Extended cavity of 445nm diode?

[The_Doctor]

When I tested my first blue diode for retroreflection death it died. As expected, but still disappointing... I'd hoped that maybe they were of tougher stuff that might cope better than red diodes do. Less disappointing was the flash of strong light that I can repeatedly get out of it that lights up my small broadband dielectric mirror so strongly that even what little gets through the mirror plane lights up the ground glass at the back like a lamp. I'm not so steady with my hands that I can do more than maintain it long enough to show that it ought to remain steady if I mounted the mirror and adjusted it carefully. So the question is: Have I created an extended cavity? Seems like one to me.. Also, if it is, might the OC of an argon 488nm laser work to make a good laser from a 445nm diode? And might various fun be had from this cavity, being that I can get at it and put things in it like crystals and other optics? If nothing else, it seems that it might be possible to deliberately burn off the facet's coating (assuming that's what they even do on those diodes) and use a much tougher OC to make the diode immune to dying from back reflection, and possibly get a very low divergence beam too.

[mixedgas]

What often happens with laser diodes is the cleaved faucet which serves as a end mirror gets blown off the diode. It turns LED at this point, as there is no feedback for lasing. So yes, you formed a crude cavity through a jagged chunk of semiconductor. Can you get a decent beam, I don't know. Is it worth 45$ plus shipping to the UK for the 442 nm HeCad OC on Ebay... I doubt it. Is it ever reproducible without Ordering a special 1000$ AR coated diode from Toptica... Nope.

Normally gratings are added to diodes as extended cavities with weak feedback. Not mirrors, as the nearly 100% feedback will almost always damage a decent diode.

Steve

[The_Doctor]

Ah, I suspected that if I had been lucky (or effortful) enough to keep it persisting longer than a few hundred milliseconds it might kill the rest of the diode. What amazed me was that I got that much out of it. I couldn't tell what the beam quality was like, though if I'd been able to look at the front surface at the same time as aiming the light into the output lens I might have done. One thing that did seem odd was that the progression of the original damage wasn't total in one go as with single modes it usually is, it's if there really might be some kind of coating being burned off rather than major damage to the facet. But if they really do that it's amazing because I thought that would be too expensive to put in projector diodes. This was an experiment I didn't try on a second diode, I want to keep the rest fresh. But eventually I'll get another shot at this. I don't expect it to be reproducible, but the bookies might get nervous if it is.

[RedlumX]

Not sure which kind of 445nm diodes you talk about, the 1W type is very robust and almost impossible to kill this way. The question rather would be why you wanted to do that. Improving spectral properties requires very delicate, stable setups.

[The_Doctor]

Not sure which kind of 445nm diodes you talk about, the 1W type is very robust and almost impossible to kill this way. The question rather would be why you wanted to do that. Improving spectral properties requires very delicate, stable setups.

Good to see you post. You don't do it very often.. As to 'why', my second line, first post gives you a clue. I got so annoyed with discoverign reds dying that way before anyone, even makers and distributers would admit that it happened! As a result, I always get the urge to try it on a new diode type at some point just to know whether I'm in for it again. Better to know than not to, says I...

Anyway, it really did die this way. It's pretty easy to deduce that if a diode is running fine on a mount whose temperature is at thermal equilibrium, powered by constant current at modest drive with no modulation, that if its output power falls dramatically, instantly, at the moment you use a mirror against it and see the initial flash of light hit back at the lens, that retroreflection death is what just hit it. Two points qualify it a bit in this case: one is that as I said, it wasn't totally destroyed, yet further afflicting it with the mirror finished the job. The diode was allegedly from the M130 projector, it's not the later and most powerful diodes. I'm not sure that makes a huge difference here. Second point is that the mirror was a tenth-wave flat >99% Newport broadband dielectric coated type, likely good enough to do what many others might not do. The first instant of the test caused power to drop by about a third, the rest over a few moments of retrying, to see what happened, (after all, once decided on such a test, what else to do?), and this is why I'm surprised at what appears to be a good extended cavity resulting. If the facet really were royally broken up, I'd not expect to repeatedly get strong glows that get through that mirror and light up the ground glass substrate like it does. I have no way to measure the strength of that light, and right now the diode is lying loose, unmounted. Could still be viable for further testing, but I don't want to do this to the new ones I still have.

What is most interesting to me is: you say the new 1W types are impossible to kill this way. That suggests you did the test, yet you ask me why. How do you know? I need to know this because my own experiment very conclusively shows that a mirror killed it like it would most other diodes I've seen of more than 50mW or so. And I'm also VERY keen to know if any new diodes are truly immune.

EDIT:
I wasn't after improving spectral properties. Not sure why you mentioned that.. The most ambitious (and entirely half-baked) thought I had about this was that a more distant OC might result in lower divergence, and assuming the output was any good at all under such conditions, a lot less vulnerable to retroreflection than most diodes.

As an afterthought, the reason why retroreflection death bothers me (quite apart from the number of expensive diodes killed to prove to a seller that this was the cause and not some other problem), is for the same reason that any DIY-er might hate a drill that burned its motor out the instant any tiny bit of metal was encountered when drilling a brick wall. Kind of hard to trust any tool that so easily fails to withstand its own driving force for more than a nanosecond. People on PL have often blamed driver failure for diode deaths, but I suspect that a combination of this reflection problem combined with high currents is the major cause. I have two first-hand statements by laser makers to that effect. Can't remember both, but one was Florian Rotter, who told me how one of his bigger headaches was dealing with early deaths caused by reflections back off PBS cubes. Even AR coated PBS cubes! So there are lots of good reasons to assess this risk.

[RedlumX]

Well I was playing a lot with extended cavity lasers also including the blue 1W types. I never killed one, but I should say that I was on the careful side and not running the things right at the edge. Overdriving a diode makes it of course much more susceptible to damage. Typically an extended cavity laser couples back in the order of magnitude of a few percent up to like 20,30%, and that won't harm if the laser is run say, at 50% or less of rated power.

For the 1W diodes, the spectral properties are anyway extremely bad at higher powers (say, higher than 50-100mW), so in order to build a clean laser, the power should not be higher than his, and at these power levels the diode cannot be damaged by any amount of feedback.

[The_Doctor]

Right, I could accept that. 50% or less for clean operation and long life is the way I want to go anyway. What really annoyed me about those Rohm reds was their tendency to die this way even at QUARTER power! (Not to mention the diabolical mode hops). No doubt this might explain Florian Rotter's headaches, and the vast numbers of those diodes that flooded eBay for a while. He wasn't the only one giving up hundreds or even thousands of those at a time... That, coupled with the earlier Opnext 80mW diodes that first taught me about this the hard way, is why I am so keen to test the limits of risk. Blue diodes do sound a lot more respectable. I'm guessing the green ones will be too, if the high Vf isn't putting them at risk from sheer exhaustion, dissipating all that heat. I haven't even got one yet, and that high Vf is already making me nervous.

As I mentioned it a couple of times recently, I ought to ask you outright: Is the stability and large clean operating region of some diode a useful gauge to its general good behaviour, at least up to maker's spec'd current if properly clamped to not too much more than 35°C or so? I've assumed that your page is good to watch for this, as it seems to correlate on some of the earlier diodes I used. My needs are far cruder than yours, but given that all the red diodes I tried can be observed to broaden linewidth visibly as a loss of specularity, a 'pastel' shade arising as speckle fades, I have found that to be a useful gauge to exceeding of safe max drive (in absence of other considerations like driver overshoot, or reflections). On this basis I assumed that your graphs of a diode's behaviour might be a good guide to picking diodes for gentle abuse. Can you tell me if this is a valid assumption?

[steve-o]

Can't remember both, but one was Florian Rotter, who told me how one of his bigger headaches was dealing with early deaths caused by reflections back off PBS cubes.

I think I know who you're talking about.. but why keep killing diodes with retroreflection? We all know that it happens and have figured ways to avoid it. It is rarely a topic around here anymore (especially since 660nm is not used too much anymore) ..

-edit- wow 2 posts whilst I was typing -- have to read now ..

[The_Doctor]

Because when I discovered it, I was apparently the only one who knew. Photonic Products Ltd were interested, but wanted proof, or at least a high degree of certainty. Once they had it, David McGuiness on his visit to Japan discussed it with Opnext, who apparently denied it. So I just lost out. When you're dumped on from a great height by an industry that can afford the loss, and might have welcomed the third party research and sent me a few replacement diodes instead of letting me lose 500 quid Sterling I could barely afford at the time, it teaches a HARD lesson in caution. I test EVERY new diode type this way. I won't do it to all of them in a batch.

EDIT. I also think it is somethign we should not blindly accept. Laser systems makers like Florian Rotter and Fleming Pedersen have dumped thousands of diodes off on eBay taking bigger losses than mine by far, over this. We ought to ask if diode makers can figure out ways to eliminate this at source. It may be tough to do, but the maker who succeeds will beat all others until they do it too.

More edit:
If you make and sell a laser system based on diodes with a lot of optics on the output, the only way to make claims on lifetime and endurance IS to make tests like these. If people aren't doing it before they ask people to spend hundreds on a new laser they're selling, they should be. I can understand the makers of such systems dumping diodes undeclared on eBay, the price is low, and it's caveat emptor. But I'd like to know what diodes passed their tests and did NOT get dumped.

[steve-o]

You test a laser diode by directly reflecting the laser light back onto the diode's facet? Doesn't that sound a little risky (and costly) for a test ?? Maybe I'm just not getting what you're trying to accomplish (?)