I was wondering if someone could give me some advice on my project?
I have 10 aquarium LED lights which consist of a constant current driver and a waterproof LED bar (constant current LED's in resin on an aluminium bar). These lights are great but are far too bright for the sensitive plants which are growing under them.
My plan was to replace the 10 drivers with a single large one that has 0-10V dimming so I can control it from the Arduino.
Not all of the LED lights are identical, some are slightly longer and have different drivers:
8x 24W 12-36V 700ma
2x 27W 12-27V 1050ma
Do I need to replace all 10 drivers with individual dimmable ones, or can I use a single driver with the 10 lights connected in parallel to it?
If I used a single large 700ma driver, would the 1050ma LED strip just be dimmer?.
You can't use 0 to 10V for LEDs, they are, as noted, constant current. You can't generally connect them in parallel either unless the current limiting device is built into the LEDS.
If I have understood you correctly the LED strips do not have built in current limiting, they rely on the driver controlling the current. That being the case each needs its own driver or you can put them in series with a much higher voltage constant current driver. If they are in series they must have the same current rating.
I have 8 LED ceiling lights with constant current drivers. I control their brightness using N channel MOSFETs in the negative supply to the lights, one MOSFET to 2 lights, PWM to the MOSFETS. This works perfectly. I can't guarantee it will work for you but maybe worth trying. Schematic below...
Here is the schematic of the light controller. Please note:
When I hand draw schematics as I did this one I use a simplified symbol for a MOSFET because it's easier to draw and I am (usually) the only person who will see it, so please, no criticism of the incorrect MOSFET symbols!
The MOSFETs shown are not logic level, this is a mistake I have been going to correct 'next week' for about a year
The part that is of interest to you is the top left corner with 4 MOSFETS. Above them are 2 LEDs to each MOSFET, these are the LED lights. Above those are the constant current power supplies, the negatives from the power supplies go to where it says 'white lamp 0V'.
The 330R resistors, diodes and LEDs connected to 'white 12V' are mimics for the main lamps. Note the diodes are needed to protect the LEDs from excessive reverse voltage from the main lamp supplies.
0-10V is a common control protocol (0-10 V lighting control - Wikipedia), which allows drivers to dim from 0 - 100% by supplying a 0-10V analog signal voltage. I found a module that converts PWM to this so can control a driver with it easily.
Correct, each LED strip is 40cm long and constant current so no limiting resistors.
Some information on your dimming method would be helpful, I have over 50 meters of voltage driven LED tape that I control with PWM on the output that's worked very well for several years, I was under the impression you could not PWM the output from a constant current driver?
This I know! I thought you were going to drive the LEDs with 0 to 10V, but clearly I misunderstood.
You can't have 1 constant current driver and LEDs in parallel from that one current source, the current won't divide evenly, it will go to whichever LEDs have the lowest Vf, even if that is only a tiny difference caused by manufacturing tolerances, difference in temperature or whatever. You might get away with it if you put low value resistors (maybe 1R) in series with each LED lamp, don't know, never tried. As noted below, I believe in experimenting...
I was under the impression you could not PWM the output from a constant current driver?
This hobby is about experimenting. I tried it and it worked. Might have ended up with smoke or flickering or plain not working, but it worked fine. No guarantee it will work with your drivers.
I'll have a play - I was lazy and used a bunch of motor controllers as they were so cheap, I removed the 555 timer and feed a PWM signal from the arduino, then switch a relay on when they are at 100% to bypass the MOSFET. Works perfectly to simulate a dawn\dust cycle on my voltage driven LED's.
Note that you cannot use PWM to externally dim a circuit including a constant-current driver. This is only possible if the driver itself is specifically designed to be controlled by a separate PWM control input.
If you attempt to use PWM to switch either the input or the output of a constant-current driver, you will have major problems - potentially quite dangerous.
Sorry Paul, but you can use PWM on the output of at least some current sources, I know, I am doing it. Has been working several years. I don't claim it will always work, just that it can work.
Well I did and nothing bad happened! I had no problems at all, it just worked exactly as I hoped it would. I did have concerns that it might not work, but I was quite comfortable trying it. Obviously I cannot promise it will always work for any constant current source, but it certainly works for some.
Have you looked at the voltages on the output of the constant current supply output on an oscilloscope? You might very well have overshooting voltages that do not show up as damage in the short term but will cause damage in the long term.
Hi Mike,
I can't remember if I did or not! However I built this several years ago and I've never had any problems with it, other than caused by not using logic level MOSFETs, a silly mistake I have not yet fixed (maybe next week...).
The point is that if you have a capacitor on the output of a constant-current driver and you switch that same output with PWM, then what happens is that with the load disconnected, the capacitor will charge to a much higher voltage so that when the load is again connected, it will draw a much higher current as the capacitor discharges so that the average current will - because it is after all, a constant-current supply - be maintained.
So the PWM will for one be ineffective in dimming the LEDs without being limited by some other malfunction in the constant-current driver (such as being unable to charge the capacitor any more since it has charged to the actual input voltage) and will be substantially over-driving the LEDs. To a certain extent, the LEDs may simply tolerate being driven with a greater current over a lesser duty cycle.
As this is generating more interest than I ever imagined I've got some oscilloscope traces for you.
These are all taken at the drains of the MOSFETs
This is 2 of the outputs at different light levels; the yellow trace is dim, note the short on pulse and the much longer off period. The blue is on about 500‰.
I checked the outputs of the constant current drivers, they have 100μF capacitors across them, so yes, a 100μF capacitor is being dumped into the LEDs, however, there is about 10m of cable between the LEDs and the capacitors, so there is some resistance to limit the instantaneous current.
I repeat what I have said, this hobby is about experimenting and I like to experiment. This circuit does what I want it to do. I knew the risks when I built it but it works as I wanted with no problems. It's been working several years. Up to you, dear reader, whether you want to copy it or not.
I suspect that the LEDs have been getting dimmer over time but as you have not measured the brightness ( you do need specialist hardware to do this ) you will not have noticed. Also from a test point of view running one without failure is some what meaningless in an attempt to calculate the MTBF ( Mean Time Between Failure ) or reliability of a design.
But then I am coming at this from the point of a mass produced consumer item that has to work in the field under many conditions. Quite understand your hobby aspect.
I think Perry's setup works because the source is also voltage limited. But there is an interesting difference in the traces: why the blue (50%) one has only about 20V peak voltage while the low duty trace has nearly 40V (excluding the ringing after turning off the MOSFET)? And what causes the step during the off period?
Grumpy_Mike:
I suspect that the LEDs have been getting dimmer over time
AFAIK all (power) LEDs get dimmer when used. IIRC it is mainly due to the excessive heat generated in the small die (too small to dissipate all the heat without damage because it is more cost effective). I expext if a given LED is driven with less average current (with reasonably low peaks) less heat will be generated and so the lifetime will be extended.
Grumpy_Mike:
But then I am coming at this from the point of a mass produced consumer item that has to work in the field under many conditions. Quite understand your hobby aspect.
Yes, well, this is a hobby forum, this is a hobby project, not one for selling.
Smajdalf:
I think Perry's setup works because the source is also voltage limited. But there is an interesting difference in the traces: why the blue (50%) one has only about 20V peak voltage while the low duty trace has nearly 40V
I noticed that when I got the traces, to the point where I was wondering if there is a fault with my oscilloscope. AFAIK they are correct. It's awkward to get the traces, the electronics is in my attic.
I purchased a dimmable driver to test, all worked as expected so I purchased 11 more which unfortunately buzz quite loudly when not at 0 or 10V signal level.
The attached pictures are of the buzzing and non buzzing, is it something that can be easily rectified or a design issue? The non buzzing unit has it's control circuitry on a daughter board (bottom right), the buzzing one has it spread over the main PCB.