Driver with PWM input

[Benm]

Those measurements are as expected - longer wavelength at higher current/temperature.

As for pulsed vs continous operation: It is possible to damage a laser driving it with a low duty cycle at its absolute maximum CW rating. The reason for this is optical damage: As a laser diode heats up, it generally becomes less efficient. For CW temperature and current are related. Using pulsed current the diode doesn't heat up as much, and the peak power could be higher compared to running hot in CW mode.

For example on that NUB07E diode: you see a drop in output beyond 5 amps CW. If you ran it at 10% duty cycle, you may not get that dip in output power because it will not heat up very much. This is obviously speculative, but could easily be tested if you dared to risk the diode.

 

[Atomicrox]

I suppose that could be accounted for on the Arduino, as long as it's properly tested first. Still gotta wait for the host, then I'll decide

 

[Benm]

Hehe, hate waiting for the mail too?

If you're not familiar with direct port manipulation you could start studying that for a bit - essential if you want to use any DAC approach

 

[Atomicrox]

International mail takes months to get here, so yeah, you could say I hate waiting for it ;P The diode's been sitting for some six months already waiting for a home, poor thing.

I'm generally good with programming and Arduino stuff, never used direct port manipulation but I have a general idea how it works. Will probably need to mess with the timers as well.

I'm thinking of an alternate strategy to avoid a DAC - essentially using one PWM pin connected to the RC you already suggested to provide analog output and another PWM pin driving a transistor between the cap's + and the 33k resistor on your schematic. Might have to change resistor values to account for the voltage drop of the transistor, but I think that'll work.


Edit: while I'm at that I can also include a thermal sensor to turn it off automatically if it overheats.

 

[Benm]

That could work, but i doubt it would really be worth it. In terms of component count you're looking at a couple of resitors, a capacitor and a transistor. A simple 32-level (5 bit) R-2R DAC would require 10 resistors and nothing else.

To each his own strategy ofcourse, but i'd favour the DAC in this case since its more elegant and also a lot faster since you don't have the RC time to account for when changing values quickly.

 

[Atomicrox]

Soldered the thing up using just the pot to control current. Used thicker wires, a beefy heatsink and the 3 extra caps you suggested: 10uF across the high-current leads, 100nF on the op-amp and 100nF between pin 3 and ground. Sense resistor is 2 x 1Ohm 3W in parallel, for a maximum of about 2.6A.

Here's the setup:

Still misbehaves when it gets to about 1.5A, though it's a bit better than before. Frequency seems to vary a lot:

Any ideas?

 

[Benm]

That's odd, but i still think it may have to do with the long wires that are sort of entangled as well. I've always built it with the opamp, transistor and sense resistors no further than about an inch apart (on breadboard and soldered protoboard). It could be that the sense line is picking up from the base drive or something like that.

Could you scope opamp pins 2, 3 and 6 while it is misbehaving like this?

 

[Atomicrox]

Sorry for the delay, didn't have time to do it this week.

1 ohm test load resistor:

Pin 2:

Pin3:

Pin6:

This was at full power (~2.7A), the shape and frequency of the wave varies with current output.

 

[Benm]

That's actually quite helpful. It seems that the output current and sense resistor voltage are matching up well, so that's good!

Pin 3 seems to be clean too.

Pin 6 got me wondering a bit though. At what supply voltage are you running the opamp power supply? I see it hitting 4.80 volts there, which could be its maximum output voltage capability at limited supply voltage.

 

[Atomicrox]

I'm running everything on my bench PSU set to 9V 3A. I have tried up to 12V before, don't think that's it.

 

[Benm]

That's not the problem then indeed, 9 volts should be ample voltage to run this circuit as it is.

Maybe this thing is too fast for its own good here, and adding a bit of dampening would actually help things. Could you connect a 100 ohms resistor between pins 6 and 2 to see if it makes any difference? This would reduce the opamp gain a bit, perhaps preventing the overshoot at higher currents.

 

[Atomicrox]

Just to be safe I tested it again with various voltages between 8V and 12V, and also with 2x18650 batts. The problem persisted.

Tried the resistor (didn't solder it up but it was making good contact) and it made the waves more triangular. I also noticed the PSU current at max power is now closer to what's to be expected (2.5A vs. 2.7A without the resistor).

Load:

Pin 2:

Pin 6:

This is pin 6 just when it starts to get crazy while I turn the pot to increase current. The load also shows a random "pulsing" but I didn't capture that.

This is some sagging I noticed on the circuit's input where it connects to the PSU. Not sure if this is to be expected even with thick leads. Also happens with batteries.

 

[Benm]

Wow, that's quite a bit of noise on the power supply line! I'm not really surprised it is acting up with a 1.6 peak-peak fluctuation on a 8 volt power rail.

First of all i would consider, just to eliminate it, to decouple the hell out of that power line. Put something like a 1000 uF elco, 100 nF multilayer and 1 nF ceramic over it just to have all frequencies covered.

If that doesn't solve it outright, can you make a scope shot with pins 2 and 6 in the same images using dual channel so i can i see the phase realitionship between them? If you have 3 channels also seeing the power supply voltage would be helpful.

Another thing that could help here is putting a capacitor across the voltage reference, maybe 100 nF or so. This would isolate the reference from power supply line fluctuation to some degree, elemininating potential problems there.