Extended cavity of 445nm diode?

[The_Doctor]

Look at it this way: You're getting paid a lot for a show.. some joyboy in the audience has brought a front surface mirror and hauled himself a few feet up some nearby gantry or pillar and thrust the mirror in a beam to see if he can send it to source. Stuff like that occurs to people, it's just sod's law. If the laser dies, so does the chance at good pay. People SHOULD be interested in this if they're making lasers to sell. Maybe it doesn't happen like that so much, but most likely only because dwell time is short, or people make it properly difficult for others to do this to a laser at a show, but the risk is real, Florian Rotter's PBS cube problems suggest that far lower degrees of reflection can do it.

I don't start out by forcing full reflection. But to test, it MUST be done. The only way to know if you have gone far enough in any new situation is to go too far. When I want to trust a diode, I have to know exactly how far I can trust it.

EDIT:
First time I discovered this, it was a metal tipped screwdriver used to adjust a collimating optic. The seller insisted I'd touched it with ESD, sending high currents though anodising on parts of the mount. I ended up shoving sparks into the collimator just to prove that my diodes can't be killed that way in my designs. Like I said, tests like these matter, or we cannot make endurance claims for lasers we sell. IF you can demonstate such ruggedness, you save money later, when people try to claim money spent on your laser after they killed it themselves. It really pays to be able to prove their claim false.

[kecked]

I had a similar thing happen when I was adjusting a pbs cube. All my diodes died at once. I blamed the supply or driver. It may well have been from this effect.

[steve-o]

Yes I've also killed laser diodes with back reflection from a PBS cube. It's an un-fun thing to have happen :/

[mixedgas]

However, screening them by blasting them is not a good idea. If your within the Raleigh range and you space the mirror the right integer multiple of half wavelengths, you stand a very strong chance of killing them. Especially if the mirror is slightly curved. If you did this with a mirror on a translation stage, you'd probably nuke 50% or more.
Steve

[The_Doctor]

It was definitely very crude. I knew it wasa lousy method, but I could never think of a better way to assess the risk of unpredictable reflections, once I learned that the tip of a chrome vanadium screwdriver could send enough light back to kill one. (Opnext 80mW. Rohm diodes are even more vulnerable). As some diodes seemed more risky that others, and as someone buying a module off me was bound to end up having accidents or worse with mirrors, I saw no answer other than to see for myself what happens before I lose money over claims I can't verify myself. Things are a bit easier now that most people know and expect to have to take care with diodes and reflections, but when I did this that was not the case. I bet most buyers on eBay still don't. Hence, I am still a bit nervous about it.

[The_Doctor]

Re PBS cubes, I haven't started setting stuff up to try yet (going to be a while before I have bought all the parts I need, work in progress.....), but can anyone save me a bit of trial and error? I have the idea that while people usually mount diodes at right angles for a cube, this may be a bad idea if done precisely. It may be better to angle the cube off a very few degrees to prevent it reflecting squarely back at the facet of the through-diode, and then adjusting the right-angled diode's entry angle such that it matches output path with the first, and then adjusting the cube angle slightly to fine tune far field alignment. What small light loss might result might be gained in big reduction of risk. Will this work ok?

[The_Doctor]

If you did this with a mirror on a translation stage, you'd probably nuke 50% or more.

Try 100%. At least, with reds I think it would. It's actually more complex to gauge, the risk relates to power and duration too, apparently independently to some extent. A Rohm diode can stand a full reflection from >99% front surface if running at rated power, during the first 24 hours or so of operation. YMMV... If hard driving, then the result is first time death. If run at HALF power for a week, the result is first time death. Better stats than these would take far more diodes than I ever want to kill. (Although instead of chucking 30 Rohm's over the wall it might have made for gruesome entertainment to gather such stats as these, but I think they were especially poor, great for learning, but people quit relying on them to make a living, hence the eBay flood). The death was instant, so I doubt the translation stage is needed to force the risk, a careful hand would maintain the required position for long enough, and more rapidly achieve it. And that is what bothered me, everyone has a hand unless very unlucky, with some diodes this really is far too easy to do with objects that don't even reflect very well at all.

If anyone knows of diodes that have a reputation for resistance, that's far more useful info than anything I can post.

[RedlumX]

.... 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?

Hi, well I have no real solid argument, but I would tend to believe that there is no close relationship. The cleanliness of the spectrum is a feature more of the low-power regime, in fact basically all diodes become terrible beyond 100mW or so in this respect. So the question is what would the mode spectrum at say, 30mW, tell about the robustness of the diode at say, 300mW - no idea!

[The_Doctor]

Fair enough. It was wishful thinking on my part really. A bit like saying 'if we can draw two buckets of clean water from a well, we might as well hope for the best till we scrape bottom." I did have one small observation that I based that hope on though.. When a red diode is driven to limit, I have seen the broadening of line width, and when increasing the gain on a resonant filter designed for a music synthesiser, I have heard it. I also know that a singer has been known to actually break a wine glass, though I don't know the details behind the story. In short, if a resonant system is driven hard enough, chaos theory kicks in. The energy in the system often bursts out of it erratically, often destructively. Same thing we see when the laminar flow in a tap stream, and the little standing waves as it hits the porcelain in a sink give way to erratic behaviour if we allow too much water to flow. All these mechanisms seem to have a common feature, and this seems to include a resistance to chaos in any system that is sufficiently clean over a large range of intensities in its 'normal' behaviour.

The problem is that once the chaos DOES kick in, all bets, of course, are off... And if this happens in diodes at surprisingly low regions of their dynamic range (power), that just doesn't help at all...

While my next thought is a question more needful of its own thread, could cooling help get clean performance at higher powers? Cooling might decrease dynamism in some aspect that favours extending this clean operation to much higher powers. I have no idea if it does or not, but you might be able to tell me. My guess is it does, if more electrons are more securely held to atoms in diode lasers at low temperatures. There seems no doubt that many diodes do at least survive three-fold power outputs that would kill them at room temperature. As far as I know, early diodes only lasted any useful time at cryogenic temperatures, a tad strange then that Mitsubishi specify as an absolute minimum a -5°C operation. Planters has apparently run them a lot cooler, even down to -196°C. Apparently they stopped improving much below -20 or so on the way down but I don't think they died.

I'm not sure what correlation or relevance this has, but I'm trying to find out things easily done that might push diodes further, safely. (For the diode, at least)... A TEC is usually cheaper and easier than combining three diodes at room temperature, so that's an incentive many here might watch intently.

One curious thing, you say nearly all diodes go agly beyond about 100 mW, and one thing I notice is that most diodes of this type seem to have in common is similarly small emitting volumes, and electrodes. I'm not sure what OPSL is like, but I suspect more powerful with cleaner behaviour. Certainly better beam profiles from what I have read. So I'm guessing that the basic facts of standard diode anatomy have problems that will not shift until some new structure becomes cheap and common to replace them with. I imagine that if just making emitters bigger, smaller, flatter, taller, or whatever, that would be done. Apart from the compromise of more power with multimode output, I've never heard of anything other than cooling which seems to make much difference.

Long.. I'll stop there before it rambles too much.

[steve-o]

It may be better to angle the cube off a very few degrees to prevent it reflecting squarely back at the facet

Yep



_____