LDD Driving more then 1000mA

[hearken]

So the search function didn't turn up anything useful for this question. Seems most people what to drive multiple strands with different LDDs. I want to drive one strand with multiple LDDs.

 

Can the LDD drivers be connected in parallel to drive more then 1000mA? For example if I want to supply 2.5A can I connect 2 1000 and a 500 to achieve this on a single strand? Since these are blackbox I don't know what the driver front end looks like. Some drivers can handle this, other's cant.

 

Specifically I'm looking to drive 12 XP-L2's in series at 2.5A.

[Horerczy]

No they don't work like that unfortunately and there are no LDD over a one amp current.

[farkwar]

If you ran 2.5 amps at a typical LED you would pop it, no?

[Horerczy]

 

If you ran 2.5 amps at a typical LED you would pop it, no?

depends on which typical chip were talking about.

[farkwar]

Do you know of a typically used LED for reef lighting that can handle 2.5 amps?

[nemosworld]

Cree XML has the ability to run over 2 amps.

[farkwar]

Would you say that LED is a typical LED over reefs right now? Is it fairly represented in most reef led lighting systems now.

[hearken]

No they don't work like that unfortunately and there are no LDD over a one amp current.

 

Is this documented somewhere or has someone attempted this? Theoretically there should be a diode on the output if it's a buck-boost and a regulator should be able to do this fine. If not inserting a diode should solve the problem. I figured these operated by using a current sense resistor in the output to control the set-point such that a constant current is achieved. If this is the case, a diode on the output before the interconnect should force each regulator to select a voltage which results in it outputting the correct current. After that it's simple circuits. I_load = I_1 + I_2.

 

If no one has tried this, I think I might. Not having >1Amp is unacceptable.

Otherwise I'm going to have to build a driver.

 

Would you say that LED is a typical LED over reefs right now? Is it fairly represented in most reef led lighting systems now.

 

Well some of the higher powered LEDs are moving to >1 Amp. The ones I want to use are rated as 3000mA or 3A maximum drive, so I want to drive at 2500mA or 2.5A. I can't account for anyone else but most of the CREE ones are >1 Amp. The Royal Blue is 1.5A max. I'd drive that at 1.3.

[jedimasterben]

 

Is this documented somewhere or has someone attempted this? Theoretically there should be a diode on the output if it's a buck-boost and a regulator should be able to do this fine. If not inserting a diode should solve the problem. I figured these operated by using a current sense resistor in the output to control the set-point such that a constant current is achieved. If this is the case, a diode on the output before the interconnect should force each regulator to select a voltage which results in it outputting the correct current. After that it's simple circuits. I_load = I_1 + I_2.

 

If no one has tried this, I think I might. Not having >1Amp is unacceptable.

Otherwise I'm going to have to build a driver.

 

 

Well some of the higher powered LEDs are moving to >1 Amp. The ones I want to use are rated as 3000mA or 3A maximum drive, so I want to drive at 2500mA or 2.5A. I can't account for anyone else but most of the CREE ones are >1 Amp. The Royal Blue is 1.5A max. I'd drive that at 1.3.

If you want higher current, I would just suggest buying different drivers. Meanwell ELN-60-27 run 9-27v and 2.3A, can be tuned to around 2.7A or so.

 

The XM-L is really inefficient at 3A, though, less than 85 lumens per watt for the neutral white. The XT-E has a similar current curve, I would not go over 1000mA on those, as once you get over that, their efficiency takes a nosedive.

[MeepNand]

 

Is this documented somewhere or has someone attempted this? Theoretically there should be a diode on the output if it's a buck-boost and a regulator should be able to do this fine. If not inserting a diode should solve the problem. I figured these operated by using a current sense resistor in the output to control the set-point such that a constant current is achieved. If this is the case, a diode on the output before the interconnect should force each regulator to select a voltage which results in it outputting the correct current. After that it's simple circuits. I_load = I_1 + I_2.

 

If no one has tried this, I think I might. Not having >1Amp is unacceptable.

Otherwise I'm going to have to build a driver.

If you build a driver, Newark sells some great driver chips.

As for the thing with diodes and current-sense resistors you mentioned, please try it. Way above me, but it would be pretty cool if it worked.

[evilc66]

LDDs cannot be run in parallel for higher current output. I've tried every conceivable setup to try and get them to work, but the drivers just shut down. The only drivers that I have seen work in parallel are ones that source the current, rather than sink it. National/TI is one of the few chip manufacturers that apply this topology (at least for internal switched drivers), but I have never seen an off the shelf solution with that setup. You're stuck either buying a high current driver, or making your own.

[hearken]

If you want higher current, I would just suggest buying different drivers. Meanwell ELN-60-27 run 9-27v and 2.3A, can be tuned to around 2.7A or so.

 

The XM-L is really inefficient at 3A, though, less than 85 lumens per watt for the neutral white. The XT-E has a similar current curve, I would not go over 1000mA on those, as once you get over that, their efficiency takes a nosedive.

 

I'm targeting the XM-L 2's and planning on driving the XT-E at no more then 1.3A. The L2's specifically have better efficency at higher current then the XT and XM-L1 series. They also cost a bit more.

 

LDDs cannot be run in parallel for higher current output. I've tried every conceivable setup to try and get them to work, but the drivers just shut down. The only drivers that I have seen work in parallel are ones that source the current, rather than sink it. National/TI is one of the few chip manufacturers that apply this topology (at least for internal switched drivers), but I have never seen an off the shelf solution with that setup. You're stuck either buying a high current driver, or making your own.

 

That is interesting. I'm planning on trying the following setup. I'm hoping that should work as there would be very minimal voltage detected by the driver at the input. I'm guessing they get pissed off if they detect current being driven backwards through the output. The diode's would stop that problem.

 

 

Unfortunately I didn't put markers on the image. But the general idea is that the drivers will be concerned only with the current through their diode and not the LED chain (this just happens, IL = I1 + I2).

 

 

Lets say the lower led is really 12 XM-L's and the bottom driver is running 700mA dropping 2.8V across each LED and 0.7 V across the diode. This would mean that the voltage across the LED chain is 33.6V and the driver is outputting 34.3V. When the top (say 500mA driver) comes on-line it will want to push 500mA. Currently it sees no voltage on the output because of the diode. To achieve 500mA through the diode it will need to force the Voltage across the diode to be 0.7V which results in the driver pushing at least 34.3V. However, now that two drivers are running the current through the LED chain has increased and so has the voltage. This means both drivers must increase the output voltage to keep the 0.7V drop across the diode to maintain their rated current output. Assuming they don't go unstable at this point both drivers should settle on something like 36.7V or there abouts. The 500mA driver will be a few millivolts less because the drop across the diode will be slightly lower due to the lower current.

 

 

Either way, I've ordered a 500 and 700 with an LED and will get the diodes at radioshack. Eventually I'll order some fast switching diodes to accommodate with the PWM frequencies.

 

[hearken]

Totally works.

 

It's times like this when I'm glad I got that Electrical Engineering degree.

 

Also, I didn't pull the 500 because it does not like to come back online if reconnected. Don't know why but I doubt anyone would be disconnecting one normally.

 

[EDIT] Wanted to add some notes. The voltage across the LED (XT-E WW) is 3.2V. The whole setup is drawing 6W. The diodes are dropping the expected 0.75V for a total of 3.9V. Assuming 100% conversion that is 6W/3.9V = 1.53A. (watts in = watts out). Now the LDDs are really only 90% efficient at the voltage I'm using and it wasn't exactly 6W so taking that into consideration we are looking at 0.9*6W/3.9V = 1.38Amp which is 700mA + 500mA as expected. Alas I can only find my pocket meter and it can't do current.

 

[EDIT 2] Buck converter efficency

Some may be wondering why the buck converters are less efficient at lower input voltages. Internally there is a power FET which is switching on and off real fast driving current into an inductor which drives the LEDs. This FET is central to the efficiency problem. There are multiple components to this but the primary is the forward voltage Vf. This is the voltage dropped by that switching FET. This voltage can vary with current but is pretty stable.

Power dissipated by a component (W) = Voltage across the component * current through component. At lower voltages the LDDs must draw more current to deliver the same output power (Pin = Pout). This means driving more current through the switching fet. So if at 10V you are drawing 1A and at 20V 0.5A and the FET drops .1V then at 10V the FET is dissipating 100mW where at 20V it is only 50mW.

 

THIS DOES NOT MEAN DRIVE AT MAX VOLTAGE FTW! There are other factors which become a problem at higher voltages, there is an ideal voltage and it is likely listed in the LDD documentation. You should use that.

 

 

[evilc66]

That's good to see that it does indeed work. I was tearing my hair out trying to get them working in parallel, but did not try that approach.

 

It looks like your circuit is a little different than the diagram you posted before. From the picture and video, it looks like you are just using the diodes on the output, and not feeding back the driver output back to the input. Doesn't look like you are using any load balancing resistors either. Can you clarify?

[hearken]

That's good to see that it does indeed work. I was tearing my hair out trying to get them working in parallel, but did not try that approach.

 

It looks like your circuit is a little different than the diagram you posted before. From the picture and video, it looks like you are just using the diodes on the output, and not feeding back the driver output back to the input. Doesn't look like you are using any load balancing resistors either. Can you clarify?

 

So those resistors are part of the LDD. I only had a symbol for a voltage controlled current source which can be used to model a fixed current supply if you put a resistor in the output. Everything before the diode is part of an LDD, ignore it.

 

Load balancing is inherent in the LDD. In my case I have a 5/7 provider ratio. The 500 supplying 5/12ths of the power while the 700 is taking 7/12ths. The LDD internal controller handles maintaining this balance via the diodes. IE each device wants to drive it's rated current through the diode attached to it. To achieve this it must overcome the voltage across the LED chain + the forward voltage of the diode. When the second device powers on, the voltage across the chain increases thus increasing the required drive voltage on the first device. The controller in each device will adjust the drive voltage to maintain it's rated output independently.

 

If people are interested I'm planning on developing a PCB board which contains the diodes (with bypasses if you don't need them, they waste power). Each output terminal will have 2 positions allowing for chaining and sockets for up to 8 LDDs. This will allow you to choose which LDDs and how they are connected in parallel. I'll likely also throw on a MCU for dimming with a header for a standard bluetooth module you can get off of amazon for 10$. I currently use one to communicate with my DIY tank controller.

 

My current design is asking for 7LDDs for 2.5, 1.2, .5, and .3 AMP drive pushing 36LEDs (12 XM-L2 (2.5A), 12 XT-E (1.2A), 6 Deep Red (.5A), 2 Hyper Red(.3A), 4 True Violet(.5A)) and will cost about $80 without the DC supply.

 

IMPORTANT NOTE - Normally when attaching buck converters in parallel you slave the switching clocks to avoid inducing low frequency ripple on the output. This is caused by minor differences in the switching frequencies, on the order of a few Hertz. I don't know what kind of filtering these devices have but it is STRONGLY RECOMMENDED that you DO NOT EXCEED 80% max capacity. This should provide enough wiggle room that if the converters start "beating" it will not blow out your LEDs.