Whitelight Diode

[buffo]

The pulse rate is a lot higher than normal, so you can scan the beam and not get a dashed-line.

As far as eye exposure goes - yeah, it's going to have higher peak power - which would make audience scanning really hard to do, but here in the US no one does it anyway because of the hoops you have to jump through to get it certified. (Bill Benner was saying that he worked with a company for nearly 2 years to get a variance for audience scanning, and I believe that is the ONLY variance for audience scanning that has been granted thus far...)

Adam

[steve-o]

Hey Adam
Well I was thinking about beam shows, mainly (no audience scanning--so if you can get 300mw pulsed out of a 160mW cw diode is it just the same brightness? Using Jenoptic for a reference they pulse for 10ps @ 80MHz. That looks like just under a 50% duty-cycle..
Steve

[buffo]

In theory, yes. In practice, I dunno. The eye's response to power is non-linear. Also, the persistance of vision is several orders of magnitude longer than the pulse durations you're talking about. Eh, you may end up with even more apparent brightness.

However, where are you going to find an 80 Mhz pulsed laser diode driver for a reasonable price?

I'm sure you could reduce the pulse rep rate to something more reasonable - say 500 Khz or so - and still be OK. Not sure how much that would cost though.

Then too, what duty cycle are those pulsed diodes rated for? If they'll make 300 mw pulsed, but only at a 10% duty cycle, you might still end up cooking it at 50%...

Adam

[steve-o]

Adam
A circuit design for a power output stage and high speed of several MHz isnt too hard--I'll see if i can dig one up
Steve

[buffo]

Dont forget the current regulation... Don't want to pop the diode.

Fred was looking into this very concept a while back, and stumbled across some surplus laser diode drivers on E-bay. He managed to win one of them for a cheap price. He decided it would be nice to have a few more, so he contacted the manufacturer to find out how much they sold for at retail. The price was *way* outside the budget of a mere hobbyist. Those drivers were rated for 2 Mhz, if I recall correctly.

Still, maybe you can come up with a cheaper alternative... Keep us posted!

Adam

[steve-o]

Adam,
Constant-current :Yeah, good point.
So were looking for a 2 MHz pulse generator (CC 0-5A adjustable)
Any Idea of the duty-cycle or pulse width?
Steve

[The_Doctor]

LM338 with IRF630 enhancement mode MOSFET to current shunt the LD? Rise/fall times are a couple of tens of nanoseconds, so this should allow modulation more than fast enough. When it's shorting the diode, the current regulator will be at reduced voltage, which recovers on applying gate voltage, as fast as the regulator can adapt, so the voltage is safely held at or below a safe value for driving current through the LD at all times. The limit to mod speed would be the regulator recovery time. There's a small radio transmitter circuit based on an LM317 (by Harry Lythall) that oscillates at 1.8 MHz outputting several watts, so it must be good for large scale voltage changes at this speed or higher, but I don't know what kind of wave shape you'd get. Peak currents when the MOSFET shunts will be high, but it can take them, having 9A continuous rating, and 36A for short pulses. A low ESR ceramic cap might add to the peak current the shunt MOSFET has to absorb, but not much, and the LD ought to at least have that much protection. (That capacitor in parallel with the LD should be the restrictor of bandwidth, all others should be on the supply line before the regulator). Just avoid inductances, those could make high energy spikes, but if those are kept minimal, it could work. LM338 and IRF630 are not cheap parts, but they can be connected very close to each other, minimising inductance in the circuit, and the drive capability is very high. It's also extremely simple, so less can go wrong that in other designs.

I haven't built this yet, so I can't verify it will work, but all research I've done suggests it should, and probably better than many other more complex circuits. This page shows a shunt modulator that can apparently handle video bandwidth, and the guy says he never popped a diode with it, so I think it's a good starting point. He shows several ways to add protection. Note, no inductances, and he doesn't even use a cap across the LD, he does nearly all of it with clamp diodes.

[steve-o]

Doc
I was thinking along similar lines except an IRF 530 (because they're sitting in a drawer right in front of me ) 630 probably be better tho..
I found this oscillator chip:
http://www.linear.com/pc/productDeta...10,C1096,P2186
But I remember an oscillator made out of a 4049 or something?
Shorting out the LD with the fet..interesting.
I'll try to post a schematic later.
It'd be nice to have variable pulse width though.
And I guess the softstart could be like the link you posted a resistor & cap or maybe a NTC
Steve

[steve-o]

Just for clarity--
Why is pulsed better than cw?
I can see how a 50% duty cycle will run the diode cooler and allow higher currents but is the apparent visual brightness higher?

[The_Doctor]

I haven't worked out a soft start yet. I used an electric double layer cap (memcap) of 1F across the LD, but that won't work for modulated diodes. If the drive is constant current but based on constant voltage fed through a resistor to the LD, you can add soft start to the regulator as described in the datasheet, but when doing it in constant current mode, it's not so easy. One way is to do it with the MOSFET, as it can do analog mod also, but it would be nice to have a way that soft starts even if TTL modulation is hammering the mod input. One way that will probably work, but is not ideal (being bulky and costly) is a series resistor feeding the regulator input, and the junction of the two being connected to one or two (or even three) 5.5V memcaps in series to ground, and each of those having a 470R in parallel to make sure they equally share the input voltage. They'll only waste a couple of milliwatts. Three allows inputs from 5 to 15 volts. Problem is, the soft start circuit costs up to 10USD or more, there has to be a better way. On the other hand, the protection against supply surges is so good that this might be a small cost to bear, given what it does. A 22µF low ESR tantalum ought to go in parallel after that series input resistor too, as the memcaps usually have lousy ESR unless you spend high.

I wouldn't use the NTC thermistors as a thermally based soft start, as they will be very hard to design in for safe upper current limit. They can be used in the constant current sense network though, for temperature compensation. I've done that with the small SMT types, 150R in parallel groups of two or three. I'm buying a lot of smaller values (22R, 33R, 47R, 100R) so I can use less of them. As they'll be thermally bonded directly to the diode mount, they can probably take more current than they're meant to, so they should be ok for DVD diode drive. Most of the current will flow through a fixed resistance anyway.

There is a circuit idea by Winfield Hill that can dupe an LM317 or LM338 constant current regulator into a linear change of current from preset maximum to near zero. I haven't built it. I tried a short-cut variant but it didn't work well, it needs a bandgap reference and a precision low current source. There might be a simplified varaint that can do soft start with very few and very cheap parts but I can't think of one. The main point is that the adjust pin tries to maintain 1.25V between it and the output pin, so if the 317 or 338 regulator is on the supply side of the LD, connecting the Adj pin to ground will drop the current below threshold. Figuring out a neat and easily predictable way to change the Adj point voltage without compromising stable main current control is beyond me so far. One hangup is that the Adj pin must cause a few mA to flow, it's not a simple high impedance input like a FET gate. I'd go with the string of Memcaps on the regulator input just avoid more difficult problems, given that the cost and bulk are still less than many dedicated diode drivers with similar levels of input protection and soft start. Just forget ultraminiaturisation if you choose that.

Re pulse vs CW at same average power, I don't know. I think the fast pulsed waves might seem brighter, but being very subjective, or at least based on human biology, it's a very hard thing to prove unless the effect is obvious. As far as I know, the effect is only obvious if the pulse rate is slow enough to make it visibly detectable, and the whole point is to make it NOT visibly detectable...

For PWM, one way is to low-pass filter the square wave from that tiny oscillator to make a triangle wave. Then, feed that to one input of an op-amp, and a variable DC offset to the other. Use a 10M resistor as positive feedback. The result will be linear control of PWM, and with the right op-amp, a very fast rise/fall time for the PWM waveform.

Edit: nm, I think the soft start CAN be based on the MOSFET gate feed entirely. All it needs is a way to make sure that the gate sees full voltage for total shunt before the LD sees any current. If this is guaranteed, it would be safe, simple, cheap, small, all that good stuff..