Here's the scenario. I'm making a laser harp and I need a constant current laser driver that can turn on or off in ~1us. My original plan was to use a laser module with an LM358 based driver but after trying it out I discovered that the rise time was ~40us. I am targeting a rise and fall time of 1us because I have a stepper motor with a mirror that is stepping at 46,080Hz (48steps/rev32usteps30rev/sec) and I would like to turn on the laser for half-step increments (~10us).
Since I will be making my own laser driver I figured I should turn up the optical power of the laser instead of accepting 5mW. For my duty cycle I can use a 14.56mW laser while still keeping the laser harp a class 3R laser product. I haven't chosen the laser diode yet but for 5mW-15mW of optical power the current must typically be 35mA-100mA. Let's say up to 250mA in case I want to go higher. The operating voltage of a laser diode is typically 2V-4V.
The laser power supply is 15V (assume 20V-10V for margins) and VCC is 3.3V.
TL;DR
Design a constant current laser driver circuit with the following specifications:
20V-10V laser supply
1us or better rise and fall time
Up to 250mA laser diode current
3.3V VCC
Attached is the circuit design I'm starting with. It's set to 60mA presently.
Use a constant current driver that's on permanently.
Switch across the diode to short it out when you want it off.
The current gets steered between the diode and the switching device as fast as the switch can handle,
the constant current source shouldn't be too upset by this.
Method 2:
Use a fast opamp (such as a video opamp). The LM358 is immensely slow even for a normal opamp,
it can't even handle audio.
A quick search found the AD8031 / AD8032 which are available in DIP and 5V compatible, fast, rail-to-rail
on inputs and outputs - many other possibilities are out there, this one may be overkill but it looks pretty
handy.
This is a little more complicated than putting a little opamp in between and switching your diode.
You mentioned about 100 mA but at what voltage - 10v to 20v @ 10 uS intervals? It would be nice to see the circuit as you imagine it. What is the actual frequency?
The LM358 is one of the first designed for single supply and the output has a crossover issue, it’s fast enough for audio but adds too much distortion for this use.
Diodes can work watch out for reverse current and reverse recovery time.
MarkT that’s a nice opamp but pretty pricey at about $6 each. I’m still not sure if we’re talking about 100k or 1 MHz with gain of 10?
There are literally hundreds or thousands of high speed opamps out there, time to get parametric searching?
Something closer to 100MHz gain-bandwidth-product would be reasonable - you want lots of gain left at the
switching speed and a suitably high slew-rate. Even audio opamps are rated 10MHz to 40MHz...
Or try a BJT current source with a fast transistor - since a current source doesn't go into saturation it should
switch pretty quickly.
You need to size the sense resistor for the current you want and I guess it will work better than your circuit. The transistor will dissipate a LOT of heat. Why do you need 15V supply voltage for driving a 2-4V diode?
Not ideal, that will waste lots of power in the sense resistor and waste 5 volts, you could use an LED as voltage
reference for the base so the sense resistor only has about 0.7V across it.
Just power the LED as normal through a resistor, it switches on the transistor and the resistor sets the current.
For 0.7V and 0.25A the sense resistor would be about 2.7 ohms, you'd run the LED at 10mA or more, so a 180
ohm current limiting resistor for it.
*** I've just noticed the laser power involved is inherently unsafe, being greater than 1mW (Class 2) That's a show-stopper to this project unless the beam size or divergence is such as to be class 2M ***
MarkT:
Not ideal, that will waste lots of power in the sense resistor and waste 5 volts, you could use an LED as voltage
reference for the base so the sense resistor only has about 0.7V across it.
But OP uses 15V for some reason unknown for me. The more power is dissipated in the resistor the better for the transistor.
That's true, add a DC-DC converter down to the minimum workable voltage would help with that. Power
resistors are generally more bulky and awkward than power transistors, though you can get TO220
power resistors they are not cheap.
LEDs for setting constant current sources and sinks is quite a popular technique as you get visual indication
for free and they are better regulators than low-voltage zeners.
With voltage to spare using the full 5V from the microcontroller is sensible though.
