How to DIM LED with Constant Current Driver?

[h2k]

How does one dim with a constant current driver, but not via PWM. There is a specific reason I do not want to use a PWM.

So it seems almost I would somehow have to adjust the constant current to a lower value. It is OK if the dimming is not linear. Basically I want the current to go from max value to variable lower values...if that is even possible.

I have access to high power pots, but not sure how that would help. Maybe the only solution is to get lower current drivers?

 

[(*steve*)]

How does one dim with a constant current driver, but not via PWM. There is a specific reason I do not want to use a PWM.

You reduce the current

So it seems almost I would somehow have to adjust the constant current to a lower value.

You beat me to it!

It is OK if the dimming is not linear.

What are you trying to dim?

If it's LEDs, you might find that the problem is your eyes. (they are not linear)

Basically I want the current to go from max value to variable lower values...if that is even possible.

If you removed the word "variable" from that sentence it would make more sense.

Do you want to be able to change, or perhaps set, the minimum lighting intensity?

I have access to high power pots, but not sure how that would help. Maybe the only solution is to get lower current drivers?

Can you describe more exactly what you're doing.

"Power pots" are very rarely the solution

 

[h2k]

I guess what I would like to do is dim the LEDs by adjusting the current (but still using a constant current drive). I know how to do it via constant voltage source and big pots, but that has a host of issues as you stated.

Lets assume that pwm is not an option. Aside from plugging in lower current drivers or physically blocking some of the light, not sure how to do it. Neither of those options is great, but they are last ditch efforts.

 

[(*steve*)]

The constant current driver will have some sort of sense resistor. If you are lucky it will have a comparator to compare the sense voltage against a reference. If you vary the reference voltage, you vary the current.

This is better than varying the sense resistor which (a) may have a very low resistance) and (b) carry the full load current.

Depending on the constant current source it may already be a switchmode device (which is essentially using PWM and an inductor to maintain a constant current)

 

[h2k]

1 - So if it is a switchmode device, will the current be PWM? I would guess not, but I want to make sure. The current itself would need to be a true constant current.

2 - You totally lost me on the rest of it. How would I vary the reference voltage. In other words, what would I need to do to actually implement a current change in the driver?

 

[(*steve*)]

Perhaps instead of talking about what-if's and maybe's, how about you show us some pictures of the constant current driver for the LEDs

 

[h2k]

Currently have this:

http://www.amazon.com/ROBERTSON-3P3...8-1&keywords=robertson+constant+current+700ma

Thinking about ordering this:

http://www.amazon.com/ROBERTSON-LD0...-14&keywords=robertson+constant+current+700ma

This is in transit:
http://www.ebay.com/itm/Mean-Well-M...990?pt=LH_DefaultDomain_0&hash=item58960dc976

But in the end, I can purchase what I need for the project. I want a total power of about 200W (min 100W) at 700mA so cost can be issue if the drivers are expensive (since it will take several).

 

[(*steve*)]

The first one seems to allow you to use a regular light dimmer to vary its output. The other two probably don't.

As they are mains powered, I would not recommend trying to add any form of adjustment to vary the current. since large parts of the internal wiring are at mains potential.

 

[h2k]

Great. I have this dimmer and tried it out with the first driver. It works just fine and the multimeter showed that the current was being adjusted from 0-700mA.

http://www.ebay.com/itm/Rotary-Dial...672?pt=LH_DefaultDomain_0&hash=item4ac8d8ddd8

I guess the questions I have are:

1) Does the driver have anything to do with pulsing of the current or is it truly constant? I am guessing it is truly constant.

2) Out of curiosity, what does this dimmer actually do? Is it a just a pot or is it something else.

3) Should the dimmer be placed between the wall outlet (US) and the driver or between the driver and LED's? I placed it between the outlet and the driver and it worked fine. If placed between the driver and LED's it was a no go. I guess that answers the question, but confirmation that this makes sense would be good.

4) I suppose this would also work? It is more compact for me:
http://www.ebay.com/itm/Rotary-Dial...672?pt=LH_DefaultDomain_0&hash=item4ac8d8ddd8

 

[(*steve*)]

1) Does the driver have anything to do with pulsing of the current or is it truly constant? I am guessing it is truly constant.

For better quality drivers the current is closer to constant. Poorer ones have more ripple

2) Out of curiosity, what does this dimmer actually do? Is it a just a pot or is it something else.

It's a lot more than a pot. It chops the AC waveform so that only part of each half-cycle goes through. As you turn the knob less and less of the half cycle goes through, and the driver reacts to this by reducing the current.

Note that not all drivers will do this. They must be "dimmable"

3) Should the dimmer be placed between the wall outlet (US) and the driver or between the driver and LED's? I placed it between the outlet and the driver and it worked fine. If placed between the driver and LED's it was a no go. I guess that answers the question, but confirmation that this makes sense would be good.

The device I had in mind goes between the incoming mains and the driver.

Similar devices can go the other side, but they would be PWM. (actually there are technical reasons why you generally wouldn't do this.

4) I suppose this would also work? It is more compact for me:
http://www.ebay.com/itm/Rotary-Dial...672?pt=LH_DefaultDomain_0&hash=item4ac8d8ddd8

It seems it would.

 

[h2k]

Is the ripple in the current detectible via multimeter? Or is it so fast (like PWM) that it may not be detectible? That way I could at least compare the drivers.

That is cool about the dimmer. I didn't realize it was that capable. The only problem I am seeing is that the dimming drivers are usually relatively low power compared to the non-dimming. This seems to be decent option, at least on paper. But not sure about the quality. I like the robertson ones since they are at least brand name, but not very much power.

http://www.ebay.com/itm/LED-Driver-...679?pt=LH_DefaultDomain_0&hash=item1c3218f11f

 

[(*steve*)]

Is the ripple in the current detectible via multimeter? Or is it so fast (like PWM) that it may not be detectible? That way I could at least compare the drivers.

No, it's too fast to be detected on a multimeter. The big test is whether you see it.

Lights that are moving (or where there is relative movement between your head and the light) can make it more obvious.

That is cool about the dimmer. I didn't realize it was that capable. The only problem I am seeing is that the dimming drivers are usually relatively low power compared to the non-dimming. This seems to be decent option, at least on paper. But not sure about the quality. I like the robertson ones since they are at least brand name, but not very much power.

http://www.ebay.com/itm/LED-Driver-...679?pt=LH_DefaultDomain_0&hash=item1c3218f11f

Dimmable drivers are also more expensive. But if that's what you need, they are the correct choice.

I can't comment on the quality of the device you've linked to.

 

[h2k]

Well, this is where it gets "interesting". The application is for light therapy (assist in the healing of skin tissue). The response of the eye (which is the typical way of judging) may not be relevant in this case.

So how much ripple would be typical (+/-10%, 40%, 100+%)? I am not sure how the tissue will respond to this ripple and the only data I have would be for a true constant current. This is also why I do not want to use PWM. With that in mind, what do you think?

I suppose the fall back plan is to use a constant voltage source with big pots, but that is not very elegant. With good heat sinks, it seems to work OK, but it is certainly a bigger pain to set up.

 

[(*steve*)]

I would suspect that for light therapy, ripple would not be a concern.

If you're using PWM, you could filter the output (use an LC filter) to vet a variable voltage output.

(And here's why it's good to know why you don't want to use PWM -- I was assuming it was due to interference)

 

[h2k]

I guess I am confused. Let me start with with PWM. Lets say I have a true 700mA constant current source (hypothetical). If it is set to have an "effective" result of reducing the light by 50%, the "effective" current is 350mA. In reality, it is just alternating 0 and 700mA to achieve something that is effectively 350mA. But how would the tissue respond to this sort of pulsing? I really don't know. If I wanted the effect of 350mA, it seems a constant 350mA source would be better than 700mA with PWM (or at least more predictable w.r.t. tissue response).

Therefore, I think it needs to be a true constant source of light (with maybe 10-20% variation). That said:

1) With a "constant current" device, how much variation in actual current could there be?

2) I don't know anything about LC filters. But assuming (1) yields an acceptable solution, it would seem that the dimmer would be OK. Of course, then one is forced to get a dimmable driver.

 

[(*steve*)]

I guess I am confused. Let me start with with PWM. Lets say I have a true 700mA constant current source (hypothetical). If it is set to have an "effective" result of reducing the light by 50%, the "effective" current is 350mA. In reality, it is just alternating 0 and 700mA to achieve something that is effectively 350mA.

Yep, PWM works like that.

But how would the tissue respond to this sort of pulsing? I really don't know.

Without someone doing an experiment, you can't really know.

I would presume it makes no difference unless it could be shown otherwise.

If I wanted the effect of 350mA, it seems a constant 350mA source would be better than 700mA with PWM (or at least more predictable w.r.t. tissue response).

Yeah, maybe.

1) With a "constant current" device, how much variation in actual current could there be?

The only way og knowing is to measure it (well, if you had the schematic you could also calculate it.)

2) I don't know anything about LC filters. But assuming (1) yields an acceptable solution, it would seem that the dimmer would be OK. Of course, then one is forced to get a dimmable driver.

I think the dimmable driver is the easiest solution, and is essentially done with off-the-shelf parts that don't require an awful lot of special skills to put together.

 

[h2k]

Steve, w.r.t. the following quotes:

------------------------------
"Depending on the constant current source it may already be a switchmode device (which is essentially using PWM and an inductor to maintain a constant current)."

"For better quality drivers the current is closer to constant. Poorer ones have more ripple"

"No, it's too fast to be detected on a multimeter. The big test is whether you see it."

1) With a "constant current" device, how much variation in actual current could there be?
The only way og knowing is to measure it (well, if you had the schematic you could also calculate it.)
-----------------------------

My concern is a general one about constant current drivers (of which I do not truly understand). You say they are essentially PWM with an inductor. Does this mean the end result is something that is nearly a real constant current or is something that is "effectively" a constant current? In other words, would a 700mA driver fluctuate no less than 600-800mA or could be something as high as 1400mA and effectively turned into 700mA.

The problem is that the eye is not a reliable sensor for this application and you say that "ripple" can't be measured via multimeter.

Similarly, is the reduced current found via a dimmable driver a "constant" type or is it some "effectively constant" current. I am OK with minor variations in current (say +/-20%), but is it possible to have something more significant than that (even without dimming) via constant current driver?

 

[(*steve*)]

If something is called a constant current driver, then one would expect close to a constant current.

It might be labelled 350mA and give you 342mA, varying only a couple of uA over the short term (say seconds), but vary by 1mA per degree C of ambient temperature.

Or It might be labelled as 350mA and vary at a rate of 20kHz between 347mA and 351mA. (and vary in some other way depending on temperature, input voltage, load, phase of moon...)

If something varies by as much as 20%, it would be absolute rubbish. But I can't say they don't exist. For example, a "very cheap" constant current source consisting of a resistor driving back to back LEDs from AC (think cheap Chinese Christmas lights) might have a nominal current of 20mA, but fall to zero and rise to almost 30mA, 100 times per second. I'm not sure if you call that 100% variation, but it's a lot and more than 20% by almost any means of calculation.

Without the circuit diagram, specs you can trust, or measurement, it is impossible to say more.

I'm not sure what else I can say.

OK, one more thing I can say. Note that I said "specs you can trust", not "specs". I would be very wary about taking on face value any specs from sellers on eBay, AliExpress, your local hardware store, supermarket, etc.

 

[h2k]

I wasn't sure if the constant current was achieved via a modulation of sorts (sort of like PWM). But I suppose not, which is a good thing!

This unit seems to be pretty good, but not much detail about it:
http://www.robertsondirect.com/LD022C070LRAP-P16748.aspx

This seems pretty good too even though it is not dimmable. But it has pretty detailed specs. I don't really get the power switching and PWM aspects of it, but as long as the end result is a constant current, I don't care too much.
http://www.ledsupply.com/docs/lpc-35-700.pdf

 

[Fish4Fun]

h2k,

I think stepping back for a moment will help. Steve has a fabulous grasp on PWM and is explaining it using precise diction, I think you need an analogy. Inductors and capacitors are similar to water tanks in that they "store something". A water tank obviously stores water while Inductors store Current and Capacitors store Voltage. Both Inductors and Capacitors are rated on their storage ability, much like a water tank is measured by its ability to store water. If a pump supplies a water tank it can be designed to begin filling the tank only when the tank is near-empty or it can be designed to maintain the tank a particular level. In both cases emptying and filling the tank take some measurable amount of time. If you want 2 gallons of water per minute to flow from the tank, it doesn't really make any difference if your pump can supply 2.1 gallons per minute or 200 gallons per minute as long as the pump keeps the water level where you want it to be it is considered "sufficient", and you get an uninterrupted supply of water at 2 gallons per minute from the tank.

In current controlled PWM there are several design considerations involving input voltage/current (pump output capacity) and load requirements (volume of water leaving the tank). Assuming your PWM driver and Input source are capable (>2 Gallons per minute in our analogy) then the output current will be constant, what will vary is "how long the pump runs". As LEDs are current dependent devices (based solely on the volume of water leaving the tank, not the pressure of that water) current controlled PWM is really the ONLY way to drive them reliably at a specific current (regardless if that current is "variable".)

Now, for the real explanation of what PWM does: A PWM cycle begins with an under-current event, this triggers a switch to energize an inductor. The time it takes for a particular amount of current to flow through an inductor is defined by L = dI/dt, or rearranged dI = Ldt. Without getting into the calculus this solves to: I = E/R * (1 -e^(-t*R/L)) (you might verify this here: http://hyperphysics.phy-astr.gsu.edu/hbase/electric/indtra.html ). This demonstrates that there is some finite period of time when the inductor is "absorbing energy" much like our water tank "absorbs water". Once the current through the inductor reaches our "set point", the switch is turned off. Over a different period of time the energy "absorbed" by the inductor is released. We won't get into the vagaries of how this occurs, but take it on faith that it can be achieved, and that the result is current now being supplied to the LEDs not by the source, but by the energy stored in the Inductor. When the current stored in the Inductor falls below a certain point the switch is again turned on, and current is supplied to the LEDs directly from the source through the inductor; meanwhile the Inductor is again "absorbing" energy (ie charging) and when the current through the LEDs again reaches "max" the switch turns back off and the current is again supplied from the energy stored in the inductor. The "difference" between the "High" and "Low" currents can be very small and is called "ripple". The "Ripple Current" is typically designed to be in the mA range for quality PWM drivers. In comparison, because of the thermal properties of LEDs,using resistors typically results in much larger ripple currents albeit over longer periods of time (seconds versus micro-seconds).

So, rest assured, PWM LED drivers are the best way to go

Fish