range from 2.5 - 8V .... i.e. it could drive 1-2 LEDs of varying specs. It will adjust the output voltage to hit the set point of 1amp. Don't worry about the exact output voltage, you just need to ensure the product covers the range of the LED(s).

Semiman

 

[ianfield]

meyoerovb,

Do not refer to it as a 3.7V, 1A buck regulator. Refer to it as a buck regulator perhaps with a regulation range from 2.5 - 8V .... i.e. it could drive 1-2 LEDs of varying specs. It will adjust the output voltage to hit the set point of 1amp. Don't worry about the exact output voltage, you just need to ensure the product covers the range of the LED(s).

Semiman

You've just described a constant current regulator.

I've just thought of a possible complication with ganging several constant current LED drivers, the buck circuit is just the same as any other, but the regulation circuit senses the current drawn by the load rather than the buck output voltage - this is usually done by a current-sense resistor (internal to the regulator) inserted in the LED cathode return.

I'm not exactly certain what effects manufacturing tolerances will have on 3 ganged bucks (a single 3A one would be far better!) but it might be as well to start by ganging just 2 and check nothing overheats before adding the third.

 

[SemiMan]

You've just described a constant current regulator.

Yes that was my point! .... the op was not quite understanding.

Some regulators will allow ganging, some will not. The issue will not be how they sense current, but the feedback loops and whether they co-operate with each other. Most will some will not. Adding a bit of extra resistance is usually all that is needed to make them work.

 

[argleargle]

You've just described a constant current regulator.

A true constant current regulator would be what we call a "buck/boost driver." We will all "get on the same page" eventually. Nobody panic! :)

I'm not exactly certain what effects manufacturing tolerances will have on 3 ganged bucks (a single 3A one would be far better!) but it might be as well to start by ganging just 2 and check nothing overheats before adding the third.

The new crop of some leds supposedly can run overdriven at 2.8A (each.) 2x1A bucks should be fine if we are talking of driving a single recent-high-tech led. This is a different story if the hypothetical experimentor is attempting to run 2 or more LED in parallel. I'm basically agreeing that 2x1a buck to feed one led COULD work, depending on a few things.

If we're talking about running a parallel-wired triple star of some recent leds, 3x1a bucks WITHOUT PULSE MODULATION should work. A lot of these stars are only "advertised" running at 100-500 mA, though.

This can easily get confusing and complicated, particularly if we are all talking about different components. The simplicity of one proper driver to one proper led is kind of nice. Some drivers support multiple output options, which is also nice.

A driver that uses PWM (pulse width modulation) would probably act really strangely when used in a gang to drive one LED.

Don't think of a LED as a "resistor" with a constant resistance. It's actually a reactive component, with its electrical properties changing based on its drive at a particular instant and even temperature and such.

Short version: When you feed several simultaneous waveforms (pulse width modulation) to a single LED, the math gets REALLY WEIRD. It's because an LED is "dynamically reactive" and not "uniformly resistive." You can't really say that an LED is "3.7V." That wouldn't be accurate.

Right? Let's have some questions. If I've said anything wrong, standard disclaimer: jump in and let's hear it.

 

[Steve K]

A true constant current regulator would be what we call a "buck/boost driver." We will all "get on the same page" eventually. Nobody panic! :)

"buck/boost" refers to a type of switching power supply that can handle input voltages that can be both above and below the voltage at the output. With a buck regulator, the output voltage must be less than the input voltage.
Whether a converter regulates voltage or current is just a matter of the feedback method.

Don't think of a LED as a "resistor" with a constant resistance. It's actually a reactive component, with its electrical properties changing based on its drive at a particular instant and even temperature and such.

a reactive component is either a capacitor or an inductor. A purely reactive component only occurs on the pages of a text book, as any real device has both real and reactive elements. For example, a review of a datasheet for a capacitor will show that it has lead inductance, lead resistance, and shunt resistance. Likewise, diodes like an LED are primarily resistive with some lead inductance, some shunt capacitance, etc. The relationship between voltage and current does change as the LED current changes, but that is not considered "reactive".... at least not in the terms of the electronics world.
To appreciate the difficulty of describing all of the weird behavior going on inside a diode, take a look at how SPICE models one. It's surprisingly complex, and is why it's fairly hard to get a good model for whatever LED you happen to be working with.

If I've said anything wrong, standard disclaimer: jump in and let's hear it.

I wouldn't say you were wrong, but a little clarification was in order. :)
When working with an audience as diverse as the typical CPF group, it's hard to explain complicated concepts without knowing the level of education of people you are talking to. In regards to the basic problem of running switching power supplies in parallel, I think the only answer is to ask the people who know.. i.e. the manufacturer of the device that you are considering using. Anything else is just speculation.

 

[argleargle]

Great post, Steve! I used the words "reactive component" and compared to a regular resistor to get the basic idea across. I wasn't about to start talking about "imaginary numbers" and such with resolving waveforms in a reactive circuit.

I just hope we didn't blow OP's mind with all of this at once! :)

 

[ianfield]

A true constant current regulator would be what we call a "buck/boost driver." We will all "get on the same page" eventually. Nobody panic! :)

You're absolutely wrong about that! A buck/boost regulator can transition between the two topolugies to cope with input below as well as above the required output voltage - such as battery equipment, like 5V TTL running off 5x NiMh, or a 3W LED in a properly designed 3D flashlight.

That's not to say a buck/boost converter can't be configured for constant current output - but I've yet to see one in a practical application.

 

[argleargle]

A true constant current regulator would be what we call a "buck/boost driver."

You're absolutely wrong about that! A buck/boost regulator can transition between the two topolugies to cope with input below as well as above the required output voltage

Wouldn't such a thing be required to regulate current regardless of source being above or below desired output? I think of a "buck" or a "boost" driver as being only half of a solution, so to speak. A buck/boost should theoretically regulate under both conditions. Yes, there are two "halves" to the design. Aren't we saying the same thing with different words? Perhaps we misunderstand each other.

That's not to say a buck/boost converter can't be configured for constant current output - but I've yet to see one in a practical application.

I am not saying that a buck/boost combination driver is required to regulate to the same value of current regardless of its current mode of operation.

 

[Steve K]

I'm open to hearing about altenative meanings for buck/boost, but I'm using what is standard for the electronics industry. For example, this is what Linear Technology uses as a header for their selection of buck/boost switching regulators:

"Linear Technology offers true buck-boost synchronous DC/DC converters. With their unique internal four MOSFET switch combination, these switching regulators can seamlessly transition from step-down mode, through 100% dropout operation and then to step-up mode to allow a fixed output voltage even when the input voltage fluctuates above or below the output."

 

[Illum]

I'm not exactly certain what effects manufacturing tolerances will have on 3 ganged bucks (a single 3A one would be far better!) but it might be as well to start by ganging just 2 and check nothing overheats before adding the third.

Unless you are dealing with linear drivers, this setup will not work. Reason is this, Switching bucks employ a sense resistor between the feedback pin and ground. Paralleling three 1A bucks would mean all three bucks have their sense resistors decreased to about 1/3 of their original value [as in the LED cathode now has three separate routes to ground]. Due to the internal variations between drivers, they will not "mutually yield" 3A in 1A partitions. Of the three bucks, the one with the lowest impedance will produce the most current. At some point the current threshold will "burn through" the internal mosfet of the chip designed to provide 1A max. The controller used to drive the Mosfet may fry trying to compensate for it. Depending on your duty cycle you might end up with a cascading failure if you tried to jolt an LED using three separate bucks.

 

[argleargle]

Thanks for mentioning cascade failure, Ilum. Excellent and on-topic points. People could do worse than to re-read what you've said, IMHO.

Steve, thanks much for your contributions to the thread (and this isn't even my thread!) I hadn't noticed your stuff before this thread, but now I know to keep an eye out. We have had some good discussion sparked here!

I just hope that we haven't scared off OP (Meyerovb.) Don't be scared! Most of us don't bite (much.) :)

Since the thread was started with a basic fundamental questions, perhaps we can get back on track with something like "tips and suggestions" while we wait for more questions? I'll start.

Tip: Set your multimeter to amps (ammeter,) and put it in series (not parallel) with your circuit when you're playing around on your test bench. The results might surprise you. Electronics people throw around terms like "1 amp" and so on, but the real measured value can be quite different. If there is a problem with the circuit, watching the current through it as you test and play with it might tip you off so you can shutdown before there's a failure. (Hopefully.) It's easy to forget that what's happening in the circuit might be different than what you think.

 

[Steve K]

Tip: Set your multimeter to amps (ammeter,) and put it in series (not parallel) with your circuit when you're playing around on your test bench. The results might surprise you. Electronics people throw around terms like "1 amp" and so on, but the real measured value can be quite different.

Good advice, but you'll only make that sort of mistake once (or you'll never tell anyone about the second time).

I'd offer the tip that there's no substitute for an oscilloscope when troubleshooting your circuit. You'll be amazed at all of the weird stuff going on that you never suspected was there.

 

[ianfield]

Unless you are dealing with linear drivers, this setup will not work. Reason is this, Switching bucks employ a sense resistor between the feedback pin and ground. Paralleling three 1A bucks would mean all three bucks have their sense resistors decreased to about 1/3 of their original value [as in the LED cathode now has three separate routes to ground]. Due to the internal variations between drivers, they will not "mutually yield" 3A in 1A partitions. Of the three bucks, the one with the lowest impedance will produce the most current. At some point the current threshold will "burn through" the internal mosfet of the chip designed to provide 1A max. The controller used to drive the Mosfet may fry trying to compensate for it. Depending on your duty cycle you might end up with a cascading failure if you tried to jolt an LED using three separate bucks.

Most implementations I've seen have a rectifier/reservoir cap arrangement, so the voltage developed across the current sense resistor for the chip's sense input, should be (mostly) DC - but there could be large ripple which could cause problems, so beware.

I did point out in an earlier post; that finding a single 3A buck would be a better idea than ganging 3x 1A bucks.

My best guess is - try 2 & see what happens, if it works then see if you can get away with 3.

 

[ianfield]

Thanks for mentioning cascade failure, Ilum. Excellent and on-topic points. People could do worse than to re-read what you've said, IMHO.

Steve, thanks much for your contributions to the thread (and this isn't even my thread!) I hadn't noticed your stuff before this thread, but now I know to keep an eye out. We have had some good discussion sparked here!

I just hope that we haven't scared off OP (Meyerovb.) Don't be scared! Most of us don't bite (much.) :)

Since the thread was started with a basic fundamental questions, perhaps we can get back on track with something like "tips and suggestions" while we wait for more questions? I'll start.

Tip: Set your multimeter to amps (ammeter,) and put it in series (not parallel) with your circuit when you're playing around on your test bench. The results might surprise you. Electronics people throw around terms like "1 amp" and so on, but the real measured value can be quite different. If there is a problem with the circuit, watching the current through it as you test and play with it might tip you off so you can shutdown before there's a failure. (Hopefully.) It's easy to forget that what's happening in the circuit might be different than what you think.

Bear in mind that the volt drop across the multimeter current shunt can be significant - especially with a regulator disigned to drive a single LED.

There might be a good reason why the results are quite different to what you were expecting.

I remember trying to measure the current on a NiCd charger - the volt-drop on the DMM test leads alone caused the charger to display a fault warning.