I'd appreciate that Kit.

OK, retic fixed.

The picture shows the constant current drive for my LED lights. I've deliberately not included the Eagle file as it does not have accurate part numbers, just whatever happened to be in the libraries with the right footprint. Use this as a guide to your own plans, I will be producing full details of my PWM LED lights once I'm finally satisfied with the design and the controlling software.

Q3 is a MOSFET with the LED string connected between the drain and +volts (via P1). Obviously the supply voltage and current rating of this device depend on the LEDs in use. The logic supply voltage is 5v.

U2 is an Op-Amp used as a comparator. I'm currently using an OPA347. Avoid high frequency designs used as video amplifiers as these don't work as comparators. I found this out the hard way.
The other chip is a LM336-2.5 2.5v reference source. D1, D3 and R10 provide additional temperature stability which may not be necessary in this application but doesn't add much cost.

R4 and R7 produce a voltage at the inverting input of the comparator which is proportional to the current through the LEDs and MOSFET. R6 is there as part of the PWM circuit not shown and is not necessary if you only want the const-I bit. Adjust these values to give about 1v at the maximum current for your choice of LED.

The other potentiometer (R8 but not labelled very clearly) presents a proportion of the reference voltage to the non-inverting input of the comparator. Adjust this pot to vary the light output of the LEDs

The circuit works as a feedback loop, controlling the current through the MOSFET to keep the two voltage inputs to the comparator exactly the same. Current stability then depends on the stability of the reference voltage source. As stated before, the brightness of a LED depends on the current through it, not on it's supply voltage. Just like any other diode, it's current will vary with temperature for a given supply voltage.

Clear as mud?

Kit

PS The heat dissipation in the MOSFET will depend on the voltage dropped across it with the required current flowing. This can be minimised by reducing the supply voltage ( if possible) with the lights on maximum brightness until the current starts to drop. Measure the voltage across the current sensing resistors to find this point then turn it up, say, 0.5v to maintain stability.

Finally got round to testing the range of control I have on my LED lamps brightness by varying the reference voltage to the constant current supply. The result was something of a surprise.

I made the test by using my camera on manual and adjusting only the exposure time until the inbuilt meter said it was correct for a white object. I adjusted the current from full brightness to the lowest output that looked stable. At very low levels there is a lot of flickering as tiny variations in the LED supply voltage have a huge effect on current when the diodes are right on the edge of their cut-off voltage.

The range of exposure times I got was from 1/20 sec to 13 sec. A whole 8 stops! This was considerably more than I'd anticipated and opens up the option for a controllable video light which does not use PWM control of brightness but adjusts the reference voltage of the constant current supply. If digital control from an Arduino , for example, was required, it could be done by using a digital to analogue converter chip in place of the manual potentiometer. Depending on exactly what electronics are mounted on the lamp, adjustment could be done over a serial link, USB or even by wireless.

Kit

Amazing result Kit! Its surprising that this isn't used on commercial video lights. Are you planning on selling boards once you have a batch made, or are you going to sell complete LED controllers or complete lights?

James,
I was very surprised at the range available from just controlling the current. At the lowest levels there is rapid noise-induced flicker which is not surprising. I'll need to get the oscilloscope out to do a more detailed study of what happens as the brightness is reduced. The whole point of this project was to produce a lamp with a very stable light output to avoid flicker in timelapse and animation use which could be dimmed by DragonFrame.

I've no plans to produce and sell boads or complete lamps. I already have a full-time job and more than enough things to occupy the rest of my time. If anybody wants to take the circuit diagram above and develop a commercial version that's fine by me. The LED modules I'm using are 40v, 40W units made by Cree, supplied by Digi-Key. They also have a mounting plate and matching reflectors.

The only problem I've had with this part of the design has been choosing a suitable op-amp for the comparator. In order to get a fast response for the PWM circuit, I originally chose a very high-frequency design intended for video amplification. This type of amp has inputs which sink a load of current if the differential input is high and are completely useless as comparators. The current amp is not as fast I'd like for the PWM function and I have another option waiting to soldered in.

For anybody wanting to digitally control the light output by varying the voltage reference, the thing you have to worry about is maintaining the long term stability of the reference. For example, if you were to use the DAC facility on the new Arduino Zero board, it would be necessary to provide a stable reference voltage to the Arduino and confirm that the DAC output is also as stable.

Kit

Is your full time job an Electrical Engineer Kit ?

I must do a test of my system again before contemplating an upgrade to something that isn't one of my hacks.

Thanks again for the circuit and explanation.

Steve

Steve,
'Electronics Technician' is nearer the mark. My specialism is high power and/or digital wireless telegraphy which tends to include everything from logic circuits for control purposes up to megawatt DC power supplies. In England I used to be able to sign myself as 'I.Eng MIET' but I haven't paid those membership fees for years.

This sort of comparator feedback loop is from 'Op-Amps 101' and is nothing original.

Earlier in this thread you mentioned using a light for video. Long term stability is less of an issue in this application since slow drift will not produce the same kind of flicker you get in timelapse or animation. In this application you might be able to use your existing PWM circuit, but simply low-pass filter the waveform to produce an adjustable constant voltage reference for the circuit above.

Kit

PS Appologies to fototripper for hijacking this thread. I hope the information is of interest.

PPS As well as being CEO, MD, Senior Animator and Tea Boy at Pipsqueak Studios, I am also head of the Engineering Division

my lights turned up here is one quick picture to give you some idea their not super bright but that is ok i actually don't intend to use them for product photography but you get the idea. i am happy enough with them and the price is right this was cross posted in my camera slider post but i figure cant hurt to have it here either. now for more experimentation

lost power the other night used a couple of those usb lights i bought with an old 12v 4 amp hour battery i had to eat dinner by probably
lasted 40 mins or so then i got bored and plugged the laptop in as well run out of juice pretty quick after that lol. then i got another
battery not long after that power came back on. but i'm thinking i'll make a little plastic box with a 300mm extension to hold the light
upright for emergency use. summer storms have a habit of taking out the power out our way.
just wondered if you guys had seen seen this cute little guy lol

Nice USB light, Chardie. Power blackouts. Love em. Don't get so many as I did as a kid but every now and then during a storm we lose it. Takes you back to basics - no computer etc. Think I was meant to be a bush kid

The occasional cyclone has been known to take the power out for days at a time up here in tropical WA. When that happens we resort to solar charged 12v batteries and a couple of LED flood lights. Cooking gets done on the gas barbie (well it would in Australia, wouldn't it?). I haven't bothered with buying a generator like many of my neighbors, there's a certain peacefulness that comes from not suffering that ever-present 50Hz buzz for a few days.