PWM drivers for lasers. Why/why not?

ossumguywill said:

You know how the power to the micro gets cut whenever you power cycle the laser?

Click to expand...

A point which i even dont tought off. i suppose its not good for powering the µC with the pen's switch. better leaving the µC in standby always powered and use a port of it for switching on the laser output/ modulation?

Xer0 said:

A point which i even dont tought off. i suppose its not good for powering the µC with the pen's switch. better leaving the µC in standby always powered and use a port of it for switching on the laser output/ modulation?

Click to expand...

For a pen laser it's doable I guess, but for a flashlight laser with a tail switch, you'd need to run an extra wire down to the negative terminal. Standby power consumption might be an issue too if you're storing it for a while, you have to pick a good low-power microcontroller and worry about turning off all the modules/peripherals, something I haven't had good experience with.

drlava said:

Hi, ok the instructions are --> HERE <--.

Note that this also allows you to use a low current switch to turn the driver on and off via the TTL control, without worrying about burning up the contacts on a clicky switch when you switch all the current on and off.

Also note that although it will appear dimmer, the eye damaging capability of the beam will not have decreased much, as it does when it is dimmed via analog means.

Click to expand...

This is a nice feature on a boost driver and fits in pretty well with something I want to try. However, this is the first time I will have dived into TTL control so I have a few questions if someone can help.What is my layout for powering the driver? Do I have power from my 3.7V 2400mAh batery hooked up in the normal way and another connection from the TTL controller? Or do I have to input the full current draw for whatever LD I'm using through the TTL controller? I think you're going to tell me that the TTL is merely a fast switch to the driver and that the power is seperate.I'm hoping it won't be the case that I need to supply the full current through the controller as the 555 won't want to sink the current I want to use!Many thanks,M

Morgan said:

This is a nice feature on a boost driver and fits in pretty well with something I want to try. However, this is the first time I will have dived into TTL control so I have a few questions if someone can help.What is my layout for powering the driver? Do I have power from my 3.7V 2400mAh batery hooked up in the normal way and another connection from the TTL controller? Or do I have to input the full current draw for whatever LD I'm using through the TTL controller? I think you're going to tell me that the TTL is merely a fast switch to the driver and that the power is seperate.I'm hoping it won't be the case that I need to supply the full current through the controller as the 555 won't want to sink the current I want to use!Many thanks,M

Click to expand...

If I'm understanding the instructions correctly, it's just a normal TTL control input, which a 555 timer can produce no problem. No need to interrupt current from the battery, hook it up as normal, cut that trace indicated, and solder your control wire to that pad.

rev0 said:

For a pen laser it's doable I guess, but for a flashlight laser with a tail switch, you'd need to run an extra wire down to the negative terminal. Standby power consumption might be an issue too if you're storing it for a while, you have to pick a good low-power microcontroller and worry about turning off all the modules/peripherals, something I haven't had good experience with.

Click to expand...

The attiny's are alread very low power (250µA at 1Mhz) and in standby goes down to even "<0,1µA" so that would be no problem. the other thing is the MicroBoost; drlava should be able to tell its "standby" power loss?

rev0 said:

If I'm understanding the instructions correctly, it's just a normal TTL control input, which a 555 timer can produce no problem. No need to interrupt current from the battery, hook it up as normal, cut that trace indicated, and solder your control wire to that pad.

Click to expand...

Excuse the daft question but a common earth is the only other TTL controller connection, correct?

M

rev0 said:

For a pen laser it's doable I guess, but for a flashlight laser with a tail switch, you'd need to run an extra wire down to the negative terminal. Standby power consumption might be an issue too if you're storing it for a while, you have to pick a good low-power microcontroller and worry about turning off all the modules/peripherals, something I haven't had good experience with.

Click to expand...

That's the point of my RC circuit... it knows how long it's been off without needing any power. And BTW AVRs go all the way down to 16khz with the internal oscillator without modifying any registers.... so if they suck 250µA at 1mhz, that would be 4µA at 16khz... I'm pretty sure that's less than the self-discharge of the battery.

ossumguywill said:

That's the point of my RC circuit... it knows how long it's been off without needing any power. And BTW AVRs go all the way down to 16khz with the internal oscillator without modifying any registers.... so if they suck 250µA at 1mhz, that would be 4µA at 16khz... I'm pretty sure that's less than the self-discharge of the battery.

Click to expand...

Well maybe that's something you can help me out with. I can't manage to get an ATtiny13V below 381µA in power down, 1MHz, all modules disabled. Nowhere near the advertised max 2µA. I did a project powered by a CR1220 where the micro was always powered and just used the sleep modes, but I'd have to replace the battery every week. I must be doing something pretty wrong.

rev0 said:

Well maybe that's something you can help me out with. I can't manage to get an ATtiny13V below 381µA in power down, 1MHz, all modules disabled. Nowhere near the advertised max 2µA. I did a project powered by a CR1220 where the micro was always powered and just used the sleep modes, but I'd have to replace the battery every week. I must be doing something pretty wrong.

Click to expand...

The idea is not to provide power to the micro with a capacitor, it's to make the cap's voltage decrease at a set rate so that on power-up, the micro can check whether the cap gives a logical high or low and use that to determine off time. But to get that low power, try running at 128khz (or less) and disable all timers and oscillators besides the watchdog timer (which is now providing the clock frequency). I'm not sure the particular registers to do this...

I still wonder why nobody makes curret regulation via µC
It's more precise than turning a pot and you can permanentyl regulate it by measuring a shunt over the ADC...

Cause most people can't set a micro.... Of course, you could use a micro AND a pot...

Xer0 said:

I still wonder why nobody makes curret regulation via µC
It's more precise than turning a pot and you can permanentyl regulate it by measuring a shunt over the ADC...

Click to expand...

Do you mean regulation via adjustable PWM duty cycle? It's been discussed before, while this does vary apparent power, it doesn't do a darn thing for safety since the peak power would still be the maximum, plus it would be useless if you wanted to hook up a lower current diode. I suppose if you found a micro with a DAC you could do some trickery to get analog modulation, but it seems like too much work. I'm starting to think there's no drivers with a µC for a reason, they're just not needed.

Edit: Just figured out what you mean, you could use one of these types of drivers, which compares feedback voltage from the current shunt to a voltage you provide. You could select the current by feeding in an analog voltage via a µC's DAC: http://laserpointerforums.com/f65/r...diode-driver-445nm-pics-53092.html#post745565

One solution is to build your own buck/boost converter from scratch, which I did once, problem is you can't get the PWM frequency or ADC sampling frequency very high and either need much larger components (inductor, smoothing capacitor) or accuracy suffers. Mine was going to be a driver for a 1W 808nm laser and the output had 200mA of ripple, which I didn't think was too acceptable.

The last option I can think of is to mix a µC with a commercial driver IC, which is what I'm trying to do (still waiting on those damn samples from National! UPS has been stuck at billing info received for a week!) and change power by altering the feedback voltage or shunt resistance using transistors.

What are you waiting on from national? I have some stuff from them hoarded....

ossumguywill said:

What are you waiting on from national? I have some stuff from them hoarded....

Click to expand...

LM3410 LED driver IC. Switch current up to 2.8 amps, it's just perfect for delivering up to 1A for a 445 and a single Li-Ion. LM3410 - 525kHz/1.6MHz, Constant Current Boost and SEPIC LED Driver with Internal Compensation 525kHz/1.6MHz, Constant Current Boost and SEPIC LED Driver with Internal Compensation

I've been using their web-based simulation suite. According to the simulations, this driver should have no overshoot, given a Vf of 4.2V and Rdynamic of 0.4 Ohms to simulate the 445nm diode, but I have no idea what other differences there are, all I know is laser diodes are a lot more sensitive and I'm a bit worried about what kind of performance I'll actually see (hopefully without destroying any laser diodes...)

Update: Got my IC samples in the mail today, and was able to assemble a prototype board with parts I had/salvaged. I took some basic measurements too, and it looks alright so far, again I'm not sure how different it will act with a real laser diode rather than a test load.

Here's the whole rather ugly looking setup:


Here's a closeup of the parts (had a bit of a magic smoke moment, thus why those two current sense resistors are pretty charred, driver is unharmed though):

The main driver IC is the tiny SOT23-5 part under the vertical inductor at the top (forgot to switch the footprint to the inductor I actually had). The ATtiny44A down on the bottom right is the brains of the operation, all it does right now is take a switch input from a pushbutton and enable the driver IC, but it provides a nice clean edge with no switch bounce. The chip next to the ATtiny is a dual MOSFET for switching power modes, basically it creates a selectable voltage divider for the feedback voltage, giving 3 ranges: high impedance (both open) gives the low current mode (400mA), a 1/1.667 divider gives a medium mode (667mA), and a 1/2.5 divider gives a high mode (1000mA).

Oscilloscope screencap of current overshoot (1mV = 1mA):


Oscilloscope screencap of ripple (1mV = 1mA):



So it appears that the driver works pretty well, overshoot is about 37.5%, which would scale up to a peak of 1375mA max if I were driving the load at full power, which for a 445nm diode seems to be fine, they've been driven higher, and that overshoot only lasts for 70µS. Current ripple is pretty manageable I guess, it's pretty unavoidable since this is a straight boost driver. Other than that, I'm a little worried about heat from the driver IC. If the specs hold true, it's got a 0.17 ohm switch, coupled with a peak current of 2.8A, so power dissipation will be <=1.33 watts. I felt the IC when it was operating at just the 640mA I have it set to now, and it was too hot to touch.

I haven't tried PWM with this IC yet, not so sure it'll be a good idea since it's got a pretty significant overshoot. I guess the switcher + linear driver is the best setup for that, but I'm going for small size, and the 445nm diodes seem pretty strong.

Can anyone tell me if measurements taken with a test load (using 1N4003 diodes) will be a lot "tamer" than with a real laser diode, should I be worried about an unforseen spike killing my 445nm diode?

Can anyone tell me if measurements taken with a test load (using 1N4003 diodes) will be a lot "tamer" than with a real laser diode, should I be worried about an unforseen spike killing my 445nm diode?[/QUOTE]

LEDs are several orders of magnitude more tolerant of spikes, the dies are huge compared to a LD die, and the current density in the LED is, much, much lower.

The 4001 test load has maybe 100 times the capacitance of a LD and is slow, it will tend not to follow spikes, but simply swap them with capacitance where a LD will try to follow the spike.

I would highly suggest a "LASORB" and a low pass filter across any Laser Diode powered by a switcher. It may be just fine steady state, but where the LED switchers are often "undefined" is startup and shutdown, and sudden powerdown.

Its not a cure all, but it will clamp most spikes fast enough to protect a LD:

LASORB - ESD absorber for laser diodes

Steve