Hey all,

I've been lurking for a few weeks but this is my first post so go easy on me

I'd like to use an Arduino (uno) and an optocoupler to TTL/PWM switch a ir LD (808 nm, 280 mA) which gets a constant current from an LM317. From my lurking here (thanks folks for all the excellent info) and in an effort to learn a little bit of eagle I've drawn up the following two circuits (which I hope upload ok!).

In both circuits the arduino drives the opto via transistor 1. In the first circuit (LD on the right) the opto then switches the supply to the LM317 via transistor 2. In the second circuit (LD in the middle) the opto switches the current through the LD via transistor 2.

I'm not an electronics engineer so I'm wondering which (if any) of these approaches makes (most) sense? Or am I going to kill the LM317 or LD...?

All the best,

Alan.

Thats a LOT of work to control a laser when you could just use a RC DAC + a npn transistor, to control current.

PWM and LM317 would not go well together, use a discrete current sink.

Just forego the LM317 and control a transistor directly. You can rig something like this up, only with a laser instead of LED.

Thanks for all the replies.

Bionic-Badger, I've been looking at that instructables circuit you linked to and I have a real newbie question for anyone with patience/time....
If I have 5V VCC, a 2.2V LD operating voltage and a 0.28A operating current.... is the following correct?
R3 = (5V-2.2V)/0.28A = 2.8V/0.28A = 10 Ohms
with a power dissipation of 0.28A*2.8V = 0.784 W
Or am I omitting something? There isn't a 0.7V drop across the Q2 NFET by any chance?

Cheers,

Alan.

@alan: in your first schematic, the transistor drived from the opto iust short-circuit your power supply on the LM317 side ..... in the second circuit, is like disconnect and reconnect theLD to the driver, and this means LD turned to LED in 50% of the cases.

Also, optocouplers uses IR leds inside, you don't need an extra transistor for drive them from the arduino board, just connect the opto to the output pin through an adequate resistor.



EDIT: and your Q2 transistor is at risk of "fluctuation" and random activation, if connected as in the first schematic, cause when the opto is not drived, the base is like left opened ..... for avoid this, the base of the drived transistor must be always tied low (to GND) with a resistor, also a 100Kohm one is ok.

Hi HIMNL9,

Ok, I think I see what you are saying, its left sortof floating. So actually if I ignore the two circuits that I posted and connect the arduino directly to the circuit that Bionic-Badger posted I should be probably ok (assuming my calculations are ok above!)? Q1 is tied to ground thru R3 and Q2 is thru R1 (on the #5 circuit image) which I guess is to ground when Arduino is low.

Thanks,

Alan.

If you wanna go full DIY, that's fine... but I just wanted to let you know... since it hasn't been mentioned in this thread yet... that you can connect the arduino directly into the TTL input of a TTL modulated driver. It only needs 2.5V to activate, so easily within the working range of the arduino.

They are pretty cheap.... Just thought I would mention it

The circuit BB posted, is a common current regulator ..... if you want to "turn-off" it (like, modulate it with a PWM signal), you can hook the transistor of the optocoupler in place where there is drawed the box with "pwm or microcontroller" words (emitter to GND and collector to the wire marked "out"), and the input of the optocoupler to the arduino board (common ground configuration if you want that the LD turn off when he arduino pin is at 1 condition, or common positive if you want the opposite)

NOT without an optocoupler, especially if you use two different power sources for the arduino and for the LD.

Ok, thanks for that HIMNL9, I'll give that a go.

Common positive: I need to connect arduino PWM to optocoupler anode (thru a resistor) and arduino ground to optocoupler cathode (no need to connect arduino 5V anywhere). Is this correct?

Thanks,

Alan.

This is common negative (but the concept is the same), and in this case, your optocouple is turned on when the arduino output is 1, turning off the LD.

For the opposite, connect the anode of the optocoupler to the +V of the arduino, and the cathode to the output pin through a resistor, so the optocoupler is turned off when the output pin is at 1, and the LD is on (this is cause that driver schematic will turn off when you close the "out" wire to GND.

Bobhaha, thanks for that. I guessed there might be but I'll try the diy route for a wee while first

HIMNL9, thanks for your suggestions. The common positive sounds good but I was thinking that if the arduino is not plugged in (and the LD is) then the LD would be on. So I've been messing with eagle again and I've added in an opto in common gnd config (I think!) to act as an on/off enable switch (image attached). What do you think... would this work?

All the best,

Alan.

Hey all,

I've been messing about with the 2-opto circuit from my last post but it doesn't work very well. When The arduino is off and optos 'open'... the laser is on... which I was trying to avoid with the second opto! I think maybe its because the fet gate is floating in this setup.

So I've modifed the design a little bit by adding a 10k resistor and removing an opto (see attachment). In this setup when the '1' input to the opto goes high then the laser is on as the FET gate sees up to ~90% of 5V VCC (...actually I've only tested with an LED so far ). When the '1' opto input is low, or not connected, the FET gate is grounded thru the 100k resistor and the laser/led is off.

I've hooked it up to pins 9 (to opto '1') and GND (to opto '2') of the Arduino Uno and the 'fade' PWM example sketch runs nicely. I'll hook it to an LD if I get a chance in the coming days and post back.

Thanks again to all who responded.

Alan.

As I said in the last post you could just drive a npn transistor like so


its the way I have my test driver set up with a npn that can handle 2A and so far I can get 2mW-1300mW on the test load with 3.3V (red diode).

The problem is that will operate any diode in pulsed mode.
Adding a inductor and diode in there will make it a proper buck converter and they can be nice and smooth. It would help to use a P-FET for highside and a N-FET to pull it's voltage down. That will allow you to run higher v inputs without much drop so that you can use a larger feedback resistor and get better current resolution.

Anyway, main point is, without a inductor it will most likely run the diode in pure pulsed mode with excessive spikes.

Are you sure about that I scoped it and its a smooth line minus the 10mV ripple.

If the driver has a cap in it, it should act against the sense resistor smoothing the voltage over it. However it may not do anything to smooth the current through the diode.

Basically, it is only possible to convert voltage by two means: inductance and charge-pumping. A LED won't mind, but a LD will as it is a lot less forgiving.

If there is a soft-start type LD driver connected on the output the situation is a bit different though. Then you are essentially using the drivers soft-start and cutting it off with the pwm to restart it, only letting it ramp up a bit. Even at low pwm and full ramp up the peak current will be limited by the driver and your power output will be adjustable "on average" but will still be full current for parts of the period and zero for the off part.

Somewhere I think there is a misundersdtanding between the two of us, and my example of arduino PWM controlled Current regulator, I'm gonna say the misunderstanding is on my part, but just in case.

I'm also assuming you are talking about my post using a LOW pass RC filter with a HIGH PWM frequency resulting in a DAC, which will control the transistor in the linear region making it act like a variable resistor, and the current sense resistor is just used (with some coding) to measure and limit how much current gets dropped across the transistor, also regulating the amount of current that the LD is getting, basicly like an over engineered LM358 current source like the examples here: Current source - Wikipedia, the free encyclopedia


(picture courtesy of Wikipedia)

Hi All,

I tried the circuit from my post #13 and it appears to work ok, though PWM values between 1 and ~90 have the same on/off times. Anyway, I built up 3 separate circuits for 3 LD's but i noticed that the R4 sense resistance required to get a particular current value (200 mA) varied a lot from one circuit to the next ...something like 10, 8 and 3 ohms where i expected 10.5 ohms ( I think, don't have my notes here). In the lowest case there was about 2V across drain-to-source of the NFET and 0.7 v across the resistor where it was roughly the inverse in the other circuits (2.2V LD, 0.55V FET, rest across R4). I rebuilt it with different components (but same LD) and got the same values.

Kmor: I haven't had a chance to try your circuit from #14 yet. I'm not really sure how it works to be honest.... does the adc somehow set the current value you want and you then pwm to go from 0A to this current value?

Kreature: Would you put the inductance and diode in series with the LD? How do you calculate a reasonable value for the inductance?

All the best,

Alan.

I believe the code would look like:

void setup() {
analogReference(DEFAULT);
pinMode(9, OUTPUT);
TCCR1A = 0X82;
TCCR1B = 25;
ICR1 = 1600;
OCR1A = 0;
}

ontime = 0;

void Loop(){

if (AnalogRead(A0) < 40){
ontime++;
}

if (AnalogRead(A0) > 40){
ontime--;
}

0CR1A = ontime;

That right there should work, and it uses timer 1 channel A, if you try and copy this code I recommend using a 10uF cap, and a 500ohm resistor for R1 and C1, the TCCR1A/B, ICR1, and 0CR1A, are the timers registers and the only way I could get teh PWM functions I wanted from the timer, to allow higher frequencies and more accurate, duty cycle control, so far I've found with that schematic, and my transistor from 700 counts to 900 counts I get rougly 1mA adjustments per count, but after that it starts to be more like 2-3mA per count and I haven't gotten all the way up yet, as for the range with this code your duty cycle control is from 0 -1600 counts, so you can see how wide of a control range you would have.

Sorry Kmor2004 you are absolutely right. I am the one misunderstanding.
I did not account for C1 as there was no value on it. I assumed it was small.
Ofcource when you use the PWM abd R/C filter to form a DAC you are correct that it will be able to linearly control the transistor although it will only work for a certain region of voltage out and V+ can not be much different from Vcc for the avr.

Ideally in that circuit the reference for the transistor should be the supply so a inversion with a NPN controlling a PNP would have given a better range of control. That said, it's actually quite clever!

My driver uses a buck converter setup but it's not done yet as it's not safe regarding disconnect and such right now. Also, the turn-on current must not overshoot or rise too fast as it can damage the LD making it a LED.

alan
I would only put inductance in the drivercircuit if I wanted to convert the voltage that way instead of dropping it over the transistor as heat.

Inductance in a lm317 based regulator will cause overshoot.
In fact, caps on the output of a lm317 reg will cause overshoot too, and excessive current upon reconnect to LD if a momentary disconnect is experienced.

There should only be a cap (1µF or so) on the LD itself together with a 10k discharge resistor and a reverse protection diode which can also be a 4.7v or 5.1v zener to also protect it a bit from excessive forward voltage.
After LEDing a few LD's I ended up with this setup based on suggestions here on the forum and simulations in ltspice.